Korea Trade and Halal Cosmetic

ISSN 1229-828X
Journal of
Korea Trade
Volume 24 Number 8 2020
The international
scholarly journal of
The Korea Trade
Research Association
Journal of Korea Trade purports to support Korea Trade Research Association
and encourage research in the area of COEX Commercial Exhibition Center #B52,
international economics, international business 513, Yeongdong-daero,
and foreign trade practices and laws. The Gangnam-gu, Seoul 06164, South Korea
Journal welcomes theoretical and empirical Tel +82-2-6000-5182, Fax +82-2-551-8465
papers in the broadly-defined international E-mail: ektra@naver.com
trade issues and policy implications in the Homepage: http://newktra.org
context of Korea Trade.
PUBLISHER
Professor Hag-Min Kim
Journal of Korea Trade Kyung Hee University, Korea
is indexed in:
Social Science Citation Index (SSCI)
EBSCO EDITOR-IN-CHIEF (until December 31, 2020)
ProQuest Professor Suk-Jae Park
KCI Woosuk University, Korea
GoogleScholar
ASSOCIATE EDITORS
ISSN 1229-828X (until December 31, 2020)
Ⓒ Korea Trade Research Association Professor Seo-Young Lee
Mokwon University, Korea
Professor Tae-Hwang Kim

Myongji University, Korea
(until December 31, 2021)

Professor In-Won Kang
Kyung Hee University, Korea
MANAGING EDITOR (until December 31, 2020)

Professor Yoon-Min Kim
Keimyung University, Korea
Guidelines for authors can be found at:

http://newktra.org
JKT Vol. 24, No. 8

Published 31 December 2020
EDITORIAL BOARD (until December 31, 2020) EDITORIAL BOARD (International)
Professor Seok-Tae Kim Professor Myeong-Hwan Kim (A.E.)
Dongkuk University, Korea Purdue University, USA
Professor Young-Hyo Ahn Professor Yongseung Han
Incheon National University, Korea University of North Georgia, USA
Professor Eun-Jae Lee Professor Adriana Grigorescu

Dankook University, Korea National School of Political Studies and Public
Administration, Romania
Professor Je-Hong Lee
Professor Victor Raul Lopez Ruiz
Chosun University, Korea
University of Castilla-La Mancha, Spain
Professor Cheong-Ghi Chun Professor Tongshui Xia
Yeungnam University, Korea Shandong Normal University, China
Professor Sang-Hyeon Han Professor Zuankuo Liu
Namseoul University, Korea Shandong Normal University, China
Professor Xingle Long
Jiangsu University, China
EDITORIAL BOARD (until December 31, 2021) Professor Jongmoo Jay Choi
Temple University, USA
Researcher Jae-Deok Kim
KIET, Korea Associate Professor Eric M. P. Chiu
National Chung-Hsing University, Taiwan
Professor Joon-Seok Oh Professor Dean Sinkovi
Sookmyung Women's University, Korea Juraj Dobrila University of Pula, Croatia
Professor Jeoung-Yul Lee Professor Peter Ractham
Hongik University, Korea Thammasat University, Thailand
Professor Ji-Soo Yi Professor Robert Butler
Hoseo University, Korea University College Cork, Ireland
Researcher Young-Sik Jeong Professor Yu Luo
KIEP, Korea Renmin University of China, China
Professor Hyuk-Soo Cho Professor John C. Yi

Chungnam University, Korea Saint Joseph's University, USA
Professor Nam-Suk Choi Professor Hongmu Lee

Jeonbuk National University, Korea Waseda University, Japan
Associate Professor Erendira Yareth
Professor Young-Jun Choi
Vargas Lopez
Kyung Hee University, Korea
University of Colima, Mexico
Professor Hong-Youl Ha Associate Professor Jie Li
Dongguk University, Korea Central University of Finance and Economics,
China
Professor Alice Y. Ouyang
Central University of Finance and Economics,
China
Associate Professor Samuel M.
Schreyer Fort Hays State University, U.S.A.
Assistant Professor Greg M.
RicheyUniversity of California Riverside, U.S.A.
ISSN 1229-828X
Journal of
Korea Trade
Volume 24 Number 8 2020
The Impact of Input and Output Tariffs on Domestic Employment across Industries:
Evidence from Korea
Yong Joon Jang ····················································································································· 1
The Impact of COVID-19 and Korea’s New Southern Policy on Its Global Value Chain
Jeong-Ho Yoo, Seul-Ki Park and In-Kyo Cheong ···································································· 19
Safeguarding Korean Export Trade through Social Media-Driven Risk Identification
and Characterization
Juthamon Sithipolvanichgul, Alan S. Abrahams, David M. Goldberg, Nohel Zaman,
Milad Baghersad, Leila Nasri and Peter Ractham ···································································· 39
Importance of Political Elements to Attract FDI for ASEAN and Korean Economy
Monthinee Teeramungcalanon, Eric M.P. Chiu and Yoonmin Kim ············································ 63
Increasing Profitability of the Halal Cosmetics Industry using Configuration
Modelling based on Indonesian and Malaysian Markets
Sara Dalir, Hossein GT Olya, Amr Al-Ansi, Alina Abdul Rahim and Hee-Yul Lee · ···················· 81
A Comparative Study on Requirements for the Buyer’s Right to Withhold Performance
for the Seller’s Actual Non-Performance under the CISG and the CESL
Byung-Mun Lee and Dong-Young Kim ················································································ 101
Effects of Market Diversity on Performance of Exporting Companies:
An Inverted U-shaped Relationship
Jungeun Lee, Chang-Bong Kim and Dong-Jun Lee ······························································· 121
Factors of Korea-China Product Trade According to GVC Changes: Focused on FTA
Su-Young Kwak, Mun-Seong Choi, Yong-Hwan Kim and Do-Hyung Lee ······························· 133
Global Productivity and Market Structure Implications of the US-China Trade War:
A CGE Modeling Approach
Jaewon Jung ······················································································································ 153
Is Text Mining on Trade Claim Studies Applicable? Focused on Chinese Cases of
Arbitration and Litigation Applying the CISG
Cheon Yu, DongOh Choi and Yun-Seop Hwang ···································································· 171
MNC Subsidiary’s Entrepreneurship and Knowledge Transfer:
Evidence from MNC Subsidiaries in South Korea
Kangmun Lee, Ji Yeon Yang and Taewoo Roh ······································································ 189
JKT
The Korea Trade
www.newktra.org
Journal of Korea Trade Vol. 24, No. 8, December 2020, 1-18 ISSN 1229-828X
https://doi.org/10.35611/jkt.2020.24.8.1 1
The Impact of Input and Output Tariffs on

Domestic Employment across Industries:
Evidence from Korea JKT 24(8)
Yong Joon Jang†
Department of International Business and Trade, Kyung Hee University, South Korea
Abstract
Purpose – This paper examines how differently output and input tariffs affect domestic employment
across industrial characteristics of comparative advantage such as labor quality and capital intensity.
Design/methodology – The paper focuses on 453 Korean industries from 2007 to 2014 because Korea
is a typical example of a natural resource-scarce open economy and experienced the transition of the
export pattern from labor intensity to technology intensity during this period.
Findings – The results show that input tariff reduction stimulated total employment, focusing on the
early 2010s, while the effects of output tariff reduction were statistically insignificant in general.
However, the stimulation effects of output tariff reduction on employment were found in comparative
advantage industries with greater labor quality and capital intensity. As for input tariff reduction, its
stimulation effects on employment were more prominent in comparative disadvantage industries with
lower labor quality and capital intensity.
Originality/value – These results provide significant implications for natural resource-scarce open
economies which are experiencing the transition of the export pattern from labor intensity to
technology intensity and the unequal distribution of income after trade liberalization: imported
intermediate inputs has become increasing important, leading to trade effects on employment and
alleviation of income inequality.
Keywords: Comparative Advantage, Employment, Imported Intermediate Inputs, Tariff

JEL Classifications: F14, F15, F16, F66
1. Introduction
The Korean government recognized the importance of international trade for its trade-
dependent small economy and has entered 16 free trade agreements (FTAs) as of 2020.
Accordingly, Korea’s international trade with FTA member countries has greatly enlarged,
and many studies have analyzed the effects of FTAs and tariff reductions on economic
performance. However, they have primarily focused on the link between exports and sector-
or firm-level productivity in final goods (Aw et al., 2000; Hahn Chin-Hee, 2004; Jang Yong-
Joon et al., 2015). In practice, imports as well as exports greatly increased in Korean FTAs;
total import share from member countries rose from 0.9 percent in 2004 to 62.8 percent in
2015 (Kim Young-Gui et al., 2017). Addressing that imports from member countries have
increased by an annual average of 12.8 percent and exports to them have increased by 10.7
percent, protectionists criticize that free trade has further widened the trade deficit with FTA
member countries and thus worsened the Korean economy. They especially attribute the rise
of unemployment to free trade policies (Bae Chan-Kwon et al., 2012).
†
First and Corresponding author: yjjang@khu.ac.kr
© 2020 Korea Trade Research Association. All right reserved. www.newktra.org
Journal of Korea Trade, Vol. 24, No. 8, December 2020
2
The protectionists’ arguments overlook the true features of Korean trade. As a country with
natural-resource scarcity, Korea mostly imports raw materials and intermediate inputs,
which make up over 70 percent of total imports, to use for domestic productions and exports.
Since the early 2010s, intermediate goods imports have further increased despite a slight
decline in total imports in Korea, deepening the global value chains (GVCs) (Kim Young-Gui
et al., 2017).1 Hence, it is very important for the Korean economy to have better access to
imported intermediate inputs as well as greater export opportunity and thus to consider the
effects of free trade on economic performances as two aspects of tariff reduction: input tariff
on intermediate goods and output tariff on final goods. The situation of Korean trade will be
in line with that of other developing countries with an open economy: Ma and Dei (2009) and
Yu (2014) for China, Fernandes (2007) for Colombia, Goldberg et al. (2010), Topalova and
Khandelwal (2011) for India, and Amiti and Konings (2007) for Indonesia. These studies find
that economic gains from input tariff reduction are much greater than those from output
tariff reduction.
This paper empirically examines how differently input and output tariff reductions on
imports affect domestic employment in Korea, depending on industrial characteristics of
comparative advantage. Unlike previous related studies, the paper focuses on employment
because recently some politicians and protectionists blamed the declines in employment on
international trade all over the world (Acemoglu et al., 2016; Rose, 2018). In Korea, the
employment issue is also at the center of political and economic dispute, stemming from
income-led growth policies and free trade policies. In addition, Whang Un-Jung et al. (2017)
argue that export expansion no longer leads to job creation and thus Korea is losing its status
as a manufacturing-based export-driven economy. They show that the export pattern of
Korea has changed from labor-intensive industry to capital (or technology)-intensive
industry under the situation that the former leads to more job creation than the latter. Also,
they show that the domestic consumption elasticity of employment is much greater than the
export elasticity of employment in input-output analysis (IOA). 2 From these results, it is
noted that the importance of exports has become eclipsed and imports of intermediate inputs
are leading to a rise in domestic consumption and in trade effects on employment. This
phenomenon will especially be more significant for a natural resource-scarce open economy
like Korea.
Also, this paper examines the heterogeneous effects of input and output tariff reductions
across industrial characteristics of comparative advantage. Amiti and Konings (2007) address
that a fall in input tariffs might have quite different effects on users of these inputs. Luong
(2011) shows that input and output tariff reductions differently affect firm productivity,
depending on the elasticity of substitution among intermediate inputs and import intensity
at the sector level. In addition to these findings, this paper focuses on various industrial
characteristics of comparative advantage such as labor quality and capital intensity which are
major production inputs. The paper finds their significant roles in the connections not only
between domestic employment and tariff reduction in imported final goods, but also between
domestic employment and tariff reduction in imported intermediate inputs.
1
Kim Young-Gui et al. (2017) show that major importing countries of Korea changed from developed
countries to natural resource-abundant countries during the period 1988-2015. Kim et al. (2011) also
show that the trade deficit in intermediate goods trade enlarged from 29 billion dollars in 2004 to 46.9
billion dollars in 2009.
2
As of 2014, the former is 11.6 employments, while the latter is 6.5 employments (Whang Un-Jung et
al., 2017). Thus, domestic consumption contributes domestic employment two times more than
exports in Korea (Whang Un-Jung, 2019). Also, Aw et al. (2000) show that the correlation between
export expansion and firm productivity is not strong in the case of Korea, unlike that of Taiwan.
The Impact of Input and Output Tariffs on Domestic Employment across Industries: Evidence from Korea
3
The rest of the paper is organized as follows. Section 2 summarizes previous studies and
sets up the research hypotheses. Section 3 provides econometric specifications and data
sources. Section 4 reports the empirical results. Finally, Section 5 provides a conclusion and
policy implications.
2. Literature Review and Research Hypotheses

Previous studies on the economic effects of imports can be categorized into two groups:
those dealing with import competition (i.e., import penetration) and the others dealing with
technology transfer (i.e., learning-by-importing) (Amiti and Konings, 2007). The former
generally focuses on imported final goods and concludes that tighter competition with
foreign firms after trade liberalization or tariff reduction can enhance productivity, economic
efficiency and job creation on one side (MacDonald, 1994; Pavnick, 2002), and on the other
side generate firm exit and job displacement as side effects (Kletzer, 2001). Accordingly, the
aggregate effects seem to be ambiguous due to the tradeoff of tighter market competition
above. Following these studies, Bernard et al. (2007) provide a theoretical framework in which
this tradeoff appears differently across industries. They find that falling trade costs render an
economy vulnerable to import penetration and generate job turnover from low productive
firms to high productive firms in all sectors. However, these reallocations of resources are
more prominent in comparative advantage industries than in comparative disadvantage
industries. The reason is that comparative advantage industries have not only the greater
export opportunity of survivors but also the greater exit of nonviable firms after trade
liberalization. However, comparative disadvantage industries have fewer opportunity to
export with tighter import penetration. Consequently, trade liberalization spurs net job
creation and a greater increase in average productivity in comparative advantage industries,
but net job destruction and a decrease in average productivity in comparative disadvantage
industries.
Meanwhile, previous studies on learning-by-importing regard the role of imported
intermediate inputs as important. Addressing better access to foreign technology and
economic efficiency through importing intermediate inputs, they conclude that trade
liberalization has a positive influence on economic performances by sourcing foreign
intermediate inputs with higher quality (Feng et al., 2016), greater variety (Kasahara and
Rodrigue, 2008), and lower price (Bernard et al., 2003; De Hoyos and Iacovone, 2013).
Accordingly, unlike ambiguous effects of imported final goods, trade liberalization in
imported intermediate inputs produces economic benefits in general. Following these studies,
Luong (2011) examines that the positive effect of imported intermediate inputs can be
different across industries: lowering input tariffs leads to a rise (or decline) in productivity in
industries with high (or low) import intensity and low (or high) elasticity of substitution
among intermediate inputs on one side. On the other side, lowering output tariffs has the
opposite effect: it has a positive (or negative) influence on productivity growth when
intermediate inputs are highly (or lowly) differentiated.
Based on these studies, this paper addresses the following research question: How do output
and input tariffs on imports affect domestic employment across industrial characteristics? In
particular, previous studies have examined the role of imported intermediate inputs on
economic performances, but ignored their heterogeneous effects across industrial
characteristics. The only exception is Luong (2011), but he simply considered import
intensity and the elasticity of import substitution as industrial characteristics. Meanwhile,
Bernard et al. (2007) considered the role of comparative advantage and comparative
4
disadvantage in the connection between output tariff reduction and aggregate productivity.
However, they did not deal with various industrial characteristics such as import intensity by
missing imported intermediate inputs in their theoretical model and thus ignoring
technology transfer through imports. Accordingly, when considering both intermediate
inputs and final goods in the analysis, it is necessary to consider various characteristics of
comparative advantage and comparative disadvantage such as the share of production inputs
in final production (Bernard et al., 2007; Whang Un-Jung et al., 2017). To fill in this gap, this
paper considers labor quality and capital intensity which are major indicators of comparative
advantage and comparative disadvantage, as a medium for the connections between a
reduction in input or output tariffs and domestic employment.
For output tariff reduction, I mainly consider Bernard et al.’s (2007) theoretical framework:
the aggregate effects on total employment seem to be ambiguous due to the tradeoff of tighter
market competition after trade liberalization, but comparative advantage industries have
greater resource reallocation and thus net job creation. Meanwhile, in addition to factor
abundance of labor and capital in the Heckscher-Ohlin model, there exist various sources of
comparative advantage such as more educated workers and better institutions (Costinot,
2009). As a skilled- and capital-abundant country, Korea also has the characteristics of
comparative advantage industries such as higher labor quality and capital intensity (Whang
Un-Jung et al. 2017)3.
With these features, I formulate the following hypotheses in Korea in line with Bernard et
al.’s (2007) arguments:
H1: The effects of output tariff reduction on employment are ambiguous in general, but the
positive effects will appear in comparative advantage industries with higher labor quality
and capital intensity.
For input tariff reduction, I mainly consider Luong’s (2011) theoretical framework with
intensity of imported intermediate inputs: the effects of a decrease in input tariffs on
productivity are positive due to technology transfer in general, and these positive effects are
more prominent in industries with high intensity of imported intermediate inputs.
Meanwhile, some basic statistics in Korea show that industries with lower labor quality and
capital intensity are more likely to depend on imports in intermediate inputs. I calculate the
sector-level average value of the import input coefficients from the Economic Statistics
System of the Bank of Korea,4 and then examine how much an industry depends on imported
intermediate inputs for final production as their import intensity in Korea. Fig. 1 reveals that
import input coefficients and two industrial characteristics, labor quality and capital intensity,
are slightly negatively related in Korea: the correlation of coefficients with import input
coefficients is -0.141 for average wage which is a proxy for labor quality and -0.122 for capital
intensity at 1% statistical significance level. This implies that imported intermediate inputs
are a substitute for production factors such as domestic skilled workers and capital for final
production. In other words, industries with lower labor quality and/or lower capital intensity
depend more on imported intermediate inputs for final production. This feature of Korea is
very similar with Liu and Qui’s (2016) finding of the substitutive relation between imported
intermediate inputs and innovation for Chinese firms. Since lower labor quality and capital
3
Whang Un-Jung et al. (2017) show that Korea’s comparative advantage has transferred from labor-
intensive industries to capital-intensive industries since the 1990’s with a rapid development of high
technology industries such as IT.
4
The import input coefficients represent the ratio of value of imported intermediate inputs in total input
value in a sector.
5
intensity are major characteristics of a comparative disadvantage industry in Korea, I
conclude that comparative disadvantage industries are more likely to depend on imports in
intermediate inputs as well as final goods.
With these features, I formulate the following hypotheses in Korea in line with Luong’s
(2011) arguments:
H2: The effects of input tariff reduction on employment are positive in general, and these
positive effects will be more prominent in industries with lower labor quality and capital
intensity.
Overall, the two hypotheses are in line with Luong (2011) and Ma and Dei (2009) in the
sense that there are opposite effects between output tariffs and input tariffs on economic
performances. Additionally, I expect that the second hypothesis will be more likely to be
correct for the early 2010s, when the ratio of intermediate inputs in total imports increased
significantly in Korea, deepening GVCs (Kim Young-Gui et al., 2017). Hence in the following
empirical analysis I break down the sample period into two parts, based on 2010.
Fig. 1. Correlations between Import Input Coefficients and Industrial Characteristics
120
100
Average Wage
80
60
40
20
0
0 1 2 3 4
Import Input Coefficient
3500
3000
Capital Intensity
2500
2000
1500
1000
500
0
0 1 2 3 4
Import Input Coefficient
Sources: Author’s calculation using KOSIS, UN COMTRADE, and BOK.

6
3. Economic Specifications and Data Sources
3.1. Econometric Specifications
Based on the literature review, I build up the following equation as a main regression model
to empirically test the effects of output and input tariffs on total employment:
ln ௜௧ ଴ ଵ ln
௜௧ ଶ ln ௜௧ ଷ ln
௜௧ ସ ln 3௜௧
ହ ln ௜௧ ଺ ln ௜௧ ௜ ௧ ௜௧ (1)
where i and t refer to sectors and years, respectively.

As the dependent variable, ln ௜௧ is the log of total employment in i at t. The two key
variables of ln ௜௧ and ln ௜௧ represent the log of output tariff rate and the log of input
tariff rate in i at t, respectively. An output tariff represents an average tax directly imposed on
import goods in i, while an input tariff represents an average tax imposed on imported
intermediate inputs, which are used to produce goods in i. As in Goldberg et al. (2010), I
calculate the weighted average value of tariffs on intermediate inputs for producing a final
good, based on an import input coefficient.5 Based on H1, I expect ହ to be ambiguous due
to the tradeoff of import penetration. Based on H2, I expect ଺ to be negative due to greater
efficiency in final production using imported intermediate inputs with lower tariff rate.
As other control variables, ln
௜௧ is the log of average wage in i at t, which is a proxy for
labor quality (Trefler, 1993; Song Sang-Yoon, 2018). ln
௜௧ is the log of average tangible
assets in i at t, which is a proxy for capital intensity (Leontief, 1953). ln ௜௧ is the log of total
export value in i at t, which is for estimating the export elasticity of employment (Whang Un-
Jung et al., 2017). ln 3௜௧ is the log of market concentration in i at t, which is a proxy for
market competition (Bernard et al., 2007; Kim Bae-Geun, 2014). It is noted that the higher
market concentration is, the lower market competition is. I expect ଵ , ଶ , and ଷ to be
positive, implying that total employment increases in response to higher labor quality, higher
capital intensity, and greater export value at the sectoral level. Meanwhile, I expect ସ to be
ambiguous, implying that the effect of market concentration on total employment is
undetermined due to the tradeoff effects of market competition (Bernard et al., 2007).
௜ is a dummy variable for i to control a sector’s innate time-invariant characteristics that
might affect an employment. ௧ is a dummy variable for t to control the macroeconomic
environment, which might affect an employment. Finally, ௜௧ denotes an error term.
In addition to (1), I consider the following econometric specification to see how the effects
of output tariff or input tariff on total employment varied in sectoral characteristics.
ln ௜௧ ଴ ଵ ln
௜௧ ଶ ln ௜௧ ଷ ln
௜௧ ସ ln 3௜௧
ହ ln ௜௧ ଺ ln ௜௧ ଻ ln ௜௧ ௜௧
଼ ln ௜௧ ௜௧ ௜ ௧ ௜௧ (2)
In (2), all variables except for the interaction terms are the same as in (1). In the interaction
terms, ௜௧ consists of the other control variables, ln
௜௧ and ln
௜௧ , and thus the effect
5
Meanwhile, Luong (2011) divides imported goods into final goods and intermediate inputs, considering
tariffs on the former as output tariffs and those on the latter as input tariffs. This is a significant
difference between this paper and Luong (2011) in considering tariffs on imported intermediate inputs.
It is noted that Goldberg et al.’s (2010) methodology with respect to input tariffs is more apt for the
objectives of this paper.
7
of output tariff or input tariff on total employment can be compared across different sectoral
characteristics. For example, considering ௜௧ as labor quality (i.e., ln ௜௧ ), if both ଵ and
଻ are negative with statistical significance, then it is concluded that output tariff reduction
increases total employment, and these positive effects were more prominent in industries
with higher labor quality. Accordingly, based on H1, I expect ଻ to be negative for the
interaction terms between output tariff and labor quality and between output tariff and capital
intensity. Based on H2, I expect ଼ to be positive for the interaction terms between input tariff
and labor quality and between input tariff and capital intensity, implying that the positive
effects of input tariff reduction on total employment were more prominent in industries with
lower labor quality and capital intensity.
In the main regression, I considered the fixed-effects model which can alleviate omitted
variable bias by eliminating ௜ . I performed the F-test and the Hausman test to verify the
reliability of the fixed-effects: the former tests the null hypothesis that sector dummies are all
together zero, while the latter tests the null hypothesis that the covariance between
independent variables and ௜ is zero. If the test results reject these null hypotheses, then the
fixed-effects model is preferred over the pooled ordinary least squares (OLS) and the random-
effects model, respectively. Also, I performed the F-test for the null hypothesis that year
dummies are all together zero to check whether there exist the year-specific effects during the
sample period in all regressions (Cameron and Trivedi, 2005).
While eliminating ௜ , the regression still may show reverse causality between a dependent
variable and independent variables. For example, a decline in employment can be logical
ground for protectionism and thus affect an increase in tariff rates (Baier and Bergstrand,
2007). To mitigate this, I considered the independent variables lagged by one year as well as
those in the current year. Also, while controlling
௧ , the global financial crisis of the late 2000s
in the sample period might significantly affect the regression results. In addition, Kim Young-
Gui et al. (2017) show that the share of imported intermediate inputs has rapidly increased
since the early 2010s in Korea. Hence, in some regressions I divided the entire sample period
into two sub-samples: the late 2000s and the early 2010s. Finally, I consider robust standard
errors clustered by sectors to control that sectoral unobservable factors inflate the statistical
significance of coefficient estimates (Aghion et al., 2003).
3.2. Robustness Check6

One of the problems in (1) and (2) is that total employment in previous years might affect
that of the current year. Also, control variables in (1) and (2) might not be strictly exogenous.
Accordingly, I consider the two-step Arellano-Bond estimator, which is a dynamic panel
model using a generalized method of moments (GMM) as a robustness check (Arellano and
Bond, 1991)7. The Arellano-Bond estimator is a very useful method to control a dynamic
panel bias which is caused by autocorrelation of employment and endogeneity problem
because it considers first differences of the variables and sets up a two-step estimation
procedure with instrument variables consisting of their lagged value. To solve an over-
identification problem caused by the fact that the number of instrument variables is greater
than that of endogenous variables, the Arellano-Bond estimator considers the GMM
estimation.
6
I referred to Cameron and Trivedi (2005) for the procedure of the Arellano-Bond estimator.
7
Arellano and Bond (1991) also applied it to the employment equations.
8
In the procedure of the Arellano-Band estimator, I performed two tests to check whether
the instruments are valid: the Arellano-Bond test for autocorrelation (AB test for AR) and the
Hansen J test. The AB test for AR(1) (or AR(2)) verifies the null hypothesis that the first-order
(or second-order) autocorrelation exists, while the Hansen J test verifies the null hypothesis
that an over-identification problem exists.
3.3. Data Collection and Summary Statistics

Table 1 lists the variables and their data sources. The database consists of 453 sectors
classified by Korea Standard Industry Code (KSIC) at a 5-digit level from 2007 to 2014 in
Korea. As a main data source, the Korean Statistical Information Service (KOSIS) of Statistics
Korea provides total numbers of employees, total salaries, and values of tangible assets at 5-
digits KSIC.8 The dependent variable in the regressions is defined as the log of total number
of employees. Average wage (AWG) is defined as the ratio of total salary in total number of
employees (i.e., a per capita salary). Capital intensity (CAP) is defined as the ratio of value of
tangible assets in total number of employees (i.e., a per capita capital). Both the monetary unit
of total salary and tangible assets are the Korean won and I considered the real values by
deflating them by the producer price index (PPI) provided by the Bank of Korea.
Total export values at the 6-digits Harmonized System (HS) code were extracted from UN
COMTRADE (United Nations Commodity Trade Statistics Database). As their monetary
unit is the US dollar, I converted them into Korean won with the annual average exchange
rate of Korean won-US dollar provided by the Bank of Korea and 5-digits KSIC with the
correlation tables provided by the KOSIS. Also, they were deflated by the PPI. The market
shares of the three largest firms (CR3) at 5-digits KSIC represent the level of market
concentration provided by the Korea Fair Trade Commission.
Trade-weighted average tariff rates at the 6-digits HS code in Korea were extracted from
the World Integrated Trade Solution (WITS) of the World Bank and also converted to a 5-
digits KSIC in order to use output tariff rates in the regressions. As in Goldberg et al. (2010),
I calculated the level of input tariff rate (ITR) as follows:
௜௧ ∑௡௝ୀଵ ௝௜ ௝௧ (3)
where j denotes an intermediate good using in sector i. ௝௜ represents the contribution ratio
of an intermediate good in i’s total production. Accordingly, ௜௧ is defined as the
contribution-weighted average tariff rate of intermediate goods in i. For ௝௜ , I considered the
import input coefficients in the input-output (IO) tables provided by the Bank of Korea,
which are defined as the ratio of value of imported intermediate inputs from j to i in total
input value in i. The import input coefficients represent the contribution of imported
intermediate inputs to produce one unit of products in i. As the IO table has its own
classification of 122 sectors, I converted it to a 5-digits KSIC with the correlation tables
provided by the BOK. When considering the log values of output tariff rates, input tariff rates,
and CR3, I added one to the original values because some of them are zero.
8
I supplemented the data of the year 2010 by the extrapolation method because the KOSIS did not collect
it.
9
Table 1. Variables and Data Sources
Variable Source
Number of Employees
Total Salary KOSIS
Tangible Assets
Export UN COMTRADE
Market Concentration (CR3) Korea Fair Trade Commission
Weighted Average Tariff Rate WITS
Import Input Coefficient Bank of Korea
Table 2 reports summary statistics. I checked correlation coefficients and found no

multicollinearity among independent variables, as they all are less than the absolute value of
0.8.
Table 2. Descriptive Statistics

Variable Obs. Mean Std. Dev. Min Max
ln ௜௧ 3,562 7.875 1.271 3.497 11.389
ln ௜௧ 3,562 3.357 0.284 2.418 4.661
ln
௜௧ 3,624 11.986 2.347 1.627 17.866
ln ௜௧ 3,562 4.652 0.751 0.946 8.028
ln 3௜௧ 3,624 3.612 0.619 0 4.615
ln ௜௧ 3,624 1.910 0.900 0 5.427
ln ௜௧ 3,624 1.904 0.726 0 4.831
4. Empirical Results
4.1. Basic Results
Table 3 reports the regression results for the entire sample period (2007-2014), the late
2000s (2007-2010), and the early 2010s (2011-2014). Columns (2), (4), and (6) show the
results with one-year lagged independent variables, while columns (1), (3), and (5) show those
with current ones. In all columns, the results from the Hausman test and the F-test show that
the fixed effects model is preferred over the random effects model and the pooled-OLS,
respectively. Also, the F-test for year-specific effects rejects the null hypothesis at 1 %
significant level, suggesting that all regressions include year dummies to control them.
The results in Table 3 are as follows. First, as expected, the coefficient estimates of ln
are positive and statistically significant in general, implying that employment increased when
worker quality rose. However, the results find no evidence when considering one-year lagged
independent variables in the samples which were divided into the late 2000s and the early
2010s. Overall, it seems that labor quality had only short-term ex post effects on employment
in Korea.
Second, the coefficient estimates of ln are also positive and statistically significant in
general, representing the positive export elasticity of employment in Korea. However, these
positive effects occurred mostly during the late 2000s rather than the early 2010s. In other
10
words, exports did not influence employment during the early 2010s, which finding is
consistent with those of Whang Un-Jung et al. (2017).9 Hence, in conjunction with Whang
Un-Jung et al. (2017), these results confirm that recently export expansion does not lead to
sufficient job creation any more in Korea since its industrial structure focuses more on capital
and technology rather than labor.
Third, the coefficient estimates of ln are positive and statistically significant in
column (1), implying that the increase in capital intensity positively affected employment
over the entire period. The results, however, find no evidence when considering one-year
lagged independent variables and dividing the sample period into the late 2000s and the early
2010s. Overall, it seems that capital intensity was not a main determinant of employment in
Korea.
Fourth, the coefficient estimates of ln 3 are negative and statistically significant in
general, implying that employment increased as market competition intensified in Korea.
These results are consistent with Kim Bae-Geun (2014), showing that a permanent rise in the
markup ratio lowered employment in Korea. In conjunction with Kim Bae-Geun (2014),
these results confirm that market competition has a positive role in an employment among
its trade-off effects in Korea. Hence the government policy as a means to reduce the degree
of market concentration (i.e., intensifying market competition level) is important in rising
employment.
Fifth, the coefficient estimates of ln
are statistically insignificant in most columns.
These results seem to come from the tradeoff between job destruction in comparative
disadvantage industries and job creation in comparative advantage industries in response to
opening the economy, as mentioned in H1. Meanwhile, in column (5) the coefficient
estimates of ln
become statistically significant with a negative sign, implying that the
pure effects of the tradeoff were positive and thus output tariff reduction rendered
employment growth during the early 2010s. These positive effects, however, were not realized
in one-year time lag, representing no long-term ex post effects.
Lastly, the coefficient estimates of ln
are statistically significant in columns (5) and (6),
implying that input tariff reduction raised an employment during the early 2010s, regardless
of time lags. Meanwhile, there were no effects of input tariff on an employment during the
late 2000s. Hence, the results are consistent with H2 only for the early 2010s, when imports
of intermediate inputs sharply increased. Hence these results are very similar with Luong
(2011) which found that the positive effects of imported intermediate inputs on economic
performance are more prominent with high import intensity.
Table 3. Main Empirical Results

Entire Period: 2007-2014 Late 2000s: 2007-2010 Early 2010s: 2011-2014
t t-1 t t-1 t t-1
(1) (2) (3) (4) (5) (6)
ln 0.510*** 0.250** 0.469*** -0.026 0.312* 0.199
(Ave.Wage) (0.116) (0.097) (0.130) (0.093) (0.160) (0.155)
ln 0.078*** 0.062*** 0.063*** 0.050*** 0.022 0.028
(Export) (0.016) (0.016) (0.018) (0.019) (0.016) (0.021)
ln
0.082** 0.000 0.061 0.000 0.064 -0.000
(Cap. Intens.) (0.039) (0.000) (0.055) (0.000) (0.016) (0.000)
9
Whang Un-Jung et al. (2017) estimated that export growth by hundred thousand dollars induced 0.19
employments in 2010, which was much lower than 3.32 in 1985 and 1.87 in 1990.
11
Table 3. (Continued)
t t-1 t t-1 t t-1
(1) (2) (3) (4) (5) (6)
ln 3 -0.127*** -0.142*** -0.094** -0.066** -0.043 -0.081***
(Mkt. Compet.) (0.032) (0.025) (0.042) (0.032) (0.032) (0.025)

-0.020 0.033 -0.029 0.056 -0.094*** -0.000
(Output Tariff) (0.051) (0.070) (0.099) (0.065) (0.032) (0.045)

-0.005 0.002 0.004 0.020 -0.080*** -0.054**
(Input Tariff) (0.022) (0.021) (0.014) (0.049) (0.027) (0.027)
Year Dummies Yes Yes Yes Yes Yes Yes
ଶ
Overall 0.274 0.362 0.258 0.280 0.108 0.310
F-Test 195.61*** 190.92*** 135.78*** 140.78*** 194.97*** 179.24***
Hausman Test 528.04*** 1,533.5*** 281.10*** 306.07*** 170.26*** 1,038.4***
Observations 3,562 3,112 1,777 1,331 1,785 1,781
Notes: 1. *, **, *** denote significance at 1%, 5%, and 10% levels, respectively.
2. Figures in parentheses are robust standard errors clustered by industry (KSIC 5 digits).
3. Year dummies are included in all regressions.
4.2. Heterogeneous Effects across Industries

Tables 4 and 5 report the regression results for interaction terms between industrial
characteristics and output and input tariffs. Table 4 shows the results with labor quality, while
Table 5 shows those with capital intensity as industrial characteristics of comparative
advantage. In all regressions, except for column (1) of Table 5, the F-test and the Hausman
test suggest that the fixed-effects model is preferred over the pooled OLS and the random-
effects model, respectively. Also, the F-test for year-specific suggests that all regressions
include year dummies to control them. Column (1) of Table 5 reports the results from the
random effects model followed by the Hausman test. Accordingly, I did the Breusch-Pagan
LM (Lagrangian Multiplier) test to check whether the random-effects model is preferred over
the pooled OLS, which suggested the former.
Table 4. Empirical Results of Interaction Terms: Labor Quality

t t-1 t t-1 t t-1
(1) (2) (3) (4) (5) (6)
ln 0.857*** 0.209 0.709*** -0.263 0.631*** 0.313
(Ave.Wage) (0.196) (0.186) (0.211) (0.282) (0.223) (0.296)
ln 0.079*** 0.061*** 0.064*** 0.048*** 0.027* 0.031
(Export) (0.016) (0.016) (0.018) (0.018) (0.016) (0.021)
ln
0.076* 0.000 0.058 0.000 0.055 -0.000
(Cap. Intens.) (0.039) (0.000) (0.053) (0.000) (0.039) (0.000)
ln
3 -0.128*** -0.142*** -0.095** -0.064** -0.044 -0.085***
(Mkt. Compet.) (0.032) (0.024) (0.041) (0.032) (0.033) (0.025)
12
t t-1 t t-1 t t-1
(1) (2) (3) (4) (5) (6)
ln 0.676*** 0.207 0.464** -0.319 0.806** 0.733
(Output Tariff) (0.254) (0.239) (0.234) (0.211) (0.346) (0.478)
-0.211*** -0.053 -0.147** 0.112* -0.273*** -0.224

(0.075) (0.071) (0.069) (0.063 (0.105) (0.147)
ln -0.060 -0.228* -0.045 -0.042 -0.323 -0.477**
(Input Tariff) (0.254) (0.126) (0.172) (0.268) (0.240) (0.232)
0.016 0.070** 0.014 0.018 0.071 0.126*

(0.045) (0.037) (0.052) (0.077) (0.068) (0.067)
ଶ
Overall 0.291 0.377 0.280 0.235 0.120 0.277
F-Test 193.32*** 185.78*** 132.65*** 137.30*** 198.11*** 177.30***
Hausman Test 228.73*** 681.99*** 281.36*** 306.15*** 245.35*** 396.03***
Observations 3,562 3,112 1,777 1,331 1,785 1,781
The results in columns (1), (3) and (5) of Table 4 show that the coefficient estimates of
ln are positive and statistically significant with current independent variables. However,
when considering a one-year time lag, all coefficient estimates of ln become statistically
insignificant. Hence the effects of output tariff on an employment were ambiguous due to the
tradeoff role of import competition, but only when considering one-year lagged independent
variables. In other words, the results imply that output tariff reduction spurs job destruction
caused by the exit of nonviable firms in the short run, but over time it is becoming weaker
since the survivors start taking the greater export opportunity and create jobs. Some previous
studies provide the ex post positive effects of trade liberalization on exports in Korea (Jang et
al., 2015, Bae et al., 2012).
Meanwhile, the coefficient estimates of ln ln
are negative and statistically
significant in current years. These results show that the positive volume of coefficient
estimates of ln are less prominent with greater labor quality and imply that the positive
effects of output tariff reduction on employment were more prominent in industries with
higher labor quality, which is consistent with H1. However, the results become statistically
insignificant in columns (2) and (6) and change to the positive sign in column (5), implying
that the different effects were not sustainable at a one-year time lag.
For ln and ln ln
in Table 4, the coefficient estimates are negative and
statistically significant in columns (2) and (6). These results imply that the positive effects of
input tariff reduction on employment were more prominent in industries with lower labor
quality at a one-year time lag, as expected in H2. In addition, the prominently positive effects
of input tariff reduction on employment in industries with lower labor quality were greater
in the early 2010s.
The results of other independent variables in Table 4 are very similar to those in Table 3.
13
Table 5. Empirical Results of Interaction Terms: Capital Intensity
t t-1 t t-1 t t-1
(1) (2) (3) (4) (5) (6)
ln 0.516*** 0.226** 0.467*** -0.026 0.257* 0.158
(Ave.Wage) (0.072) (0.094) (0.132) (0.086) (0.137) (0.135)
ln 0.112*** 0.060*** 0.065*** 0.050** 0.022 0.026
(Export) (0.011) (0.016) (0.018) (0.019) (0.016) (0.021)
ln
0.273*** -0.000 0.118 0.000 0.294*** -0.000
(Cap. Intens.) (0.065) (0.000) (0.088) (0.000) (0.101) (0.000)
ln
3 -0.172*** -0.146*** -0.090** -0.065* -0.046 -0.087***
(Mkt. Compet.) (0.024) (0.025) (0.041) (0.034) (0.032) (0.026)
ln 0.513*** 0.066 0.077 0.067 0.567* 0.161
(Output Tariff) (0.134) (0.133) (0.244) (0.307) (0.296) (0.166)
-0.103*** -0.007 -0.147** -0.002 -0.129** -0.032

(0.021) (0.023) (0.069) (0.060) (0.057) (0.032)
ln 0.050 -0.155 0.108 0.038 -0.075 -0.359***
(Input Tariff) (0.066) (0.100) (0.099) (0.322) (0.162) (0.130)
-0.011 0.034 0.014 -0.004 -0.001 0.064**

(0.021) (0.022) (0.052) (0.065) (0.033) (0.026)
ଶ
Overall 0.331 0.372 0.277 0.272 0.138 0.274
F-Test
9,663.8*** 185.78*** 134.18*** 139.83*** 199.56*** 182.05***
(or BPLM)
Hausman Test 4.33 227.42*** 382.03*** 213.53*** 445.93*** 182.77***
Observations 3,562 3,112 1,777 1,331 1,785 1,781
3. Year dummies are included in all regressions. 4. In column (1) the result of the Breusch-
Pagan LM (BPLM) test are presented, not that of F-test.
The results in Table 5 show that the coefficient estimates of ln are very similar to those
in Table 4. Meanwhile, the coefficient estimates of ln ln
are negative and
statistically significant in columns (1), (3) and (5). These results imply that the positive effects
of output tariff reduction on employment were more prominent in industries with higher
capital intensity, which is consistent with H1. Again, however, these different effects were not
sustainable at a one-year time lag.
The result in column (6) of Table 5 shows that the positive effects of input tariff reduction
on employment were more prominent in industries with lower capital intensity at a one-year
time lag in the early 2010s. In conjunction with lower labor quality, they all represent the
characteristics of comparative disadvantage industries in Korea. Consequently, these results
are all consistent with H2, especially for the early 2010s when the ratio of intermediate inputs
in total imports increased significantly in Korea, deepening GVCs.
The results of other independent variables in Table 5 are very similar to those in Tables 3
and 4.
4.3. Robustness Results

Tables 6 and 7 report the regression results of the Arellano-Bond estimator to support the
main results shown in Tables 3, 4 and 5, respectively. Columns (1), (2), and (3) of Table 6
14
report the results for the entire sample period, the late 2000s, and the early 2010s, respectively.
Table 7 reports the results of interaction terms between output or input tariffs and two control
variables: columns (1)-(3) for labor quality and columns (4)-(6) for capital intensity. In all
regressions, the AB test for AR and the Hansen J test ensure that the instruments starting with
2 lags are jointly valid and do not have an over-identification problem.
The results in Table 6 show that the decrease in input tariffs stimulated total employment
in the early 2010s in Korea, which is consistent with the results shown in Table 3. The
coefficient estimates of output tariff, however, become statistically insignificant in the
Arellano-Bond estimator. Hence, in conjunction with the results from the fixed-effects
model, I conclude that the effects of output tariff reduction on total employment were
ambiguous in general, which is consistent with H1. The coefficient estimates of other control
variables, except for market concentration, from the Arellano-Bond estimator are very similar
to those from the fixed-effects model. The coefficient estimates of ln 3, however, are
statistically insignificant. 10 Finally, the results show that employment in the current year
decreased while increasing in the preceding year.
Table 6. Main Empirical Results in the Arellano-Bond Estimator

Entire Period Late 2000s Early 2010s
(1) (2) (3)
ln ௧ିଵ -0.225** 0.008 -0.872**
(Employment at t-1) (0.103) (0.041) (0.357)
ln 0.333*** 0.386*** -0.225
(Ave.Wage) (0.099) (0.115) (0.829)
ln
0.022* 0.066* 0.011
(Export) (0.013) (0.025) (0.017)
ln 0.053** 0.311*** 0.098
(Cap. Intens.) (0.026) (0.106) (0.090)
ln 3 -0.011 -0.023 -0.011
(Mkt. Compet.) (0.017) (0.023) (0.046)
-0.023 0.022 -0.020
(Output Tariff) (0.014) (0.018) (0.061)
-0.003 0.006 -0.073**
(Input Tariff) (0.099) (0.010) (0.035)
Year Dummies Yes Yes Yes
# of Groups 448 446 447
# of Instruments 27 17 17
AB Test for AR(1) 2.23** 3.53*** 1.94*
AB Test for AR(2) -1.46 -1.20 -1.61
Hansen J Test 9.22 4.17 4.96
Observations 2,663 1,331 1,332
2. Figures in parentheses are robust standard errors.
3. AB represents Arellano-Bond.
10
Alternatively, I considered the Herfindahl-Hirschman Index (HHI) instead of ଷ , but it did not
change the results.
15
The results in Table 7 are also very consistent with those in Tables 4 and 5. The positive
effects of output tariff reduction on an employment were more prominent in industries with
higher labor quality and higher capital intensity. Especially, these prominent effects occurred
mainly in the late 2000s for industries with higher labor quality. Also, as expected, the positive
effects of input tariff reduction on total employment were more prominent in industries with
lower labor quality and lower capital intensity, mainly in the early 2010s. Hence, I conclude
that all results in the main regressions and the robustness checks generally support the
hypotheses 1 and 2.
Table 7. Empirical Results of Interaction Terms in the Arellano-Bond Estimator

Labor Quality Capital Intensity
Entire Late Entire Late Early
Early 2010s
Period 2000s Period 2000s 2010s
(1) (2) (3) (4) (5) (6)
ln ௧ିଵ -0.779** -0.001 -1.164*** -0.273*** 0.005 -0.757***
(Employment t-1) (0.313) (0.062) (0.313) (0.104) (0.064) (0.249)
ln 0.798* 0.669*** -0.397 0.294*** 0.353*** 0.204
(Ave.Wage) (0.470) (0.154) (0.521) (0.091) (0.122) (0.174)
ln
0.012 0.046* 0.005 0.024* 0.041 0.003
(Export) (0.089) (0.025) (0.018) (0.012) (0.035) (0.015)
ln -0.022 0.318*** 0.055 0.228*** 0.323 0.059
(Cap. Intens.) (0.288) (0.115) (0.041) (0.041) (0.269) (0.310)
ln 3 0.007 -0.024 0.224 -0.009 -0.002 0.147
(Mkt. Compet.) (0.138) (0.024) (0.157) (0.017) (0.021) (0.119)
ln 1.131** 0.629** 0.774 0.436*** 0.534 0.780
(Output Tariff) (0.501) (0.262) (0.560) (0.100) (0.684) (0.817)
-0.335** -0.181** -0.208 -0.084*** -0.090 -0.137
(0.155) (0.077) (0.148) (0.018) (0.122) (0.138)
ln -0.070 -0.053 -1.166** 0.069 1.113 -0.432*
(Input Tariff) (0.163) (0.099) (0.535) (0.051) (0.128) (0.226)
0.022 0.018 0.321** -0.015 0.023 0.086*
(0.057) (0.030) (0.155) (0.011) (0.027) (0.048)
# of Groups 448 446 447 448 446 447
# of Instruments 17 19 24 29 19 24
AB Test for AR(1) 2.46** 2.90*** 2.49** 2.57** 2.50** 2.42**
AB Test for AR(2) -0.06 -1.16 -1.47 -1.15 -0.99 -1.32
Hansen J Test 1.55 3.28 4.89 9.17 4.89 9.61
Observations 2,663 1,331 1,332 2,663 1,331 1,332
2. Figures in parentheses are robust standard errors.
3. AB represents Arellano-Bond.
16
5. Conclusion
This paper empirically examines how differently output and input tariffs affect domestic
employment across industrial characteristics. Korea which is a natural resource-scarce open
economy has pursued FTAs with other natural resource-abundant countries and thus greatly
increased imports of intermediate inputs. Accordingly, better access to imported
intermediate inputs as well as export expansion is important when analyzing trade gains in
Korea. Based on the Bernard et al.’s (2007) and the Luong’s (2011) theoretical frameworks,
this paper sets up two research hypotheses regarding the heterogeneous roles of industrial
characteristics in the stimulation effects of reductions in output and input tariffs on domestic
employment. The sample consists of 453 sectors from 2007 to 2014 in Korea.
The empirical results show that input tariff reduction stimulated total employment,
focusing on the early 2010s, while the effects of output tariff reduction were statistically
insignificant, especially in the long run. In spite of its ambiguous effects in the entire
industries, the stimulation effects of output tariff reduction on employment were found in
comparative advantage industries in which labor quality and capital intensity were relatively
greater. For input tariff reduction, however, the results were the opposite: its stimulation
effects on employment were more prominent in comparative disadvantage industries with
lower labor quality and capital intensity. These opposite results between input and output
tariffs look very similar to Ma and Dei’s (2009) for wage inequality and Luong’s (2011) for
firm productivity. All these results, including this paper, imply that better access to imported
intermediate inputs through free trade policies improves imbalanced welfare across
industries as well as production efficiency.
Consequently, this paper provides significant implications for other resource-scarce open
economies as well as Korea, which are experiencing the transition of the export pattern from
labor intensity to technology intensity and the unequal distribution of income after trade
liberalization. For these countries this paper addresses that imported intermediate inputs has
become increasing important, leading to trade effects on employment growth and alleviation
of income inequality. In particular, the economic benefits from better access to imported
intermediate inputs will be greater when considering their linkage effects on service
industries. I will leave this topic on industrial linkage between manufacturing and service
industries to future work.
References
Acemoglu, D., D. Autor, D. Dorn, G. Hanson and B. Price (2016), “Import Competition and the Great
US Employment Sag of the 2000s”, Journal of Labor Economics, 34(1), 141-198.
Amiti, M. and J. Konings (2007), “Trade Liberalization, Intermediate Inputs and Productivity:
Evidence from Indonesia”, American Economic Review, 97(5), 1611-1638.
Arellano, M. and S. Bond (1991), “Some Tests of Specification for Panel Data: Monte Carlo Evidence
and an Application to Employment Equations”, Review of Economic Studies, 58(2), 227-297.
Aw, B., S. Chung and M. Robers (2000), “Productivity and Turnover in the Export Market: Micro-
level Evidence from the Republic of Korea and Taiwan (China)”, World Bank Economic Review,
41(1), 65-90.
Bae, Chan-Kwon, J. Kim, H. Keum and Y. Jang (2012), The Impact of Free Trade Agreements on
Economic Performance in Korea (Policy Analyses, No. 12-03), Sejong, Korea: Korea Institute
17
for International Economic Policy.
Baier, S. and J. Bergstrand (2007), “Does Free Trade Actually Increase Members’ International
Trade?”, Journal of International Economics, 71(1), 72-95.
Bernard, A., J. Eaton, B. Jensen and S. Kortum (2003), “Plants and Productivity in International
Trade”, American Economics Review, 93(4), 1268-1290.
Bernard, A., S. Redding, S. and P. Schott (2007), “Comparative Advantage and Heterogeneous
Firms”, Review of Economic Studies, 74(1), 31-66.
Cameron, A. and P. Trivedi, P. (2005), Microeconometrics: Methods and Applications, Cambridge,
UK: Cambridge University Press.
Costinot, A. (2009), “On the Origins of Comparative Advantage”, Journal of International
Economics, 77(2), 255-264.
De Hoyos, R. and L. Iacovone (2013), “Economic Performance under NAFTA: A Firm-Level
Analysis of the Trade-Productivity Linkages”, World Development, 44(C), 180-193.
Feng, L., Z. Li and D. Swenson (2016), “The Connection between Imported Intermediate Inputs and
Exports: Evidence from Chinese Firms”, Journal of International Economics, 101(c), 86-101.
Fernandes, A. (2007), “Trade Policy, Trade Volumes and Plant-level Productivity in Colombian
Manufacturing Industries”, Journal of International Economics, 71(1), 52-71.
Goldberg, P., A. Khandelwal, N. Pavcnik and P. Topalova, P. (2010), “Imported Intermediate Inputs
and Domestic Product Growth: Evidence from India”, Quarterly Journal of Economics, 125(4),
1727-1767.
Hahn, Chin-Hee (2004), Exporting and Performance of Plants: Evidence from Korean Manufacturing
(NBER Working Paper, No. 10208), Cambridge, MA: NBER.
Jang, Yong-Joon, M. Cho and H. Kim (2015), “Trade Liberalization and Firm Productivity: Evidence
from Korea”, Journal of Korea Trade, 19(4), 21-41.
Kasahara, H. and J. Rodrigue (2008), “Does the Use of Imported Intermediates Increase Productivity?
Plant-level Evidence”, Journal of Development Economics, 87(1), 106-118.
Kim, Bae-Geun. (2014), “An Empirical Analysis of the Effects of a Change in Market Structure on
Production and Employment in the Korean Economy”, Journal of Economic Theory and
Econometrics, 25(2), 1-39.
Kim, Young-Gui, J. Kang, H. Kim and H. Hyun (2011), A Study on the Determinants and Productivity
Spillover Effects of Korea’s Intermediate Goods Trade (Policy Analyses, No. 11-13), Sejong,
Korea: Korea Institute for International Economic Policy.
Kim, Young-Gui, H. Park, H. Keum and S. Lee (2017), “Determinants of Korea’s Import Structure
and Its Effects on Firm’s Distribution”, World Economy Brief, 7(21).
Kletzer, L. (2001), Measuring the Costs of Trade-Related Job Loss, Washington, DC: Peterson
Institute for International Economics.
Leontief, W. (1953), “Domestic Production and Foreign Trade: The American Capital Position Re-
examined”, Proceedings of the American Philosophical Society, 97, 332-349.
Liu, Q. and L. Qiu (2016), “Intermediate Input Imports and Innovations: Evidence from Chinese
Firms’ Patent Filings”, Journal of International Economics, 103(c), 166-183.
Luong, T. (2011), “The Impact of Input and Output Tariffs on Firms’ Productivity: Theory and
Evidence”, Review of International Economics, 19(5), 821-835.
Ma, Y. and F. Dei (2009), “Product Quality, Wage Inequality and Trade Liberalization”, Review of
International Economics, 17, 244-260.
MacDonald, J. (1994), “Does Import Competition Force Efficient Production?”, Review of Economics
and Statistics, 76(4), 721-727.
Pavcnik, N. (2002), “Trade Liberalization, Exit and Productivity Improvements: Evidence from
Chilean Plants”, Review of Economic Studies, 69(1), 245-276.
18
Rose, S. (2018), Is Foreign Trade the Cause of Manufacturing Job Losses?, Washington, DC: Urban
Institute.
Song, S. (2018), Establishment Size and Wage Inequality: The Roles of Performance Pay and Rent
Sharing (BOK Working Paper, No. 2018-4), Seoul: Bank of Korea.
Topalova, P. and A. Khandelwal (2011), “Trade Liberalization and Firm Productivity: The Case of
India”, Review of Economics and Statistics, 93(3), 995-1009.
Trefler, D. (1993), “International Factor Price Differences: Leontief Was Right!”, Journal of Political
Economy, 101(6), 961-987.
Whang, Un-Jung, S. Lee, H. Kim and Y. Kang (2017), The Employment Effects of Exports (Policy
Analyses, No. 17-18), Sejong, Korea: Korea Institute for International Economic Policy.
Whang, Uu-Jung (2019), “Exports and Job Creation in South Korea: industry-level analysis”, Journal
of Korea Trade, 23(1), 2-18.
Yu, M. (2014), “Processing Trade, Tariff Reductions and Firm Productivity: Evidence from Chinese
Firms”, Economic Journal, 125(585), 943-988.
https://doi.org/10.35611/jkt.2020.24.8.19 19
The Impact of COVID-19 and Korea’s New

Southern Policy on Its Global Value Chain
Jeong-Ho Yoo JKT 24(8)
Trade, Investment and Innovation Division, United Nations ESCAP, Thailand
Seul-Ki Park
Department of FTA Policy and Business Consulting, Inha University, Korea
In-Kyo Cheong†
Department of International Trade, Inha University, Korea
Abstract
Purpose – The Korean government has been promoting the New Southern Policy (NSP) prior to the
onset of the COVID-19 pandemic, which damage global value chain (GVC). The purpose of this paper
is to emphasize that the NSP should be developed to provide tangible support in corporate GVC
adjustment, away from diplomatic activities in order to offset GVC losses due to COVID-19 and
expand export capabilities.
Design/methodology – Two research methodologies are combined for this paper: A computational
general equilibrium (CGE) model is used to estimate the impacts of the COVID-19 pandemic and
NSP on Korea’s exports, and the decomposition methodology (Wang, Wei and Zhu, 2013) to evaluate
the stability of GVC. The conventional CGE model was modified to obtain an estimate for
decomposition. The research methodology adopted in this study was attempted for the first time, and
it can be widely used in future GVC research.
Findings – Results found the effects of COVID-19 reduced Korea’s total exports by 27% and GVC by
more than 30%. In particular, VA in Korea’s exports to the NSP region was found to have a huge
impact in heavy industries and textiles, and its exports to Vietnam seemed to suffer the largest loss in
GVC among ASEAN countries. If the NSP is implemented properly, it appears that it could offset
much of the negative impacts of COVID-19, implying the importance of the effectiveness of the NSP.
Originality/value – Many papers have assessed the NSP descriptively, and the GVC has been a topic
for many publications. However, the impact of COVID-19 on Korea’s GVC with the NSP countries
has not been quantitatively studied. This paper emphasizes that the NSP should be pursued based on
the results of quantitative analysis. In addition, the research methodology of this paper can be used
for other GVC research with relevant modifications.
Keywords: COVID-19 Pandemic, Global Value Chain (GVC), GVC Adjustment, New Southern
Policy (NSP), Vertical Specialization Trade
JEL Classifications: F13, F14, F17
1. Introduction
As the US-China trade conflict intensified, Korea’s heavy dependence on China in terms
of its exports and production networks needed to be diversified and adjusted. As a result, the
Korean government decided to promote the New Northern Policy (NNP) and New Southern
Policy (NSP) in 2018. The NSP seeks to expand and deepen economic relations with ASEAN
countries and India. To this end, the Korean government established a new special committee
for the New Southern Policy. ASEAN countries and India can be good partners for Korea in
†
Corresponding author: inkyoc@gmail.com
20
adjusting its global value chain (GVC), which is the international division of production tasks
for the reduction of production costs and the creation of high value added (VA), based on
improved efficiency and market access for intermediate goods and final assembly.1
The NSP faces the unexpected environmental change with the outbreak of COVID-19 this
year. Korea, which followed China as the country with the most serious COVID-19 crisis
early this year, had immediate success in overcoming the COVID-19 pandemic. However,
Korea is currently facing a second crisis of COVID-19 that started in mid-August 2020. As a
result, the Bank of Korea (2020) lowered economic growth forecasts to -1.3% for this year. In
order to prevent further deterioration in economic growth, Korea should reorganize its GVC
to be more resilient as quickly as possible and establish an export system that can adapt
appropriately to changes in the trade environment. In this respect, the NSP has great
implications for companies and the national economy.
Many countries blocked entry as a preventative measure after the declaration of the
pandemic by the World Health Organization (WHO) on March 11, 2020. People’s
movements were restricted, and production facilities were shut down, including the NSP
region. As a result, demand and supply were damaged on a global scale. Many research
institutes are projecting that COVID-19 will push the reorganization of GVC (Foreign Policy,
2020; McKinsey Global Institute, 2020; OECD, 2020b; UNCTAD, 2020; World Bank, 2020;
WTO, 2020a), in addition to severe economic depression globally.
GVC is a concept related to a company’s production activities, so it is very difficult to
analyze macroeconomically. This is due to the difficulty in establishing a data system at the
national level containing the information on the change in the procurement method of
individual companies’ parts and intermediate goods (WTO, 2017).2 This is why the number
of papers on the quantitative analysis of GVC is relatively low, while there is a lot of
descriptive literature on GVC. In addition to data, the analysis of the COVID-19 crisis on
Korea’s GVC requires a set of complex simulation tasks. The effect of COVID-19 on GVC
should be estimated first, and these results should be decomposed to several VA indices. The
same methodology will be applied for Korea’s NSP in order to evaluate the impact on its GVC
with NSP countries.
Although the Korean government is promoting the NNP and the NSP, the NNP has not
made the expected progress. The NSP, which sets ASEAN and India as targets for
cooperation, has had positive results. Korean companies consider ASEAN countries, which
are geographically close and active in economic cooperation, as an alternative production
base for China (KITA, 2020). Indeed, trade and investment with ASEAN have increased in
recent years seen in Kwak Sung-Il (2020), and a series of KIEP reports such as Lee Jae-Ho
(2019). India is also an NSP target area, and many Korean companies are operating
production facilities there.
It has long been pointed out that Korea may be the biggest victim of the US-China conflict.
This is because Korea’s trade dependence on the two countries is approximately 40%, and
Korea is one of the countries with the highest trade dependence in the world. The NSP was
designed to escape from this trap. The outbreak of the COVID-19 pandemic has created poor
external conditions for Korea’s GVC. The purpose of this paper is to emphasize that the NSP
should be developed to provide tangible support in corporate GVC adjustment to offset GVC
1
One-third of world production is made by Multinational Corporations (MNCs), and cross-border
transactions by MNCs account for half of world trade (OECD, 2018). In addition, 70% of international
trade is trade of raw materials and intermediates required for production, implying GVC trade (OECD,
2020a).
2
Although it is deeply related to the actual business, it is difficult to systematically construct the related
data, so the vast GVC literature uses descriptive research methods.
21
losses resulting from COVID-19. This paper attempts to estimate the impact of the COVID-
19 pandemic and Korea’s NSP on its GVC with the NSP region.
2. Literature Review
Following the COVID-19 pandemic declaration, countries blocked entry, ordered
containment and self-isolation, and the world economy came to a halt. Since then, economic
activity has partially recovered, but many countries still do not allow cross-border movement.
Research on the effects of the NSP under COVID-19 is currently lacking, although the
literature on the effects of COVID-19 on GVC is rich. The relationship between COVID-19
and GVC is first reviewed and then summarized on the NSP and Korea’s GVC with the NSP
countries.
2.1. GVC after the Covid-19 Outbreak

In the current situation, there is a limit to grasping the true impact of the pandemic. It is
predicted that the world before COVID-19 will be markedly different from the world after
COVID-19. Foreign Policy (2020) reported the views of world-renowned scholars such as
Richard N. Haass, chairman of the Council for Foreign Relations (CFR), regarding the
question “How the Economy Will Look After the Coronavirus Pandemic?” All respondents
expect massive economic losses due to the shut-down and blockade. Professor John Ikenberry
of Princeton University suggested strengthening competition between countries, the strategic
decoupling of the US and China, and the negative impacts of globalization in addition to the
economic damage. Dr. Gita Gopinath of the IMF warned that the international movement of
people could be brought back to the situation in the 1970s, and this will have a huge impact
on GVC. Many countries may pursue domestic economic activity due to supply chain
vulnerability, and supply chains may be localized rather than global outsourcing. Chairman
Haass of the CFR predicted that many countries would pursue selective self-sufficiency, and
Robin Niblett, director of Chatham House, also raised questions about maintaining long-
distance supply chains.
After the global financial crisis in 2008, GVC weakened, and structural changes appeared
in the weakening of the manufacturing GVC, the strengthening of the service GVC, the
reduced importance of labor costs in GVC, and the increase of GVC participation in the
knowledge-intensive sector (WTO, 2018). Also, the localization of GVC was remarkable. The
WTO (2020a) is concerned about rising trade costs due to the pandemic, implying severe
damage to GVC. Trade costs may reduce after the coronavirus is overcome, but some may
persist due to changes in the policy environment or market dynamics. In particular, the WTO
expects that the cost of travel and air transport will not decrease for a considerable period.
COVID-19 ravaged production and demand globally through its global supply chains
(Baldwin and Tomiura, 2020). In the early phase of the pandemic, the supply of major parts,
mainly in the automobile and electronics industries, was delayed or restricted, resulting in
turmoil in the global supply chain (Haren and Simchi-levi, 2020).
Researchers have a variety of opinions on the impact and response of COVID-19 on GVC.
According to Urata (2020), automobile production in Japan declined by 10% in February
compared to last year due to supply disruptions from China, and by 45% in April because of
the expansion of the virus in Japan. The OECD (2020b) is proposing reshoring to home
countries or configuring various intermediate suppliers to make GVC more resilient to
external shocks. These proposed options are also supported by the studies by Inoue and Todo
22
(2017) and Zhu, Ito and Tomiura (2016) that studied the GVC of companies in affected areas
of the 2011 earthquake in East Japan. Bonadio et al. (2020) pointed out that the
nationalization of GVC may worsen economic performance in a pandemic situation, and
Baldwin and Freeman (2020) refers to the situation in which global production activities were
suspended, as governments took measures.
GVC enables an optimal production configuration, lowering the unit cost of products, and
splitting production tasks over several countries; that is, production is fragmented, depending
on the international competitiveness of each production task (OECD, 2013). In addition, the
improvement in the competitiveness of certain production tasks can attract multinational
corporations (MNCs), so GVC plays a pivotal role in economic development and poverty
eradication in developing countries (World Bank, 2019).
2.2. Korea’s New Southern Policy and COVID-19

The government of Korea adopted the “New Southern Policy” for increasing the country’s
focus on ASEAN and Southern Asian countries such as India. In order to cope with the
reorganization of the global economic order caused by US-China tension while reducing
excessive dependence on China, the Moon Jae-in administration recognized that it became
necessary to diversify trade and foreign investment. Korea is now oriented to have better
diplomatic relations in coping with the challenges that threaten national security.
Korea’s NNP, which consists of 16 key tasks such as the development of the Arctic logistics
route, may have a great potential economic effect, but many political variables pose risks. In
addition, due to delays in improving relations with North Korea, it may take time to
commercialize. On the other hand, the NSP is currently forming a close trade-investment
relationship and is a region of interest for Korean companies due to the spread of global
protectionism and conflicts between the US and China. Moreover, the Korean government
has upgraded the organization and staff at the Korean Embassies in the region. President
Moon has also visited most ASEAN countries to support ASEAN-Korea cooperation.
Research on economic cooperation is also more frequent in the NSP than in the NNP.
Although Jeong Jae-Won (2020) examined the economic effects of the NNP, its feasibility is
low. But many studies are found for the NSP. Lee Keun, Szapiro and Mao (2018) researched
ways to use ODA to expand the GVC with the NSP regions. Shim Jae-Hee (2019) analyzed
the Trade Complementarity Between Korea and NSP Countries, suggesting closer economic
cooperation as a way to minimize the damage of the US-China conflict. Cho Choong-Jae et
al. (2018) discusses measures to stimulate Korea-India economic cooperation to realize the
NSP. Na Seung-Kwon, Lee Sung-Hee and Kim Eun-Mee (2018) analyzed Japanese and Chinese
foreign policies in the ASEAN region from the perspective of soft power and suggested
implications for the NSP. Lee Jae-Ho (2019) has examined Korea’s investment in the ASEAN
region since the announcement of the NSP, and Han Hyoung-Min et al. (2019) analyzed
foreign direct investment (FDI) by country, trying to provide policy implications for Korea’s
NSP.
Comparing the progress of economic cooperation so far, the NSP is more dominant than
the NNP (Kim Jong-Bub, 2019), although the former has made larger improvements. Many
Korean companies exited China and have reestablished their production bases in other
ASEAN countries, including Vietnam and India. As of the end of 2018, Korea’s cumulative
investment in ASEAN amounted to US$ 61.9 billion, and the number of Korean companies
in the region reached 14,680. In 2018, the year after the NSP was announced, investment in
ASEAN nations was made by 1,291 Korean companies, and the investment amount was $6.13
billion, with growth ratios of 14.1% and 16.7%, respectively. As investments in ASEAN
increased, Korean financial institutions increased local businesses. According to Kim Jeong-
23
Han and Seo Byeong-Ho (2018), domestic banks are relatively active in opening their offices
in ASEAN nations. At the end of 2017, the number of overseas branches of domestic financial
companies was 431, of which 157 were operating in the ASEAN region.
2.3. Quantitative Studies on Korea’s GVC with NSP Countries

Research on the GVC with NSP countries has been actively conducted in Korea in recent
years. Most studies were conducted prior to the COVID-19 outbreak, so most do not reflect
any COVID-19 effects. In addition, many studies focused on GVC analysis with ASEAN
nations, and it is difficult to find papers that quantitatively examined GVC with India. The
GVC analysis method in Wang, Wei and Zhu (2013, abbreviated as WWZ) has been used in
many studies, and the multi-regional input-output (ADB-MRIO) framework developed by
the ASEAN Development Bank (ADB) has been widely used as a database. Taking the results
of the five studies mentioned, the ASEAN-Korea GVC is still weak. Although there are
differences depending on the study, it can be seen that Korea has formed one of the strongest
GVCs with Vietnam among the 10 ASEAN member countries.
Table 1. Summary of Studies on Korea’s GVC with NSP Countries

Methodology and
Researcher Major findings
data base
KIEP(2019) WWZ(2013), Korea’s deep GVC with some of NSP countries,
ADB-MRIO database especially Vietnam and Singapore
KIEP(2018) WWZ(2013), Only Vietnam among ASEAN countries has a
ADB-MRIO database deep GVC with Korea.
Yoo Jeong-Ho WWZ(2013), The foreign VA share of intermediate goods is
and Lee Jun- OECD-ICIO database relatively high in ASEAN countries.
Yeop (2019)
KOTRA(2019) WWZ(2013), Korea has high GVC structure with Malaysia,
EoRA-MRIO Thailand and Indonesia.
The Korea Institute for International Economic Policy (KIEP) published two volumes of
GVC research with NSP regions. KIEP (2018) suggested that only Vietnam among the
ASEAN countries has a deep GVC with Korea. Laos and Cambodia have a deep GVC
structure with China, while the Philippines and India have a strong GVC structure with the
US. Japan is known to have developed a high GVC with Indonesia. KIEP (2019) showed that
the ASEAN region showed a significant increase in domestic VA exports, and it was analyzed
that Korea’s GVC participation was higher with ASEAN than other regions, such as RCEP
and NAFTA. In particular, the use of GVC in the NSP region is steadily improving, and the
regional value chain participation is increasing. This has led to the expansion of regional
production divisions.
Yoo Jeong-Ho and Lee Jun-Yeop (2019) found that ASEAN countries are relatively located
in downstream industries, importing capital goods, and exporting consumer goods. From the
viewpoint of VA trade, the foreign VA of intermediate goods is relatively high in ASEAN
countries. In addition, the degree of GVC participation by ASEAN countries varies
considerably, and the proportion of vertical specialization in Singapore, Vietnam, and
Malaysia is high. KOTRA (2019) found that the GVC participation of the NSP countries was
centered on the forward GVC and that Korea’s regional value chain with NSP countries was
deepened over time. It was found that Korea’s participation is high in forward GVC with
Malaysia and Thailand, and backward GVC with Indonesia.
24
3. Methodology and Simulation Scenarios
The CGE model is widely used to analyze the impacts of a specific policy. The impacts of
COVID-19 and the NSP on international trade can also be analyzed with the CGE model. In
order to analyze the impacts of these issues on GVC, a GVC decomposition methodology is
required, and the CGE model must also be modified to produce data suitable for GVC
analysis. Here, the research methodology is briefly presented first, and the GVC analysis
methodology is then presented. This is followed by the modification of the conventional CGE
model.
3.1. Methodology
Research methods can be roughly divided into two parts. First, the impacts of COVID-19
and the NSP on trade is estimated, and then the resulting estimates are analyzed using the
GVC decomposition method. The whole research methodology is presented in Fig. 1. By
combining the CGE model and the GVC methodology, the impacts of the COVID-19 and
the NSP on GVC are estimated. The data derived includes domestic VA exports (DVA),
foreign VA exports (FVA), re-import after exports of intermediate goods (RDV), and double
accounting (PDC). The effects on GVC are examined. Korea’s exports are shown in Section
3.2. Although the CGE model and GVC methodology are built with many equations, this
paper has presented the main ideas briefly so that readers can grasp the methodology.
Fig. 1. Structure of Research Methodology
GVC analysis cannot be performed with gross trade data, which can be seen in export and
import statistics. This is because gross trade data differs from value added (VA). Korea’s
exports in 2019 were $542.4 billion, but a part of it was Korea’s VA, and the remaining VA is
provided to foreign suppliers. Therefore, a method of extracting VA from trade statistics is
needed.
Intermediate inputs in production can be provided domestically or imported. A value chain
can be formed when imported intermediates are used. The importance of GVC is evident,
but the method of measuring GVC is extremely limited due to the difficulty of obtaining data.
In view of these difficulties, Hummels, Ishii and Yi (2001) devised the concept of vertical
specialization (VS). It is the value of imported intermediate goods included in producing
exports, and country j’s VS ratio in total exports can be calculated as follows:
25
ೕ ∑೔ ೕ೔ ∑೔ೕ೔ /ೕ೔ ⋅ೕ೔ ೕ೔ ೕ೔
∑೔ ೕ೔
∑೔ ೕ೔
∑
⋅ (1)
ೕ ೕ ೕ೔
ೕ

/
/ (2)
ೕ
where u : 1×n vector of ones, AM : n×n imported coefficients, X : n×1 vector of exports, Xj :
country j’s total exports, I : n×n identity matrix, AD : n×n domestic coefficients, n : number of
sectors.
The VS ratio can be calculated either by using a weighted average of VS shares across
sectors (eq. 1) or with a matrix of imported coefficients (domestic coefficients) (eq. 2). This is
useful for analyzing the GVC of a country. Johnson and Noguera (2012), Wang, Wei and Zhu
(2013), Koopman, Wang and Wei (2014) and others have devised methods to extend the VS
of Hummels to a multi-country framework. Their research develops an accounting
framework that can calculate DVA, FVA, RDV, and PDC in the context of production
fragmentation across several countries, thereby enabling international comparison of GVC.
For this, major matrix calculations are done as follows:
Gross exports can be divided into intermediate and final goods’ exports.

(3)

where X : gross export vector of Country s, : final demand vector that gives demand in
Country r for final goods produced in s,
: IO coefficient matrix, giving intermediate use
in r of goods produced in s.
Gross output is absorbed according to the following networks.
௥ 1 ீ ିଵ
௥௥ ௥௥ 1 ீ ିଵ
௥௥ ௥௦ 1 ீ ିଵ
௥௦ ௦௦ 1 ீ ିଵ
௥௦ ௦௥ (4)
whereX : gross output vector of Country s,

: n×n Global IO coefficients , Y : demand
vector of Country r, sourced from Country s.
Finally, the VA included in exports is decomposed as follows;

(5)
whereA : Input-Output coefficient, when Country s export to Country r, ∗ : Leontief vector
of Country r, when Country * export,
Equation (5) presents the decomposition as VA = DVA + FVA + RDV + PDC.
This paper estimates the effects of COVID-19 and the NSP using the GTAP CGE model.
The GTAP model was developed by a research team led by Professor Tom Hertel of Purdue
University and spread internationally. It is the most widely used CGE model worldwide, as it
provides a CGE model and related database.3 The GTAP model is a global economic model
and defines the behavioral equations of major economic actors (consumers, companies,
governments) as the method used in economics (Leontief, Cobb-Douglas: CD, Constant
Elasticity of Substitution: CES). A country’s income is distributed as consumption,
government expenditure, and savings in the CD method, and demand for goods and services
is determined by the CES specification. In the case of production, the Leontief (fixed
coefficient) structure is adopted for the highest tier of decision making, and the CES is applied
to the demand for production factors or intermediate inputs (Fig. 2). The GTAP model is
3
Regarding the value of GTAP database, refer to McDonald and Thierfelder (2004), Rutherford (2005)
and Aguiar et al. (2019).
26
described in detail in Hertel and Tsigas (1997).
For CGE users, the GTAP database is very useful. This is because, in addition to data on
trade, consumption, government spending, and taxes, the Social Accounting Matrix (SAM)
required for CGE is assembled and systemized in a single system. The database, which may
take several years in assembling, is systematically compiled by the GTAP Center and is
designed to be convenient for users (Hertel and Tsigas, 1997). Users of the CGE model other
than GTAP also use this database (Aguiar et al., 2019; Rutherford, 2005). Version 10 of the
GTAP database was used, which includes 141 countries and 65 industries.
Fig. 2. Structure of the GTAP CGE Model
National economic structure Income distribution
Structure for consumer’s demand Top structure for producer
Source: Various sources such as Hertel and Tsigas (1997).
For this study, the regions and industries of the GTAP database mentioned earlier are
aggregated, as shown in Table 3. Korea, individual ASEAN countries, India, China, and the
US were classified as individual countries, the 28 EU member states were classified as the EU,
and the remaining countries were classified as rest of the world (ROW). Since there are as
many as 16 countries/regions, CGE simulation is facilitated only when the number of
industries is reduced as much as possible. The economy of a region is divided into 10
industries, which is commonly disaggregated.
Table 2. Aggregation Scheme for Country/Region and Sector

Country/Region (16) Sector (10)
Korea, individual members of Agriculture, Textile-apparel, chemicals,
Classification
ASEAN countries, India, China, high-tech industries, steel, machinery,
USA, EU, ROW automobiles, heavy industry, light industry
In the WWZ (2013) method of analyzing GVC, trade statistics have one blind spot. That
is, the use of goods imported by a country is unknown. For example, when a billion dollars
worth of auto parts are imported, it is not known how much would be used to manufacture
exports in most cases. GTAP uses the Armington (1969) specification to determine the
amount of imports for the simplification of the model, and these imports are distributed
among consumption (household and government) and intermediate goods (company) using
27
the CES function. Therefore, the GTAP model itself cannot generate the necessary country-
specific import data for manufacturing exports that is required in analyzing the impact on
GVC.
Peter, Andrew and Lennox (2011) provide a useful module in extending the data of the
GTAP using proportional assumption (Jonson and Noguera, 2012; Lejour, Rojas-Romagosa
and Veenendaal, 2014). 4 Following Lejour, Rojas-Romagosa and Veenendaal (2014), this
paper borrows the module of GTAP-VA based on the proportional assumption in calculating
necessary data for GVC analysis from the simulation of the GTAP model,5 decomposing VA
for the GVC analysis following WWZ (2013). Detailed information can be found in Peter,
Andrew and Lennox (2011), Antimiani and Fusacchia (2018).
3.2. Evaluation of Korea’s Exports with GVC Decomposition

The difference-in-differences (DID) analysis results of Korean exports of its major
intermediate goods (electrical-electronic parts, textiles, auto parts), as shown in Figure 3,
found that exports of electrical-electronic parts to China and auto parts to the US decreased
by 7.8% and 10.7%, respectively, comparing levels before and after the COVID-19 outbreak.
However, the export reduction of intermediate goods to ASEAN nations was relatively small
compared to those of the US and China.
Fig. 3. DID Results on Korea’s Major Exporting Intermediate Goods (%)
Note: Korea’s monthly export data used.

Source: Korea Trade Statistics Promotion Institute.
In order to consider the background for this finding, it is necessary to decompose Korea’s
GVC for the NSP region. For this, the GVC decomposition methodology of WWZ (2013)
was applied to the data of the 2017 ADB-MRIO, which includes India and six ASEAN
countries. The VA included in the total exports is decomposed into DVA, FVA, RDV, and
PDC, as presented in Fig. 1.
4
Early versions of the GTAP model in 1990s were built on extended version of data using proportional
assumption, which assumes imported goods to be distributed in equal ratios across sectors.
5
Alternative option to derive the additional data is to calculate bilateral trade ratios of intermediate and
final goods by classifying intermediate and final goods based on the UN BEC codes, as Walmsley, Hertel
and Hummels (2014) and Aguiar et al. (2016). Yet, this still has a limitation, since it needs an
assumption for the ratios, which is similar with proportional assumption.
28
For Korea’s GVC by country, the DVA ratio varies greatly from country to country. The
DVA for ASEAN nations (average) was lower than for China, the US, and India. However,
the FVA of Korea’s exports to ASEAN nations was 18.8% higher than that of other regions
that ranged from 7.3% to 12.9%. That is, the input ratio of imported parts is high for Korea’s
exports of intermediate goods to ASEAN nations. Linking this to the regional trade
performance shown in Figure 3 suggests that despite the novel coronavirus outbreak in the
first half of this year, Korea had little difficulty in procuring foreign intermediate goods
necessary for the production of intermediate goods for export to ASEAN countries.
Table 3. GVC Decomposition of Korea’s Exports to Major Countries

DVA FVA RDV PDC
ASEAN 66.4% 18.8% 0.51% 14.6%
Indonesia 86.5% 9.01% 1.22% 4.67%
Malaysia 61.0% 20.8% 0.19% 22.2%
Philippines 79.0% 12.3% 0.22% 8.09%
Singapore 43.4% 23.7% 0.21% 31.6%
Thailand 67.6% 17.9% 0.19% 10.5%
Viet Nam 61.0% 29.6% 0.19% 18.3%
India 77.5% 12.9% 0.77% 8.86%
China 75.6% 8.99% 7.71% 7.84%
US 84.7% 7.30% 3.85% 4.16%
Source: Author’s estimation based on ADB-MRIO.
RDV accounts for the VA included in re-imported intermediate goods, which were
exported after processing in a third country. The higher the RVC ratio, the longer the length
of GVC. In this regard, RVC can be interpreted as an indicator of how long GVC is spread
bilaterally. According to Table 2, the ratio of RDV in exports to ASEAN and India was
significantly lower than that of China and the US. This means that although Korea is
developing GVC with NSP countries, it remains at a shallow level that includes simple
processing. This can be seen as consistent with the survey results from the existing literature,
such as KIEP (2018), Yoo Jeong-Ho and Lee Jun-Yeop (2019). This implies that if production
contracts due to COVID-19, Korea’s GVC risk may be relatively small regarding the trade
with ASEAN nations and India. In other words, the GVC decomposition suggests that the
damage caused by COVID-19 can be directly seen in trade with China and the US in terms
of Korea’s GVC, while it is relatively easy to manage the risk in trade with ASEAN nations
and India. In this context, the reason why the COVID-19 shock to Korea’s exports of
intermediate goods to ASEAN nations, shown in Table 2, was relatively lower than that of
China, and the US can be understood in this context.
3.3. Scenario
As discussed in the previous section, the impact of COVID-19 on GVC varies by region. It
is expected that companies will adjust GVC to manage the risk caused by COVID-19. The
adjustment of the GVC is highly likely to take place by focusing on the NSP region currently
being promoted by the government of Korea. Therefore, the scenarios of this study not only
measures the damage of COVID-19 but also consider the effects of the NSP as a post-COVID-
19 strategy.
29
In the case of the NSP, there is heavy criticism that there is only a policy, and a concrete
plan of action has not been implemented yet. The three communities (people, win-win
prosperity, and peace) proposed by the Presidential Committee on New Southern Policy and
the 16 detailed strategies are similar to the grand development plan by ASEAN, and the NSP
lacks in specific measures to induce the interest of ASEAN countries (Choi Young-Jong,
2019).
As of the end of August 2020, the number of COVID-19 confirmed cases worldwide has
exceeded 25 million, and the death toll is approaching 85,000 (WHO COVID-19 Dashboard).
The number of new confirmed cases was about 285 thousand per day, and it is difficult to
predict the end of the COVID-19 pandemic. In April of this year, the IMF (2020) predicted
that the global economy would decrease 2.1% due to COVID-19 risk, and then increased its
predicted decrease to 4.9% in a subsequent report two months later. As in the worst-case
scenario presented in the IMF report, global economic growth could decrease by 10% or
more.
Since COVID-19 loss is difficult to predict, this paper considers three scenarios. Table 4
presents the 3 scenarios with variables for demand (y : income) and supply (ams : logistics
efficiency, afe : reduction in labor productivity),6 referring to the forecast in ADB (2020), PwC
(2020), and ILO (2020).7 In addition, the Korean government’s NSP strategy is included in
these scenarios. From an economic perspective, Korea’s NSP objective will be to adjust GVC
for risk management. ASEAN nations are active in strengthening economic cooperation with
Korea.8 The adjustment of GVC will lead to a continued increase in Korea’s investment in the
ASEAN region. Scenario 1 (S1) assumes a relatively low damage outcome, while scenario 3
(S3) sets the worst-case scenario. Currently, S2, which is between S1 and S3, is predicted as
having a likelihood of occurring.
Overseas investment and overseas M&A disruptions are serious due to the coronavirus
risk. However, if NSP can solve these problems, it can be of great help in extending GVC. In
general, investment occurs in countries with low risk, and in the GTAP model, the slack
variable for investment (cgdslack) is swapped with the rate of return (rore) to analyze the
investment effect of risk reduction (Malcolm, 1998). In this study, it is assumed that the GVC
risk caused by the COVID-19 outbreak could be reduced through GVC adjustment with the
NSP countries, and this is included in the scenarios.9
6
Considering the negative effects of COVID-19 on the supply side, exogeneous variables of productivity
(afe) and trade efficiency (ams) were adopted to be shocked. In the CGE model used for this paper,
these variables could be relevant for the lockdowns and cross-border restrictions on the movement of
natural persons, which result in the losses in productivity and trade efficiency. In addition to these,
income reduction was reflected with the variable of income (y).
7
In estimating the unprecedented impact of COVID-19 on GVC, it was very difficult for authors to
predict the impact of COVID-19 on the global economy. Accordingly, they prepared 3 simulation
scenarios and related shocks based on the economic outlooks by international economic institutions in
the middle of 2020 when this paper was written. It may appear somewhat arbitrary, but they chose
scenarios and shocks with the best judgment.
8
ASEAN is promoting high value-added industries through investment attraction and technology
transfer from Korea (ASEAN briefing, 2014).
9
The variable rore representing future capital investment returns is the sum of risk and rorg, which is the
global average return on investment. In the GTAP model, when RORDELTA is 1, it can be expressed
as rore(r) = rorg(r) + cgdslack(r). At this time, cgdslack is an investment risk.
30
Table 4. Summary of Simulation Scenarios
S1 S2 S3
Damage of Supply Productivity (afe) -3% -5% -10%
COVID-19 Trade efficiency (ams) -5% -10% -15%
Demand Income (y) -5% -10% -15%
New Investment Risk reduction (rore) -10% -15% -20%
Southern Spillover Labor productivity (afeall) 10% 15% 20%
Policy
Next, Korea’s continued investment in the ASEAN region could create a spillover effect.
Deng, Falvey and Blake (2008) explain that FDI causes a spillover effect, which in turn
contributes to productivity growth. In particular, permitting the priority transfer of essential
personnel to ASEAN countries, which the Korean government is pursuing after the COVID-
19 outbreak, will also have a positive effect on Korean companies’ investment into the ASEAN
region. Cross-border movement of essential personnel has become an important issue in
enhancing the productivity of investment in host countries (Javorcik, 2004; Liu et al., 2000;
Meyer and Sinani, 2009). In this study, this effect is reflected in the scenarios through the
productivity variable (afeall).
4. Simulation Results and Interpretation

Many estimates are produced by calculating the effects of COVID-19 and the NSP on
Korea’s GVC over several steps using the modified GTAP model and VA decomposition
method. However, in this paper, only the core content is presented, focusing on the VA
decomposition due to the limitation of pages.
4.1. Impact of COVID-19

The possible impact of COVID-19 on Korea’s total exports is presented in Table 5. Korea’s
total exports are expected to decrease from 9.4% to 27.2%. The estimates in this study are
close to the projections of the WTO (2020b) that global trade could decrease by 12.9%-31.9%
this year. For reference, the total export volume of Korea in the first half of this year was
$246.4 billion, which is down by 11.3% from the same period last year. The impact by
country/region shows a similar result as in Fig. 3. The decline in Korea’s exports to China and
the US is expected to be larger than the decline in total exports, while the shock on exports to
ASEAN nations, India, and the EU is expected to be relatively low.
Table 5. Impact of COVID-19 on Korea’s Total Export by Country/Region (Unit: %)

Total Export
S1 S2 S3
ASEAN -8.8 -16.7 -25.5
India -7.8 -14.3 -22.1
China -11.5 -21.5 -33.0
USA -15.5 -29.3 -44.8
EU -6.8 -12.8 -19.6
Total -9.4 -17.8 -27.2
31
The impact on export VAs was decomposed into DVA, FVA, and PDC. In the case of
domestically procured VA (DVA), depending on the scenario, it decreases by 8.9%-25.7%,
and looking at the shocks by country/region, exports to the US have the largest shock,
followed by 33.9% in Vietnam, and 31.7% in China.
It was noted that Hummels and his colleagues developed VS as an indicator to analyze the
impact on GVC. In Table 6, VS is calculated as the sum of FVA and PDC. It can be seen that
the imported VA (FVA) of Korea’s exports to the world can decrease by up to 29.3% (Scenario
3), and the double-counted VA (PDC) can decrease by up to 37.6%. Also, FVA and PDC are
expected to decline significantly in exports to the US. As the trade environment deteriorates
due to COVID-19, the results suggest that the GVC of producing goods for exports to the US
could be the most exposed to risk. The biggest GVC damage could happen to Vietnam,
excluding the US and China. This outlook is in line with Vietnam’s economic performance
after COVID-19.10
Table 6. Decomposition of COVID-19 Impacts on VA (Unit: %)

DVA FVA PDC
S1 S2 S3 S1 S2 S3 S1 S2 S3
ASEAN -8.4 -16.0 -24.4 -9.2 -17.6 -26.9 -12.0 -23.3 -35.3
- Cambodia -6.8 -12.8 -19.6 -6.7 -13.0 -19.7 -9.8 -18.4 -28.2
- Indonesia -8.1 -14.7 -22.8 -9.1 -17.1 -26.2 -9.9 -19.0 -28.9
- Laos -0.6 -0.6 -1.3 -0.0 -0.8 -0.8 -2.4 -4.8 -7.1
- Malaysia -6.5 -12.3 -18.8 -7.6 -14.6 -22.2 -9.8 -19.9 -29.7
- Philippines -4.5 -8.8 -13.2 -5.0 -9.9 -14.8 -7.4 -15.6 -23.0
- Singapore -7.5 -14.7 -22.2 -9.1 -17.8 -26.9 -11.3 -22.9 -34.2
- Thailand -7.6 -14.6 -22.2 -8.8 -16.9 -25.7 -11.4 -22.0 -33.3
- Vietnam -11.7 -22.2 -33.9 -12.2 -22.9 -35.1 -15.8 -29.7 -45.5
India -7.2 -13.0 -20.2 -8.6 -15.8 -24.5 -10.3 -19.9 -30.1
China -11.1 -20.7 -31.7 -11.9 -22.4 -34.3 -15.0 -28.5 -43.5
USA -14.9 -28.4 -43.3 -16.4 -30.9 -47.3 -19.7 -36.8 -56.5
EU -6.7 -12.7 -19.5 -6.8 -13.0 -19.8 -9.5 -18.9 -28.4
Total -8.9 -16.8 -25.7 -10.1 -19.2 -29.3 -12.8 -24.8 -37.6
This can be interpreted as showing the vulnerability of Korean exports. The production
bases of Korean companies established in China and Vietnam use many intermediate goods
made in third countries, and a large portion of the products produced are exported to the US.
If any of the interlocking GVC systems is disturbed, the operation of the entire production
system may experience significant slow-downs, as shown in Table 6.
The VA impact on the NSP countries in Table 6 is decomposed by industry in Table 7.
First, looking at the impact on the DVA indicator, the industries that are expected to have the
greatest impact across all scenarios are in the order of heavy industry, textile, pharmaceuticals,
10
According to Bloomberg (2020), Vietnam has successfully quarantined COVID-19, but economic
growth has deteriorated significantly due to a decline in demand in major export markets such as the
US. Samsung Electronics (Vietnam), which produces 20% of Vietnam’s total exports, also lowered its
export target this year from $45.5 billion in 2019 to $32 billion.
32
and chemicals. In the case of FVA, the greatest impacts are expected to occur in the order of
heavy industry, textile, pharmaceuticals, and light industry. In the case of the PDC indicator,
the shocks are expected to be largest in the order of heavy industry, textile, light industry, and
pharmaceuticals. From this table, it can be seen that if the COVID-19 risk prolongs in the
future, Korea’s heavy industry and textile will have a significant impact on exports to the NSP
countries, and the procurement of intermediate goods (FVA, PDC) required for production
is the most vulnerable. Based on this, it can be said that the NSP authority should hurry to
seek effective supply chain support measures for these industries.
Table 7. Impacts of COVID-19 on Korea’s Exports to NSP Region by Sector (Unit: %)

DVA FVA PDC
S1 S2 S3 S1 S2 S3 S1 S2 S3
Agriculture -12.2 -23.7 -35.9 -11.8 -23.0 -34.8 -13.7 -26.9 -40.5
Heavy Industry -19.3 -32.8 -52.0 -17.9 -30.9 -48.8 -23.1 -42.3 -65.4
Textile-Apparel -14.3 -27.0 -41.3 -14.6 -27.2 -41.8 -18.6 -34.2 -52.8
Light industry -11.1 -21.0 -32.1 -11.3 -21.2 -32.4 -14.4 -27.1 -41.5
Chemical -11.7 -22.5 -34.2 -10.5 -20.2 -30.7 -12.6 -24.2 -36.8
Pharmaceutical -13.8 -26.0 -39.8 -12.0 -22.8 -34.9 -14.1 -27.2 -41.3
Metal -8.0 -14.7 -22.8 -9.1 -16.7 -25.8 -11.7 -22.0 -33.8
High-tech -7.4 -14.3 -21.6 -7.7 -15.0 -22.7 -11.5 -22.6 -34.0
Machinery -2.6 -3.9 -6.5 -4.2 -6.8 -11.0 -7.6 -14.3 -21.9
Transport -1.7 -2.8 -4.5 -3.2 -5.7 -9.0 -6.2 -12.4 -18.7
4.2. VA Decomposition Including NSP Effects

The expansion effect of the NSP on Korea’s export VA is not separately reported due to
space restrictions, and the effects of COVID-19 and the NSP are presented in Table 8 for easy
comparison. The minimum and maximum values of VA effects are separated by commas in
[ , ]. Focusing on DVA, in the case of ASEAN as a whole, DVA in Korea’s exports could
decrease by 8.4% to 24.4% due to the effects of COVID-19. On the other hand, the negative
impacts can be partially offset by the NSP. Depending on the investment promotion policy,
the recovery can be 2.8%~7.1%, and the productivity improvement policy based on the
spillover effect is expected to offset 4.6%~10.1%. In addition, when both policies are used
simultaneously, the export DVA recovery effect ranges from 7.4% to 17.3%. The countries
with the highest growth in DVA through the NSP are Thailand and Singapore, which are
expected to increase up to 26.3% and 25.3%, respectively.
Table 8. Impacts of COVID-19 and NSP on Domestic Value Added (DVA) (Unit: %)
Damage of Recovery effects of the NSP
COVID-19 Investment Spillover Total NSP
ASEAN [-8.4, -24.4] [2.8, 7.1] [4.6, 10.1] [7.4, 17.3]
- Cambodia [-6.8, -19.6] [1.7, 3.4] [0.2, -0.9] [1.8, 2.4]
- Indonesia [-8.1, -22.8] [2.9, 7.0] [2.2, 4.0] [5.1, 11.0]
33
Table 8. (Continued) (Unit: %)
- Laos [-0.6, -1.3] [3.8, 10.1] [2.7, 7.7] [6.5, 17.8]
- Malaysia [-6.5, -18.8] [2.9, 6.7] [4.2, 8.8] [7.0, 15.5]
- Philippines [-4.5, -13.2] [3.3, 8.3] [4.5, 9.7] [7.8, 18.0]
- Singapore [-7.5, -22.2] [3.5, 10.4] [6.1, 14.9] [9.6, 25.3]
- Thailand [-7.6, -22.2] [4.7, 12.9] [5.6, 13.5] [10.3, 26.3]
- Vietnam [-11.7, -33.9] [1.4, 3.2] [5.2, 11.3] [6.6, 14.5]
India [-7.2, -20.2] [2.2, 5.7] [3.1, 6.1] [5.3, 11.7]
Note: number means [min, max] effects under each scenario.
The effects of COVID-19 and the NSP on Korea’s VS (FVA+PDC), which is the GVC
index, are summarized in Table 9. The effects of spillover are bigger than those of investment.
This is clear for India. In the case of ASEAN as a whole, VS exports could decrease from 9.3%
to 26.9% due to COVID-19. However, depending on the NSP scenarios, negative effects are
expected to recover by 7.8%-14.3%, although net effects remain negative. By country, exports
to Singapore can recover by up to 17.0%. The COVID-19 impact on Korea’s exports to India
is expected to be at a level similar to that of ASEAN as a whole, and export VA is expected to
recover by 15% if the NSP is realized under scenario 3.
Table 9. Impacts of COVID-19 and NSP on Vertical Specialization (VS) (Unit: %)

ASEAN [-9.3, -26.9] [2.7, 7.1] [5.2, 7.2] [7.8, 14.3]
- Cambodia [-6.8, -19.8] [1.8, 4] [-0.9, -3.1] [0.9, 0.9]
- Indonesia [-9.1, -26.3] [2.6, 6.4] [0.4, -0.3] [3.0, 6.1]
- Laos [0.0, 0.8] [4.3, 11.8] [-0.5, -1.2] [3.8, 10.6]
- Malaysia [-7.6, -22.3] [2.6, 6.4] [2.2, 4] [4.8, 10.4]
- Philippines [-5.0, -14.9] [3.1, 8.3] [2.1, 3.4] [5.2, 11.8]
- Singapore [-9.1, -27] [3.8, 11.3] [3.1, 5.7] [6.9, 17.0]
- Thailand [-8.8, -25.7] [4.3, 11.9] [1.9, 3] [6.2, 14.9]
- Vietnam [-12.2, -35.2] [1.4, 3.4] [4.3, 8.9] [5.7, 12.4]
India [-8.6, -24.5] [0.8, 2.9] [5.2, 12.2] [6.0, 15.0]
Note: number means [min, max] effects depending on the scenarios; VS effects are calculated as
summation of FVA and PDC.
The effects of COVID-19 and the NSP on RDV11, which is an international VA network
indicator, are reported in Table 10. In the case of ASEAN nations as a whole, it is analyzed
that the RDV index in exports to ASEAN nations could decrease by 10.7%-29.9% due to
COVID-19. This table suggests that the promotion of the NSP will be able to fully recover
from the negative effects of the virus. Korea’s exports to Indonesia, Malaysia, and the
11
RDV is drawn from the FVA_rfl in GTAP-VA model, which is defined as the VA embedded as re-
imported intermediates that reflect back to source after processing in the third country.
34
Philippines showed high recovery rates, showing a difference from the GVC indicator VS in
Table 10. Similar results are found for Korea’s exports to India.
Table 10. Impacts of COVID-19 and NSP on RDV Index (Unit: %)

Damage of Recovery effects of NSP
ASEAN [-10.7, -29.9] [3.4, 12.3] [11.1, 20.6] [14.5, 33.0]
- Cambodia [-11.7, -33.1] [-1.4, -3.4] [4.1, 8.3] [2.8, 4.8]
- Indonesia [-8.2, -24.6] [4.1, 16.4] [9.8, 22.5] [13.9, 38.9]
- Laos [0.0, -16.7] [0.0, 0.0] [16.7, 33.3] [16.7, 33.3]
- Malaysia [-7.7, -22.5] [4.6, 14.2] [7.7, 17.9] [12.3, 32.1]
- Philippines [-5.1, -17.3] [3.2, 10.3] [7.7, 17.3] [10.9, 27.6]
- Singapore [-13.8, -38.0] [3.2, 9.5] [4.0, 7.3] [7.3, 16.8]
- Thailand [-9.2, -26.7] [3.6, 10.8] [3.2, 6.0] [6.8, 16.7]
- Vietnam [-18.7, -47.2] [0.7, 2.5] [5.7, 12.3] [6.4, 14.7]
India [-9.1, -24.5] [0.3, 0.8] [9.9, 23.5] [10.2, 24.2]
Note: number means [min, max] effects depending on the scenarios.
5. Conclusion and Policy Implications

The post-COVID-19 trade environment can be characterized by intensifying conflict
between the US and China, incapacitating the multilateral trade system led by the WTO,
‘Nation First,’ restructuring GVC, digital transformation, and others. The COVID-19 crisis
served to disrupt the WTO-led multilateral system, and countries recognized the importance
of active trade policies. Considering the current increase in protectionism and bilateralism,
countries should prepare strategies to respond. Relations between the US and China have
been deteriorating since the enactment of the Hong Kong Security Act following the dispute
over COVID-19 liability, and the US is recommending its own companies to withdraw from
China.
According to the results of this paper, COVID-19 was found to reduce Korea’s total exports
by 27% and GVC by more than 30%. Korea’s exports have a GVC structure that is vulnerable
to COVID-19, and considering the conflict between the US and China, Korea’s exports are
bound to suffer a big blow. If decoupling between the US and China, which accounts for
approximately 40% of Korea’s total exports, becomes a reality, Korea’s exports and its GVC
will be severely affected, as discussed in Chapter 4 of this paper.
However, if the NSP develops to the point where it works in terms of investment and
spillover, it appears that it could offset much of the negative impacts of COVID-19. In this
respect, the NSP has important policy implications. This chapter attempts to suggest several
measures for the NSP.
The government should support companies’ supply chain adjustment with visible
incentives. The NSP is a representative foreign trade policy of Korea, and is operated in a
format supported by relevant ministries such as the Ministry of Foreign Affairs and the
Ministry of Trade, Industry, and Energy, and led by the NSP Special Committee under the
Blue House. Until now, the major programs of the NSP have been focusing on diplomatic
activities such as summit meetings. The level of Korea’s international cooperation has not
improved compared to 10 years ago. In addition, the level of GVC with the NSP countries has
35
not changed much, except that Korean companies have invested heavily in Vietnam in recent
years.
The US, EU, and Japan are announcing massive support to companies that exit from China.
Japan is supporting 70% of the cost of factory relocation for reshoring. The Korean
government should establish and support companies that are in a similar situation. The KITA
(2020) argues that the ASEAN region should be focused as a stable production base to replace
China. Korean companies are seeking to establish production bases in third countries, rather
than returning production to Korea due to the nation’s labor costs and regulations. While
ASEAN countries are also offering support measures to attract companies in relocation, the
government of Korea should provide tangible support for companies moving to the NSP
region even if they are not reshoring to Korea.
As cross-border human movement is prohibited due to the coronavirus, companies are
experiencing great difficulties in investing in the NSP region and local marketing. Therefore,
the government should address these problems in terms of trade policy. The government
should be active in expanding special entry procedures for entrepreneurs, facilitating trade
and investment, and revising the FTA and reforming regulations. Regional cooperation
organizations, such as ASEAN+6, the Regional Comprehensive Economic Partnership
Agreement (RCEP), and the Asia-Pacific Economic Cooperation (APEC), can be good
channels to discuss the introduction of a “Special Passport for Business Persons” system. As
Tables 8-10 suggest, the spillover effects from allowing international movement of people
were slightly higher than the investment effect. As the COVID-19 situation can be extended
indefinitely, the effect of the special passport system can be significant.
COVID-19 makes overseas M&A difficult, which involves long bargaining and many
negotiations. The Korea Trade-Investment Promotion Agency (KOTRA), which has offices
in 127 cities around the world, should expand its role in supporting corporate M&A activities.
Domestic companies must acquire chemical materials and electronic materials companies
with high technical capabilities to develop core intermediary procurement capabilities. This
will contribute to stabilizing GVC with the NSP region and help cope with Japanese export
regulations.
Due to the coronavirus risk, where overseas business trips are difficult, exporting
companies have no choice but to rely on “non-face-to-face” marketing. Moreover, marketing
in ASEAN and Indian markets is not easy. It is difficult to enter overseas markets without
digital competency and online marketing. Recently, in the ASEAN region and India, online
platform transactions such as e-commerce and social media are increasing rapidly. It is
necessary for the government to step up and build platforms to advance into the NSP region.
Right now, the trade authorities, including KOTRA, should support companies to improve
the accessibility of Korean companies to platforms such as Lazada and Shopee, which are
widely used in the ASEAN market.
Finally, the limitations of this paper should be mentioned. First of all, if the COVID-19
becomes much worse than it is today, there is a possibility that the methodology of this study
may not properly reflect these changes. In this case, a new analytical methodology may be
required. Next, setting up a scenario is always controversial. The authors would like to clarify
that this is a scenario that they have adopted with substantial consideration. Although
simulation estimates differ depending on the scenarios, the research results of this paper
clearly show the importance of the effectiveness of the NSP. The modified GTAP CGE model
and VA decomposition methodology consist of systems of many equations and matrix
operations. The authors are regretful about the brief presentation of the main ideas in this
paper due to space limitations. However, relevant content can be provided upon the request
of the reader.
36
References
ADB (2020), An Updated Assessment of the Economic Impact of COVID-19 (ADB briefs, No.133),
Manila: Asian Development Bank.
Aguiar, A., C. Carrico, T. Hertel, Z. Hussein, R. McDougall and B. Narayanan (2016), Extending the
GTAP Framework for Public Procurement Analysis (GTAP Working Paper, No. 82), West
Lafayette, IN: Purdue University.
Aguiar, A., M. Chepeliev, E. Corong, R. McDougall and D. van der Mensbrugghe (2019), “The GTAP
Data Base: Version 10”, Journal of Global Economic Analysis, 4(1), 1-27.
Antimiani, A., I. Fusacchia and L. Salvatici (2018), “GTAP-VA: An Integrated Tool for Global Value
Chain Analysis”, Journal of Global Economic Analysis, 3(2), 69-105.
ASEAN briefing (2014), Understanding ASEAN’s Free Trade Agreements. Available from:
https://www.aseanbriefing.com/news/understanding-aseans-free-trade-agreements/ (accessed
August 2, 2020)
Baldwin, R. and R. Freeman (2020), “Supply Chain Contagion Waves: Thinking Ahead on
Manufacturing ‘Contagion and Reinfection’ from the COVID Concussion”, VoxEU.org.
Available from: https://voxeu.org/article/covid-concussion-and-supply-chain-contagion-waves
(accessed August 2, 2020)
Baldwin, R. and E. Tomiura (2020), “Thinking Ahead about the Trade Impact of COVID-19”. In R.
Baldwin and B. W. D. Mauro (Eds.), Economics in the Time of COVID-19, London: Centre for
Economic Policy Research.
Bank of Korea (2020), The Impact of Covid-19 on GVC and Implications (BOK Issue Note, No. 2020-
10), Seoul: Bank of Korea.
Bloomberg (2020, August 26), “Three-Decade Economic Boom Comes to a Sudden Halt in Vietnam”.
Available from: https://www.bloombergquint.com/global-economics/three-decade-economic-
boom-comes-to-a-sudden-halt-in-vietnam
Bonadio, B., Z. Huo, A. Levchenko and N. Pandalai-Nayar (2020), Global Supply Chains in the
Pandemic (NBER Working Paper, No. w27224), Cambridge, MA: National Bureau of
Economic Research.
Cho, Choong-Jae, Young-Chul Song, Jung-Mi Lee, Do-Yeon Kim, Na-Youn Park and Chi-Hyun Yun
(2018), India-Korea Economic Cooperation for the New Southern Policy, Sejong, Korea: Korea
Institute for International Economic Policy.
Choi, Young-Jong (2019), “South Korea’s New Southern Policy and the Search for a New Regionalist
Strategy in the Emerging Indo-Pacific Era”, Journal of Northeast Regional Studies, 34(1), 5-35.
Deng, Z., R. Falvey and A. Blake (2008), “Productivity Spillover of Foreign Direct Investment: A
Computable General Equilibrium Model of China”, Presented at the 11th Annual Conference
on Global Economic Analysis, Helsinki, Finland.
Foreign Policy (2020, April 15), “How the Economy Will Look After the Coronavirus Pandemic?”. Available
from https://foreignpolicy.com/2020/04/15/how-the-economy-will-look-after-the-coronavirus-
pandemic/
Han, Hyoung-Min, Jeong-Gon Kim, Young-Chul Song, Chi-Hyun Yun (2019), The Growth of
Investment into India and the Implications for the New Southern Policy (World Economy Today,
No. 19-03), Sejong, Korea: Korea Institute for International Economic Policy.
Haren, P. and D. Simchi-Levi (2020, February 28), “How Coronavirus Could Impact the Global
Supply Chain by Mid-March”, Harvard Business Review. Available from
https://hbr.org/2020/02/how-coronavirus-could-impact-the-global-supply-chain-by-mid-march
(accessed August 5, 2020)
Hertel, T. and M. Tsigas (1997), “Structure of the GTAP Model”. In T. Hertel (Ed.), Global Trade
Analysis, New York, NY: Cambridge University Press.
37
Hummels, D., J. Ishii and K. M. Yi (2001), “The Nature and Growth of Vertical Specialization in
World Trade”, Journal of international Economics, 54(1), 75-96.
ILO (2020), COVID-19 and the world of work: impact and policy responses, Geneva, Switzerland:
International Labour Organization.
IMF (2020), The Great Lockdown: Worst Economic Downturn since the Great Depression,
Washington, DC: Author.
Inoue, H. and Y. Todo (2017), Propagation of Negative Shocks through Firm Networks: Evidence
from Simulation on Comprehensive Supply Chain Data (RIETI Discussion Paper, No. 17-E-
044), Tokyo: Research Institute of Economy, Trade and Industry.
Javorcik, B. S. (2004), “Does Foreign Direct Investment Increase the Productivity of Domestic Firms?
In Search of Spillovers through Backward Linkages”, American Economic Review, 94(3), 605-
627.
Jeong, Jae-Won (2020), Analysis of the New Northern Policy: Korea-Russia/EAEU FTA (KERI
Insight, No. 20-02), Seoul: Korea Economic Research Institute.
Johnson, R. C. and G. Noguera (2012), “Accounting for Intermediates: Production Sharing and Trade
in Value Added”, Journal of International Economics, 86(2), 224-236.
KIEP (2018), Korea’s ODA Strategy to Strengthen Global Value Chains in New Southern Regions
(Long-term Trade Strategies Study Series, No.18-02), Sejong, Korea: Author.
KIEP (2019), Global Value Chain Analysis in the New Southern Region, Korea’s Trade Expansion
and Upgrading Strategy (Policy Analyses, No. 19-14), Sejong, Korea: Author.
Kim, Jong-Bub (2019), Evaluation of the New Southern Policy and the New Northern Policy and
Prospects (KAPS Winter Issue, No.59), Seoul: The Korean Association for Policy Studies.
Kim, Jeong-Han and Byeong-Ho Seo (2018), The New Southern Policy and Korean banks’ business
in the region (KIF VIP Report, No.18-05), Seoul: Korea Institute of Finance.
Koopman, R., Z. Wang and S. J. Wei (2014), “Tracing value-added and Double Counting in Gross
Exports”, American Economic Review, 104(2), 459-494.
KOTRA (2019), Utilization Strategy of Value Chain in Major New Southern Policy Countries (Global
Market Report, No. 19-100), Seoul: Author.
Kwak, Sung-Il (2020), Planning the Future of Korea’s New Southern Policy, Washington, DC: Asia
Pacific Bulletin, 516.
Lee, Jae-Ho (2019), The Trend of Korea’s Investment into ASEAN since the New Southern Policy
(World Economy Focus, No. 19(05)), Sejong, Korea: Korea Institute for International Economic
Policy.
Lee, Keun, M. Szapiro and Z. Mao (2018), “From Global Value Chains (GVC) to Innovation Systems
for Local Value Chains and Knowledge Creation”, The European Journal of Development
Research, 30(3), 424-441.
Lejour, A., H. Rojas-Romagosa and P. Veenendaal (2014), Identifying Hubs and Spokes in Global
Supply Chains using Redirected Trade in Value Added (ECB Working Paper, No. 1670),
Frankfurt, Germany: European Central Bank.
Liu, X., P. Siler, C. Wang and Y. Wei (2000), “Productivity Spillovers from Foreign Direct Investment:
Evidence from UK Industry Level Panel Data”, Journal of International Business
Studies, 31(3), 407-425.
Malcolm, G. (1998), Modeling Country Risk and Capital Flows in GTAP (GTAP Technical Paper,
No. 13), West Lafayette, IN: Purdue University.
McDonald, S. and K. Thierfelder (2004), Deriving a Global Social Accounting Matrix from GTAP
Versions 5 and 6 Data (GTAP Technical Paper, No. 22), West Lafayette, IN: Purdue University.
McKinsey (2020), COVID-19: Briefing Materials. Available from
https://www.mckinsey.com/~/media/McKinsey/Business%20Functions/Risk/Our%20IInsight/
COVID%2019%20Implications%20for%20business/COVID%2019%20May%2202/COVID-
19-Facts-and-Insights-June-1-vF.pdf (accessed August 1, 2020)
38
Meyer, K. E. and E. Sinani (2009), “When and Where does Foreign Direct Investment Generate
Positive Spillovers? A Meta-analysis”, Journal of International Business Studies, 40(7), 1075-
1094.
Na, Seung-Kwon, Sung-Hee Lee and Eun-Mee Kim (2018), Soft Power Policy of Japan and China
and Policy Implications for the New Southern Policy (World Economy Today, No. 18(30)),
Sejong, Korea: Korea Institute for International Economic Policy.
OECD (2013), Interconnected Economies: Benefiting from Global Value Chains, Paris: Author.
OECD (2018), Multinational Enterprises in the Global Economy: Heavily Debated but Hardly
Measured, Paris: Author.
OECD (2020a), Trade Policy Implications of Global Value Chains (OECD Trade Policy Papers, No.
161), Paris: OECD.
OECD (2020b), COVID-19 and Global Value Chains: Policy Options to Build More Resilient Production
Networks, Paris: Author.
Peter, G. P., R. Andrew and J. Lennox (2011), “Constructing an Environmentally-Extended Multi-
Regional Input–Output Table Using the GTAP Database”, Economic Systems Research, 23(2),
131-152.
PwC (2020), The Possible Economic Consequences of a Novel Coronarivurs (COVID-19) Pandemic:
Australia Matters. Available from https://www.pwc.com.au/publications/australia-matters/
economic-consequences-coronavirus-COVID-19-pandemic.pdf (accessed August 5, 2020)
Rutherford, T. F. (2005), GTAP6inGAMS: The Dataset and Static Model. Available from
http://www.mpsge. org/gtap6/gtap6gams. pdf (accessed August 19, 2019)
Shim, Jae-Hee (2019), “A Study on the Trade Complementarity Between Korea and New Southern
Countries”, The Journal of Humanities and Social science, 10(6), 943-956.
UNCTAD (2020), Global Trade Impact of the Coronavirus (COVID-19) Epidemic, Geneve,
Switzerland: United Nations Conference on Trade and Development.
Urata, S. (2020), “Reimagining Global Value Chains after COVID-19”, ERIA. Available from
https://www.eria.org/news-and-views/reimagining-global-value-chains-after-covid-19/,
Accessed August 12, 2020)
Walmsley, T. L., T. Hertel and D. Hummels (2014), “Developing a GTAP-based Multi-region, Input–
Output Framework for Supply Chain Analysis”. In B. Ferrarini and D. Hummels (Eds.), Asia
and Global Production Networks, London: Edward Elgar Publishing.
Wang, Z., S. J. Wei and K. Zhu (2013), Quantifying International Production Sharing at the Bilateral
and Sector Levels (NBER Working Paper, No. w19677), Cambridge, MA: National Bureau of
Economic Research.
World Bank (2019), World Development Report 2020: Trading for Development in the Age of Global
Value Chains, Washington, DC: Author.
World Bank (2020), East Asia and Pacific: Countries Must Act Now to Mitigate Economic Shock of
COVID-19, Washington, DC: Author.
WTO (2017), Measuring and Analysing the Impact of GVCs on Economic Development, Geneva,
Switzerland: Author.
WTO (2018), World Trade Statistical Review 2018, Geneva, Switzerland: Author.
WTO (2020a), Trade Costs in the Time of Global Pandemic, Geneva, Switzerland: Author. Available
from https://www.wto.org/english/tratop_e/covid19_e/trade_costs_report_e.pdf
WTO (2020b, April 8), “Trade Set to Plunge as COVID-19 Pandemic Upends Global Economy”.
Available from https://www.wto.org/english/news_e/pres20_e/pr855_e.htm
Yoo, Jeong-Ho and Jun-Yeop Lee (2019), “New Southern Policy from the Value-Added Trade
Perspective of Korea and China”, Korean-Chinese Social Science Studies, 52(0), 89-110.
Zhu, L., K. Ito and E. Tomiura (2016), Global Sourcing in the Wake of Disaster: Evidence from the
Great East Japan Earthquake (RIETI Discussion Paper, No. 16-E-089), Tokyo: Research
Institute of Economy, Trade and Industry.
https://doi.org/10.35611/jkt.2020.24.8.39 39
Safeguarding Korean Export Trade through

Social Media-Driven Risk Identification and
Characterization
JKT 24(8)
Juthamon Sithipolvanichgul
Department of Accounting, Thammasat Business School, Thammasat University, Bangkok, Thailand
Alan S. Abrahams
Department of Business Information Technology, Pamplin College of Business, Virginia Tech, Blacksburg, VA, USA
David M. Goldberg
Department of Management Information Systems, Fowler College of Business, San Diego State University,
San Diego, CA, USA
Nohel Zaman
Department of Information Systems and Business Analytics, College of Business Administration,
Loyola Marymount University, Los Angeles, CA, USA
Milad Baghersad
Department of Information Technology and Operation Management, College of Business, Florida Atlantic University,
Boca Raton, FL, USA
Leila Nasri
Department of Business Information Technology, Pamplin College of Business, Virginia Tech, Blacksburg, VA, USA
†
Peter Ractham
Centre of Excellence in Operations and Information Management, Thammasat Business School,
Thammasat University Bangkok, Thailand
Abstract
Purpose – Korean exports account for a vast proportion of Korean GDP, and large volumes of Korean
products are sold in the United States. Identifying and characterizing actual and potential product
hazards related to Korean products is critical to safeguard Korean export trade, as severe quality issues
can impair Korea’s reputation and reduce global consumer confidence in Korean products. In this
study, we develop country-of-origin-based product risk analysis methods for social media with a
specific focus on Korean-labeled products, for the purpose of safeguarding Korean export trade.
Design/methodology – We employed two social media datasets containing consumer-generated
product reviews. Sentiment analysis is a popular text mining technique used to quantify the type and
amount of emotion that is expressed in the text. It is a useful tool for gathering customer opinions
regarding products.
Findings – We document and discuss the specific potential risks found in Korean-labeled products
and explain their implications for safeguarding Korean export trade. Finally, we analyze the false
positive matches that arise from the established dictionaries that were used for risk discovery and
utilize these classification errors to suggest opportunities for the future refinement of the associated
automated text analytic methods.
Originality/value – Various studies have used online feedback from social media to analyze product
defects. However, none of them links their findings to trade promotion and the protection of a specific
country’s exports. Therefore, it is important to fill this research gap, which could help to safeguard
export trade in Korea.
Keywords: Electronic Commerce, Electronic Word of Mouth, Risk Management, Social Media,
Sentiment Analysis
JEL Classifications: D12, F14, O53
†
Corresponding author: Peter@tbs.tu.ac.th
40
1. Background: Korean Exports and Online Reviews
South Korea was the United States’ sixth largest source of imports in 2018. This totaled
$74.3 billion, including billions of dollars of electronics and food products (OUSTR, 2020). A
significant number of these Korean products are sold online or reviewed online by American
consumers. A reputation for producing high quality products facilitates premium product
pricing (Shapiro, 1983), and a reputation for providing poor quality products, relative to other
countries, can result in price discounts (Agrawal and Kamakura, 1999), and thus, in the long
run, may negatively impact Korean export trade.
Often, international buyers must rely on online reviews as credible sources to make
decisions on product quality, brand recognition, and aftersales services (Giri and Thapa,
2016; Dwivedi et al., 2009). These sources can range from user reviews on e-commerce
websites (e.g., Amazon.com) to product-related online videos and video comments (e.g., on
YouTube.com), to product reviews on social media sites (such as Facebook.com), and to
product reviews posted on blogs and online forums. These reviews play important roles in
creating an electronic form of word-of-mouth (eWom), where users who have used products
or services share their thoughts and feedback through electronic channels. eWom can help
companies to increase both their sales and market recognition. In the past decade, companies
have relied on eWom as a tool to build trust with customers who increasingly rely on the past
experiences reported by other customers who have used the products or services, instead of
traditional marketing platforms, like print media or TV ads. For example, online reviews from
Amazon, eBay and YouTube have become very popular with online consumers (Von
Helversen et al., 2018).
The effects of eWom on companies and their products can be multifold, and the feedback
can be positive or negative. In short, if products garner positive reviews, then the associated
companies can benefit from the eWom. However, companies can also receive negative
responses from their customers if their products or services have unfavorable reviews (Pee,
2016). Thus, in addition to interacting with their customers through social media, companies
can analyze users’ feedback from social media too (Boon-Long and Wongsurawat, 2015).
Conventionally, companies have received customer feedback from various channels. However,
this feedback often relies on official channels, such as business partners, distributors, or other
storefronts – both offline and online – to report the issues to the parent companies. However,
the turnaround time for information about defective products may be weeks or months, and
it can take additional time for companies to collect and analyze as there are significant time
lags between manufacturing, distribution, purchase, and consumption. Furthermore,
product defect reports are often not directly received by the manufacturer, since many
customers do not report defective products through the official channels. To obtain data on
product feedback or defects, companies can analyze customers’ feedback through the online
reviews that are readily available through various social media channels. By identifying and
characterizing product issues through online reviews, companies can get current, up-to-date
information on product defects. Companies can use this information to improve their
product development for future products. Using data extracted from online feedback can
greatly help to reduce the various risks that come with manufacturing and distributing
defective products.
Regarding product compliance risks, product recalls are a major concern for Korean
companies (Huh and Choi, 2016). The Korean government enforces hundreds of product
recalls annually, with approximately two-thirds of recalled products originating from China,
and the remaining one-third originating primarily from Korea (Nottage, 2020). Korean
products have often suffered calamitous recalls in international markets – such as the 1.9-
Safeguarding Korean Export Trade through Social Media-Driven Risk Identification and Characterization
41
million-unit Samsung Galaxy Note 7 recall (CPSC, 2016a/2017), and the 2.4-million-unit
Samsung top-loader recall (CPSC, 2016b). The recent major Korean enoki mushroom recall
involved three companies, and was associated with 31 hospitalizations and 4 deaths (CDC,
2020). Product recalls may have substantial negative effects on the wealth of sellers (Jarell and
Peltzman, 1985), severely damaging their corporate image and investor confidence (Zhao et
al., 2014). They can also have negative trade policy implications, such as potential trade
restrictions and severely negative trade consequences. For instance, Liu, Kerr and Hobbs
(2009) document that there has been substantial negative media attention, strong pressure to
tighten regulatory barriers, outright bans on Chinese imports, serious declines in consumer
confidence in Chinese goods, and consequent declines in Chinese exports following signifi-
cant product safety incidents with Chinese exports. These declines were so severe that Liu,
Kerr and Hobbs (2009) contend there was even substantial collateral damage, with reductions
in export sales even for Chinese firms with no safety problems. Safeguarding Korean exports
by avoiding damaging highly public quality incidents is thus critical.
Globally, product risk characterization has been an import concern of both regulators and
corporations. Formal product risk assessment methodologies have been developed and
documented by the United States Consumer Product Safety Commission (US CPSC), Health
Canada (2016), and the European Commission (2015). All of these incorporate an evaluation
of both the severity of the potential injury to consumers and the likelihood of injuries
occurring. Risk assessment is informed by heterogeneous data from a variety of sources,
including but not limited to physical product tests and inspections, hospitalization reports
and death certificates, and unsolicited consumer complaints via multiple modalities, such as
telephones, mail, and the Internet.
Various studies have used sentiment analysis to analyze online feedback from social media
to analyze product defects, such as Abrahams et al. (2012) and Bergstrand and Finlaw (2011).
Prior research has, for instance, used online reviews to analyze defects in home appliances
(Goldberg and Abrahams, 2018; Law et al., 2017), children’s toys and baby products
(Mummalaneni et al., 2018; Winkler et al., 2016), electronics (Abrahams et al., 2015) and
automobiles (Abrahams et al., 2013). However, none of these studies link their findings to the
conceptual frameworks of risk management in the context of a specific country that could
potentially lead to the improvement of production within that specific country. Therefore, it
is important to fill this research gap, which could lead to improvements in the manufacturing
of different products being sold online in the source country.
2. Related Work
In this section, we describe related work in the areas of eWom, brand crisis management,
risk management, and product development.
2.1. eWom and Brand Crisis Management

eWom has garnered much attention from companies who sell products through e-
commerce channels. In the past, companies usually relied on traditional marketing channels,
such as TV commercials or printed ads (Geraghty and Conway, 2016). However, with the
proliferation of the internet and digital technologies over the past few decades, companies
have switched their attention to online advertising to increase the reach to their customers
(Nwokah and Ngirika, 2017). The increase in the popularity of e-commerce and m-
commerce transactions has transformed consumer behavior and prompted consumers to be
42
more receptive to information being presented in digital forms (Teeramungcalanon, 2020).
Customer feedback through online product reviews has become an important source that
consumers use to search for opinions and shared experiences before they make decisions
about buying products (Liu and Zhang, 2010). In addition, if products have negative reviews,
such as product defects or hazardous parts, it could indicate potential future consumer action
or regulatory action, such as possible lawsuits or product recalls. Product recalls are a
“product crisis” for companies, which can lead to them developing negative reputations. This
makes these companies vulnerable to future sales losses, especially where the firm has
diminished control of social media content about its product (Lee et al., 2015). In some
serious cases, lawsuits can result from product defects that result in property damage, or
consumer injury or harm, especially if reported by many consumers. If these negative physical
or emotional impacts are intensified through online channels, the risks can harm companies’
reputations in both the short and long term, which could lead to an international product
being recalled and hinder trade activities in the long run. Hence, there has been a need for
brand crisis management (Souiden and Pons, 2009). This may involve the use of information
obtained from eWom to prevent, identify, and manage the negative effects on a company’s
reputation, as well as other negative impacts on companies who rely on eWom from online
product reviews.
2.2. Risk Management and Product Development Life Cycle

One strategy for managing brand crises is to rely on gaining valuable customer insights
from product reviews through social media channels. Companies can benefit from analyzing
the data to gain an understanding of what customers want (Anshari et al., 2019;
Sithipolvanichgul, 2018). By leveraging the available data from social media channels,
companies can also use the findings to improve future versions of the products. Ultimately,
the Product Development Life Cycle (PDLC) can be improved by having the available data to
help improve the understanding of customer requirements, prevent future design and
manufacturing errors, and shorten the development time between the releases of new product
versions (Hines et al., 2006). The PDLC stages are Plan, Develop, Evaluate, Launch, Assess,
Iterate or Kill. Planning usually involves the pre-production processes where analyses of
customers and competitors are the main concerns for companies. By identifying consumer
needs through data retrieved from social media channels, companies can get insight
information from their customer demographic and from competitors’ products when users
make product comparisons in their reviews. In the Develop and Evaluate stages, companies
can gain valuable information by identifying and classifying products’ defects and errors to
help companies have both the direct interaction and feedback necessary to assess and further
improve their products. However, if the products’ defects cause damages, either in terms of
the company’s reputation or financially through lawsuits, then companies can decide to
conduct product recalls or even terminate the product development completely.
Hence, online reviews through social media can play a pivotal role to help inform the
product development process. The information that can be shared through social exchanges
between consumers and the manufacturer is important for product improvement (Piller et
al., 2012). Companies who collect information efficiently from customers can use that
information for their benefit to reduce the risk of them investing their resources in developing
problematic or undesirable products in future versions (Ogawa and Piller, 2006).
Prior research has leveraged social media to both find and assess hazards across a broad
array of consumer products (Lockett et al, 2006). Social media data sources include product
reviews from online retailers (Goldberg and Abrahams, 2018; Mummalaneni et al., 2018;
43
Winkler et al., 2016; Zaman, 2019), discussion boards (Abrahams et al., 2015), and online
videos (Nasri et al., 2018). While it is not an exhaustive source of risk reports, social media
data presents a convenient public repository of an extensive assortment of consumer-
generated product hazard reports.
Past studies of product safety concerns mentioned in social media have been agnostic to
country-of-origin and so have not provided the methods needed to direct attention to
products sourced from a specific country. In this study, we develop country-of-origin-based
product risk sensing methods for social media, with a specific focus on products manu-
factured in the Republic of Korea. Such a focused analysis is helpful for Korean manufac-
turers, trade representatives, and trade advisors, as it facilitates both product quality man-
agement and reputation management for Korean brands singularly and Korea collectively.
3. Methodology
We employed two social media datasets containing consumer-generated product reviews:
(1) textual product reviews from Amazon, the world’s largest e-commerce retailer, and (2)
videos from YouTube, the world’s largest online video sharing platform.
Sentiment analysis is a popular text mining technique that is used to quantify the type and
amount of emotion expressed in the text. It is a useful tool for gathering customer opinions
regarding products. Sentiment analysis uses each individual word context and a dictionary of
positive and negative words to determine context-aware sentiment. There has been much
previous research using sentiment analysis used to study online reviews, e.g., Cheng and Jin
(2019); Fang and Zhan (2015); Fan et al. (2017); Jo and Oh (2011). We employed smoke-term
analysis – a specific adaptation of sentiment analysis, shown to be more effective for product
defect discovery (Abrahams et al., 2015) – for our analysis.
3.1. Amazon Product Reviews

We obtained a dataset of approximately 233 million product reviews from Amazon.com
from Ni et al. (2019), which covered the period from May 1996 through to Oct 2018. We
filtered the dataset to recognize (without case-sensitivity) the word “Korea” in both the
product title and description. We removed Books, Movies and TV shows, Software, and
Video Games from the dataset, as these items are unlikely to cause hazards and could lead to
an excessive number of false positive matches for terms like “hospital”, “dangerous”, “fire”,
and so forth, which may indicate potential product hazards in other categories. We arrived at
a subset of 99,807 reviews that mentioned “Korea”, as shown in Table 1, which indicates the
number of reviews per product category. As Table 1 indicates, four categories (Cell Phones &
Accessories [27% of reviews], Grocery [18%], Amazon Home [14%], and Tools & Home
Improvement [12%]), accounted for most of the reviews [total 70% of all reviews] for Korean
products on Amazon during the 22-year period under review.
Table 1. Distribution of Korean Product Reviews by Product Category

Product category Number of reviews
All Beauty 1,560
All Electronics 1,647
Amazon Home 13,589
Arts, Crafts & Sewing 1,673
Automotive 5,120
44
Product category Number of reviews
Baby 100
Camera & Photo 5,240
Car Electronics 452
Cell Phones & Accessories 26,726
Collectibles & Fine Art 1
Computers 2,732
GPS & Navigation 21
Grocery 17,766
Health & Personal Care 514
Home Audio & Theatre 1,640
Industrial & Scientific 240
Musical Instruments 662
Office Products 2,184
Pet Supplies 383
Sports & Outdoors 2,720
Tools & Home Improvement 12,063
Toys & Games 2,774
We employed established “smoke-term” mechanisms to identify the reviews that mentioned

potential risks. A “smoke term” is a word or phrase determined by statistical and information
retrieval methodologies that are more prevalent in reviews that mention a safety concern than
they are in reviews that do not. To filter the Amazon product reviews, we applied two distinct
smoke-term dictionaries, sourced from Goldberg and Abrahams (2018) – see Table 2, Panel
(a) – and Zaman (2019) – see Table 2, Panel (b) and Appendix A. For the former dictionary,
all terms were equally weighted, meaning that the total smoke score for the Amazon review
was incremented uniformly by 1 point each time a matching term was found. For the Zaman
dictionary, each term had its own weight, meaning a match for a high-weight term (that is, a
term highly prevalent in Zaman’s dataset of historic safety concern reports) would increment
the total smoke score for the review by a higher amount than for a low-weight term (a term
less prevalent in the historic safety concern reports). We manually curated the full-term list
from Zaman (2019) to remove terms deemed unlikely to relate to hazard reports, and arrived
at an abbreviated smoke term list from Zaman (2019), which we report here. For brevity, Table
2, Panel (b) shows only the top 95 single words from the Zaman smoke-term dictionary, and
the remaining terms (94 two-word and 100 three-word phrases) are included in Appendix A.
Term weights have been omitted for brevity, but terms have been listed in order from highest-
to lowest-weight single-words, highest- to lowest-weight 2-word phrases, and highest- to
lowest-weight 3-word phrases, so the more important terms appear first in each list.
Table 2. Smoke Terms Used to Filter Amazon Product Reviews Containing “Korea”
(a)
(b)
Goldberg and
From Zaman (2019); continued in Appendix A
Abrahams (2018)
burning choking burned slips flames shaky
caught fire irritation face bruises violently
dangerous fell blood toxicity smack harm
fire head swallowed lead painful tipping
hazard cut unstable intestine slip suffer
injuries sharp hazards gash bruise injury
45
(a)
(b)
Goldberg and
From Zaman (2019); continued in Appendix A
Abrahams (2018)
safety fall mouth sparks horrid emergency
unsafe chemicals hot respiratory vomiting pneumonia
caught fire toxic asleep gas dermatitis mold
fire hazard hazard secure accidentally arsenic poisoned
not recommend smoke stitches bleeding injuring warns
on fire hit cutting burned asthma carcinogenicity
a fire hazard burning fumes slippage burn pulmonary
caught on fire choke chin hospital cutting suffocated
neck magnets terrified rash suffocating
rash throat cuts dangerous esophagus
fallen hurt exposed died sores
wobbles wobbly cancer punched crash
slippery allergic sick bromine agony
3.2. YouTube Product Review Videos

We obtained a dataset of 16,402 consumer product review videos from YouTube, from
Nasri et al. (2018), including the manufacturer’s name, product type, video URL, and video
title, description, and publication date. This dataset spanned the period from March 2006
through to January 2017. Nasri et al.’s video sample was scored and filtered using Nasri et al.’s
list of 200 hazard-indicative words to boost the proportion of hazard mentions and was
manually coded to annotate the existence or absence of a potential safety concern in each
video. The dataset covered 70 distinct international manufacturers. For each manufacturer,
we manually ascertained the headquartering country of the corporation to determine whether
it was based in the Republic of Korea. We extended the procedure from Nasri et al. (2018) by
incorporating a modified snowball (chain referral) sample (Goodman, 1961; Biernacki and
Waldorf, 1981), using both YouTube’s “Up Next” recommendations, as well as a manual
search on the product’s name or product type and incident-type keywords (e.g. “top loader
explode”) to roll related observations (videos) into the dataset. For all the items, we have
reported the consumer’s verbatim content, and made no claims as to the veracity of the
consumer’s claims or the legitimacy of safety concerns – these require further investigation
by competent authorities.
4. Results
We report here the results for products associated with the word “Korea”, in both the
Amazon product reviews dataset and the YouTube product-review video dataset.
4.1. Amazon Product Reviews

For the Goldberg and Abrahams dictionary, 73 of the Amazon reviews for products labeled
“Korea” in our dataset matched two or more smoke terms, and a further 454 reviews matched
one smoke term from the Goldberg dictionary.
46
For the Zaman dictionary, 9,737 Amazon reviews for products labeled “Korea” in our
dataset matched one or more smoke terms. Fig. 1 shows the distribution of total smoke scores
for Amazon reviews for products for “Korea”, with one or more matches when scored with
Zaman’s weighted smoke-term dictionary.
Fig. 1. Overall smoke score for Amazon Product Reviews containing “Korea”, scored with
abridged Zaman (2019) dictionary
Table 3 lists an illustrative sample of Amazon reviews for products that included “Korea”
in their product title or description, and contained terms in our smoke-term dictionaries.
Table 3 includes the manufacturer, Amazon’s unique product identifier (ASIN) with the date
when the review was posted, and the consumer’s comment that contained the matching term.
Term matches are highlighted in bold for any terms matching the Goldberg and Abrahams
(2018) or the abridged Zaman (2019) smoke-term dictionaries.
Table 3. Illustrative Amazon Product Reviews for Product Descriptions Containing “Korea”
and Matching Smoke-Term Dictionaries
Manufacturer ASIN, Date Product Type Comment with Match
Gens ace B017B4NPOK Battery pack “DANGEROUS... Second time I charged it, the
7/10/2017 battery caught on fire and exploded. It lit my
workbench on fire and almost burned down my
garage.”
Dongsuh B00FRECXFO Tea “After drinking this tea, I was hospitalized for severe
5/4/2018 nausea… I was totally amazed when the hospital
blood tests revealed Marijuana!! The hospital and I
believe the tea was spiked with Marijuana.”
Ceptics B007SBGBDO Travel adapter “Caught on fire the 1st time I used it!”
11/18/2014 plug
Ceptics B00BLTCHN2 Travel adapter “Made loud popping noise, then flash of light, then
3/11/2015 plug really bad burning smell.”
47
Aketek B0053DDNW6 Dashcam “I use it for one day, the unit started to smoke... UNIT
2/20/2013 IS DANGEROUS!!!”
8/13/2014 “battery started smoking as soon as I put it in the
camera and it melted the plastic.”
11/2/2015 “The second I plugged in the power cord it started
smoking. Thank god I unplugged it before it caught
fire.”
1/3/2018 “When I plugged it in, the screen flashed once and
then I saw SMOKE... yes, SMOKE coming from the
interior of the camera; so I quickly unplugged the
camera. After unplugging the camera, the smoke
lasted from approximately 2 minutes; and for the
entire time I clearly smell a burnt substance... this
product could be categorized as being unsafe; and
could potential cause fire.”
2/26/2013 “When I installed it in the car and connected the
power to it, the thing began to sizzle and spark and
smoke almost instantly.”
4/18/2013 “bought 4 in a row... 3rd one it has smoke come out
when plugged in.”
Cheengoo B00N40P4P4 Baby rattle “the tags start to come loose after a couple washes.
They are easy to remove but definitely a choking
2/18/2018
hazard for baby.”
Goldengulf B00CC8OZ8Q Tablet “this is a dangerous product!... attempting to charge
10/2/2013 computer it, it became EXTREMELY hot and then was dead... it
got so hot when I attempted to see how it was
charging that it actually stuck to the leather topped
table it was sitting on!!!”
Buddha Teas B00NC5P9S8 Chaga Tea “My husband was highly allergic to whatever is in this
product... He broke out in welts and throat and lips
swelled up. Had to take Benadryl. Very scary
situation.”
Gwangcheon B01DI1R3UC Korean “I had an allergic reaction to these. I‘m not at all
Joyang Food 1/30/2018 Seaweed allergic to seaweed (I eat it all the time), so it leads me
Co Ltd to believe that something that they used when
processing this was the cause.”
Annie Chun’s B000E148MG Packaged food “Corn starch! ... We are allergic to corn starch, and...
10/14/2009 this DOES have corn starch. If you are allergic, or
sensitive to corn, do not order this product. The corn
is not listed, but believe me, it is there – and once you
get your package you will see it on the label.”
B000E123IC “It contains soy sauce despite ingredient label. The

6/7/2016 ingredients in the image show this product does not
contain soy sauce. The product I received does
contain soy sauce. I am allergic to soy.”
48
B01EWVORPG “Contains Coconut Oil!... But I always felt unwell after
8/14/2018 eating them. I’m allergic to coconut... spoiler alert,
apparently these have coconut oil in them! So what,
you say? Well, I’d carefully read the ingredients list on
Amazon, and there’s no coconut oil listed... Long
story short, five stars for taste, but they seriously need
to address the ingredients list on here.”
B000E123IC “... while they seem to have changed the recipe, they
7/19/2015 haven‘t changed their information on the packaging.
It‘s listed as having a mild rating for spiciness, doesn‘t
mention any peppers in the ingredients... The 2 I‘ve
used, both were extremely hot, both had large
quantities of small dried red peppers. This incorrect
labeling could result in serious health issues for
anyone with allergies to peppers.”
B000E148MG “an order arrived with two of the noodle packets

6/3/2011 punctured with the noodles growing green, fuzzy
mold all over. Disgusting… Then the most recent
order, June 2012, arrived with rancid sauce and rancid
peanuts unfit for consumption.”
A review of the manufacturers listed in Table 3 revealed that they may:

a) be based in Korea (the Dongsuh website shows a Seoul business address) and/or
manufactured in Korea (Gwangcheon Joang Food Co specifies its seaweed as
“Product of Korea”).
b) be a subsidiary or affiliate of a Korean corporation (e.g. Annie Chun’s is a brand of
CJ CheilJedang, a Korean public company).
c) source raw materials from Korea (Gens ace specified “Superior Japan and Korea
Lithium Polymer” materials).
d) use an ingredient with traditional Korean roots (“Chaga” is a Korean medicinal herb
– Shirakabatake – but the distributor, Buddha Teas, sources their product from
Canada).
e) distribute to Korea or have products intended for Korean use (e.g. Aketek dashcam
provides Korean as one of the 10 available languages; one customer mentioned that
the Aketek product was made in China; Ceptics travel adapter plugs lists Korea as a
compatible zone, though the Amazon listing specifies that manufacturer is a United
States limited liability corporation; the Goldengulf tablet computer lists Korean as a
supported language).
f) have a Korean name, Korean influence or inspiration (e.g. “Cheengo” means
“friend” in Korean, though the product is designed in America and was made in
China).
Items under a) through to d) (above) may present reputational risks for Korean companies.
Items under e) (above) may present risks to Korean consumers, as well as Korean companies
that distribute these products to Korean consumers. Items under f) may be false positives as
they may be neither manufactured in Korea nor distributed there.
Table 3 indicates that even if specific consumer comments are not specific risks, online
49
reviews may be helpful in identifying systemic failures or systemic risks. For example, the
United States Food Allergen Labeling and Consumer Protection Act of 2004 (FALCPA)
requires milk, eggs, fish, shellfish, tree nuts, peanuts, wheat, and soy to be explicitly declared
on food box labels. In the illustrative Annie Chun’s reviews in Table 3, the customers mention
that the online product listing omits the allergen, but the physical box declares the allergen:
there is no clear regulatory violation here as the law requires an allergen declaration only on
the box label, rather than the online product listing. Further, while soy and coconut (classified
as a tree nut) fall into the FALCPA “major food allergen” categories, one reviewer mentions
a corn allergy, which is not classed as a “major food allergen”. Nevertheless, the allergen-
related comments directed at Annie Chun’s products in Table 3 indicate systemic risks
(possible product recall due to labeling errors, and consumer dissatisfaction due to ingredient
listing errors) for the corporation. In Feb 2020, Annie Chun’s issued a voluntary recall for its
Japanese-Style Teriyaki Noodle bowls due to undeclared peanuts (FDA, 2020), indicating that
the systemic risk of recall due to labeling errors is indeed substantial.
4.2. YouTube Product Review Videos

Samsung was the only Korean-headquartered corporation in the dataset, and 130 videos
with unique titles matched the manufacturer’s name Samsung, as well as one of the 200 smoke
words in Nasri et al.’s smoke word list. Of the 200 smoke words employed by Nasri et al.
(2018), only 26 unique smoke words matched a YouTube product review for a Korean
product in our YouTube video dataset. These words are listed in Table 4.
Table 4. Distinct Smoke Words from Nasri et al. (2018) Matching Korean Products in our
YouTube Video Dataset
explode horrible smell
malfunction defects causing
beware damage careful
smoking leakage burn
problem abnormal injure
fault crap incident
risk shit secure
fire safety injured
cause concern
An analysis of the matching videos indicated that they were constituted of videos:
(a) related to prior product recalls;
(b) that claimed potential safety concerns which had not yet been addressed in prior
product recalls; or
(c) were clear false positives, unrelated to safety concerns.
We address (a) and (b) here and return to (c) in the Limitations and Future Work section.
(a) Videos related to prior product recalls

One hundred and one (101) matching videos were related to product recalls. These
constituted:
50
 73 videos relating to CPSC recall number 16-266 (“Samsung Recalls Galaxy Note7
Smartphones Due to Serious Fire and Burn Hazards”) and the follow-on expanded
recall 17-011; and
 28 videos relating to CPSC recall number 17-028 (“Samsung Recalls Top-Load
Washing Machines Due to Risk of Impact Injuries”).
From a risk-management perspective, tracking the proliferation of these videos clarifies the
genesis and severity of the negative publicity related to the recall. However, videos relating to
prior product recalls are unhelpful in detecting emergent product risks, as the emergent risk
would have already been investigated by the corporation and regulators prior to the recall
announcement.
(b) Videos claiming potential safety concerns, unaddressed by recalls

Table 5 lists an illustrative sample of consumer safety claims for Korean products that were
found in YouTube videos, either prior to the announcement of a product recall, or not known
to be associated with a product recall. The superscript O in Table 5 indicates the observations
that were obtained through snowball (chain referral) sampling using YouTube’s “Up Next”
feature. Consumer claims are not necessarily indicative of bona fide product safety issues, and
all claims require further investigation to verify their legitimacy. Some videos involved
consumer misuse (e.g. the use of an unauthorized or potentially counterfeit product or a
third-party replacement part, such as 3rd party battery).
However, in at least five cases in Table 5 (videos where consumers claim their top load
washing machine “exploded”), it appears that the consumer’s claim was verified by a
subsequent product recall, some 1 – 3 years later. This indicates that, though hazard reports
are rare, online videos may indeed be an early warning signal of an emergent product risk.
Under the presumption that the video incident report is for the same issue that prompted the
eventual top loader recall, it is possible that, had Samsung responded to the reported top
loader issue at the time of the earliest online video report (in 2013), rather than 3 years later,
it could have substantially reduced the number of defective units in circulation (2.8 million
units were recalled), product failure incidents1, associated unit recall cost, and its reputational
damage of these incidents.
An analysis of the matching videos indicated that they were constituted of videos:
(a) related to prior product recalls;
(b) that claimed potential safety concerns which had not yet been addressed in prior
product recalls; or
(c) were clear false positives, unrelated to safety concerns.
We address (a) and (b) here and return to (c) in the Limitations and Future Work section.
(a) Videos related to prior product recalls

One hundred and one (101) matching videos were related to product recalls. These
constituted:
1
CPSC recall 17-028 narrative states that: “Samsung has received 733 reports of washing machines
experiencing excessive vibration or the top detaching from the washing machine chassis. There are
nine related reports of injuries, including a broken jaw, injured shoulder, and other impact or fall-
related injuries.”
51
 73 videos relating to CPSC recall number 16-266 (“Samsung Recalls Galaxy Note7
Smartphones Due to Serious Fire and Burn Hazards”) and the follow-on expanded
recall 17-011; and
 28 videos relating to CPSC recall number 17-028 (“Samsung Recalls Top-Load
Washing Machines Due to Risk of Impact Injuries”).
From a risk-management perspective, tracking the proliferation of these videos clarifies the
genesis and severity of the negative publicity related to the recall. However, videos relating to
prior product recalls are unhelpful in detecting emergent product risks, as the emergent risk
would have already been investigated by the corporation and regulators prior to the recall
announcement.
(b) Videos claiming potential safety concerns, unaddressed by recalls

Table 5 lists an illustrative sample of consumer safety claims for Korean products that were
found in YouTube videos, either prior to the announcement of a product recall, or not known
to be associated with a product recall. The superscript O in Table 5 indicates the observations
that were obtained through snowball (chain referral) sampling using YouTube’s “Up Next”
feature. Consumer claims are not necessarily indicative of bona fide product safety issues, and
all claims require further investigation to verify their legitimacy. Some videos involved
consumer misuse (e.g. the use of an unauthorized or potentially counterfeit product or a
third-party replacement part, such as 3rd party battery).
However, in at least five cases in Table 5 (videos where consumers claim their top load
washing machine “exploded”), it appears that the consumer’s claim was verified by a
subsequent product recall, some 1 – 3 years later. This indicates that, though hazard reports
are rare, online videos may indeed be an early warning signal of an emergent product risk.
Under the presumption that the video incident report is for the same issue that prompted the
eventual top loader recall, it is possible that, had Samsung responded to the reported top
loader issue at the time of the earliest online video report (in 2013), rather than 3 years later,
it could have substantially reduced the number of defective units in circulation (2.8 million
units were recalled), product failure incidents , associated unit recall cost, and its reputational
damage of these incidents.
Table 5. Illustrative Examples of Video-Based Consumer Safety Claims for Korean Products
Eventual Recall Date
Consumer Original
Video ID2 Product Type Product Age (Time Elapsed)
Safety Claims Video Date
[CPSC Recall ID]
XiFELuvnslg 1670 Printer 1 day “Smoking” Jan 25, None.
2013
O SROrqiil1zs M2020 Unknown “Burn Dec 9, None.
Printer smoking”3 2016
jpH-3cxm9O0 Top Loader 1 day “Exploded!” Oct 12, Nov 4, 2016
Washing 2013 (1-3 years later)
Machine [17-028]
2
Source video can be found by appending Video ID to the URL: https://www.youtube.com/watch?v=
3
Video poster later comments product is “normal” (not dangerous). Video comments claim: “Mine is
doing similar (M2835DW) ... something lit up as if something is on fire or glowing with heat”... “Mine
does that too!”... “Mine just did that last week. Now I'm afraid to use it.”
52
Table . (Continued)
Original Eventual Recall Date
Consumer Safety
Video ID Product Type Product Age Video (Time Elapsed)
Claims
Date [CPSC Recall ID]
Nc4ps-dPoI0 “Brand new” “Top Load washer Apr 6,
explodes” 2015
gIn23nqN3ew < 1 year old “Samsung Sep 27,

washing machine 2015
explodes”
O 5 separate “Consumers Nov 3,

i1UFVKuqxGk incidents claim some 2015
Samsung washing
machines
explode”
O +- 1 year old. “Aftermath of Nov 16,

HKaxB8Wb4gE Samsung Washer 2015
Explosion/Malfun
ction”
Nq7Omb_r6eU Front Loader 8 years “Smoking ready Jan 13, None.

Washing to catch fire” 2014
Machine
gzVOtZaeDLU S3 mini Unknown “Swollen battery: June 30, None.

phone potential risk of 2014
fire or explosion”
uGgE48Ll2Zw Galaxy S6 Unknown “Naked Samsung Sep 16, None.

S6 explodes” 2016
DyVXnVEA9Ko ME21H706M About 2 years “Runs by itself Sep 2, None.

QS with the door 2016
microwave closed”
O skzG4DPs6-4 SMH1816S Unknown “Running while May 30, None.

Microwave door open” 2019
Our findings confirm the observations from the prior research (Abrahams et al., 2012) that
sentiment analysis (finding words associated with negative emotion) is insufficient to
discover safety concerns, since these can be expressed in non-emotive factual language. For
example, the phrases of interest for reviews DyVXnVEA9Ko and skzG4DPs6-4 suggest
potential safety issues (a microwave running with its door open, or running by itself without
the user pressing the control panel button) and contain no negative sentiment words.
In the absence of further investigation, which is beyond the scope of this research, it is
difficult to verify the legitimacy of consumer safety claims. For example, the printer smoke
53
reported in videos XiFELuvnslg and SROrqiil1zs may not result in the ignition of a
combustible, and could simply be a water vapor cloud resulting from high in-room humidity.
Nevertheless, consumer statements that are possibly not legitimate hazards can still be
damaging to the brand, and risk management may be necessary – e.g. the manufacturer may
reduce the negative revenue impact by timeously countering the consumer’s claims through
an explanation or response on the third-party video platform.
Though the authenticity and severity of each consumer claim require further investigation,
the data suggests that online video can serve as an early indicator of emergent product risks
and emergent brand risks. As highlighted by Nasri et al. (2018), video-based reviews are an
alternative to text-based product reviews, in the context of safety hazard detection. Videos
can provide more information about the underlying problem. Manufacturers can review the
videos and identify the exact technical problems. Also, it is less likely that online videos are
faked, compared to text-based reviews as reviewers show the problems in their videos. Despite
these advantages, few studies have been conducted on video-based reviews being used to
identify safety hazards.
5. Discussion
It is important to note that the consumer claims reflected here and throughout this research
are not necessarily legitimate safety concerns, as verification would require an investigation
beyond the scope of this work. Threats to safety-hazard legitimacy include, inter alia:
 The review may be fake or could exaggerate the seriousness of the issue: for example,
if it were posted by a disgruntled consumer or a zealous competitor.
 The product acquired may be counterfeit or non-OEM (not produced by the Original
Equipment Manufacturer), in which case the legitimate manufacturer may not be at
fault. For example, exploding batteries may be knock-offs that do not comply with
electronics manufacturing standards or have not undergone rigorous pre-shipment
testing. Counterfeit and non-OEM products (e.g. manufactured in China) may be
cannibalizing Korean exports. Identifying counterfeits or non-OEM products through
hazard reports can allow counterfeits to be removed from online marketplaces and can
restore export sales to Korean OEM manufacturers.
 The safety hazard may arise from consumer misuse rather than a design or
manufacturing flaw. For instance, a fire hazard complaint relating to travel adapters
could result from a consumer mistakenly inserting the adapter into an outlet with the
incorrect voltage.
 The product may have been damaged during storage or shipment, which may have
been caused by improper distributor or retailer handling or product tampering rather
than manufacturer error. For example, “came with cut seals on the side and no seal for
the lotion. Which seems a little offsetting... had to return the item!” or “This thing
came in dirty!” or “Looks used and dirty which is concerning since it‘s a FACE
PRODUCT!!!!!”
Though the claims reported here, if verified, could represent a product safety risk, they may
represent a brand-reputation or country-of-origin reputation risk, even in the absence of
verification, since they may be perceived as being legitimate by potential customers reading
online reviews prior to purchasing. Prior research has demonstrated that online reviews have
a demonstrable effect on sales (Floyd et al., 2014) and extreme ratings, which many safety
hazard reports are, have a disproportionately large effect on consumer perceptions (Filieri et
54
al., 2018). Regardless of their veracity, these reviews could therefore discourage purchases of
that item or that brand, or embolden customers contemplating a product return, thus
prompting additional costly refunds.
By identifying and characterizing the possible product risks from this research, Korean
companies can use the findings as a guideline to improve their product quality throughout
the PDLC processes. Ultimately, by looking thoroughly at customer feedback and eWom
from popular online channels, Korean companies can work to prevent negative actions (e.g.
product returns, additional bad reviews, or lawsuits) from dissatisfied customers, or costly
regulatory enforcement actions from government agencies. Hence, Korean companies can
improve their brand reputations overseas by assuring their international customers of their
product’s safety and quality, as well as their responsiveness to customer needs.
We employed established “smoke-term” mechanisms to identify reviews that mentioned
potential risks. Many English words are polysemous (have multiple meanings) and can thus
be found in different contexts with different meanings. “Smoke terms” are also highly
category-specific, with some terms being significantly associated with safety concern reports
in some product categories, but not others. False positives are thus common and warrant
further investigation to determine how to mitigate false positives (reduce the number of false
positives). Table 4 itemizes a representative list of common false positives, with smoke term
matches highlighted in bold, and suggests strategies for mitigating these false positives in
future research. In all cases, we recommend manual review by human coders as a false-
positive mitigation strategy, since humans are more accurate at distinguishing true from false
positives than current machine intelligence approaches. Various parallel-user collaborative
tagging systems – such as Amazon Mechanical Turk, Amazon Sage Maker Ground Truth,
Figure Eight (formerly CrowdFlower), CloudFactory, LabelBox, Datasaur.ai, Heartex,
DefinedCrowd, and Virginia Tech’s PamTag – allow large teams of humans to be employed
at a relatively low cost, with managed inter-rater reliability, to label data (e.g. mark items as
false positives after a further manual review). With sufficient human-labeled data, deep
learning artificial intelligence techniques can potentially be employed to reduce false positives
further.
Our current approach uses literal tokens, so it matches only exact phrases. Thus, the smoke
phrase “cut my finger” would match “it cut my finger”, but would not match sentences with
similar semantics, such as “it cut her finger”, “it cut his finger”, “it cut my toe”, or “it cut my
lip” (italics indicate semantically similar entities that are not exact literal token matches).
Furthermore, the current approach looks for exact spellings, so it would overlook
misspellings, such as “fnger” or “cutt”. Misspellings are common in online postings. This
limitation leads to false negatives, where actual hazards may not be detected due to strict
literal matching. To reduce false negatives, future work should employ additional strategies,
such as word embeddings and automated spellchecking. Word embeddings, e.g. Glove
(Pennington et al., 2014), would allow us to computationally discover words in the same
category (my/his/her are pronouns referring to a person; finger/toe/lip are human body parts),
so that the semantic entity may be used in matching, rather than just the literal token.
Automated spellcheckers (including open-source options, such as Hunspell, GNU Aspell,
and JOrtho) would allow us to determine if a particular word is a misspelling of a matching
token. This would allow matching to be done on tokens that are similar but not identical and
reduce false negatives. Spellcheckers are imperfect, though, so they could also create false
positives. Achieving an acceptable balance between false-positives and negatives during the
introduction of spellchecking requires further research.
55
Table 6. Illustrative List of False Positives and Potential Mitigation Strategies
False Positive Example Potential Mitigation Strategies
“makes my car look sick.” Employ Word Sense Disambiguation, for example, using
“hurt to look at”; “it hurts so Harvard General Inquirer, as in Abrahams et al. (2012), to
good!” distinguish, for instance, the literal from the figurative sense of
“mic [microphone] cut in and out.” the word. The figurative sense is more likely to be a False
“scared to death” (of spicy food) Positive.
“it‘s a good spicy and a good fire.”
(spicy food) Multiple false positives result from a consumer using “fire” to
“Beware! These noodles are… used describe spicy Korean food.
for… The Fire Noodle Challenge”;
“fire in my mouth”, “fire in my Distinguish the noun (“lead” ingredient) from the verb (“lead”
tongue.” to).
“they aren‘t kidding on the spicy
front. Tissues were immediately
needed as my sinuses were cleared
out by fire.”
“she fell in love with it.”
“I was lead to believe.”
“you have to cut it to fit”, “I cut it Employ Part of Speech Tagging and Named Entity
to match the center tooth”, “I cut Recognition to determine which type of item is the linguistic
that piece off”, “I cut meat”, object of the verb. If a human or human body part is the object
“my mom… is using [this of the harm-related verb (e.g. “cut”), the item is more likely to
mandolin] for almost all types of be a True Positive. If an inanimate object is the linguistic
cutting”, “I always cut my pieces object of the harm-related verb (e.g. “it cut the apple
up when juicing”, “I cut my drying beautifully”), the item is more likely to be a False Positive.
time in half”, “I cut down my coffee
intake.” Distinguish common use (“will burn and stick”) for cookware,
“Great cutting board.” versus unreasonable risk and a substantial product hazard (US
“The apple [logo] is cut out CPSC, 2012).
perfectly!”
“get a piece of cardboard to cut.”
“cheap filters that may hurt your
machine.”
“It is beautifully made and all the
cut outs are in the correct positions
to operate the controls.”
“I started having problems with it”,
“I started having poor signals”,
“will burn and stick.”
“if I cut myself shaving it shows the Identify hypotheticals (e.g. “if”, “unless”, “would”); not the
blood”, “if I am in an emergency same as actuals.
and someone needs to call my
family”, Estimate severity of the hypothetical situation: multiple
“I‘m not sure if I can handle this…! reports of “would cut my fingers” are likely to be a minor
yes you can, unless you‘re allergic discomfort and not necessarily a substantial incision risk.
to the fish listed.”
“I got these… for 20 bucks… retail
price states 70… If I had paid 70
bucks for these things, I would have
used them to cut my own wrists”,
“would cut my fingers.”
56
False Positive Example Potential Mitigation Strategies
“wrist doesn’t hurt anymore”, Identify negations (e.g. “not”, “doesn’t”, “prevent”, “resistant”)
“not hurt your skin”, that reverse the meaning of harm-related words.
“it‘s NOT a deathtrap”,
“many safety features to prevent
accidental cuts”,
“came with… cut resistant gloves”,
“I chose it because it does not have
plastic parts that a baby can chew
off and choke on.”
“I love this lotion. It is one of the
few I am not allergic to.”
“My old juicer hurt my ears!” Determine if the linguistic subject of the complaint is the
“I’m allergic to latex… this is current focal product, or another product (e.g. previous
definitely the best latex free brand.” product that the customer owned). Filter complaints unrelated
to the focal product.
“the edges” Remove phrases generating vast volumes of innocuous false
positives in the target product category from the smoke-term
dictionary (e.g. “the edges” appears in many reviews for phone
cases).
Video post-dates product recall Compare the video date, manufacturer, and product type to
announcement. the CPSC product recall database. Demote the ranking of
videos that are posted directly subsequent to the recall
announcement, since these videos are likely to be in direct
response to the recall announcement, and thus do not cover a
previously unknown risk.
“Another well built fake is available Video is a review of a counterfeit item. Though it’s not a
that you should beware of.” legitimate product hazard, counterfeits damage revenue and
the brand, cannibalize the export sales of the Korean OEM
manufacturer, and subvert post-manufacture safety
inspections and certifications.
Finally, using the word “Korea” as a filter is insufficiently precise – for example, some
products in our dataset use a Korean-inspired recipe. Future work should explore additional
techniques for determining whether the product is manufactured in, or distributed to, Korea.
6. Limitations and Future Work

Expanding the data sources to cover more modalities is still needed in future work.
Consumer hazard claims may appear on corporate product discussion forums4 or on third-
4
For example consumer hazard claims like the following have appeared on Samsung’s corporate product
community forums:
• a front loader making a “bang” and emitting “smoke”
(https://us.community.samsung.com/t5/Washers-and-Dryers/Front-Load-Washer-Bang-and-then-
moke/td-p/238421/page/10)
• “NE58F9500SS Range fire hazard… knobs will turn with the slightest bump… had multiple fires and
melted plastic on my stove due to this defect” ( https://us.community.samsung.com/t5/Kitchen-and-
Family-Hub/NE58F9500SS-Range-fire-hazard/td-p/143698),
57
party consumer complaint websites5. Though these claims may or may not be legitimate, the
ability to scour rapidly for their existence and rapidly counter their legitimacy, or launch
investigations into their veracity, is important for risk management.
The mass-scale transcription of online videos, via automated Artificial Intelligence or
managed human transcription teams, may also be a fruitful path of future research, since
smoke terms can appear in the audio narrative, rather than in the textual video title or
description typed by the person posting the video.
A further potential data source for emergent risks is internet user search queries – see Fig.
2. Dotson et al. (2017) contend that search query data “holds promise for assessing brand
health”. Panel (a) in Fig. 2 – a screen capture from the Google Trends web search query
analysis tool – shows a large user query spike for “Samsung recall” around September 2016.
This spike indicates a potentially severe impact on the Samsung brand due to the September
2016 Samsung Galaxy Note 7 recall. Major search engines also provide consumers with
“related search” suggestions that may be helpful to manufacturers for detecting new risks. For
example, in Fig. 2, Panel (b), related searches on Google for the key phrase “Samsung recall”
indicate that consumers had been searching not just for existing recalls (such as Samsung’s
phone recall and top-loader recall), but also to determine if there have been recalls for
Samsung’s washer mold, washer DC errors, and washer rust – all potential new areas of risk.
Aside from the consumer-facing “related search” suggestions, major search engine providers
– such as Google Ads, Microsoft Advertising, and so on – all provide keyword selection tools
for merchants and manufacturers that show advertisers-related keywords and search volumes
in more detail. These advanced, professional advertiser tools could help manufacturers to find
recall-related searches. For instance, by typing in “[trademark name] recall” in the advertiser
tool, the manufacturer can see a detailed list of related search terms and volumes and conduct
a more sophisticated risk assessment of the nature (related topics) and severity (topic search
volumes) of the risks. Yom-Tov (2017) contends that web searches can predict recalls, though
tests this hypothesis only on medical drug recalls.
For consumer products, we expect that web searches may predict recall advocacy, if not
actual recalls. For example, Panel (a) in Fig. 2 indicates that “clothes dryer” is a related topic
for “Samsung recall” and Panel (b) indicates that “washer rust” is a related topic for “Samsung
recall”. Though we were unable to find an official recall for these issues, Panel (c) reveals that
consumers or consumer advocacy groups have been agitating for recalls for both issues, and
class-action lawsuits have been either filed or are under consideration. For the topic “samsung
washer dc error recall”, Panel (d) indicates that multiple consumers reported their shared
frustration on Facebook, and these consumers could potentially push either individually or
collectively for eventual civil lawsuits or regulatory action. Future work is needed to examine
the extent to which search engine queries and related links can be useful in both risk
identification and characterization.
“Very dangerous stove turns on just leaning to get something... very dangerous appliance?!
NE58k9430SS” (https://us.community.samsung.com/t5/Kitchen-and-Family-Hub/Fire-hazard-stove-
recalled-yet/td-p/811453).
5
For example, claims like the following have appeared on a third-party consumer complaint site:
• “Fire hazard - burners turn on too easily”, “fire/burn hazard. If you just bump the burner controls
they immediately go to high heat”, “I have melted multiple plastic bags and set several things on fire
because all the time the knob gets bumped on when you walk by it”, “At least twice we have had
dishes explode due to the knobs turning from just leaning against them”,
• “Turned on the broiler element … less than 5 min later there was an explosion”,
“After a little over a year the top glass exploded” (See multiple pages of reviews at: https://www.
consumeraffairs.com/homeowners/samsung-stove-oven-range.html)
58
Fig. 2. Potential Alternative Web Search Query Data Sources for Emergent Risk Discovery
(a) (b)
(c)
59
Fig. 2. (Continued)
(d)
7. Conclusions
Our research employed smoke-term dictionaries and hazard discovery text-analytic
procedures, which were adapted from established methods to filter large volumes of online
review data for products associated with Korea, from both textual and video sources. Our
research demonstrates that social media sources can be a fertile reservoir of early warning
signals that could be used to detect emergent product risks for Korean-labeled products. Our
analysis of multiple false positive cases indicates that both human- and machine-based
mitigation strategies can be effective in reducing irrelevant matches and reducing the data
review burden for product risk managers. The automated methods that we have
demonstrated and proposed here can help Korean manufacturers to save time, decrease the
risk of injuries, recalls, and reputation damage, as well as more efficiently resolve product
problems.
Our methods may help to safeguard Korean export trade and export in multiple ways.
Promptly identifying and mitigating product risks can maintain consumer confidence in
Korean products: quality surveillance protects the reputation of both the specific Korean
manufacturer and Korean manufacturers in general as producers of high-quality products.
In addition to facilitating premium pricing, safeguarding Korea’s quality reputation could
avoid negative trade policy decisions, global consumer confidence issues, and export trade
declines that, for instance, Liu, Kerr, and Hobbs (2009) found befell Chinese manufacturers
after the occurrence of severe Chinese product recalls. Finally, hazard incidents may point to
the existence of counterfeit products that are cannibalizing Korean OEM export sales, and
Korean export trade may be safeguarded by the delisting of these counterfeits from popular
online marketplaces, or simply by actively discouraging non-OEM purchases by highlighting
the risks of non-OEM products. Social media surveillance of Korean-labeled products thus
presents a novel and promising means of safeguarding Korean export trade.
60
References
Abrahams, A. S., W. Fan, G. A. Wang, Z. Zhang and J. Jiao (2015), “An Integrated Text Analytic
Framework for Product Defect Discovery”, Production and Operations Management, 24(6),
975-990.
Abrahams, A. S., J. Jiao, W. Fan, G. A. Wang and Z. Zhang (2013), “What‘s Buzzing in the Blizzard
of Buzz? Automotive Component Isolation in Social Media Postings”, Decision Support
Systems, 55(4), 871-882.
Abrahams, A. S., J. Jiao, G. A. Wang and W. Fan (2012), “Vehicle Defect Discovery from Social
Media”, Decision Support Systems, 54(1), 87-97.
Agrawal, J., and W. A. Kamakura (1999), Country of Origin: A Competitive Advantage? International
Journal of Research in Marketing, 16(4), 255-267.
Anshari, M., M. N. Almunawar, S. A. Lim and A. Al-Mudimigh (2019), “Customer Relationship
Management and Big Data Enabled: Personalization & Customization of Services”, Applied
Computing and Informatics, 15(2), 94-101.
Bergstrand, F. and E. Finlaw (2011), Enhancing Current Feedback Processes through Social Media
Monitoring: An Exploratory Study of Social Media and Social Media Monitoring Practices
within an MNC Looking to Combine New Practices with Traditional Customer-centric
Processes (Master’s Thesis), Växjö, Sweden: Linnaeus University.
Biernacki, P. and D. Waldorf (1981), “Snowball Sampling: Problems and Techniques of Chain
Referral Sampling”, Sociological methods & research, 10(2), 141-163.
Boon-Long, S. and W. Wongsurawat (2015), “Social Media Marketing Evaluation Using Social
Network Comments as an Indicator for Identifying Consumer Purchasing Decision
Effectiveness”, Journal of Direct, Data and Digital Marketing Practice, 17(2), 130-149.
Centers for Disease Control (CDC) (2020), Outbreak of Listeria Infections Linked to Enoki
Mushrooms. Available from https://www.cdc.gov/listeria/outbreaks/enoki-mushrooms-03-
20/index.html (Accessed July 21, 2020)
Cheng, M. and X. Jin (2019), “What Do Airbnb Users Care About? An Analysis of Online Review
Comments”, International Journal of Hospitality Management, 76, 58-70.
Constantinides, E. and N. I. Holleschovsky (2016), “Impact of Online Product Reviews on Purchasing
Decisions”, International Conference on Web Information Systems and Technologies,
SCITEPRESS, 2, 271-278.
European Commission (EC), EU General Risk Assessment Methodology (Document No. 2015-IMP-
MSG-15). Available from https://ec.europa.eu/docsroom/documents/17107 (Accessed July 21,
2020)
Dwivedi, Y. K., A. Papazafeiropoulo, D. H. Brown and L. Kaewkitipong (2009), “Relative Size and
Complexity: E-business Use In Small and Medium Sized Tourism Enterprises in Thailand”,
Journal of Enterprise Information Management, 22(1), 212-231.
Dotson, J. P., R. R. Fan, E. M. Feit, J. D. Oldham and Y. H. Yeh (2017), “Brand Attitudes and Search
Engine Queries”, Journal of Interactive Marketing, 37, 105-116.
Fan, Z. P., Y. J. Che and Z. Y. Chen (2017), “Product Sales Forecasting Using Online Reviews and
Historical Sales Data: A Method Combining the Bass Model and Sentiment Analysis”, Journal
of Business Research, 74, 90-100.
Fang, X. and J. Zhan (2015), “Sentiment Analysis Using Product Review Data”, Journal of Big Data,
2(1), 5.
Filieri, R., F. Mcleay, B. Tsui and Z. Lin (2018), “Consumer Perceptions of Information Helpfulness
and Determinants of Purchase Intention in Online Consumer Reviews of Services”, Information
& Management, 55(8), 956-970.
Floyd, K., R. Freling, S. Alhoqail, H. Y. Cho and T. Freling (2014), “How Online Product Reviews
61
Affect Retail Sales: A Meta-Analysis”, Journal of Retailing, 90(2), 217-232.
Food and Drug Administration (FDA) (2020), CJ Foods Voluntarily Recalls Annie Chun’s Japanese-
Style Teriyaki Noodle Bowls Due to Undeclared Peanuts. Available from
https://www.fda.gov/safety/recalls-market-withdrawals-safety-alerts/cj-foods-voluntarily-
recalls-annie-chuns-japanese-style-teriyaki-noodle-bowls-due-undeclared-peanuts (Accessed
April 18, 2020)
Geraghty, G. and A. T. Conway (2016, June 16-17), “The Study of Traditional and Non-traditional
Marketing Communications: Target Marketing in the Events Sector”, Paper Presented at the
12th Annual Tourism and Hospitality Research in Ireland Conference, Limerick, Ireland.
Giri, S. and K. Thapa (2016), “A Study of Customer Satisfaction on After Sales Service of Two
Wheelers in Kathmandu Valley”, Journal of Business and Social Sciences Research, 1(1), 1-21.
Goldberg, D. M. and A. S. Abrahams (2018), “A Tabu Search Heuristic for Smoke Term Curation in
Safety defect discovery”, Decision Support Systems, 105, 52-65.
Goodman, L. A. (1961), “Snowball Sampling”, The Annals of Mathematical Statistics, 148-170.
Hånell, S. M., E. R. Nordman, D. Tolstoy and N. Özbek (2019), “It’s a New Game Out There”: E-
Commerce in Internationalising Retail SMEs. International Marketing Review, 37(3), 515-531.
Health Canada (2016), Overview of Health Canada-S Consumer Product Safety Program Risk
Characterization Methodology. Available from https://www.canada.ca/en/health-
canada/services/ consumer-product-safety/legislation-guidelines/guidelines-policies/overview-
health-canada-consumer-product-safety-program-risk-characterization-method.html (Accessed
May 20, 2020)
Hines, P., M. Francis and P. Found (2006), “Towards Lean Product Lifecycle Management”, Journal
of Manufacturing Technology Management, 17(7), 866-887.
Holleschovsky, N. I. and E. Constantinides (2016), “Impact of Online Product Reviews on Purchasing
Decisions”, The 12th International Conference on Web Information Systems and Technologies,
2, 271-278.
Huh, K. and C. Choi (2016), “Product Recall Policies and Their Improvement in Korea”,
Management and Production Engineering Review, 7(4), 39-47.
Jarrell, G., and S. Peltzman (1985), “The Impact of Product Recalls on the Wealth of Sellers”, Journal
of Political Economy, 93(3), 512-536.
Jo, Y. and A. H. Oh (2011, February), “Aspect and Sentiment Unification Model for Online Review
Analysis”, Paper Presented at the 4th ACM International Conference on Web Search and Data
Mining, 815-824.
Law, D., R. Gruss and A. S. Abrahams (2017), “Automated Defect Discovery for Dishwasher
Appliances from Online Consumer Reviews”, Expert Systems with Applications, 67, 84-94.
Lee, L. F., A. P. Hutton and S. Shu (2015), “The Role of Social Media in the Capital Market: Evidence
from Consumer Product Recalls”, Journal of Accounting Research, 53(2), 367-404.
Liu, H., W. A. Kerr and J. E. Hobbs (2009), “Product Safety, Collateral Damage and Trade Policy
Responses: Restoring Confidence in China’s Exports”, Journal of World Trade, 43, 97.
Liu, R. R. and Zhang, W. (2010), “Informational Influence of Online Customer Feedback: An
Empirical Study”, Journal of Database Marketing & Customer Strategy Management, 17(2),
120-131.
Lockett, N., D. H. Brown and L. Kaewkitipong (2006), “The Use of Hosted Enterprise Applications
by Smes: A Dual Market and User Perspective”, Electronic Markets, 16(1), 85-96.
Mummalaneni, V., R. Gruss, D. M. Goldberg, J. P. Ehsani and A. S. Abrahams (2018), “Social Media
Analytics for Quality Surveillance and Safety Hazard Detection in Baby Cribs”, Safety Science,
104, 260-268.
Nasri, L., M. Baghersad, R. Gruss, N. S. W. Marucchi, A. S. Abrahams and J. P. Ehsani (2018), “An
Investigation into Online Videos as a Source of Safety Hazard Reports”, Journal of Safety
62
Research, 65, 89-99.
Ni, J., J. Li and J. Mcauley (2019, November), “Justifying Recommendations Using Distantly-
Labeled Reviews and Fine-Grained Aspects”, Proceedings of the 2019 Conference on Empirical
Methods in Natural Language Processing and the 9th International Joint Conference on Natural
Language Processing (EMNLP-IJCNLP), 188-197.
Nottage, L. (2020), “Improving the Effectiveness of the Consumer Product Safety System: Australian
Law Reform in Asia-Pacific Context”, Journal of Consumer Policy, 1-22.
Nwokah, N. G. and D. Ngirika (2017), “Online Advertising and Customer Satisfaction of E-Tailing
Firms in Nigeria, Intelligent Information Management, 10(1), 16-41.
Ogawa, S. and F. T. Piller (2006), “Reducing the Risks of New Product Development”, MIT Sloan
Management Review, 47(2), 65.
Office of the United States Trade Representative (OUSTR) (n.d.), Korea. Available from
https://ustr.gov/countries-regions/japan-korea-apec/korea (Accessed August 19, 2020)
Pee, L. G. (2016), “Negative Online Consumer Reviews: Can the Impact Be Mitigated?”,
International Journal of Market Research, 58(4), 545-567.
Pennington, J., R. Socher and C. D. Manning (2014), “Glove: Global Vectors for Word
Representation”, Proceedings of the 2014 Conference on Empirical Methods in Natural
Language Processing (EMNLP), 1532-1543.
Piller, F. T., A. Vossen, and C. Ihl (2012), “From Social Media to Social Product Development: The
Impact of Social Media on Co-Creation of Innovation”, Die Unternehmung, 65(1), 1-22.
Shapiro, C. (1983), “Premiums for High Quality Products as Returns to Reputations”, The Quarterly
Journal of Economics, 98(4), 659-679.
Sithipolvanichgul, J. (2018), “The Determinant of Enterprise Risk Management Implementation:
Evidence in Thailand and Malaysia”, Thammasat Review, 21(1), 1-25.
Souiden, N. and F. Pons (2009), “Product Recall Crisis Management: The Impact on Manufacturer‘s
Image, Consumer Loyalty and Purchase Intention”, Journal of Product & Brand Management,
18(2), 106-114.
Teeramungcalanon, M. (2020), “An Analysis of Consumer Preference Towards Thai Products on
Chinese E-Commerce Platform”, Thammasat Review, 23(2), 1-21.
Von Helversen, B., K. Abramczuk, W. Kopeć and R. Nielek (2018), “Influence of Consumer Reviews
on Online Purchasing Decisions in Older and Younger Adults”, Decision Support Systems, 113,
1-10.
Winkler, M., A. S. Abrahams, R. Gruss and J. P. Ehsani (2016), “Toy Safety Surveillance from Online
Reviews”, Decision Support Systems, 90, 23-32.
Yom-Tov, E. (2017), “Predicting Drug Recalls from Internet Search Engine Queries”, IEEE Journal
of Translational Engineering in Health and Medicine, 5, 1-6.
Zaman, N. (2019), Online Review Analytics: New Methods for Discovering Key Product Quality and
Service Concerns (Doctoral Dissertation), Blacksburg, VA: Virginia Tech.
Zhao, X., Y. Li, B. B. Flynn and S. Ng, (2014), “Impact of Product Recall Announcements on
Shareholder Wealth in China”, Global Supply Chain Quality Management: Product Recalls and
Their Impact, 142(1), 197-217.
https://doi.org/10.35611/jkt.2020.24.8.63 63
Importance of Political Elements to Attract

FDI for ASEAN and Korean Economy
Monthinee Teeramungcalanon JKT 24(8)
Pridi Banomyong International College, Thammasat University, Thailand
Eric M.P. Chiu†

Graduate Institute of National Policy and Public Affairs, National Chung-Hsing University, Taiwan
Yoonmin Kim
Department of Economics and Finance, Keimyung University, South Korea
Abstract
Purpose – Recent empirical studies have shown that FDI is expected to be strongly associated with
democratic governance, political stability, and sound macroeconomic conditions of the host country.
We attempt to take it a step further to see if governments implement a major change in institutional
characteristics, will the institutional reform toward better governance have a substantive effect in
enhancing FDI inflows. This paper thus aims to analyze the importance of good governance as an
important factor in the attractiveness of FDI inflows in ASEAN+3 (Korea, China, Japan) countries.
Design/methodology – To determine the effects of good governance on FDI inflows across ASEAN+3
countries recorded between 1996-2018, the Worldwide Governance Indicators (WGI) are used to
investigate the impact of good governance on FDI inflows. The model has been estimated by using
fixed effects to show the robustness of the results.
Findings – Our main findings can be summarized as follows: Political Stability, Rule of Law, and Voice
and Accountability have a statistically significant impact on the inflow of FDI in the ASEAN+3
Countries, especially for Korean economy. Moreover, GDP growth continue to exert their positive
influence. However, Regulatory Quality, Government Effectiveness and Control of Corruption,
though equally important, are insignificant to attract FDI inflows. The key finding is that good
governance has a significant impact on inward FDI in the ASEAN+3 countries.
Originality/value – Existing studies focus on the impact of political factors on FDI across countries.
This paper instead attempts to investigate which type of good governance is the most important in
promoting FDI inflows across ASEAN+3 countries, which is essential for multinationals to consider
when choosing a foreign site as a possible FDI destination.
Keywords: ASEAN+3 Countries, Foreign Direct Investment, Good Governance, Korean Economy
1. Introduction
Over the past few decades, Foreign Direct Investment (FDI) has played an important role
as a key driver in economic development as well as the most important driving force of
globalization. It contributes to economic development through raising capital accumulation,
increasing the transfer of technology, enhancing productivity, creating employment
opportunities, and fostering macroeconomic stability.
ASEAN, the third-largest economy in Asia presents a compelling investment destination
for many investors (Kirk, Ractham and Abrahams, 2016). FDI inflows and exports have long
†
Corresponding author: ericchiu@nchu.edu.tw
64
since been important contributors to ASEAN’s economy, which, over the past few years, has
seen an inflow of FDI reaching US$137 billion in 2017. Similarly, the ASEAN+3 (Korea,
China, Japan) countries established in the 1990s has become one of the most dynamic
markets in the world and is the key to regional economic growth (Kirk, Abrahams and
Ractham, 2016). This region has developed into mechanisms of great significance in
strengthening and deepening regional cooperation, particularly in economic, social, and
political areas.
Although there were challenges resulting from uncertainties in the global economy, in 2017
for example, the total trade between ASEAN+3 countries amounted to US$ 813.5 billion,
which accounted for 31.6% of ASEAN’s total trade. In addition, the total foreign direct
investment (FDI) flows from Korea, China, and Japan countries into ASEAN reached
US$ 29.9 billion, accounting for 21.8 % of the total FDI inflow to ASEAN as shown in Fig. 1
below.
Fig. 1. FDI as % of GDP in ASEAN+3 Countries (1996-2018)
90
80
70
60
50
40
30
20
10
0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
Source: Word Bank (2019).
In recent years, good governance has emerged as a key factor in the globalization process
and international capital movement (Angkinand and Chiu, 2011). Moreover, countries with
high comparative advantages, locational attractions with immobile natural resources, and
benefits of internalizing cross-border intermediate product markets are more likely to attract
MNCs to engage in the FDI (Dunning and Lundan, 2008; Suh Jae-Hyun, 2020). In addition,
countries with a risk-free environment and a good quality institution can facilitate investment
by reducing the transactions, manufacturing, and production costs, which can, in turn,
increase the profitability and ultimately, influence international investors to engage in cross-
border transactions (North, 1990). Further, the quality of good institutions can help attract
FDI through good governance, which represents an important factor in the attractiveness of
foreign investment (North et al., 2008). Thus, governments that seek to attract FDI can create
more favorable conditions for multinational enterprises through the improvement of political
institutions and economic policies that can help stimulate FDI inflows (White et al., 2015).
65
Fig. 2. FDI as % of GDP in Korea, China, and Japan (1996-2018)
0
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
-1
China Japan Korea
Source: Word Bank (2019).
Previous literature has shown that, for instance, FDI is expected to be strongly associated
with democratic governance, political stability, and sound macroeconomic conditions of the
host country (Bitzenis and Žugić, 2016). We attempt to take it a step further and ask if
governments implement a major change in institutional characteristics (Sayruamyat and
Nadee, 2020), will the institutional reform toward better governance have a substantive effect
in enhancing FDI inflows? The next and more important question is which type of good
governance is the most important in promoting FDI inflows? We studied these questions
based on a sample of 13 AESAN+3 countries from 1996–2018 and found that the
improvement of the political and institutional environments can attract more inward FDIs in
ASEAN+3 countries.
This paper is organized as follows: Section 2 begins with a brief literature review on the
relationship between good governance and FDI inflows; Section 3 presents the state of
governance in ASEAN+3 countries; Section 4 discusses the methodology and describes the
data used; Lastly, empirical results are presented in Section 5; and the conclusion is presented
in Section 6.
In this section, the conceptual and theoretical frameworks of this research is outlined
followed by the empirical studies that focus on the relationship between governance and FDI.
2.1. The Concept of Good Governance

In recent years, the concept of “good governance” has been widely used to explain the
process of decision-making; the process by which decisions are either implemented or not
(Kaufman et al., 1999), and the process that when implemented in its totality, leads to
sustainable development and change (Aminuzzaman, 2006). Governance can be defined as a
system of government focusing on effective and accountable institutions and the relationship
between government and society (Halfani et al., 1994; McCarney et al.,1996; Tim et al., 2013).
66
The emergence of social media as boundary objects in crisis response: a collective action
perspective. Similarly, governance is also a process by which the decisions stakeholders are
implemented, interest is articulated, and decision-makers are held accountable.
According to the United Nation Development Bank, good governance is defined as a
framework of public management based on a fair and efficient system of justice, the rule of
law, and involvement in the process of governing and being governed (UNDP, 1997). It can
be seen as the authority exercised to manage a country’s economic, political, and admini-
strative affairs to foster social cohesion and integration and improve the well-being of citizens
(UNDP, 1998). In addition, good governance can help eradicate poverty, create employment,
and enhance living conditions with sustainable environmental standards (UNDP, 1997).
According to UNDP, good governance can be measured by the following nine core
characteristics: participation, consensus orientation, accountability, rule of law, effectiveness
and efficiency, transparency, responsiveness, equity, and strategic vision (Mehta, 2000).
Similarly, the World Bank defined good governance as the process by which governments
are selected and replaced, the capacity of a government to effectively formulate and
implement policies, and the respect of citizens and the state for their institutions that govern
interactions among them (Kaufman et al., 1999). Further, good governance can help foster
and establish strong institutions to support sustained economic and social development in a
country. In this paper, we focus on the effect of policymakers’ decisions on improving
domestic institutional environments in terms of attracting FDI inflows. Following the work
of Kaufmann et al. (1999/2009) in particular, we categorize the institutional effects of good
governance into six types: voice and accountability, political stability, government efficiency,
regulatory quality, rule of law, and control of corruption. The primary reason for the focus
on these types of good governance is due to most emerging market countries in Southeast
Asia usually experience increasing but somewhat volatile, trade and capital mobility while
having less mature political systems, greater instability, and generally a weaker legal and
administrative system, which would be more susceptible to political influences on a
macroeconomic level.
2.2. Linkages between Good Governance and FDI Inflows

Recently, many researchers identified the concept of good governance as an important
factor for the attractiveness of foreign investments. A number of determinants are identified
in the empirical literature that affects inward FDIs. Additionally, several studies have
investigated various institutional aspects of governance such as democracy, corruption, and
rule of law (Bailey, 2018; Blonigen, 2005; Vijayakumar et al., 2010; Wei and Scheleifer, 2002)
while others have focused on only one of the institutional factors that affect FDI inflows
(Ahlquist, 2006; Jensen, 2003). This section presents an overview of the latter, focusing on the
relationship of the impact of good governance on FDI inflows.
Theoretically various types of institutions tend to affect and being associated with the
health of national economy. Efficient institutions such as a more transparent democratic
government will signal market agents (both domestic and foreign investors) about future
economic fundamentals, and can thereby shape market expectations on the inflows of foreign
investment. Likewise, contract enforcement and protection of private property rights as well
as stable political environment that provide stable investment environment are necessary to
improve market efficiency in transaction since they are likely to lower transaction costs and
barriers to foreign investors. In other words, as Angkinand and Chiu (2011) have suggested,
institutions that heavily corresponding with bad economic fundamentals will tend to
destabilize market expectations and increase market uncertainty, therefore reducing the
67
incentives for foreign trade. Moreover, it is the role of government to encourage FDI inflows
into the sectors that are in line with the country’s overall economic structural adjustments in
order to enhance national competitiveness and reduce social and economic inequality
(Teeramungcalanon and Chiu, 2020).
Many studies have empirically shown that political and institutional factors are necessary
determinants of FDI movements in developing countries. For example, a study on 52
countries from 1985–2002 found that low corruption, a developed banking sector, access to
information, and efficient bureaucracy were essential factors for FDI inflows (Bénassy-Quéré
et al., 2007). Similarly, political instability, corruption, and non-transparent institutional
factors affected the FDI inflows of multinational companies and their subsidiaries negatively
(Wang and Swain, 1995). Another study showed that corruption and bad governance lead to
an increase in administrative costs and therefore, reduced FDI inflows (Morisset, 2002).
Meanwhile, for the Middle Eastern and North African countries from 2002–2007, a better
quality of governance would have had a positive impact on FDI inflows (Samimi and Ariani,
2010), whereas for African countries, a good infrastructure, less corruption, political stability,
and a reliable legal system had a positive impact on FDI inflows (Asiedu, 2006).Moreover,
empirical studies employed the six good governance indicators that were developed by
Kaufmann et al. (1999) to investigate the impact of good governance on FDI inflows through
voice and accountability, political stability, government effectiveness, regulatory quality, rule
of law, and control of corruption.
From a study of 50 African countries, political stability, control of corruption, and
regulatory quality were found to have a significant impact on FDI inflows (Gangi and
Abdrazak, 2012), whereas low regulatory law and political stability enhanced FDI inflows of
45 developing countries in Africa, Asia, and Latin America (Bissoon, 2011). A study of 103
countries between 2001–2007 revealed that good governance indicators showed mixed results
in both effect and significance (Spatafora and Luca, 2012), which is in line with the result from
a study of 124 countries between 1996–2009 (Gordon et al., 2012). From 1995–1997, voice
and accountability, government effectiveness, regulation, and control of corruption had
significant positive effects on 115 developing and developed countries (Globerman and
Shapiro, 2003). Additionally, a study of 15 Asian countries from 1996 to 2007 revealed that
political stability, rule of law, government effectiveness, control of corruption, and absence of
violence were the primary factors affecting FDI inflows (Mengistu and Adhikary, 2011).
Similarly, political stability and regulatory quality had a positive impact on FDI inflows, while
corruption, absence of governance effectiveness and rule of law had a negative influence
(Shah and Samdani, 2015). This is in line with the study of 122 developing countries from
2002 to 2017 that revealed that governance effectiveness and regulatory quality were essential
factors in attracting FDI inflows (Ross, 2019). In addition, political stability had a significant
positive effect in the case of developing countries (Baek and Qian, 2011; Zheng, 2011), while
in the case of emerging countries, control of corruption has had a positive and significant
effect on inward FDI (Mathur and Singh, 2013). Therefore, governance indicators were found
to have a positive and significant effect on the inward FDI (Ali et al., 2010; Globerman and
Shapiro, 2003; Hur et al., 2011; Muhammad et al., 2011; Wernick et al., 2009). Based on the
above literature, two hypotheses are derived as follows:
H1: Institutional structures of good governance have a significant impact on FDI inflows,
ceteris paribus.
H2: The magnitude of the effect of each type of good governance is statistically and
significantly different from the other, ceteris paribus.
68
3. State of Governance in ASEAN+3 Countries
Table 1 shows the results of calculations of the governance situation in ASEAN+3 countries
for the period 1996–2018; we computed the average value of the indicators for each country
assuming that the maximum value of the indicator is 2.5 (strong governance), and the lowest
value is -2.5 (weak governance). The results of the calculation show that, the governance
situation in the ASEAN+3 countries for the period 1996–2018 was poor with the exception
of a few countries as governance indicators for the majority of the countries fell below the
zero level. The exceptions include Korea, Brunei, Malaysia, Singapore, and Japan. Among the
six governance indicators, the worst performance was in the VAA indicators with negative
values for 10 countries. This implies that poor governance results in the lack of democracy,
lack of freedom of expression, association, and free media.
Fig. 3 shows the comparison of governance indicators between ASEAN and Korea, China,
and Japan during the period of 1996-2018 (Teeramungcalanon, 2020). It can be seen Korea
and Japan showed positive values for all the governance indicators, which reflect good
governance, while ASEAN show negative values for the governance indicators. Among the
ASEAN countries, Indonesia, Cambodia, Lao PRD, and Myanmar showed negative values
for all the governance indicators, which reflect very poor governance.
In the case of Korea, the rule of law is well established, as Korea has puts efforts to
strengthen the rule of law as one of the pillars of its foreign policy such as international rule-
making that responds to issues the global economy face. The legal and regulatory systems of
Korea are fair and balanced, subsidies, tax credits, and other incentives are offered to attract
foreign investment. However, in terms of managing political stability and control of
corruption, Korea still has room to improve. Japan has a low level of corruption with a close
relationship among companies, politicians, and government agencies that foster an inwardly
cooperative business climate. Korea, on the other hand, has been suffering from political
instability and corruption which hinders the inflows of FDI in recent years.
Table 1. Governance Indicators in ASEAN+3 Countries (1996–2018)

Voice and Political Government Regulatory Rule Control of
Country Average
Accountability Stability Effectiveness Quality of Law Corruption
Brunei -0.77 1.19 0.90 0.99 0.56 0.51 0.56
Indonesia -0.19 -1.16 -0.30 -0.33 -0.62 -0.76 -0.56
Cambodia -0.93 -0.45 -0.84 -0.41 -1.12 -1.13 -0.81
Lao PRD -1.60 -0.07 -0.76 -1.06 -0.97 -1.02 -0.91
Myanmar -1.79 -1.20 -1.34 -1.80 -1.44 -1.29 -1.47
Malaysia -0.39 0.22 0.99 0.60 0.45 0.24 0.35
Philippines 0.10 -1.18 -0.04 -0.04 -0.36 -0.55 -0.35
Singapore -0.05 1.22 2.15 2.00 1.58 2.17 1.51
Thailand -0.26 -0.54 0.28 0.23 0.10 -0.31 -0.08
Vietnam -1.39 0.27 -0.27 -0.60 -0.42 -0.56 -0.49
Korea 0.70 0.37 0.95 0.81 0.96 0.47 0.71
China -1.56 -0.42 0.06 -0.27 -0.47 -0.40 -0.51
Japan 1.02 1.07 1.39 1.03 1.36 1.35 1.20
Source: Authors’ calculation using WGI data.
69
Fig. 3. Governance Indicators between ASEAN and Korea, China, and Japan (1996-2018)
1.5
0.5
0
Voice and Political Government Regulatory Rule of Law Control of
-0.5
Accountability Stability Effectiveness Quality Corruption
-1
-1.5
-2
Korea China Japan ASEAN
Source: Authors’ calculation using WGI data.
4. Research Design
To determine the effects of good governance on FDI inflows, the governance indicators are
employed based on the World Bank Worldwide Governance Indicators developed by
Kaufmann et al. (1999). Experts using an unobserved components model base this dataset on
a wide variety of across-country surveys and polls. In accordance with the previous studies
(Cornia, 2015; Mihaylova, 2015), the model also includes several control variables that are
expected to have an impact on FDI (Table 2). The empirical analysis is based on a panel of 13
countries over the period 1996–2018. The following equation is estimated to investigate the
impact of good governance on FDI:
∝ , (1)
According to the equation, the dependent variable FDIit is measured by annual inflows of
FDI (millions USD at current price) for region i in the period t. GOVit is a vector of good
governance indicators in the host country, and CVit is a vector of control variables. The fixed-
effects (FE) model is applied to check for the robustness of the results.
Foreign Direct Investment (FDI): When multinationals enter a foreign market, either by
acquiring a firm or broadening the existing business activities in the host country, it is called
foreign direct investment. This study will use FDI inflows as the dependent variable of 13
ASEAN+3 countries to gauge the influence of good governance on inward FDIs.
The following six worldwide governance indicators represent different aspects of
institutional quality in a country. The empirical analysis will use two sets of data: the first
dataset is derived from the estimate of governance (ranging from approximately -2.5 [weak]
to 2.5 [strong] governance performance), and the second dataset is derived from the
percentile rank among all countries ranging from 0 (lowest) to 100 (highest) rank.
Voice and Accountability (VAA): This factor measures the ability of a country’s citizens to
participate in selecting their government as well as the freedom of expression, association,
and free media (Kaufmann et al., 2009). Through accountability, the public can have
70
increased access to information regarding the government’s performance and the
opportunity to raise their voice (Albritton and Bureekul, 2009). Accountability also enables
the state and its institutions (the public sector) to work toward developing effective strategies
through strong monitoring mechanisms (Cheema and Rondinelli, 2007). In this context,
studies have shown that voice and accountability can have a positive effect on FDI (Blanton
and Blanton, 2007; Doces, 2010; Zheng, 2011), while others have shown that voice and
accountability has a negative effect on FDI (Guerin and Manzocchi, 2009; Jadhav, 2012).
Therefore, FDI can be affected by voice and accountability through inclusion or exclusion of
public opinion on investments, which in turn can allow or deter foreign investment (Gani,
2007).
Political Stability (POL): This measures perceptions of the likelihood of political instability
(Kaufmann et al., 2008). Studies have shown that political stability is a significant factor that
influences an investor’s decision to establish their business (Martinez-Zarzoso, 2011) and also
has a positive, significant impact on the inflow of FDI (Baek and Qian, 2011; Busse et al., 2011;
Wang, 2009). In addition, political instability or unexpected changes to the government or
country’s institutions can affect FDI inflows and economic growth (Schneider and Frey, 1985;
Tuman and Emmert, 2004). Therefore, multinationals would prefer countries with steady
policies that ensure the continuity of policies and support international investments and
businesses.
Government Effectiveness (GOV): This reflects perceptions regarding the quality of public
services, civil service and the degree of its independence from political pressures, the quality
of policy formulation and implementation, and the credibility of the government’s
commitment to such policies (Kaufmann et al., 2008). The quality of a government’s
regulatory effectiveness and practices have an impact on FDI inflows (Rammal and
Zurbruegg, 2006) because its effectiveness plays a major role in attracting foreign firms into
the host country (Rodrick, 2008). In addition, effective government infrastructure and
favorable conditions for domestic and foreign investors are important determinants in FDI
inflows. Moreover, as a stable government guarantees the continuation of policies, a positive
relationship between government effectiveness and FDI inflows is expected.
Regulatory Quality (RQ): This reflects perceptions regarding the ability of the government
to formulate and implement policies and regulations that permit and promote private sector
development (Kaufmann et al., 2008). Further, a country’s quality of a regulation system
depends on the level to which regulations are made and society accepts them (Lee and Tan,
2006). Hence, it is the responsibility of the government to create legislation to regulate the
economy, frame the competitive environment and factor endowment, and establish a
regulatory environment by which business is conducted (Henisz, 2000; Rugman and
Verbeke, 2000). To boost economic development by attracting foreign investments, a
government must ensure that the economy is operated by employing strong regulations and
policies (Rodrik, 1999). If the regulatory quality is high, a positive influence on overseas
investment is expected as multinationals seek locations where the regime is market-friendly.
Rule of Law (ROL): This factor reflects perceptions regarding the extent to which agents
have confidence in and abide by the rules of society and in particular, the quality of contract
enforcement, property rights, the police, and the courts as well as the likelihood of crime and
violence (Kaufmann et al., 2008). Moreover, government officials and institutions that respect
and obey the legal frameworks are the most effective way of promoting the rule of law
(Shivute, 2008), and a country with a strong rule of law protection plays a positive role in
attracting FDI (Lee and Mansfield, 1996; Jadhav, 2012; Nunenkamp and Spatz, 2004).
Therefore, countries with poor property rights protection and weak legal institutions will
attract fewer investors to invest at risk and vice versa.
71
Controls of Corruption (COC): This reflects perceptions regarding the extent to which
public power is exercised for private gain. This includes both petty and grand forms of
corruption as well as the “capture” of the state by elites and private interests (Kaufmann et al.,
2008). Countries that manage to decrease the level of corruption and improve the rule of law
and property protection can help attract more FDI inflows (Vittal, 2001); otherwise,
corruption can lower the confidence and trust people have in the economic system, politics,
institutions, and their leaders (Rohwer, 2009). In this context, studies have shown that
corruption affects inward FDI (Al-Sadig, 2009; Goodspeed et al., 2011; Gordon et al., 2012)
and can lead to a decrease in the profitability of investment and an increase in the uncertainty
of the host country. Likewise, it can also lead to an increase in investment (Teksoz, 2006).
Therefore, corruption is an important measure of governance as it has an important influence
on investments. In addition to the good governance variables, this model also includes several
controlled variables that are likely to have an impact on FDI inflows (Angkinand and Chiu,
2011; Chiu and Willet, 2019).
Trade Openness (TRADE): The degree to which a country is open to the outside world is
one of the most important stimulants to attract FDI. This variable uses the ratio of Trade to
GDP as an indicator of the degree of openness. Previous studies have shown a positive effect
of trade liberalization on FDI because the openness of the host countries can help
demonstrate the ease with which the goods can be imported and exported (Govil, 2013;
Ranjan and Agrawal, 2011; Shah and Samdani, 2015).
GDP Growth (GDP): Such growth can reflect the growing market of the host country and
is considered the most crucial factor in attracting MNCs (Kok and Ersoy, 2009). Additionally,
a high GDP growth rate not only indicates the effectiveness of government institutions and
the stability of economic policy but also shows how well-off consumers are in a country
(Globerman and Shapiro, 2003).
Inflation Rate (INF): This variable uses the Consumer Price Index (CPI) as the annual
inflation rate. However, an increase in inflation may decrease the purchasing power of the
poor more than that of the rich. Nevertheless, it can also lead to an increase in investment
that results in higher employment creation thus improving income distribution (Kai and
Hamori, 2009).
Table 2. Variable Definitions

Variables Variable Definition Source
FDI Foreign Direct Investment, Net Inflows % of GDP World Bank
VAA Voice and Accountability Index WGI
POL Political Stability and Absence of Violence Index WGI
GOV Government Effectiveness Index WGI
RQ Regulator Quality Index WGI
ROL Rule of Law Index WGI
COC Control of Corruption Index WGI
TRADE Trade openness World Bank
GDP Market Size World Bank
INF Consumer Price Index World Bank
72
To reduce the problem of reverse causality, all of the independent variables are lagged by
one year which at the same time deals with the problem of a crisis forcing a change in capital
controls. Of course, there is the potential for our approach to suffer from omitted variable
bias and deeper endogeneity problems (Achen 1986; Maddala 1983; Timpone 2001). One way
to deal with endogeneity problems with a dichotomous dependent variable and endogenous
explanatory variables is to use two stage probit least squares (2SPLS). However, as pointed
out by Timpone 2001, even if we obtain a consistent estimator via this two-stage approach,
the standard errors remain inaccurate and their correction is extremely difficult in practice.
Thus, the full model to investigate the effects of good governance on FDI in the ASEAN +
3 countries can be expressed as follows:
∝
(2)
,
This section presents the descriptive statistics for all variables followed by the empirical
results of the effects of good governance on FDI inflows across the ASEAN+3 countries over
the period 1996–2018.
5.1. Descriptive Statistics

The descriptive statistics shown in table 3 reveals that FDI as % of GDP vary from country
to country with a sample range of -2.76% (minimum) and 28.02% (maximum).
Moreover, the six good governance indicators also vary largely from country to country
with a huge range. For instance, government effectiveness and regulatory quality vary from a
minimum of -1.62 and -2.34 respectively to a maximum of 2.44 and 2.26 respectively, while
voice and accountability vary from a minimum of -2.23 to a maximum of 1.11. Similarly,
control of corruption also varies from a minimum of -1.67 to a maximum of 2.33. This
indicates that Singapore that can manage to decrease the level corruption can attract more
FDI inflows, while country like Myanmar with low level of control of corruption tend to
attracts less FDI inflows. Political stability that ensure the continuity of policies and support
international investments and businesses also vary from a minimum of -2.09 to a maximum
of 1.59. However, the rule of law index shows a minimum average fluctuation among the six
components of good governance that varies from a minimum of -1.74 to a maximum of 1.84.
In addition, the controlled variables also exhibit a wide variation in the sample period. GDP
growth rate which is a crucial factor in attracting investors shows a minimum growth rate of
-13.13 to a maximum growth rate of 14.53 with a average fluctuation of 4.88. This implies that
most of the countries such as Singapore have achieved a high growth rate, while some
countries have experienced low performance and even negative GDP growth. Trade openness
which is one of the important stimulants to attract FDI is found to be 107.73% on average,
with a wide range of 0.17% (minimum) and 437.3% (maximum). Similarly, the inflation rate
also varies from -2.31% in some countries to about 125.27% in other countries. The above
indicators indicate that ASEAN+3 countries have a wide range with regard to the quality of
governance, the macro-economic policy measure, and the ability to attract FDI. Therefore,
countries with good governance and steady economic policy can help to attract international
investment and business.
73
Variables Observation Mean Std. Dev. Minimum Maximum
FDI 299 4.30 5.064 -2.76 28.02
VAA 299 -0.55 0.887 -2.23 1.11
POL 299 -0.05 0.908 -2.09 1.59
GOV 299 0.24 0.972 -1.62 2.44
RQ 299 0.09 0.986 -2.34 2.26
ROL 299 -0.03 0.932 -1.74 1.84
COC 299 -0.10 0.993 -1.67 2.33
TRADE 299 107.73 87.667 0.17 437.33
GDP 299 4.88 3.631 -13.13 14.53
INF 299 5.46 11.414 -2.31 125.27
Source: Authors’ calculation using WGI and World Bank data.
5.2. Regression Results

The empirical results of the effects of good governance on FDI inflows across the ASEAN+3
countries over the period 1996–2018 by using the FE model is presented in Table 4. The
dataset is derived from the estimate of governance (ranges from approximately -2.5 [weak] to
2.5 [strong] governance performance).
The empirical results reveal that political stability and rule of law have a positive and
statistically significant relationship with FDI inflows at the 1% level of significance, while
voice and accountability have a negative and statistically significant relationship with FDI
inflows at the 1% level of significance. Meanwhile, government effectiveness, regulatory
quality, and control of corruption do not have a significant relationship with FDI inflows. The
result is robust with regard to the addition of other controlled variables, which implies that
political stability and rule of law are key elements of good governance that are crucial in
stimulating FDI inflow in the ASEAN+3 countries.
It is worth noting that political stability has a positive influence on inward FDI for
ASEAN+3 countries, demonstrating that they have steady policies that ensure the continuity
of policies and support international investment and business, which in turn attracts more
foreign investors to the host country. The result is in line with the previous studies of Baek
and Qian (2011), Busse et al. (2011) and Wang (2009) pointed out that political stability has
a great influence on investors’ decisions in establishing their businesses in a host country.
Hence, host countries should maintain a level of political stability that will foster confidence
among foreign investors.
Moreover, rule of law is also an important component of good governance infrastructure
that plays a positive role in facilitating FDI inflows to a host economy. This reveals that
countries with an effective and transparent legal system not only increases domestic
investment but also attracts foreign investors for long-term investment. In addition,
institutionalizing an effective rule of law is also significantly influential in protecting
intellectual property and individual rights thereby promoting innovation and invention. The
result is in line with the findings of Jadhav (2012) and Nunenkamp and Spatz (2004) who
pointed out that a country’s strong rule of law protection played a positive role in attracting
FDI.
Alternatively, voice and accountability has a negative effect on FDI inflows. In the early
stage of a country’s development, the freedom of expression of citizens might influence the
government’s decision-making in foreign investment policy. As people have more access to
74
information regarding the government’s performance and opportunity to raise their voice,
this eventually has an impact on the FDI inflows. The result is in line with the finding of
Guerin and Manzocchi (2009) and Jadhav (2012), which showed that voice and
accountability have a negative effect on FDI. This implies that voice and accountability affect
FDI by the inclusion of public opinion on investments, which, in turn, deters foreign
investment (Gani, 2007).
Table 4. Impact of Good Governance on FDI Inflows-Fixed Effects (Estimate)

(1) (2) (3) (4)
VAA -2.2694 *** -1.6614 ** -1.6602 ** -1.5536 **
(-2.96) (-2.21) (-2.20) (-2.01)
POL 1.5268 *** 1.8267 *** 1.8165 *** 1.8662 ***
(3.18) (3.90) (3.85) (3.91)
GOV -0.3324 -0.9269 -0.9219 -0.9798
(-0.39) (-1.11) (-1.11) (-1.17)
RQ 0.8855 0.9272 0.9060 0.8845

(1.08) (1.17) (1.13) (1.10)
ROL 3.7666 *** 2.0541 * 2.0487* 1.8829 *

(3.29) (1.76) (1.75) (1.57)
COC 0.1034 0.1239 0.1430 0.0693
(0.09) (0.12) (0.13) (0.06)
GDP 0.0001 *** 0.0001 *** 0.0001 ***
(4.57) (4.56) (4.59)
INF -0.0024 -0.0026
(-0.20) (-0.21)
TRADE 0.0045
(0.67)
Observations 299 299 299 299
R-Squared Within 0.1600 0.2185 0.2186 0.2198
Note: *p<0.1, **p<0.05, ***p<0.001.

Beside the good governance indicators, GDP growth rate is another important variable that
exerts a significant positive impact on FDI inflows implying that the economic growth of
ASEAN+3 countries shows a signal of economic stability and favorable investment climate
with a high degree of market demand that is more attractive to foreign investors. This result
is in line with the study by Globerman and Shapiro (2003) that shows that a country with a
stable GDP growth rate not only indicates the stability of economic policy and the
effectiveness of the government institutions but also shows how well off the consumers are in
a country. Furthermore, other variables such as inflation rate and trade openness are not
statistically significant with respect to FDI inflows in the ASEAN+3 countries.
Table 5 presents the results of good governance on FDI inflow using the second dataset
derived from the percentile rank of governance (ranges from 0 [lowest] to 100 [highest] rank).
This dataset is used to check for the robustness of the results. The empirical results reveal that
among the governance indicators and political stability are positively influential on FDI
75
inflows, implying that the ASEAN+3 countries have strong political stability that helps boost
foreign investors’ confidence in investing in the host country. However, voice and
accountability have a negative and statistically significant relationship with FDI inflows.
Similarly, the result is robust with the addition of other controlled variables, which implies
that political stability and rule of law are the key elements of “good governance” that are
crucial in stimulating FDI inflow in the ASEAN+3 countries, whereas voice and
accountability seem to lower the inward flow of FDI.
Table 5. Robustness Check on the Impact of Good Governance on FDI Inflows-Fixed Effects
(Rank)
(1) (2) (3) (4)
VAA -0.9855 *** -0.0692 *** -0.0663 *** -0.0672 ***
(-3.81) (-2.71) (-2.56) (-2.59)
POL 0.0807 *** 0.0821 *** 0.0828 *** 0.0817 ***
(4.92) (5.21) (5.24) (5.13)
GOV 0.0085 -0.0025 -0.0056 -0.0065
(0.30) (-0.09) (-0.20) (-0.23)
RQ 0.0247 0.0230 0.0205 0.0019
(0.75) (0.73) (0.65) (0.56)
ROL -0.0388 * -0.0013 * -0.0029 * -0.0019
(-0.93) (-0.03) (-0.07) (-0.05)
COC 0.0427 0.0366 0.0361 0.0380
(1.44) (1.28) (1.26) (1.32)
GDP 0.0001 *** 0.0001 *** 0.0001 ***
(4.90) (4.95) (4.93)
INF 0.0049 0.0050
(0.76) (0.77)
TRADE -0.0081
(-0.64)
Observations 299 299 299 299
R-Squared Within 0.1446 0.2124 0.2140 0.2151
Note: *p<0.1, **p<0.05, ***p<0.001.
In addition, macroeconomics indicators such as GDP growth also show significant effects
on FDI inflows. In using the two datasets, the findings reveal that political stability, voice and
accountability, and rule of law, are the three important determinants of FDI inflows. The
above findings lend strong support for the significance of good governance to FDI inflows in
the ASEAN+3 countries. Therefore, the ASEAN+3 countries that are deficient in political
stability and rule of law may attract less FDI.
6. Conclusion Remarks
Many researchers have explored the impact of good governance as an important factor in
the attractiveness of foreign investment in developing countries. This paper thus aimed to
76
analyze the impact of good governance on FDI inflows across the ASEAN+3 countries over
the period of 1996–2018. The model has been estimated by using fixed effects to show the
robustness of the results. The findings reveal that political stability, rule of law, and voice and
accountability are the three crucial factors that have a significant impact on FDI inflows of
the ASEAN+3 countries. This implies that these countries have a stable political atmosphere
that helps foster confidence in foreign firms and has a strong rule of law protection that plays
a positive role in attracting FDI. For the future research, a game theoretical framework could
be considered in better understanding the interactions between governments and economic
developments.
In conclusion, good governance is one of the crucial factors that multinationals consider
when choosing a foreign site as a possible FDI destination, especially in the ASEAN+3
countries. Therefore, countries that strive to attract foreign capital should pay attention not
only to enhancing good governance environments, effective rule of law, and the stability of
economic policies to attract overseas investors but also to improve and maintain various
aspects of good governance to maximize economic welfare in the long run.
References
Achen, C. H. (1986), The Statistical Analysis of Quasi-Experiments, Berkeley, CA: University of
California Press.
Ahlquist, J. S. (2006), “Economic Policy, Institutions, and Capital Flows: Portfolio and Direct
Investment Flows in Developing Countries”, International Studies Quarterly, 50(3), 681-704.
Albritton, R. B. and T. Bureekul (2009), Are Democracy and “Good Governance” Always
Compatible? Competing Values in the Thai Political Arena (Working Paper, No. 47), Montreal,
Canada: International Political Science Association.
Ali, F. A., N. Fiess and R. Macdonald (2010), “Do Institutions Matter for Foreign Direct Investment?”,
Open Economies Review, 21(2), 201-209.
Al-Sadig, A. (2009), “The Effects of Corruption on FDI Inflows”, Cato J., 29, 267.
Aminuzzaman, S. M. (2006), “Governance and Development: An Overview”, Governance and
Development–Bangladesh and Regional Experiences, 11-32.
Angkinand, A. P. and E. M. Chiu (2011), “Will Institutional Reform Enhance Bilateral Trade Flows?
Analyses from Different Reform Aspects”, Journal of Economic Policy Reform, 14(3), 243-258.
Asiedu, E. (2006), “Foreign Direct Investment in Africa: The Role of Natural Resources, Market Size,
Government Policy, Institutions and Political Instability”, World Economy, 29(1), 63-77.
Baek, K. and X. Qian (2011), “An Analysis of Political Risks and The Flow of Foreign Direct
Investment in Developing and Industrialized Economies”, Economics, Management and
Financial Markets, 6(4), 60-91.
Bailey, N. (2018), “Exploring the Relationship Between Institutional Factors and FDI Attractiveness:
A Meta-Analytic Review”, International Business Review, 27(1), 139-148.
Bénassy-Qu’éré, A., M. Coupet and T. Mayer (2007), “Institutional Determinants of Foreign Direct
Investment”, The World Economy, 30(5), 764-782.
Bissoon, O. (2011), “Can Better Institutions Attract More Foreign Direct Investment (FDI)? Evidence
from Developing Countries”, International Conference on Applied Economics ,1(1) 59-70.
Bitzenis, A. P. and V. P. Žugić (2016), “Barriers for FDI in Serbia: Evidence from the Manufacturing
Sector”, Global Business and Economics Review, 18(2), 227-246.
Blanton, S. L. and R. G. Blanton (2007), “What Attracts Foreign Investors? An Examination of
Human Rights and Foreign Direct Investment”, The Journal of Politics, 69(1), 143-155.
77
Blonigen, B. A. (2005), “A Review of the Empirical Literature on FDI Determinants”, Atlantic
Economic Journal, 33(4), 383-403.
Busse, M., P. Nunnenkamp and M. Spatareanu (2011), “Foreign Direct Investment and Labour Rights:
A Panel Analysis of Bilateral FDI Flows”, Applied Economics Letters, 18, 149-152.
Cheema, G. S. and D. A. Rondinelli (Eds.). (2007), Decentralizing Governance: Emerging Concepts
and Practices, Washington, DC: Brookings Institution Press.
Chiu, E. M. and T. D. Willett (2019), “Capital Controls and Currency Crises Revisited: A Political
Economy Analysis”, Emerging Markets Finance and Trade, 1-21.
Cornia, G. A. (2015), Income Inequality in Latin America: Recent Decline and Prospects for its
Further Reduction (WIDER Working Paper, No. 2015/020), Helsinki: United Nations
University World Institute for Development Economics Research.
Doces, J. A. (2010), “The Dynamics of Democracy and Direct Investment: An Empirical Analysis”,
Polity, 42, 329-351
Dunning, J. H. and S. M. Lundan (2008), Multinational Enterprises and the Global Economy,
Cheltnam, England: Edward Elgar Publishing.
Gangi, Y. A. and R. S. Abdulrazak (2012), “The Impact of Governance on FDI Flows to African
Countries”, World Journal of Entrepreneurship, Management and Sustainable Development, 8,
162-169.
Gani, A. (2007), “Governance and Foreign Direct Investment Links: Evidence from Panel Data
Estimations”, Applied Economics Letters, 14(1), 753-756.
Globerman, S. and D. Sharpiro (2003), “Global Foreign Direct Investment Flows: The Role of
Governance Infrastructure”, World Development, 30(11), 1899-1919.
Globerman, S. and D. Shapiro (2003), “Governance Infrastructure and US Foreign Direct
Investment”, Journal of International Business Studies, 34(1), 19-39.
Goodspeed, T., J. Martinez-Vazquez, and L. Zhang (2011), “Public Policies and FDI Location:
Differences Between Developing and Developed Countries”, Public Finance Analysis, 171-191.
Gordon, L. A., M. P. Loeb and W. Zhu (2012), “The Impact of IFRS Adoption on Foreign Direct
Investment”, Journal of Accounting and Public Policy, 31(4), 374-398.
Govil, M. (2013), “Determinants of Inward FDI in Developing Countries: Evidence from a Time
Series Data Analysis”, Anvesha, 6(4), 22-29.
Guerin, S. S. and S. Manzocchi (2009), “Political Regime and FDI From Advanced to Emerging
Countries”, Review of World Economics, 145(1), 75-91.
Halfani, M. P., P. McCarney and A. Rodriguez (1994), Towards an Understanding of Governance:
The Emergence of An Idea, Toronto: University of Toronto.
Henisz, W. J. (2000), “The Institutional Environment for Multinational Investment”, The Journal of
Law, Economics, and Organization, 16(2), 334-364.
Hur, J., R. A. Parinduri and Y. E. Riyanto (2011), “Cross-Border M&A Inflows and Quality of
Country Governance: Developing Versus Developed Countries”, Pacific Economic Review,
16(5), 633-655.
Jadhav, P. (2012), “Determinants of Foreign Direct Investment in BRICS Economies: Analysis of
Economic, Institutional and Political Factor”, Procedia – Social and Behavioral Sciences, 37, 5-
14.
Jensen, N. M. (2003), “Democratic Governance and Multinational Corporations: Political Regimes
and Inflows of Foreign Direct Investment”, International Organization, 587-616.
Kai, H. and S. Hamori (2009), “Globalization, Financial Depth and Inequality in Sub-Saharan
Africa”, Economics Bulletin, 29(3), 2025-2037.
Kaufmann, D., A. Kraay and M. Mastruzzi (2008), Governance Matters VII: Aggregate and
78
Individual Governance Indicators 1996-2007, New York, NY: World Bank.
Kaufmann, D., A. Kraay and P. Zoido-Lobatón (1999), “Aggregating Governance Indicators (Vol.
2195), New York, NY: World Bank.
Kirk, K., A. Abrahams and P. Ractham (2016), “E-progression of Nonprofit Organization Websites:
US versus Thai Charities”, Journal of Computer Information Systems, 56(3), 244-252.
Kirk, K., P. Ractham and A. Abrahams (2016), “Website Development by Nonprofit Organizations in
An Emerging Market: A Case Study of Thai Websites”, International Journal of Nonprofit and
Voluntary Sector Marketing, 21(3), 195-211.
Kok, R. and A.B. Ersoy (2009), “Analysis of FDI Determinants in Developing Countries”,
International Journal of Social Economics, 36(1/2), 105-123.
Lee, H. H. and H. B. Tan (2006), “Technology Transfer, FDI and Economic Growth in the ASEAN
Region”, Journal of the Asia Pacific Economy, 11(4), 394-410.
Lee, J. Y. and E. Mansfield (1996), “Intellectual Property Protection and US Foreign Direct
Investment”, The Review of Economics and Statistics, 181-186.
Maddala, G. S. (1983), Limited-Dependent and Qualitative Variables in Econometrics, New York,
NY: Cambridge University Press.
Martínez-Zarzoso, I. (2011), “The Log of Gravity Revisited”, Applied Economics, 45(3), 311-327.
Mathur, A. and K. Singh (2013), “Foreign Direct Investment, Corruption and Democracy”, Applied
Economics, 45(8), 991-1002.
McCarney, P. L. (1996), “Considerations on the Notion of Governance. New Directions for Cities in
the Developing World”, Cities and Governance: New Directions in Latin America, Asia and
Africa, 3-20.
Mehta, D. (2000), Urban Governance: Lessons from Asia, Pathumthani, Thailand: Regional Office
for Asia and the Pacific, 317-340.
Mengistu, A. A. and B. K. Adhikary (2011), “Does Good Governance Matter for FDI Inflows?
Evidence from Asian Economies”, Asia Pacific Business Review, 17(3), 281-299.
Mihaylova, S. (2015), “Foreign Direct Investment and Income Inequality in Central and Eastern
Europe”, Theoretical & Applied Economics, 22(2).
Morisset, J. (2002), “Foreign Direct Investment in Africa: Policies also Matter”, New horizons for
Foreign Direct Investment, 548, 167.
Muhammad, A., K. Hashim, H. A. Imran, A. H. Mushtaq and C. M. Irfan (2011), Institutions,
Macroeconomic Policy and Foreign Direct Investment: South Asian Countries Case (MPRA
Paper, No. 32480), Available from https://mpra.ub.uni-muenchen.de/32480/1/Institutions_
macroeconomic_policy_and_foreign_direct_investment-South_Asian_countries_case.pdf.
North, D. (1990), Institutions, Institutional Change and Economic Performance, Cambridge,
England: Cambridge University Press.
North, D., D. Acemoglu, F. Fukuyama and D. Rodrik (2008), Governance, Growth, and Development
Decision-Making (No. 44186), New York, NY: World Bank, 1-44.
Nunnenkamp, P. and J. Spatz (2004), “Intellectual Property Rights and Foreign Direct Investment: A
Disaggregated Analysis”, Review of World Economics, 140(3), 393-414.
Rammal, H. G. and R. Zurbruegg (2006), “The Impact of Regulatory Quality on Intra-Foreign Direct
Investment Flows in the ASEAN Markets”, International Business Review, 15(4), 401-414.
Ranjan, V. and G. Agrawal (2011), “FDI Inflow Determinants in BRIC Countries: A Panel Data
Analysis”, International Business Research, 4(4), 255.
Rodrik, D. (1999), “Where did All the Growth Go? External Shocks, Social Conflict, and Growth
Collapses”, Journal of Economic Growth, 4(4), 385-412.
Rohwer, A. (2009), “Measuring Corruption: A Comparison Between the Transparency International's
79
Corruption Perceptions Index and the World Bank's worldwide Governance Indicators”, CESifo
DICE Report, 7(3), 42-52.
Ross, A. G. (2019), “Governance Infrastructure and FDI Flows in Developing Countries”,
Transnational Corporations Review, 11(2), 109-119.
Rugman, A. M. and A. Verbeke (2000), “Multinational Enterprises and Public Policy”, International
Business, London: Palgrave Macmillan, 21-43.
Samimi, A. J. and F. Ariani (2010), “Governance and FDI in MENA Region”, Australian Journal of
Basic and Applied Sciences, 4(10), 4880-4882.
Sayruamyat, S. and W. Nadee (2020), “Acceptance and Readiness of Thai Farmers Toward Digital
Technology”. In Y. D. Zhang, T. Senjyu, C. So-In and A. Joshi (Eds.), Smart Trends in
Computing and Communications, Springer: Singapore, 75-82.
Schneider, F. and B. S. Frey (1985), “Economic and Political Determinants of Foreign Direct
Investment”, World Development, 13(2), 161-175.
Shah, M. H. and S. Samdani (2015), “Impact of Trade Liberalization on FDI Inflows to D-8
Countries”, Global Management Journal for Academic & Corporate Studies (GMJACS), 5(1),
30-37.
Shivute, P. (2008), “The Rule of Law in Sub-Saharan Africa - An Overview”, Human Rights and the
Rule of Law in Namibia, 213-215.
Spatafora, M. N. and M. O. Luca (2012), Capital Inflows, Financial Development, and Domestic
Investment: Determinants and Inter-Relationships (No. 12-120), Washington, DC: International
Monetary Fund.
Suh, Jae-Hyun (2020), “The Cost of a Sudden Stop during the Global Financial Crisis”, Thammasat
Review, 23(1), 1-13.
Teeramungcalanon, M. (2020), “An Analysis of Consumer Preference Towards Thai Products on
Chinese E-Commerce Platform”, Thammasat Review, 23(2), 1-21.
Teeramungcalanon, M. and E. M. Chiu (2020), “The Effects of Foreign Direct Investment on Income
Inequality of Thailand”, Southeast Asian Journal of Economics, 8(1), 107-138.
Teksoz, S. U. (2006), Corruption and Foreign Direct Investment: An Empirical Analysis, Munich:
Munich Graduate School of Economics.
Tim, Y., L. Yang, S. L. Pan, L. Kaewkitipong and P. Ractham, (2013), The Emergence of Social Media
as Boundary Objects in Crisis Response: A Collective Action Perspective. Available from
https://aisel.aisnet.org/icis2013/proceedings/ResearchInProgress/27/ (accessed October 20, 2020)
Timpone, R. J. (2001), “Concerns with Endogeneity in Statistical Analysis: Modeling the Inter-
dependence between Economic Ties and Conflict”, Economic Interdependence and Interna-
tional Conflict, Ann Arbor, MI: University of Michigan Press, 289-310.
Tuman, J. P. and C. F. Emmert (2004), “The Political Economy of US Foreign Direct Investment in
Latin America: A Reappraisal”, Latin American Research Review, 9-28.
UNDP, A. (1998), UNDP and Governance: Experiences and Lessons Learned, New York, NY:
Author.
UNDP (1997), Governance for Sustainable Human Development, New York, NY: UNDP.
Vijayakumar, N., P. Sridharan and K. C. S. Rao (2010), “Determinants of FDI in BRICS Countries:
A Panel Analysis”, International Journal of Business Science & Applied Management
(IJBSAM), 5(3), 1-13.
Vittal, N. (2001), “Corruption and the State”, Harvard International Review, 23(3), 20.
Wang, M. (2009), “Manufacturing FDI and Economic Growth: Evidence from Asian Economies”,
Applied Economics, 41(8), 991-1002.
Wang, Z. Q. and N. J. Swain (1995), “The Determinants of Foreign Direct Investment in Transforming
80
Economies: Empirical Evidence from Hungary and China”, Weltwirtschaftliches Archiv, 131(2),
359-382.
Wei, S. J. and A. Shleifer (2000), “Local Corruption and Global Capital Flows”, Brookings Papers on
Economic Activity, (2), 303-346.
Wernick, D. A. and J. Haar (2009), “Do Governing Institutions Affect Foreign Direct Investment
Inflows? New Evidence from Emerging Economies”, International Journal of Economics and
Business Research, 1(3), 317-332.
White III, G. O., A. Chizema, A. Canabal and M. J. Perry (2015), “Legal System Uncertainty and FDI
Attraction in Southeast Asia”, International Journal of Emerging Markets, 10(3), 572-597.
Zheng, Y. (2011), “Credibility and Flexibility: Political Institutions, Governance, and Foreign Direct
Investment”, International Interactions, 37(3), 293-319.
https://doi.org/10.35611/jkt.2020.24.8.81 81
Increasing Profitability of the Halal

Cosmetics Industry using Configuration
Modelling based on Indonesian and JKT 24(8)
Malaysian Markets*
Sara Dalir
University of Surrey, Guildford, Surrey, United Kingdom
Hossein GT Olya
Director of Research for Marketing & CCI, Sheffield University Management School, United Kingdom
Amr Al-Ansi
College of Hospitality & Tourism Management, Sejong University, South Korea
Alina Abdul Rahim

Faculty of Science and Technology, Universiti Sains Islam Malaysia, Malaysia
†
Hee-Yul Lee
Food Service Planning & Franchise Management Department, Sejong Cyber University, South Korea
Abstract
Purpose – Based on complexity theory, this study develops a configurational model to predict the
profitability of Halal cosmetics firms in the Indonesian and Malaysian markets. The proposed research
model involves two level configurations—industry context and selling strategies—to predict high and
low scores of a firm’s profitability. The industry context configuration model comprises industry
stability, product homogeneity, price sensitivity, and switching cost. Selling strategies include
customer-focused, competitor-focused, and margin-focused approaches.
Design/methodology – This is the first empirical study that calculates causal models using a combination
of industry context and selling strategy factors to predict profitability. Data obtained from the marketing
managers of cosmetics firms are used to test the proposed configurational model using fuzzy-set
qualitative comparative analysis (fsQCA). It contributes to the current knowledge of business marketing
by identifying the factors necessary to achieve profitability using analysis of condition (ANC).
Findings – The results revealed that unique and distinct models explain the conditions for high and
low profitability in the Indonesian and Malaysian halal cosmetic markets. While customer-focused
selling strategy is necessary to attain a higher profit in both the markets, margin-focused selling
strategy appears to be an essential factor only in Malaysia. Complexity of the interactions of selling
strategies with industry factors and differences between across two study markets confirmed that
complexity theory can support the research configurational model. The theoretical and practical
implications are also illustrated.
Originality/value – Despite the rapid growth of the global halal industry, there is little knowledge
about the halal cosmetic market. This study contributes to the current literature of the halal market
by performing a set of asymmetric analytical approaches using a complex theoretical model. It also
deepens our understating of how the Korean firms can approach the Muslim consumer’s needs to
generate more beneficial turnover/revenue.
Keywords: Industry Stability, Product Homogeneity, Price Sensitivity, Switching Cost, Selling
Strategy, Profitability
* This work was supported by the Ministry of Education of the Republic of Korea and the National
Research Foundation of Korea (NRF-2017S1A22039266).
†
Corresponding author: hylee02@sjcu.ac.kr
82
1. Introduction
In recent years, consumers have been spending a high proportion of their disposable
income on cosmetics, mainly because of growing online sales, easy access to the internet and
the undeniable impact of beauty bloggers via social media channels. The growth rate of the
cosmetics industry was estimated to touch 13.67% in 2019 compared with the 2014
performance (Ray, 2017). In the last few years, global cosmetic firms have continued to focus
their efforts on product innovation in order to attract potential consumers with an appetite
for new, unique, superior, and premium products, as well as to keep existing consumers loyal
to specific brands (L’Oréal, 2018).
The quality and authenticity of cosmetic products as important drivers of sustainable
profitability required to be considered in decision making and manufacturing stages by firms
(Kaul and Luo, 2018). This might vary in different target groups of customers. For instance,
using animal ingredients (e.g., pork and lard) is a serious issue in the markets where religion
(e.g., Islam and Judaism) is emphasized. Muslim consumers are only permitted to use halal
products and services. “Halal” is an Arabic word which means “permissible” or “lawful” and
“Haram,” the opposite of halal, means forbidden under Islamic law (Shah Alam and Sayuti,
2011; Wilson and Liu, 2010). Halal and haram have been applied to all aspects of the Islamic
way of life, ranging from food products, cosmetics and pharmaceuticals, leather products to
even entertainment (Abd Rahman, Asrarhaghighi and Ab Rahman, 2015). In addition to the
above-mentioned products and services, it is advised that accompanying cosmetic accessories
such as brushes, bottles, containers, packaging, and mirrors must also be produced and
delivered according to the Islamic law, called Sharia.
The number of studies on modelling a firm’s profitability has increased in recent years (e.g.,
Hirsch and Schiefer, 2016; Lee and Alnahedh, 2016; Hongwei et al., 2017; Islam and Khan,
2019). Some of them focused on cultural/religion-oriented markets from consumer
perspective (Aisyah, 2017; Mustafar et al., 2018; Nawawi, Roslin and Hamid, 2018; Shahid,
Ahmed and Hasan, 2018; Sriminarti and Nora, 2018; Putri, Daryanti and Ningtias, 2019).
However, there is a paucity of research attempts modelling the profitability of halal markets
from firm and industry perspectives. Drawing on the existing firm–industry interaction
effect, this study aims to fill the research gap by modelling the profitability of halal cosmetics
firms in the two key halal markets of Malaysia and Indonesia. This study responds to
Leischnig and Kasper-Brauer’s (2016) study that called for modelling firms profitability in a
specific industry. A two-level configurations of industry factors (i.e., industry stability,
product homogeneity, price sensitivity, and switching cost) and firm factors (i.e., customer-
focused selling, competitor-focused selling, and margin-focused selling) is used to predict the
firms’ profitability in halal cosmetics industry.
This empirical study attempts to address the following research questions. First, how to
increase the profitability of halal cosmetics firms based on industry conditions? Second, what
patterns of pricing strategies contribute to the profitability of halal cosmetics firms? Third,
what firm and industry conditions are necessary to increase profitability? Fourth, how do the
causal models for stimulating the high profitability of Malaysian halal cosmetics firms differ
from models for the profitability of Indonesian firms? Using the data obtained from the sales
managers of halal cosmetics firms in Malaysia and Indonesia, the proposed configurational
model is tested by fuzzy set Qualitative Comparative Analysis (fsQCA) and Analysis of
Necessary Condition (ANC).
To the best of the authors’ knowledge, this is the first empirical study that investigates the
necessity for predictors of profitability. These case-oriented analytical approaches are
considered as relatively new and pragmatic techniques in management studies, which help to
Increasing Profitability of the Halal Cosmetics Industry using Configuration Modelling
based on Indonesian and Malaysian Markets
83
generate new knowledge through exploring causal algorithms leading to an expected
outcome (e.g., profitability) (Woodside, 2014; Huang and Huarng, 2015; Leischnig and
Kasper-Brauer, 2016; Stroe, Parida and Wincent, 2018; Tho, 2018; Franklin and Marshall,
2019).
2. Theoretical Background
2.1. Halal Cosmetics
Generally, there are rigorous regulations that oblige cosmetics firms to produce safe
cosmetics for consumers and service providers such as hairdressers and beauticians, prior to
marketing as the products may contain various chemical ingredients (The EU Cosmetics
Regulation, 2009; US Food and Drug Administration, 2018). Halal cosmetics products are
not exempt from these worldwide regulations, and a cosmetic product is considered halal if
it is deemed safe (DSM, 2008). Detecting halal and non-halal ingredients is essential to verify
cosmetics and personal care products as halal. Maintaining halal is crucial for some
consumers as they may perceive psychological and health risks while consuming non-halal
products and services (Olya and Al-Ansi, 2018). Halal cosmetics must be free from any of the
following ingredients:
Ingredients derived from any part of non-halal animals, such as dog and pig
including lard, gelatin, collagen, glycerin, and allantoin. It is permitted to use the
gelatin and collagen of halal animals, such as cows or sheep.
Any harmful ingredients to consumers, such as mercury, lead, and hydroquinone.
Any ingredients from the human body, such as human stem cells, urine, placenta,
blood, amniotic fluid, vomit, and pus.
2.2. Religion-oriented Markets

A huge upturn in Islamic business marketing has been observed in the last few years (GIER,
2018). This has encouraged scholars to study halal markets of different industries, including
food, fashion and art, cosmetics and pharmaceuticals, finance, supply chain management,
logistics services, marketing, and hospitality and tourism (Abd Rahman et al., 2015; Ahmed,
2008; GIER, 2018; Han et al., 2019; Hassim, 2014; Jamal and Sharifuddin, 2015; Musa, 2014;
Olya and Al-Ansi, 2018; Wilson, 2012). Halal as a cultural element, presents an opportunity
to gain a competitive advantage in emerging source markets (Olya and Al-Ansi, 2018). Halal
cosmetics branding is identified as an opportunity for international industry players to
compete on a wider scale (Aoun and Tournois, 2015).
Customer awareness of non-food halal products such as cosmetics has been increasing,
especially among Muslim customers who are living in multi-religious societies. Considerable
population of Muslims worldwide is willing to pay premium prices for halal certified beauty
products. According to the State of the Global Islamic Economy (SGIE) report 2019/20,
Muslims’ global spending on cosmetics was estimated at USD 64 billion in 2018, which
constituted 8% of the global spend on cosmetics, and is forecasted to reach USD 95 billion by
2024. Briliana and Mursito (2017) found that attitude and religiosity affect the intention of
the Indonesian Muslim youth to purchase halal cosmetics products. Halal cosmetics have all
the attributes to fulfil a broader social role by addressing the growing global consumer
demand for ethical, vegan, and eco-friendly products. Hence, halal cosmetics is evidently a
potentially lucrative market with numerous opportunities for growth.
84
India, Indonesia, Malaysia, and Bangladesh are among the top 10 spenders of halal beauty
products. The total value of cosmetics exported from China to the Organization of Islamic
Cooperation (OIC) countries amounted to USD 0.8 billion in 2018 (SGIE, 2020). According
to the Mintel Global New Product Database, 73% of halal cosmetics products sold around the
world between 2014 and 2016 were from the Asia-Pacific region (Spencer, 2017). The
investment of manufacturers in halal-certified products has been increasing due to the
perceived high profitability of halal cosmetics market. Many cosmetics firms from around the
world strive to obtain halal certification for their products which is a prerequisite to penetrate
the OIC cosmetics markets. There are collaborations between different countries in the Asia-
Pacific region that apply innovative strategies for production and marketing of halal
cosmetics. Within this region, Southeast Asia as an emerging in the halal cosmetics industry
accounted for 61.2% of the total halal cosmetics market in Asia (USD 1.4 billion) in 2015
(Future Market Insight, 2015).
The expansion of halal industry is critical for economic growth and development of
Malaysia and Indonesia. Malaysia is listed by Global Islamic Economy Report (GIER) (2018)
as the leading country with highest economic growth in terms of halal industry, that is mainly
triggered by strong global trends and policies implemented by the Malaysian authorities over
the past decade. From different segments of halal industry, cosmetics manufacturing has
flourished rapidly due increase in demand and it is estimated that the total halal cosmetics
expenditure reaches to USD 3.1 billion annually (GIER, 2018). The Halal Industry
Development Corporation (HDC) and the Department of Islamic Advancement of Malaysia
were major contributors in enhancing the performance quality and stability of the halal
industry, locally and globally. The Malaysia International Halal Showcase (MIHAS), which is
an annual four-day trade exhibition held in Kuala Lumpur, hosted by the Ministry of
International Trade and Industry created a bridge between local and international businesses
to identify further opportunities in different sectors of the halal industry including cosmetics.
It creates a valuable platform for new ventures by cosmetic companies from developed
countries such as South Korea to target this market. For instance, about 30,000 traders and
visitors from 88 countries gathered in the MIHAS in 2019 that resulted in a total trade value
of over USD 380 million (MIHAS, 2020).
Similar to the Malaysian market, the Indonesian halal cosmetics market records a high
performance in the halal industry due to its large Muslim population of 229 million, which
accounts for about 13% of the world’s Muslim population (World Population Review, 2020).
The Indonesian cosmetics market was ranked the second among Muslim countries with
highest halal cosmetics expenditure of USD 4 billion per year (GIER, 2018). Both public and
private sectors in Indonesia aim to accelerate the market growth to achieve a high level of
business value. The market can benefit significantly from the economic standing position of
being among the global top 20 largest economies, according to the Indonesia halal lifestyle
center (IHESR, 2018). In this regard, Indonesian government also tried to improve the
reachability of the current halal standards to more enterprises worldwide by establishing the
Halal Product Assurance Agency (Jefriando and Suroyo, 2019).
There are numerous international beauty brands and foreign cosmetics companies that
have been trying to penetrate the Malaysian and Indonesian halal markets from France, the
United States, China, Taiwan, Japan, South Korea, Singapore, and Thailand. The
representatives of the brands usually attend trade shows, business events, workshops, and
exhibitions to evaluate the market demand and understand the assessment process of halal
standards (IHESR, 2018). Specifically, South Korea as a leading global player in the cosmetics
industry with USD 6.26 billion value, has been consulting with local cosmetics distributors in
Malaysia and Indonesia to understand the needs of the halal market (Korea Times, 2016).
85
The number of halal certified manufacturers in South Korea is constantly increasing. For
instance, recently 11 manufacturers including Kolmar Korea and Cos Nine have allocated
several production lines to halal certified cosmetics for both local and international markets.
Moreover, the South Korean government has signed an agreement with the HDC to produce
halal cosmetics in Malaysia. To support halal cosmetics exports, the South Korean Ministry
of Food and Drug Safety is also providing free training sessions on halal cosmetics
certification (SGIE, 2020). Compliance with halal standards, understanding the target market,
particularly certain selling strategies are vital steps in ensuring sustainable profitability of halal
cosmetic firms.
2.3. Industry Configuration

An accurate assessment of an industry conditions in terms of stability, product
homogeneity, price sensitivity and switching cost guides a company to model its market entry
decision. Drawing on Leischnig and Kasper-Brauer’s (2016) study, industry stability, price
sensitivity, product homogeneity, and switching cost are four key determinant factors of an
industry’s position. Industry stability denotes the level of changes in a market’s products,
customer preferences and features as well as the market’s responses to the changes (Reimann,
Schilke and Thomas, 2010a). Usually, an unstable industry is associated with high product-
innovation rates (Reimann et al., 2010b), that provides an opportunity to sell premium
products with higher price (Leischnig and Kasper-Brauer, 2016). Product homogeneity also
indicates customer perception towards the substitutability level of products from different
firms within an industry (Homburg, Klarmann and Schmitt, 2010). Since similar technology
has been typically used in the production of homogenous products in a market (Li, Lu and
Guo, 2017; Weiss and Heide, 1993), customer intimacy (Crittenden & Crittenden, 2014;
Mulia et al., 2020) and customer relationship management (Foltean et al., 2019; Reimann et
al., 2010a) are vital strategies to achieve profitability in a market with highly homogenous
product.
Normally, low switching cost and lack of clear differentiation of a product from other
competitor’s products result in high price sensitivity (McGahan and Porter, 1997; Vastani
and Monroe, 2019). Price sensitivity refers to the degree of variation in customer demand for
a product with respect to price changes (Chiu et al., 2019; Tellis, 1988), and switching cost is
defined as customers’ cost of switching from one producer to another (Burnham, et al., 2003;
Cheng and Lee, 2020). Low switching cost may lead to less profit in the long run where firms
need to strengthen their relationship with customers and highlight the distinct features of
their products in comparison to similar products available in the market (Christino et al.,
2020; Reimann et al., 2010a).
2.4. Selling Strategies

As an important component of business strategies, undertaking a suitable selling strategy
is essential to penetrate the potential markets. An effective selling strategy may lead to higher
efficiency, and then generating value and profit. The adopted three types of selling strategy
factors at firm level include customer-focused selling, competitor-focused selling, and
margin-focused selling (Leischnig and Kasper-Brauer, 2016; Terho et al., 2015). These
strategies focus on customer needs, rivals in a market and their strengths and weaknesses, as
well as profit margins in sales activities, respectively. Customer-focused selling strategy offers
mutual interaction between a firm and a customer. A firm’s salesforce can extend business
relationships by targeting a segmented market and retaining customers through validating
their preferences and priorities (Hartmann et al., 2020; Pekovic and Rolland, 2016).
86
Customer-focused selling approach and segmentation are also identified as critical
components of strategic sales planning in business-to-business trade (Mangus and Ruvio,
2019).
Competitor-focused selling strategy refers to understanding the present condition of
market players in terms of market size, customer profile, and business marketing philosophy
(Slater, Olson and Eibe Sørensen, 2012). A firm’s decision on whether to enter a new market
could be affected by competitors practice, customer orientation and market value. Identifying
competitors’ goals, strategies, resources, and capabilities enables firms to assess their own
potential to explore a new market (Goh et al., 2019; Kim, Min and Chaiy, 2015). Margin-
focused selling strategy refers to net margin profits obtained from sales practices and business
activities. Firms sales margin is considered as an effective approach to boost firm’s
profitability and increase perceived value by customers (Benoit et al., 2020). This strategy
would be particularly useful when a firm is targeting new markets, such as halal market (Jamal
and Sharifuddin, 2015). To this end, we applied the selling strategies as predictors of
profitability of halal cosmetics firms in Malaysia and Indonesia.
2.5. Profitability
Profitability is a crucial element that measures the firm’s current performance and
competitive edge, which contributes in market expansion and productivity. Ball et al. (2015)
defined profitability as “earnings,” and Novy-Marx (2013) stated that profitability is a
significant measurement indicator of returns to a firm and leads the business value strategy.
The selling strategy approaches are developed to guide a firm in achieving the most beneficial
outcome in terms of profitability. In this study, profitability is interpreted as the gross profit,
returns, and earnings of the halal cosmetics industry in Malaysia and Indonesia.
A review of literature shows that predicting profitability of firms is complex considering
the influence of internal (e.g. internal marketing strategies) and external (e.g. industry
characteristics) factors (Čirjevskis, 2020; Erimia, 2018). Furthermore, the concept of halal is
complex as there are diverse religiosity laws in Islamic countries (Olya and Al-Ansi, 2018).
Complexity theory is frequently used to explain complex combinations of independent
variables in predicting the desired outcome within dynamic and uncertain environments
(Han et al., 2019; Olya, Altinay and De Vita, 2018; Wu et al., 2014). With this realization, this
study applies complexity theory that describes complex interactions of the independent
factors in predicting the profitability of halal cosmetic firms. Specifically, this study combines
three selling strategies of firms with four industry factors to predict profitability of halal
cosmetics in complex and dynamic markets of Malaysia and Indonesia.
2.6. Configurational Model

The proposed research model involves two configurations of industry factors and selling
strategies within the halal cosmetics industry context (Fig. 1). Industry configuration
comprises industry stability, product homogeneity, price sensitivity, and switching cost.
Selling strategies include customer-focused, competitor-focused, and margin-focused selling
approaches. These two configurations are used as predictors of conditions leading to high
and low scores of profitability. Using fsQCA, causal models leading to a low scored outcome
(profitability) are predicted, which are not necessarily a mirror of those for a high scored
outcome. Necessity of industry conditions and selling strategies are also identified using
ANC.
87
Fig. 1. Proposed Configurational Model
3. Method
3.1. Measurement and Data
This study used questionnaire to measure variables of industry configuration, selling
strategy and profitability. The questionnaire consisted of two sections: the first part was
dedicated to the background and purpose of the survey, and the second part included scales
related to industry factors and selling strategies. The scale items were extracted from validated
variables used by Leischnig and Kasper-Brauer (2016), which are represented in Table 1. The
responses to questions were measured using 7-point Likert scales, ranging from 1 (extremely
disagree) to 7 (extremely agree). At the end of questionnaire, four open-ended questions were
asked to capture views of the marketing managers on most popular halal cosmetic brands,
influential marketing communication tactics, distribution channels, and key contemporary
challenges of the cosmetic market. The questionnaire was initially developed in English, and
thereafter back translated to Bahasa Melayu and Bahasa Indonesia. The translated survey
questionnaire was reviewed and verified by three academic professionals who are native
speakers and actively work in the field of marketing and in the halal industry. A pilot study
was also conducted with 11 members of sales and marketing teams of Malaysian companies
to ensure the clarity and comprehensibility of the survey questions. Based on the pilot study
results, no revisions were required, as all questionnaire items were clear.
Data were obtained from the managers of sales and marketing departments of halal
cosmetics firms in Malaysia and Indonesia. Of these, 13% were at age between 20 and 30, 37%
were 31–40, 41% were 41-50, and 9% were over 50 years old. In terms of gender of
respondents, 66% were male and 34% female. Regarding education, 28% held a college degree,
59% held a postgraduate degree, and 13% had basic education. The majority of them had 5-
10 years of work experience (44.6%), 37.4% had over 10-year, and 18.7% and less than 5 years’
88
work experience. As the survey was conducted face-to-face, the response rate was relatively
high (73%). From 93 invited managers, 68 of them (39 respondents from Malaysia and 29
respondents from Indonesia) responded to all questionnaire items. Two filter questions were
asked prior to inviting the directors’ participation in the survey to ensure that participants
met the required criteria for the subsequent questions in the survey: “Do you produce halal
cosmetics?” and “Is your company certified by a Halal Authority?”
3.2. Analytical Approaches

Initially, a screening and scanning process was applied to select valid cases for data analyses.
Two principal steps were utilized to investigate the data analyses. First, the reliability and
validity of the measures were assessed using a set of measurement modelling metrics (Hair et
al., 2016). Second, configurational modelling using fsQCA 2.5 software was used to test the
proposed configurational model for both the Indonesian and Malaysian datasets. The
sufficient causal combination of the antecedents (i.e., industry configuration and selling
strategy configuration) were explored to measure the desired outcomes (i.e., profitability).
The fsQCA is a powerful analytical approach for small to medium-sized datasets that can
model complex phenomena for marketing (Olya and Altinay, 2016; Zaefarian et al., 2017) in
general and halal marketing in specific (Han et al., 2019; Olya and Al-Ansi, 2018).
The fsQCA uses Boolean algebra to explore sufficient and consistent conditions leading to
the expected outcome and includes three steps of calibration, truth table algorithms, and
counterfactual analysis. In calibration, crisp data are transformed to fuzzy data, and all
possible models describing the conditions for stimulating the outcome appear as truth tables,
which are refined based on two probabilistic metrics. Sufficient and consistent models are
selected in the counterfactual section. The sufficiency and consistency of models are
measured using coverage (Formula 1) and consistency (Formula 2), which are analogous to
R2 and correlation in symmetrical analysis, respectively (Ragin, 2009; Wu et al., 2014).
Coverage:௜ ௜ ∑
, ⁄∑ (1)
Consistency: ௜ ௜ ∑
, ⁄∑ (2)
where Xi is case i’s membership score in the predictor set (X) and Yi is case i’s membership
score in the expected outcome (Y: profitability) (Ragin, 2009). In addition, ANC was used to
identify the necessary factors for high profitability.
4. Results
4.1. Measurement Model Testing
A confirmatory factor analysis (CFA) using SmartPLS was conducted to check the
psychometric properties of the measures. Table 1 depicts the standardized factor loadings and
descriptive statistics of the scale items. According to the CFA results, all items were
sufficiently and significantly loaded under the assigned construct (loading value >.5, p<.001).
The magnitude of kurtosis and skewness values falls within the recommended range (±3),
which indicates normality of the data. Harman’s single-factor test was performed to check for
common method variance (CMV). According to the results, no major factor accounted for
the largest variance among all components below 20%, which means CMV was not a serious
threat in this study (Podsakoff et al., 2003).
89
Table 1. Results of Descriptive Statistics and Factor Analysis
Scale items SLV Mean SD Kurtosis Skewness
Industry stability
There are no frequent changes of targeted .837*** 4.250 1.612 -.261 -.117
segments in halal cosmetics market.
There are no frequent changes in halal .786*** 3.716 1.543 -.186 .242
cosmetics products.
Change in the halal cosmetics industry is .694*** 3.207 1.730 -.348 .632
slow.
Fluctuations in the halal cosmetics .692*** 4.118 1.334 .308 -.183
industry are predictable.
Obsolescence of halal products .664*** 3.912 1.502 -.299 -.325
marketing is slow.
Product homogeneity
Most halal products have no difference .783*** 4.090 1.484 -.178 -.242
from competing halal cosmetics
companies.
Halal products are highly standardised. .757*** 5.212 1.430 -.229 -.575
Homogeneity of halal products is high. .611*** 5.000 1.467 .710 -.737
Many halal products are identical in .723*** 4.397 1.903 -.880 -.329
quality.
Price sensitivity
Customers check prices for value for .788*** 5.265 1.511 -.427 -.517
money for halal cosmetics products.
Customers buy the lowest price halal .683*** 4.897 1.601 .168 -.686
cosmetics products that suit their needs.
Customers rely heavily on price when .668*** 4.765 1.775 -.556 -.425
they need to choose halal cosmetics
products.
Switching cost
Customer costs in switching to another .857*** 4.147 1.364 .616 -.131
cosmetic product (switching cost: cost of
purchasing product from another brand)
are high.
Selling halal products would not be easy .828*** 3.956 1.528 -.210 .278
to middle and low-income class
customers.
The process of switching to new halal .751*** 4.388 1.445 -.203 .138
cosmetics products is not quick and easy
for the customers.
Switching to another cosmetic product .585*** 4.791 1.635 -.391 -.430
does bear general risks for the customers.
Profitability
Return on sales in the firm in which I .928*** 5.176 1.271 .284 -.297
work is high.
My firm reaches planned financial goals. .569*** 5.426 1.240 -1.195 -.064
Our business is profitable. .682*** 5.382 1.295 -.924 -.212
Notes: 1. SLV: standardized loading value; SD: Standard Deviation.
2. ***p<.001.
90
The results from reliability and validity tests are presented in table 2. The three measures
of Cronbach’s Alpha (α), rho_A, and composite reliability were used to check internal
consistency among the scale items. The values of the three measures were above the
recommended level of .7, indicating the five constructs were reliable. The average variance
extracted for all factors was above .5, proving the convergent validity of study measures. In
terms of discriminant validity, the heterotrait-monotrait ratio of correlation values were
smaller than the commonly accepted level of .85 (Hair et al., 2016; Henseler, Hubona and
Ray, 2016).
Table 2. Results of Reliability and Validity Tests

Convergent Discriminant validity
Reliability
Constructs validity (HTMT)
α rho_A CR AVE 1 2 3 4
1 Industry stability .713 .773 .721 .544
2 Price sensitivity .767 .763 .758 .512 .325
3 Product homogeneity .757 .769 .744 .521 .630 .462
4 Profitability .789 .823 .779 .550 .465 .201 .347
5 Switching cost .734 .744 .739 .583 .788 .545 .530 .533
Note: α: Cronbach’s Alpha, CR: composite reliability, AVE: average variance extracted, HTMT:
heterotrait-monotrait ratio of correlation, Fit statistics: Standardized Root Mean Square
Residual (SRMR) = .013 (below .08: recommended level).
4.2. Configurational Model Testing

Table 3 illustrates the fsQCA results for industry factors predicting high and low scores of
the halal cosmetics firms’ profitability in Indonesian and Malaysian markets. Two causal
models emerged for each country (i.e., Indonesia coverage: .685, consistency: .990, and
Malaysia coverage: .660, consistency: .981). Ragin (2008) recommended .2 and .8 as
thresholds for coverage and consistency, respectively. In Indonesia, recipe 1 indicates that the
combination of low industry stability, high product homogeneity, and high switching cost
leads to high profitability of halal cosmetics companies. While in Malaysia, low industry
stability, high product homogeneity, and high price sensitivity result in high profitability.
According to the recipe 2, high product homogeneity, high price sensitivity, and high
switching cost lead to high profitability of the firms in both countries.
As shown in Table 3, one recipe emerged that predicts a low score of firm profitability (i.e.,
Indonesia coverage: .593, consistency: .776, and Malaysia coverage: .683, consistency: .751).
The recipe shows that the combination of low industry stability, low product homogeneity,
low price sensitivity, and high switching cost leads to low profitability of the halal firms in
Indonesia. Whereas, low scores of industry stability, price sensitivity, and switching cost
result in low profitability of the Malaysian firms. The XY plots of recipe 1 and recipe 2 for
high profitability in Indonesia and Malaysia are also depicted in Table 3, which indicates the
asymmetric relationship between the models and outcome of profitability.
The fsQCA results of testing selling strategies to predict the high and low scores of firms’
profitability are shown in Table 4. The table presents causal models, which emerged for
achieving high profit in the both countries. In Indonesia, a combination of customer-focused,
competitor-focused, and margin-focused selling strategies leads to a high profitability of halal
cosmetics firms (Indonesia coverage: .723, consistency: .972). In Malaysia, two models lead
to high profitability (coverage: .514, consistency: .927). The first recipe indicates that the firms
Table 3. Results from Industry Context Configuration for High and Low Profitability of Halal Cosmetics Companies in Indonesia (left) and Malaysia (right)
Indonesia RC UC Con Malaysia RC UC Con
prf = f(ins, hem, psen, swct) prf = f(ins, hem, psen, swct)
R1: ~ins*hem*swct .619 .048 .995 R1: ~ins*hem*psen .467 .070 .988
R2: hem*psen*swct .636 .065 .989 R2: hem*psen*swct .589 .192 .979
solution coverage: .685 solution coverage: .660
solution consistency:.990 solution consistency:.981
~prf = f(ins, hem, psen, swct) ~prf = f(ins, hem, psen, swct)
R1:~ins*~hem*~psen*swct .593 .593 .776 R1: ~ins*~psen*~swct .683 .683 .751
solution consistency:.776 solution consistency:.751
Note: RC: raw coverage, UC: unique coverage, Con: consistency, ~: negation, prf: profitability, ins: industry stability, hem: product homogeneity, psen: price sensitivity, swct: switching cost.
91
92
that stress more on a margin-focused selling strategy than a competitor-focused strategy, may
generate higher profit margins. In the second recipe, more emphasis was placed on a margin-
focused selling strategy than customer needs to attain more profit. According to the causal
models shown in Table 4 for low profit in Indonesia (coverage: .705, consistency: .701) and
Malaysia (coverage: .625, consistency: .864), Indonesian firms that focus less on competitors
and margin-focused selling strategies and more on customer-focused selling strategy gain low
profit scores. Furthermore, low profit in Malaysian firms result from less focus on customer-
focused and competitor-focused selling strategies and more on a margin-focused selling
approach.
Table 4. Results from Selling Strategy Sonfigurations for High and Low Profitability of Halal
Cosmetics Companies in Indonesia (left) and Malaysia (right)
Indonesia RC UC Con Malaysia RC UC Con
Recipes for high profitability: prf = f(cfs, comp, mfs) prf = f(cfs, comp, mfs)
R1: cfs*comp*mfs .723 .723 .972 R1: ~comp*mfs .448 .207 .938
solution coverage: .723 R2: ~cfs*mfs .307 .066 .928
solution consistency:.972 solution coverage: .514
solution consistency: .927
Recipes for low profitability: ~prf = f(cfs, comp, mfs) ~prf = f(cfs, comp, mfs)
R1: cfs*~comp*~mfs .705 .705 .701 R1: ~cfs*~comp*mfs .625 .625 .864
solution consistency:.701 solution consistency: .864
Note: RC: raw coverage, UC: unique coverage, Con: consistency, ~: negation prf: profitability, cfs:
customer-focused selling approach, comp: competitor-focused selling approach, mfs: margin-
focused selling approach.
Table 5 illustrates how a combination of industry context and selling strategy

configurations could lead to high profitability of halal cosmetics firms in Indonesia
(coverage: .566, consistency: .994) and Malaysia (coverage: .477, consistency: .995). A
combination of low industry stability, high product homogeneity, high switching cost, and
applying the three selling strategies by the Indonesian firms (i.e., customer-focused,
competitor-focused, and margin-focused strategies) result in high profitability. The firms that
are experiencing high industry stability, product homogeneity, and switching cost in Malaysia
are most likely to secure higher profit if they apply the all selling strategies (Table 5).
Table 5. Results from a Combination of Industry Context and Selling Strategy Configurations
for High Profitability of Halal Cosmetics Companies in Indonesia (left) and Malaysia
(right)
Indonesia Raw coverage Unique coverage Consistency
prf = f(ins, hem, psen, swct, cfs, comp, mfs)
R1:~ins*hem*swct*cfs*comp*mfs .566 .566 .994
solution coverage: .566
Malaysia Raw coverage Unique coverage Consistency
prf = f(ins, hem, psen, swct, cfs, comp, mfs)
R1:ins*hem*psen*swct*cfs*comp*mfs .477 .477 .995
Note: ~: negation, prf: profitability, ins: industry stability, hem: product homogeneity, psen: price
sensitivity, swct: switching cost, cfs: customer-focused selling approach, comp: competitor-
focused selling approach, mfs: margin-focused selling approach.
93
The recipes resulting in low profitability, based on a combination of industry and selling
strategy configurations, are presented in Table 6. According to the fsQCA results, the
Indonesian firms that work in an industry with low stability, low price sensitivity, high
product homogeneity, and high switching cost will benefit less if they apply all three selling
strategies. Whilst, low industry stability, product homogeneity, price sensitivity, and
switching cost, with lesser emphasis on a customer-focused selling approach than
competitor-focused and margin-focused selling strategies, lead to low profit in Malaysia
(Table 6).
Table 6. Results from Combination of Industry Context and Selling Strategy Configurations
for Low Profitability of Halal Cosmetics Companies in Indonesia (left) and
Malaysia (right)
Indonesia Raw coverage Unique coverage Consistency
~ prf = f(ins, hem, psen, swct, cfs, comp, mfs)
R1:~ins*hem*~psen*swct*cfs*comp*mfs .654 .654 .661
solution consistency:.661
Malaysia Raw coverage Unique coverage Consistency
~ prf = f(ins, hem, psen, swct, cfs, comp, mfs)
R1:~ins*~hem*~psen*~swct*~cfs*comp*mfs .495 .495 .870
Note: RC: raw coverage, UC: unique coverage, Con: consistency, ~: negation, prf: profitability, ins:
industry stability, hem: product homogeneity, psen: price sensitivity, swct: switching cost, cfs:
customer-focused selling approach, comp: competitor-focused selling approach, mfs: margin-
focused selling approach.
Finally, the results of the ANC are shown in Table 7. A consistency value above .9 represents
the condition necessary to attain the expected outcome (Feng et al., 2019). A customer-
focused selling strategy appears as a necessary predictor of the profitability of halal cosmetics
firms in both Indonesia and Malaysia. Employing a margin-focused selling strategy in
Malaysia’s halal cosmetics industry is necessary to gain high profit whereas it is not necessary
in the Indonesian market.
Table 7. Results of the Necessary Condition Analysis

Outcome: Profitability
Necessary predictor Indonesia Malaysia
Consistency Coverage Consistency Coverage
Industry stability .538 .982 .685 .951
~Industry stability .735 .961 .575 .913
Product homogeneity .826 .930 .780 .947
~ Product homogeneity .420 .992 .485 .923
Price sensitivity .766 .891 .824 .899
~ Price sensitivity .445 .983 .410 .948
Switching cost .760 .954 .704 .963
~Switching cost .515 1.000 .576 .933
Customer-focused .910 .894 .903 .891
selling approach
~Customer-focused .294 1.000 .307 .914
selling approach
94
Table 7. Results of the Necessary Condition Analysis
Outcome: Profitability
Necessary predictor Indonesia Malaysia
Consistency Coverage Consistency Coverage
Competitor-focused .815 .934 .816 .960
selling approach
~Competitor-focused .428 .972 .462 .926
selling approach
Margin-focused selling .854 .935 .913 .913
approach
~Margin-focused .398 1.000 .344 .987
selling approach
Note: bold value represents necessary factor (consistency >.9).
5. Discussions and Conclusion

5.1. Theoretical Implications
This study contributes to the current knowledge of industrial marketing, by developing and
testing configurational models to predict the halal cosmetic firms’ profitability in Indonesia
and Malaysia. The halal cosmetics industry is a flourishing and lucrative market that attracts
a wide range of brands with different sizes of operation and equity. This study helps
Indonesian and Malaysian halal cosmetics firms to understand how to adapt with the industry
conditions and develop effective selling strategies to secure profitability. The fsQCA
calculated sufficient and consistent algorithms (i.e., recipes) for low profitability, which is
different from the opposite mirror of algorithms for high profitability. This empirical study
is also among the first research to identify the necessary conditions for firms’ profitability.
The different strategies to achieve high profitability are explored in the context of
Malaysian and Indonesian markets by adopting two configurations. The first configuration
encompassed industry factors (i.e., industry stability, product homogeneity, price sensitivity,
and switching cost), which revealed slightly different conditions in Malaysia and Indonesia.
Attaining high profitability in the Malaysian market entails low industry stability, high
product homogeneity, and high price sensitivity while the Indonesian market involves low
industry stability, high product homogeneity, and high switching cost. This indicates that in
the both markets high product homogeneity, high price sensitivity, and high switching cost
lead to high profit for firms. This recipe directs national and international firms (e.g., South
Korean companies) that can increase their profits by following the recommended recipes. On
the other hand, low industry stability, low product homogeneity, and low price sensitivity,
along with high switching cost, lead to low profitability in the Indonesian market. In Malaysia,
low industry stability, low price sensitivity, and low switching cost result in low profitability.
This shows that low product homogeneity would not lead to low profitability in the Malaysian
market.
The results of the second configuration models indicated that the recipes of selling
strategies leading to both high and low scores of firms’ profitability are different in Malaysia
and Indonesia. Therefore, halal cosmetics firms ought to adjust their selling strategies to avoid
conditions for low profitability. In Indonesia, high profits of halal cosmetics firms result from
applying customer-focused selling, competitor-focused selling, and margin-focused selling
strategies. However, the Malaysian market offers two different recipes in terms of selling
strategies, to earn high profits. To achieve this, firms can either apply margin-focused selling
strategy with less emphasis on competitor-focused selling strategy or use margin-focused
95
selling strategy with less emphasis on customer-focused selling strategy. In contrast, more
emphasis on customer-focused selling strategy than competitor-focused and margin-focused
selling strategies result in low profitability in the Indonesian market. In the Malaysian market,
the conditions are different, as less emphasis on customer-focused and competitor-focused
selling strategies with high focus on margin-focused selling strategy cause low profitability.
The findings obtained from a combination of industry and selling strategy configurations
confirm the complexity of the industry as different recipes explain the conditions to attain
high and low profit. In Malaysia, the combination of high industry stability, product
homogeneity, price sensitivity, and switching cost along with the application of the three
selling strategies (customer-focused selling, competitor-focused selling, and margin-focused
selling) lead to high profitability. On the contrary, low industry stability, product
homogeneity, and switching cost together with the application of three selling strategies
describe a condition leading to a high level of profitability in Indonesia. Industry instability
in Indonesia, unlike Malaysia, can contribute to higher profitability. Therefore, this is an
opportunity for South Korean start-ups and SMEs to generate more profit by penetrating the
Indonesian market.
This empirical study also extends the current knowledge of firm profitability by identifying
the necessary conditions for two different markets. According to the ANC results, a
customer-focused selling strategy is a necessary predictor of profitability in both Malaysia and
Indonesia. Thus, firms need to conduct an in-depth contextual analysis to understand the
preferences of customers and develop an effective customer-focused selling strategy. As
Mangus and Ruvio (2019) recommended Koran firms can apply compatibility-based
assimilation and value-added differentiation as two strategies to improve their relationship
with buyers of halal cosmetics in Malaysia and Indonesia. Findings indicated that margin-
focused selling strategy appeared as a necessary predictor of profitability in Malaysia. Hence,
the firms should emphasize on margin-focused selling strategy along with customer-focused
selling strategy to attain high profitability in Malaysia. In accordance with Benoit et al. (2020),
margin-focused selling strategy can increase the firm profit when customers have willingness
to pay more for halal cosmetics which is most likely among customers with a strong religious
background.
5.2. Practical Implications

Many countries have regulations concerning halal product assurance. For instance, the
Indonesian parliament issued the Halal Product Assurance Law in October 2014, to reassure
Indonesian Muslim citizens that the products they consume are halal. As noted in Article 1
of the law, “halal products are goods and/or services that are related to food, beverage,
medicine, cosmetics, chemical products, biological products, genetically engineered
products, as well as consumer goods that are worn, used, or utilized by the public.” The
government regulation No. 31/2019 stresses on the requirement for all the products defined
in Article 1 to be halal certified, except for those products which originate from haram
(forbidden) material (McDonald, 2019).
In Southeast Asia, Indonesia and Malaysia are key markets for halal beauty products
because of the growing number of middle class and affluent Muslim consumers who are halal
conscious. Since 2010, Malaysia’s economy has been experiencing annual growth of about
5.4%, and is expected to transition from an upper middle-income to a high-income economy
by 2024 (World Bank, 2020a). Similarly, Indonesia has had an average annual growth rate of
5.6% in GDP over the last 50 years, which made the country a middle-income economy. The
consumption of the middle-class population, as a major driver of economic growth, has
increased by 12% annually since 2002 and now constitutes almost 50% of all household
consumption in Indonesia (World Bank, 2020b). It was estimated that the Indonesian and
96
Malaysian cosmetics and beauty market is worth USD 6.03 billion and USD 5 billion in 2019,
respectively (Lim, 2019).
We develop practical implications based on findings from marketing managers’ views on
popular halal cosmetic brands, influential marketing communication tactics, distribution
channels, and key contemporary challenges of the cosmetic market. Warda Cosmetics and
Sariayu are two most popular brands in the Indonesia market. Indonesian customers are
interested in more functional products such as body care and hair care. According to the
marketing managers view, distribution channels of halal cosmetics are complex and involve
several intermediaries (e.g. drug stores, department stores, retailers, and supermarkets).
There is an opportunity for Korean firms to simplify this distribution channel by developing
an integrated digital marketing mix. To penetrate the Indonesia market and compete with
local competitors, Korean firms can use digital marketing and TV advertising as two
influential communication channels. Korean firms can also follow these channels to
understand industry conditions and selling strategies of competitors and use them along with
celebrity branding to communicate sale promotion campaigns. The reason we suggest
celebrity branding and sale promotion as supplementary marketing communication tactic is
that price and quality are two key criteria that Indonesian customers are considering to make
the decision to purchase halal cosmetics. Nonetheless, TV commercials are among effective
channels to increase awareness of the customers about Korean brands as it has very large
convergence considering the pollution of Indonesia.
Sugarbelle, Pretty Suci, Sorfina Hal and Forest Colour are among the most popular halal
cosmetics brands in Malaysia. In Malaysia there are demands for both general and functional
cosmetics. Distribution channel in Malaysia, which is more direct compared to the Indonesia
channel, mainly involves online shopping, events and expo where customers can directly
purchase halal cosmetics from the brands. TV and radio advertising and digital marketing
using social media are main marketing communication channels to promote cosmetics in
Malaysia. Diversity of religions in Malaysia is more than Indonesia. Lack of awareness of halal
cosmetics, needs for innovation in halal cosmetics, the competitive environment and demand
for low prices are key challenges that firms might face in Malaysia. Korean firms should take
the challenges into account for making decisions about the type of products and targeted
group of customers. They can target young female customers from middle class income as
they are among top spenders on halal cosmetics.
This study modelled the profitability of firms in the halal cosmetics industry; it may not be
possible for the outcome to be generalized to another industry or country. As findings
indicated that the results for the Malaysian and Indonesian markets are not similar, the
models must be adjusted according to the conditions of targeted industry and internal
marketing strategies. Although, a larger sample size might help with generalizability of the
study outcomes. Another limitation is that cross-sectional data are used to test the proposed
configurational models. Therefore, a longitudinal study is recommended to understand
whether the explored models would change over time. This study focuses on the supply side
of halal cosmetics products, an empirical study will be required to examine the demand side
(customers) and identify whether models for achieving desired behavioral responses (e.g.,
loyalty, willing to pay more for halal cosmetics) vary across different countries.
References
Abd Rahman, A., E. Asrarhaghighi and S. Ab Rahman (2015), “Consumers and Halal Cosmetic
Products: Knowledge, Religiosity, Attitude and Intention”, Journal of Islamic Marketing, 6(1),
148-163.
Ahmed, A. (2008), “Marketing of Halal Meat in the United Kingdom: Supermarkets versus Local
97
Shops”, British Food Journal, 110(7), 655-670.
Aisyah, M. (2017), “Consumer Demand on Halal Cosmetics and Personal Care Products in
Indonesia”, Al-Iqtishad: Jurnal Ilmu Ekonomi Syariah, 9(1), 125-142.
Aoun, I. and L. Tournois (2015), “Building Holistic Brands: An Exploratory Study of Halal
Cosmetics”, Journal of Islamic Marketing, 6(1), 109-132.
Ball, R., J. Gerakos, J. T. Linnainmaa and V. V. Nikolaev (2015), “Deflating Profitability”, Journal of
Financial Economics, 117(2), 225-248.
Benoit, S., M. Kienzler and C. Kowalkowski (2020), “Intuitive Pricing by Independent Store
Managers: Challenging Beliefs and Practices”, Journal of Business Research, 115, 70-84.
Briliana, V. and N. Mursito (2017), “Exploring Antecedents and Consequences of Indonesian Muslim
Youths’ Attitude towards Halal Cosmetic Products: A Case Study in Jakarta”, Asia Pacific
Management Review, 22(4), 176-184.
Burnham, T. A., J. K. Frels and V. Mahajan (2003), “Consumer Switching Costs: A Typology,
Antecedents, and Consequences”, Journal of the Academy of marketing Science, 31(2), 109-126.
Cheng, S., and S. J. Lee (2020), “Understanding Users’ Intention to Switch Mobile Instant
Messaging”, International Journal of Mobile Communications, 18(4), 462-483.
Chiu, Y. P., S. K. Lo, A. Y. Hsieh and Y. Hwang (2019), “Exploring Why People Spend More Time
Shopping Online than in Offline Stores”, Computers in Human Behavior, 95, 24-30.
Christino, J., T. Silva, L. R. Moura and L. H. Fonseca (2020), “Antecedents and Consequents of Brand
Love in the Smartphone Market: An Extended Study of the Impact of Switching Cost”, Journal
of Promotion Management, 26(3), 301-321.
Čirjevskis, A. (2020), “Managing Competence-based Synergy in Acquisition Processes: Empirical
Evidence from the ICT and Global Cosmetic Industries”, Knowledge Management Research &
Practice, 1-10.
Crittenden, V. L. and W. F. Crittenden (2014), “Market Orientation in Commodity Marketing: The
Necessary Link Between Marketing and Production”. In M. Enke, A. Geigenmüller and A.
Leischnig (Eds.), Commodity Marketing, Wiesbaden, Germany: Springer Gabler, 421-430.
Department of Standards Malaysia (DSM) (2008), Malaysian Standard MS 2200: Part 1:2008
Islamic Consumer Goods – part 1: Cosmetic and personal Care - General Guidelines,
Cyberjaya, Malaysia: Author, 1-6.
Erimia, C. L. (2018), “Harmonization of the National Legislation on Cosmetic Products in the Context
of the Standardization of the European Legislation in the Field”, European Journal of Medicine
and Natural Sciences, 2(3), 46-50.
Feng, K., L. Altinay and H. Olya (2019), “Social Well-being and Transformative Service Research:
Evidence from China”, Journal of Services Marketing, 33(6), 735-750.
Foltean, F. S., S. M. Trif and D. L. Tuleu (2019), “Customer Relationship Management Capabilities
and Social Media Technology Use: Consequences on Firm Performance”, Journal of Business
Research, 104, 563-575.
Franklin, D. and R. Marshall (2019), “Adding Co-creation as an Antecedent Condition Leading to
Trust in Business-to-business Relationships”, Industrial Marketing Management, 77, 170-181.
Global Islamic Economy Report (GIER) (2018), An Inclusive Ethical Economy, State of the Global
Islamic Economy 2018/19 report, in Collaboration with Dinar Standard: Thomson Reuters.
Available from https://haladinar.io/hdn/doc/report2018.pdf
Goh, W. H., Y. N. Goh, S. K. Ariffin and Y. Salamzadeh (2019), “How Green Marketing Mix
Strategies Affects the Firm’s Performance: A Malaysian Perspective”, International Journal of
Sustainable Strategic Management, 7(1-2), 113-130.
Hair Jr., J. F., G. T. M. Hult, C. Ringle and M. Sarstedt (2016), A Primer on Partial Least Squares
Structural Equation Modeling (PLS-SEM), Thousand Oaks, CA: Sage publications. Available
from https://doi.org/10.1016/j.lrp.2013.01.002
Han, H., A. Al-Ansi, H. G. Olya and W. Kim (2019), “Exploring Halal-friendly Destination Attributes
in South Korea: Perceptions and Behaviors of Muslim Travelers toward a Non-Muslim
98
Destination”, Tourism Management, 71, 151-164.
Hartmann, N. N., H. Wieland, S. L. Vargo and M. Ahearne (2020), “Advancing Sales Theory through
a Holistic View: How Social Structures Frame Selling”, Journal of Personal Selling & Sales
Management, 40(4), 221-226.
Hassim, N. (2014), “Hijab and the Malay-Muslim woman in media”, Procedia-Social and Behavioral
Sciences, 155, 428-433.
Henseler, J., G. Hubona and P. A. Ray (2016), “Using PLS Path Modeling in New Technology
Research: Updated Guidelines”, Industrial Management & Data Systems, 116(1), 2-20.
Hirsch, S. and J. Schiefer (2016), “What Causes Firm Profitability Variation in the EU Food Industry?
A Redux of Classical Approaches of Variance Decomposition”, Agribusiness, 32(1), 79-92.
Homburg, C., M. Klarmann and J. Schmitt (2010), “Brand Awareness in Business Markets: When Is
it Related to Firm Performance?”, International Journal of Research in Marketing, 27(3), 201–
212.
Huang, C.W. and K. H. Huarng (2015), “Evaluating the Performance of Biotechnology Companies
by Causal Recipes”, Journal of Business Research, 68(4), 851–856.
Indonesia Halal Economy and Strategy Roadmap (IHESR) (2018), A Preview Comprehensive
Roadmap by Indonesia Halal Lifestyle Center (IHLC) in Partnership with Dinar Standard.
Available from https://www.bi.go.id/id/ekonomi-dan-keuangan-syariah/materi/Bahan-
Sosialisasi/Documents/Indonesia-Halal-Report-and-Strategy-2018.pdf
Islam, M. S. and M. S. Khan (2019), “The Determinants of Profitability of the Pharmaceutical
Industry of Bangladesh: A Random Effect Analysis”, International Journal of Financial
Research, 10(2), 68-74.
Jamal, A. and J. Sharifuddin (2015), “Perceived Value and Perceived Usefulness of Halal Labeling:
The Role of Religion and Culture”, Journal of Business Research, 68(5), 933-941.
Jefriando, M. and G. Suroyo (2019, February 15), “Indonesia to Postpone Halal Label Deadline amid
Industry Concerns”, REUTERS. Available from https://www.reuters.com/article/us-indonesia-
economy-halal/indonesia-to-postpone-halal-label-deadline-amid-industry-concerns-
idUSKCN1Q40MZ
Kaul, A. and J. Luo (2018), “An Economic Case for CSR: The Comparative Efficiency of For-profit
Firms in Meeting Consumer Demand for Social Goods”, Strategic Management Journal, 39(6),
1650-1677.
Kim, N., S. Min and S. Chaiy (2015), “Why do Firms Enter a New Product Market? A Two-
Dimensional Framework for Market Entry Motivation and Behavior”, Journal of Product
Innovation Management, 32(2), 263-278.
Korea Times (2016, October 18), “Cosmetics Firms Seek for Halal Certification”. Available from
http://www.koreatimes.co.kr/www/tech/2016/10/129_216321.html
Lee, G. K. and M. A. Alnahedh (2016), “Industries’ Potential for Interdependency and Profitability:
A Panel of 135 Industries, 1988–1996”, Strategy Science, 1(4), 285-308.
Leischnig, A. and K. Kasper-Brauer (2016), “How to Sell in Diverse Markets? A Two-level Approach
to Industry Factors and Selling Factors for Explaining Firm Profitability”, Journal of Business
Research, 69(4), 1307-1313.
Li, Y., Q. Lu and G. Guo (2017), “Effects of Social Network on Water-saving Irrigation Technology
Adoption: Homogeneity or Heterogeneity?”, Research of Agricultural Modernization, 38(6),
978-986.
Lim, A. (2019), Revealed: Three Markets in APAC to Watch for Cosmetics Growth in 2019. Available
from https://www.cosmeticsdesign-asia.com/Article/2019/01/07/Revealed-Three-markets-in-
APAC-to-watch-for-cosmetics-growth-in-2019
L’ORÉAL (2018), Annual Report. Available from https://www.loreal-
finance.com/sites/default/files/2019-10/LOreal_2018_Annual_Report_0.pdf
Mangus, S. M. and A. Ruvio (2019), “Do Opposites Attract? Assimilation and Differentiation as
99
Relationship-building Strategies”, Journal of Personal Selling & Sales Management, 39(1), 60-
80.
McDonald, G. (2019), GOI Issues New Implementing Regulation on Halal Product Assurance (GAIN
Report, No. ID1913), Washington, DC: U.S. Department of Agriculture. Available from
https://apps.fas.usda.gov/newgainapi/api/report/downloadreportbyfilename?filename=GOI%2
0Issues%20New%20Implementing%20Regulation%20on%20Halal%20Product%20Assuranc
e%20_Jakarta_Indonesia_6-14-2019.pdf.
McGahan, A. M. and M. E. Porter (1997), “How Much does Industry Matter, Really?”, Strategic
Management Journal, 18(S1), 15-30.
MIHAS (2020), “Malaysian International Halal Showcase”. Available from http://mihas.com.my/
Mulia, D., H. Usman and N. B. Parwanto (2020), “The Role of Customer Intimacy in Increasing
Islamic Bank Customer Loyalty in Using E-banking and M-banking”, Journal of Islamic
Marketing (Forthcoming).
Musa, R. (2014), “Factors Influencing Attitude towards Halal Cosmetic among Young Adult Urban
Muslim Women: A Focus Group Analysis”, Procedia-Social and Behavioral Sciences, 130,
129-134.
Mustafar, M., R. M. Ismail, S. N. Othman and R. Abdullah (2018), “A Study on Halal Cosmetic
Awareness among Malaysian Cosmetics Manufactures”, International Journal of Supply Chain
Management, 7(5), 492-496.
Nawawi, S. B., R. B. M. Roslin and N. B. A. Hamid (2018), “Customers’ Intention to Repurchase
Halal Personal Care Products: The Moderating Role of Religiosity”, In Proceedings of the 2nd
Advances in Business Research International Conference, Singapore: Springer, 39-54.
Novy-marx, R. (2013), “The Other Side of Value: The Gross Profitability Premium”, Journal of
Financial Economics, 108(1), 1-28.
Olya, H. G. and A. Al-ansi (2018), “Risk Assessment of Halal Products and Services: Implication for
Tourism Industry”, Tourism Management, 65, 279-291.
Olya, H. G. and L. Altinay (2016), “Asymmetric Modeling of Intention to Purchase Tourism Weather
Insurance and Loyalty”, Journal of Business Research, 69(8), 2791-2800.
Olya, H., L. Altinay and G. De Vita (2018), “An Exploratory Study of Value Added
Services”, Journal of Services Marketing, 32(3), 334-345.
Pekovic, S. and S. Rolland (2016), “Customer Orientation and Firm’s Business Performance: A
Moderated Mediation Model of Environmental Customer Innovation and Contextual
Factors”, European Journal of Marketing, 50(12), 2162-2191.
Podsakoff, P. M., S. B. MacKenzie, J. Y. Lee and N. P. Podsakoff (2003), “Common Method Biases
in Behavioral Research: A Critical Review of the Literature and Recommended
Remedies”, Journal of Applied Psychology, 88(5), 879.
President Republic of Indonesia (2014), Law Concerning Halal Product Assurance. Available from
http://extwprlegs1.fao.org/docs/pdf/ins139990.pdf
Putri, I. S., S. Daryanti and A. R. Ningtias (2019), “The Influence of Knowledge and Religiosity with
Mediation of Attitude Toward the Intention of Repurchasing Halal Cosmetics”, In 12th
International Conference on Business and Management Research (ICBMR 2018).
Ragin, C. C. (2009), Redesigning Social Inquiry: Fuzzy Sets and Beyond, Chicago, IL: University of
Chicago Press.
Ray, A. (2017, April), Halal Cosmetics Market by Product Type, Application, and Distribution
Channel: Global Opportunity Analysis and Industry Forecast, 2014–2022. Available from
https://www.alliedmarketresearch.com/halal-cosmetics-market
Reimann, M., O. Schilke and J. S. Thomas (2010a), “Customer Relationship Management and Firm
Performance: The Mediating Role of Business Strategy”, Journal of the Academy of Marketing
Science, 38(3), 326-346.
Reimann, M., O. Schilke and J. S. Thomas (2010b), “Toward an Understanding of Industry
Commoditization: Its Nature and Role in Evolving Marketing Competition”, International
100
Journal of Research in Marketing, 27(2), 188–197.
Shah Alam, S. and N. Mohamed Sayuti (2011), “Applying the Theory of Planned Behavior (TPB) in
Halal Food Purchasing”, International Journal of Commerce and Management, 21(1), 8-20.
Shahid, S., F. Ahmed and U. Hasan (2018), “A Qualitative Investigation into Consumption of Halal
Cosmetic Products: The Evidence from India”, Journal of Islamic Marketing, 9(3), 484-503.
Slater, S. F., E. M. Olson and H. Eibe Sørensen (2012), “Creating and Exploiting Market Knowledge
Assets”, Journal of Business Strategy, 33(4), 18-27.
Spencer. N. (2017, March 22), “Halal Producers Target Non-Muslim and Ethical Lifestyle
Consumers”. Available from https://www.cosmeticsdesign-asia.com/Article/2017/03/22/Halal-
producers-target-non-Muslim-and-ethical-lifestyle-consumers
Sriminarti, N. and L. Nora (2018), “The Role of Product Knowledge and Attitudes toward Purchasing
Intentions: The Moderating Effect of Religiosity on Halal Cosmetics”, KnE Social Sciences,
518-529.
Stroe, S., V. Parida and J. Wincent (2018), “Effectuation or Causation: An fsQCA Analysis of
Entrepreneurial Passion, Risk Perception, and Self-efficacy”, Journal of Business Research, 89,
265-272.
Tellis, G. J. (1988), “The Price Elasticity of Selective Demand: A Meta-analysis of Economic Models
of Sales”, Journal of Marketing Research, 25(4), 331–341.
Terho, H., A. Eggert, A. Haas and W. Ulaga (2015), “How Sales Strategy Translates into Performance:
The Role of Salesperson Customer Orientation and Value-based Selling”, Industrial Marketing
Management, 45, 12-21.
The EU Cosmetics Regulation (2009), Regulation (Ec) No 1223/2009 of the European Parliament
and of the Council of 30 November 2009 on Cosmetic Products. Available from https://eur-
lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32009R1223&from=EN
Tho, N. D. (2018), “Firm Capabilities and Performance: A Necessary Condition Analysis”, Journal
of Management Development, 37(4), 322-332.
US Food and Drug Administration (2018), Cosmetics Guidance & Regulation. Available from
https://www.fda.gov/cosmetics/cosmetics-guidance-regulation
Vastani, S. F. and K. B. Monroe (2019), “Role of Customer Attributes on Absolute Price
Thresholds”, Journal of Services Marketing, 33 (5), 589-601.
Weiss, A. M. and J. B. Heide (1993), “The Nature of Organizational Search in High Technology
Markets”, Journal of Marketing Research, 30(2), 220-233.
Wilson, J. A. (2012), “The New Wave of Transformational Islamic Marketing: Reflections and
Definitions”, Journal of Islamic Marketing, 3(1), 5-11.
Wilson, J. A. and J. Liu (2010), “Shaping the Halal into a Brand?”, Journal of Islamic Marketing, 1(2),
107-123.
Woodside, A. G. (2014), “Embrace•perform•model: Complexity Theory, Contrarian Case Analysis,
and Multiple Realities”, Journal of Business Research, 67(12), 2495–2503.
World Bank (2020a), The World Bank in Malaysia. Available from
https://www.worldbank.org/en/country/malaysia/overview#1
World Bank (2020b), Aspiring Indonesia: Expanding the Middle Class. Available from
https://www.worldbank.org/en/country/indonesia/publication/aspiring-indonesia-expanding-
the-middle-class
World Population Review (2020), Muslim Population By Country 2020. Available from
https://worldpopulationreview.com/countries/muslim-population-by-country/
Wu, P. L., S. S. Yeh and A. G. Woodside (2014), “Applying Complexity Theory to Deepen Service
Dominant Logic: Configural Analysis of Customer Experience-and-outcome Assessments of
Professional Services for Personal Transformations”, Journal of Business Research, 67(8),
1647-1670.
Zaefarian, G., C. Thiesbrummel, S. C. Henneberg and P. Naudé (2017), “Different Recipes for
Success in Business Relationships”, Industrial Marketing Management, 63, 69-81.
https://doi.org/10.35611/jkt.2020.24.8.101 101
A Comparative Study on Requirements for

the Buyer’s Right to Withhold Performance
for the Seller’s Actual Non-Performance JKT 24(8)
under the CISG and the CESL*
Byung-Mun Lee
Professor, Department of Global Commerce, Soongsil University, South Korea
Dong-Young Kim†
Ph.D. Candidate, Department of International Trade, Graduate School, Soongsil University, South Korea
Abstract
Purpose – The buyer’s right to withhold performance is a useful and important self-help remedy to
protect himself from the seller’s breach of contract, and it is also the coercive means to induce the
seller to perform his part of contract. However, the buyer’s exercise of such a right often exposes
himself to the risk of breaching the contract. This is generally due to his ignorance when he is entitled
to the right and also uncertainties inherent in the law. Therefore, the purpose of this paper is to
examine what the requirements should be fulfilled before the buyer exercises the right for the seller’s
actual breach of contract.
Design/methodology – In order to achieve the purposes of the study, it executes a comparative study
of the rules as to the requirements for the buyer’s right to withhold performance for the seller’s actual
non-performance under the CISG and the CESL. It mainly focuses on performance due, the seller’s
non-performance, the buyer’s readiness to perform and the requirement of notice.
Findings – The main findings of this comparative study can be summarized as follows: Although the
CISG has no expressive provision for the buyer’s general right to withhold performance for the seller’s
actual non-performance, it may be inferred from the general principles the CISG underlies,
synallagmatic nature of the contract. In addition, it can be drawn by analogy from relevant provisions
of the CISG. On the other hand, the CESL expressively provides that the buyer has a general right to
withhold performance where the seller fails to tender performance or perform the contract. Therefore,
it seems that the position of CESL is rather easier and more apparent to allow the buyer to withhold
performance for the seller’s non-performance.
Originality/value – Most of the existing studies on the right to withhold performance under the CISG
have centered on the right to withhold performance for an anticipatory breach of contract. On the
other hand, there have been few prior studies on the right to withhold performance for the actual non-
performance during a contractual period of performance. Therefore, this paper examined the
requirements for the buyer’s right to withhold performance under the CISG and the CESL in a
comparative way for the seller’s actual breach of obligation. In this conclusion, it may provide practical
and legal considerations and implications for business people who are not certain about the right to
withhold performance.
Keywords: Breach of Contract, Requirement for Withholding of Performance, Right to withhold

Performance, Withholding of Performance
JEL Classifications: K12, K40
†
Corresponding author: ddesigner77@gmail.com
102
1. Introduction
In case of a contract for the international sale of goods which is premised on the bilateral
obligations of the parties, the buyer undertakes to perform his part of the contract in the
expectation that the seller will also likewise perform (Schwenzer, Hachem and Kee, 2012,
548). However, if the seller is anticipated to fail or actually fails to perform his obligation, the
buyer will face the difficult problem of having to decide whether to perform his obligation
such as taking delivery of goods or paying the price (Schwenzer, Hachem and Kee, 2012, 548).
If he decides to perform the contract, he may be exposed to the risk of the non-performance
of the seller, and if he does not, he may be in breach of contract (Schwenzer, Hachem and
Kee, 2012, 548). As such, where the seller fails or is anticipated to perform the contract, the
right to withhold performance is an important remedy that the buyer protects himself from
the seller’s breach of contract (Schwenzer, 2016, 1003), and it is also a means to coerce the
seller to do his contractual obligations without putting an end to the contract (Nyer, 2006,
41). From the buyer’s perspective, the main functions and benefits of the right to withhold
performance are as follows; First, as the buyer withholds his obligation without terminating
a contract, the right to withhold performance is to preserve the contract and also to offer the
buyer a reasonable and convenient means of protecting himself (Kröll, Mistelis and
Viscasillas, 2018, 887; Schulze, 2012, 514-515). Second, the buyer can prevent from incurring
unnecessary costs or losses in the future by withholding his performance (Kröll, Mistelis and
Viscasillas, 2018, 888), and also can save money compared to solving problems through
courts and other dispute-solving bodies (Kröll, Mistelis and Viscasillas, 2018, 887). Third, the
buyer can induce or encourage the seller who has failed or is anticipated to perform the
contract to do his obligation by exercising the right to withhold performance, such as refusal
of taking delivery of goods, or payment (Bianca and Bonell, 1987, 390; Kröll, Mistelis and
Viscasillas, 2018, 888). Forth, the buyer can enjoy an immediate effect by responding quickly
to the seller’s violation through the right to withhold performance instead of using judicial
remedies or other methods (Lee Byung-Mun, 2002, 256). Finally, the right to withhold
performance can solve problems by providing opportunities for amicable agreements
through cooperation between the parties (Kröll, Mistelis and Viscasillas, 2018, 887), and if an
amicable agreement is difficult, it makes the termination of the contract be made more clearly
and easily. Furthermore, it can also help to proceed with compensation issues more effectively
(Kröll, Mistelis and Viscasillas, 2018, 887-888; Schulze, 2012, 514).
Until now, most of the previous studies on the right to withhold performance under the
UN Convention on Contracts for the International Sales of Goods (hereinafter referred to as
“the CISG”)1 have centered on the right to withhold performance for an anticipatory breach
of contract. On the other hand, there have been few prior studies on the right to withhold
performance for the actual non-performance during a contractual period of performance.
This seems due to the fact that unlike the right to withhold performance for the anticipatory
breach of contract, there is no expressive provision for the actual non-performance under the
CISG. It has caused a lot of confusion in legal and business practice when the breached party
is entitled to exercise the right in the case where the breaching party fails to perform a
1
The CISG is an international unification norm that governs the International Sales of Goods, which is
the basis of international trade, being considered as the successful international treaty (Suk, Kwang-
Hyun, 2010). Korea has joined the CISG on February 17, 2004, and became the contracting state of the
CISG from March 1, 2005 (Oh, Won-Suk and Kang-Hun Ha, 2013). As of March 20, 2020, 93 countries
all over the world including USA, China, Germany, France, and Canada have joined the CISG
(https://uncitral.un.org/en/texts/salegoods/conventions/sale_of_goods/the CISG/status) and it is
playing a role as the most famous international unification law.
A Comparative Study on Requirements for the Buyer’s Right to Withhold Performance for the Seller’s
Actual Non-Performance under the CISG and the CESL
103
contract.
Having said that, this study purports to examine in particular the requirements for the
buyer’s right to withhold performance for the seller’s actual non-performance under the
CISG; performance due, the seller’s non-performance, the buyer’s readiness to perform, and
the requirement of notice. In addition to the examination of those requirements under the
CISG, it attempts to compare them with the Common European Sales Law (hereinafter
referred to as “the CESL”) and evaluate them in light of the discipline of comparative law.2
One of the reasons why the CESL is chosen for the law to be compared with the CISG is that
it is the most recently unified law in the area of sale of goods and has a clear and expressive
provision for the right to withhold performance. Furthermore, on the basis of this
comparative study, it may render practical and legal considerations and implications to those
who are not certain about the right to withhold performance.
2. Performance Due
2.1. The CISG
The first requirement for the buyer’s right to withhold performance for the seller’s actual
breach of contract is that the time for the buyer’s performance in the contract becomes due
(Schwenzer, Hachem and Kee, 2012, 550). Although such a requirement is not expressively
stipulated under the CISG, it is a generally accepted requirement because otherwise there
would be no breach from the side of the buyer which does not in turn need us to protect him. 3
The time of the buyer’s performance should be at the same time with or after the seller’s
performance (Schwenzer, Hachem and Kee, 2012, 551). For the former case, it may be
inferred from the provision for the time of payment in the CISG (CISG Art. 58(1)). It provides
that if the buyer is not obliged to pay the price at any other specific time, the buyer must pay
when the seller places either the goods or documents controlling their disposition at the
buyer’s disposal in accordance with the contract and the CISG (CISG Art. 58(1)). In other
words, the buyer’s obligation of payment can be seen as a concurrent obligation of the seller’s
delivery of either goods or documents controlling their disposition. As such, it is a generally
recognized principle that the parties’ respective obligations to deliver and to pay are
2
In addition to the CISG, there have been various discussions for the unification of law in the area of
international trade, and one of the most recent and representative works is the CESL, which can be the
result of efforts for unification of sales law within the European Union. (We, Kye-Chan, 2015, 1647-
1648). The values of the study on the CESL can be argued as follows. First, the CESL was introduced as
the unified law in the area of contracts for the sale of goods in the European Union which was intended
to be a governing law based on the parties’ voluntary agreement (Ha Kyung-Hyo et al., 2014, 2-5).
Second, on the one hand, since the success of the CISG in harmonizing the law in the area of
international sale of goods, the Unidroit Principles of International Commercial Contracts (hereinafter
referred to as ‘the PICC’) and the Principles of European Contract Law (hereinafter referred to as ‘the
PECL’) or the Draft Common Frame of Reference (hereinafter referred to as ‘the DCRF’) have been
introduced as either model laws or products of the academic research project. On the other hand, the
CESL takes the form of firm law which was proposed for adoption by the European Commission. In
this respect, the value of the study on the CESL is significant considering the ripple effect of the PICC,
the PECL, and the DCFR on business and academic fields (Paek, Kyoung-Il, 2013, 436). Third, it can
be provided as a valuable practical reference for Korean companies that are already in EU market or
have a plan to advance to the market at the time when the trade is gradually increasing since the
conclusion of FTA between Korea and EU (Shim, Gap-Young and Chong-Seok Shim, 2015, 54).
3
The requirement of performance being due should be also same for the seller’s performance because
the seller’s actual breach assumes it.
104
concurrent obligations unless otherwise agreed in the contract (Schwenzer, Hachem and Kee,
2012, 551).
Despite of this principle, the parties can fix order of performance freely in their contract
which allocates the risks between them (Schwenzer, Hachem and Kee, 2012, 551). Therefore,
depending upon the contractual order of performance, either the seller or the buyer takes the
risk of the other party’s non-performance. This means that if the buyer agrees to pay the price
before the seller’s performance, the buyer is not generally allowed to exercise the right to
withhold performance on the basis that the seller has not performed his part of contract.
However, he may exceptionally do so only if it becomes apparent that the seller will commit
an anticipatory breach of contract which amounts to non-performance of a substantial part
of his obligations (CISG Art. 71(1)). All in all, the buyer’s right to withhold performance may
be granted if the time of performance in the buyer’s part becomes due where the buyer is
supposed to perform his obligation at the same time as or after the seller’s performance
(Schwenzer, Hachem and Kee, 2012, 548).
2.2. The CESL

There is no explicit provision as to a requirement for the buyer’s right to withhold
performance in the CESL that the time for the buyer’s performance in contract becomes due.
Nevertheless, it can be inferred from articles 113 and 126 of the CESL. The former clearly
provides that the buyer who is supposed to perform his obligation at the same time as or after
the seller performs can withhold his performance until the seller’s performance has been
rendered or the seller has performed in accordance with the contract (CESL Art. 113(1)). This
explains that if the buyer wants to exercise his right to withhold performance, it must be
foremost satisfied with a requirement that the time for his performance must become due.
The time for the buyer’s performance is generally arranged by the contract and by virtue of
such arrangement, it may become due at the same time as or after the seller has rendered
performance or has performed. Where the contractual arrangement requires the buyer to
perform his part of contract before the seller’s performance, the buyer is not generally allowed
to withhold his performance even if the time for the buyer’s performance becomes due.
However, the buyer may be exceptionally entitled to exercise the right to withhold
performance if it is reasonable for the buyer to believe that the seller will not perform the
contract when the seller’s performance becomes due.
In addition to article 113(1) of the CESL, the requirement that the buyer’s performance
become due can be also inferred from article 126 of the CESL. It provides a default rule for
the time of performance that the time when the buyer has to pay the price comes due at the
time of the seller’s delivery of goods (CESL Art. 126(1)). Therefore, unless otherwise agreed
by the parties, the buyer may be entitled to withhold to pay the price because the seller’s non-
delivery of the goods causes the time for the buyer’s payment not to become due (Schulze,
2012, 515). This seems to underly the synallagmatic nature of the contract as a basis for the
buyer’s right to withhold performance in which the seller’s delivery of goods and the buyer’s
payment are reciprocal to one another.
2.3. Comparison and Assessment

As shown above, it is natural to say that there is a requirement for the buyer’s right to
withhold performance that the time for the buyer’s performance becomes due. This is because
if the time of performance has not come, no obligation of the buyer to perform is raised, and
thus the buyer does not need to exercise the right to withhold performance (Schwenzer,
Hachem and Kee, 2012, 550). Notwithstanding no expressive provision both under the CISG
105
and the CESL, the requirement that the time for the buyer’s performance becomes due has
been found by analogical interpretation of the related provisions under the CISG and the
CESL. However, it is deemed that the CESL which has a general right to withhold
performance for the seller’s non-performance is rather clearer than the CISG for us to infer
the requirement. It is because, unlike the CISG, the CESL provides that the general right to
withhold performance is granted to the buyer only when he has to perform at the same time
as or after the seller’s performance. This explains that the buyer is entitled to the general right
to withhold performance where the time for the buyer’s performance becomes due at the
same time as, or after the seller’s performance.
3. Non-Performance by the Seller

There is no doubt in that the buyer is entitled to the right to withhold perform the contract
where the seller has totally failed to perform some of his obligations like obligations to deliver
the goods, to pass the property of goods, to hand over the documents relating to the goods
and some others. However, unlike those cases of the total failure to perform, the awkward
question arises where the seller delivers goods that are not in conformity with the contract in
terms of quality, quantity, or packaging (CISG Art. 35), or which are not free from any right
or claim of a third party (CISG Arts. 41, 42). The same situation arises where the seller fails
to perform some other incidental obligations (Schwenzer, Hachem and Kee, 2012, 551). The
following accounts discuss those intractable cases, first, where the goods delivered or the
documents tendered are not in conformity with the contract, and second, where they are not
free from a third party’s right or claim, third, where the seller fails to perform any of his
incidental obligations.
3.1. Non-Conformity and Third Party’s Rights or Claims

3.1.1. Existence of Non-Conformity of Goods or Third Party’s Rights or Claims
a) The CISG
The CISG provides for the buyer’s remedies for the seller’s breach of contract (CISG Art.
45) and expresses that the buyer is entitled to the right to refuse to take the delivery of goods
only in specific cases such as early delivery and the delivery of the excess quantity (CISG Art.
52). As such, while the CISG recognizes the buyer’s right to withhold performance only in
those specific cases, it is not clear whether it grants a general right to withhold performance
to the buyer where the seller delivers defective goods, especially the ones with defects in
quality, packing, lack of quantity or defects in title (Schwenzer, Hachem and Kee, 2012, 549;
Bianca and Bonell, 1987, 390).4 However, some commentaries argue that a right to withhold
performance can be derived by analogy from the related provisions under the CISG (Bianca
and Bonell, 1987, 389; Enderlein and Maskow, 1992, 229; Jafarzadeh, 1998, 121). Their
arguments are as follows.
First, it is argued that the buyer’s right to withhold performance can be inferred through
4
Many of countries of civil law systems that have basis on Roman law do not recognize a right to
withhold performance for the seller’s delivery of non-conforming goods but for the case of impossibility
of performance and delay. On the other hand, traditionally structured civil law legal systems and
modernized civil law systems seem to recognize the buyer’s right to withhold performance if the seller
performs improper performance, for example in case of delivering non-conforming goods (Schwenzer
et al., 2012, 552).
106
the bilateral relations, although not specifically stated, between articles 30 and 53 of the CISG
that provide the main obligations of the parties under the CISG (Jafarzadeh, 1998, 123; Lee
Byung-Mun, 2002, 262; Lee Byung-Mun and Kim Dong-Young, 2020, 805). This is based on
the general principle of the CISG that the seller should deliver the goods that conform to the
contract and the CISG and if they were otherwise, the buyer is not obliged to take delivery of
the goods (Jafarzadeh, 1998, 122). In other words, the buyer is obliged to pay the price and
take delivery of the goods only when the seller is ready to deliver goods, has the intention to
do so, and actually performs in accordance with the contract and the CISG (Jafarzadeh, 1998,
123). This means that if the seller does not perform his duty under article 30, it can be assumed
that the buyer is allowed to refuse to perform his obligations under article 53 of the CISG
(Jafarzadeh, 1998, 123).
Second, although article 58 of the CISG empowers the buyer to refuse the payment only
where the seller does not deliver the goods or documents controlling their disposition, it is
asserted that the right to withhold performance applies to the other obligations by virtue of a
general principle drawn from article 71 of the CISG (Schwenzer, 2016, 882; Kröll, Mistelis
and Viscasillas, 2018, 817). Article 71 of the CISG provides that a party may withhold his
performance if it becomes apparent that the other party will not perform a substantial part of
his obligations. Here the right to withhold performance under article 71 of the CISG is
generally understood to be applied to some other obligations of the party than delivery and
payment (Schwenzer, 2016, 1004-1006). Therefore, as long as the anticipatory breach of those
obligations allows the buyer to withhold performance, he should be also allowed to withhold
performance for the actual breach (Schwenzer, 2016, 882 et seq.; Kröll, Mistelis and
Viscasillas, 2018, 817).
Third, they also argue that the buyer’s right to withhold performance can be inferred
through article 86(1) of the CISG (Bianca and Bonell, 1987, 390; Enderlein and Maskow, 1992,
229 et seq.). They presuppose that the buyer’s right to reject the goods under article 86 of the
CISG is basically same as the right to refuse to take delivery of the goods as in article 52 of the
CISG (Bianca and Bonell, 1987, 390; Enderlein and Maskow, 1992, 229). In addition, they
maintain that the right to refuse to take delivery of the goods can be drawn by analogy from
article 58 which provides that the buyer’s payment is subject to the seller’s delivery of the
goods which are in accordance with the contract (Bianca and Bonell, 1987, 391; Enderlein
and Maskow, 1992, 229). Furthermore, they assert that article 46(1) and 47(1) of the CISG
are to recognize the buyer’s right to refuse to take delivery under certain circumstances
(Bianca and Bonell, 1987, 390; Enderlein and Maskow, 1992, 229). This argument seems
plausible in the sense that it answers the question of how the buyer’s general right to withhold
performance for non-conformity of goods can be inferred from the existing provisions under
the CISG (Lee Byung-Mun, 2002, 259).5
Fourth, Article 52(2) of the CISG provides the buyer’s right to refuse to take delivery of the
excess quantity, whereas the CISG does not express for whether the buyer has a right to refuse
to take delivery in the case of a shortage in quantity. Nevertheless, it is argued that the buyer’s
right to withhold performance can be inferred from Article 51(2) of the CISG which provides
that the buyer can avoid the contract as a whole if the seller’s non-performance or non-
conformity amounts to a fundamental breach of the contract (Jafarzadeh, 1998, 124). Namely,
5
On the other hand, some scholars contend that the right to reject the goods in article 86 is in its nature
different from the right to refuse to take delivery of the goods (Jafarzadeh, 1998, 122; Lee Byung-Mun,
2002, 259). Their argument is based on the former can arise regardless of whether the goods have been
taken by the buyer in the first place. This explains article 86 is not justifiable as a provision which
deduces a general right to refuse to take delivery of the goods as in articles 52 or 71 (Jafarzadeh, 1998,
122; Lee Byung-Mun, 2002, 259).
107
it can be assumed that if the shortage in quantity amounts to a fundamental breach of the
contract under Article 25 of the CISG, the buyer may refuse to take delivery of the goods and
to pay the price until the seller tenders complete delivery pursuant to the contract (Jafarzadeh,
1998, 124; Schwenzer, Hachem and Kee, 2012, 548). Then here the questions remain in the
case where the shortage in quantity does not amount to a fundamental breach of contract. In
this context, it may be controversial concerning the right to withhold performance being
available to the buyer, and if it is, to what extent of the buyer’s part of performance the buyer
can withhold. It is maintained that even in the case of shortage in quantity, the buyer’s right
to withhold the performance can be similarly derived by analogy from the bilateral relations
between articles 30 and 53 of the CISG (Jafarzadeh, 1998, 123; Lee Byung-Mun and Kim
Dong-Young, 2020, 805). And there are some arguments that even if the less quantity is
trivial, the right to withhold performance should be recognized because it is the remedy to
protect the aggrieved party from the other party’s breach of contract, and is a means of
coercing the other party into full performing (Nyer, 2006, 41). Assuming that the buyer has
such a right, it is commonly recognized that the buyer’s right to withhold performance in the
case of shortage in quantity must be exercised by the ratio (Jafarzadeh, 1998, 124; Lee Byung-
Mun and Kim Dong-Young, 2020, 805; Schwenzer, Hachem and Kee, 2012, 551).
Lastly, the CISG does not also express on whether the buyer’s right to withhold
performance is rendered in the case where the seller delivers the goods which are not free
from a third party’s right or claim. However, it can be inferred in the same way as non-
conformity of goods based on the bilateral relations between articles 30 and 53 of the CISG
(Jafarzadeh, 1998, 123; Lee Byung-Mun and Kim Dong-Young, 2020, 805). In other words,
the buyer is allowed to refuse to pay the price and take delivery of the goods as long as the
seller has delivered the goods which are not in accordance with the contract and the CISG.
The seller’s delivery of goods which are not in accordance with the contract and the CISG
undoubtedly include the cases where the seller delivers the goods which are not free from a
third party’s right or claim. Therefore, based on the general principle of the CISG, if the seller
fails to deliver the goods which are not free from a third party’s right or claim as required in
articles 41 and 42 of the CISG, it can be assumed that the buyer is not obliged to take delivery
and to pay the price (Bianca and Bonell, 1987, 392; Lee Byung-Mun and Kim Dong-Young,
2020, 805).
b) The CESL
The CESL explicitly provides that the buyer can withhold performance until the seller has
tendered performance or has performed (CESL Art. 113(1)). However, it is not clear whether
this explicit provision can be applied even for the cases where the seller delivers the goods
which are not in conformity with the contract in terms of quality, quantity, or packaging, or
not free from the third party’s rights or claims. Even though there is no explicit provision in
the CESL for those cases, it is considered that the right to withhold performance should be
accepted for the following reasons.
First, the commentary book of the CESL states that the buyer can exercise the right to
withhold performance for the complete failure to perform. In addition, it says that the right
can be extended to the cases of the partial non-performance or the non-conformity of the
goods (Schulze, 2012, 516).
Second, recognizing there is the buyer’s right to withhold performance for non-conformity,
the CESL provides that the buyer can withhold the whole or part of his duty to the extent
justified by the seller’s non-performance and the buyer may withhold his performance
partially where the seller’s obligations are to be performed in separate parts or are otherwise
divisible (CESL Art. 113(3)). This seems to justify that the buyer’s right to withhold
108
performance for non-conformity is not limited to the complete failure to perform, but
extended to other non-conformity.
Third, although not specifically stated under the CESL, one can draw bilateral relations
from articles 91 and 123 of the CESL which provide respectively the seller’s and the buyer’s
main obligations. On the basis of the bilateral relations, the right to withhold performance
under article 113 of the CESL should be interpreted in a way that renders it applies to the
cases of non-conformity of the goods. Thus, if the seller fails to ensure that the goods conform
to the contract, the buyer may be allowed to exercise the right to withhold performance
pursuant to article 113 of the CESL (Schulze, 2012, 516). This interpretation can be similarly
applied to the cases where the seller delivers the goods which are not free from third party’s
rights or claims. The reason for that can be found in that the concept of non-conformity
under the CESL includes not only the cases where the seller delivers the goods which are not
in conformity with the contract in terms of quality, quantity, description, or packaging, not
supplied with accessories or instructions (CESL Art. 99(1)), but also the cases where the seller
delivers the goods which are not free from the third party’s rights or claims (CESL Art. 102). 6
Furthermore, as regards the non-conformity of the goods in quantity, the CESL also
explicitly provides about the buyer’s right to refuse to take delivery not only for the excess
quantity but also for the lack of quantity (CESL Art. 130(2), (3)). However, one must note
that the buyer may refuse the lack of quantity only where he has a legitimate interest (CESL
Art. 130(2)). Otherwise, the buyer shall take delivery of the goods that are lack in quantity
(We, Kye-Chan, 2015, 1670). If the buyer refuses to take delivery of a quantity of the goods
less than that of the contract, the buyer can withhold an entire payment under the CESL. On
the other hand, if the buyer decides to take the delivery of them, he may withhold the payment
only in relation to the non-delivered part of goods (Schulze, 2012, 516; CESL Art. 130(2)).
c) Comparison and Assessment

As discussed above, although the CISG explicitly expresses the buyer’s right to withhold
performance for certain cases, it does not provide a general right to withhold performance
(CISG Art. 52). Due to the lack of an expressive provision under the CISG, it caused
uncertainties as regards the question of whether the right to withhold performance is
recognized as a general right where the seller delivers the goods which are not in conformity
with the contract in terms of quality, packing and quantity and are not free from third party’s
rights and claims. Despite the uncertainties, many scholars have tried to draw the buyer’s
general right to withhold performance by virtue of gap-filling rules under the CISG (CISG
Art. 7). That is, they seem to treat the matter of uncertainties as one of the matters which is
not expressly settled under the CISG, but can be done by virtue of a general principle to be
drawn from relevant provisions of the CISG. On this basis, they have struggled to introduce
the buyer’s general right to withhold performance within the existing regulatory framework
of the CISG in various ways. The first and most important way to infer the right under the
CISG was to find out the bilateral relations between articles 30 and 53 of the CISG which
6
In this case, the buyer can withhold the whole or part of his performance within the scope which can
be justified pursuant to the degree of the seller’s non-performance (CESL Art 113(3); Schulze, 2012,
516). In case the seller’s non-performance can be divisible, the buyer can refuse to pay the price only
for the relevant part. 6 If there is a partial non-conformity of goods, the buyer should assess the
significance of the non-conformity and may withhold to pay the price to the extent justified by the
significance of the non-conformity of goods (Schulze, 2012, 516). The entire payment can be also
withheld depending on the contents of the seller’s non-performance (Schulze, 2012, 516). In addition,
the scope of withholding performance by the buyer will be determined in relation to the remedies that
the buyer exercises (Schulze, 2012, 516).
109
provide the main obligations of the parties. In addition to this, it was argued that on the basis
of a general principle drawn from article 71 of the CISG, the buyer’s right to refuse the
payment under article 58 of the CISG could apply to the cases of the seller’s failure to perform
other obligations than the delivery of goods and handing over of documents controlling their
disposition. Furthermore, it was also argued that the buyer’s right to refuse to take delivery as
another type of the right to withhold performance can be derived by analogy from articles 58,
46(1), 47(1) of the CISG another right on the premise that the buyer’s right to reject the goods
under article 86 of the CISG is equivalent to his right to refuse to take delivery under article
52 of the CISG.
In contrast with the position of the CISG, the CESL expressively provides that the buyer
has a general right to withhold performance where the seller fails to tender performance or
perform the contract. It does not specifically deal with the cases of all kinds of lack of
conformity. However, as we examined above, it was not difficult to see the buyer’s right to
withhold performance in the case of non-conformity of the goods. This is evidenced by the
fact that the CESL posits the specific provision for the general right to withhold performance
at the forefront of such right and it is followed by other specific provision for the buyer’s right
to withhold performance in the case of the seller’s partial or divisible performance, early
delivery, shortage and excess delivery in quantity. This explains the buyer’s right to withhold
performance should be, as clearly stated in its commentary book, also applied to the partial
non-performance or the lack of conformity.
As such, the position of the CESL which has a specific provision for a general right to
withhold performance was rather easier and more apparent to allow the buyer to withhold
performance in the case of the lack of conformity. The position of the CISG also recognizes
such right on the basis of analogical interpretation of relevant provisions under the CISG but
may be open to criticism due to the absence of a specific provision for a general right to
withhold performance and unnecessary and inefficient analogical interpretation. In this
regard, it is submitted that it may cause the unnecessary misunderstanding and disputes
between the contractual parties about the buyer’s right to withhold performance, particularly
where the seller delivers the goods which are not in conformity with the contract and not free
from the third party’s rights or claims. Therefore, in order for the contractual parties to avoid
those situations, they must fully keep in mind and it is desirable to clearly specify in their
contract in what cases they have a right to withhold performance.
3.1.2. Non-Conformity of Documents

a) The CISG
It is not clearly provided under the CISG whether the buyer can exercise his right to
withhold performance where the seller tenders the defective documents (Jafarzadeh, 1998,
126). However, it can be argued that this problem may be solved by inferring from article
58(1) of the CISG, which provides that the buyer must pay when the seller places the
documents that control the disposition of goods at the buyer’s disposal in accordance with
the contract and the CISG (Jafarzadeh, 1998, 126). According to this article, as the buyer is
obliged to pay only when the seller places the documents that control the disposition of goods
at the buyer’s disposal if the seller does not do so, it means that the buyer is entitled to refuse
payment (Jafarzadeh, 1998, 126). However, one must note that this article is not applied to all
kinds of documents but limits to the documents controlling the disposition of goods
(Jafarzadeh, 1998, 126). If so, the issue is whether the buyer can refuse to take delivery of
documents or withhold the payment if the seller tenders the documents that do not conform
to the requirements of the contract in the case where they are not kinds of the documents
110
controlling the disposition of goods (Lee Byung-Mun, 2002, 262 et seq.; Lee Byung-Mun and
Kim Dong-Young, 2020, 806). In this regard, insofar as the documents concerned in article
58(1) of the CISG are those controlling the disposition of the goods, one may assume the
buyer is not allowed to refuse to pay the price where the seller tenders a defective document
which does not control the disposition of the goods. However, it is submitted that the
documents stated in article 58(1) of the CISG include not only documents of the title but also
those used for the seller to perform his obligation to deliver the goods pursuant to articles 30
and 34 of the CISG (Enderlein and Maskow, 1992, 219 et seq.; Schwenzer, 2016, 878 et seq.;
Kröll, Mistelis and Viscasillas, 2018, 815 et seq.). This interpretation seems to extend the
buyer’s right to refuse to pay the price to the cases of some other documents than documents
of title. In addition, it is argued that, even if the documents tendered are not kind of those
controlling the disposition of the goods, the buyer is still entitled to reject a non-conforming
document and refuse to pay the price. This argument is based upon the bilateral relations
between articles 30 and 53 of the CISG that provide the main obligations of the parties under
the CISG (Jafarzadeh, 1998, 127; Lee Byung-Mun, 2002, 263 et seq.). Furthermore, as it is
provided in the CISG, the refusal of documents which do not control the disposition of the
goods and conform to the contract accords to usages and practices that are widely applied
and accepted between the contractual parties (CISG Art. 9). That is, if the documents
tendered by the seller are not in accordance with the contract, the buyer may in practice treat
them in the same way as if the goods themselves do not conform to the contract and thus
reject the documents and refuse to pay the price (Bianca and Bonell, 1987, 428 et seq.;
Enderlein and Maskow, 1992, 219 et seq.; Lee Byung-Mun, 2002, 264; Lee Byung-Mun and
Kim Dong-Young, 2020, 806).
b) The CESL
Article 91(e) of the CESL provides that one of the seller’s main obligations is to deliver
documents representing or relating to the goods or documents relating to the digital content
as may be required by the contract. Like its counterpart, article 123(1)(c) of the CESL
provides that the buyer is under the obligation to take over such documents representing or
relating to the goods as may be required by the contract. These two provisions provide the
main obligations of both parties and these obligations shall be interpreted pursuant to their
bilateral contractual relation (Schulze, 2012, 534). Based on this bilateral contractual relation
and the buyer’s right to withhold performance under article 113 of the CES, it should be
interpreted that the buyer is allowed to withhold his performance until the seller tenders
documents representing or relating to the goods.
However, the question remains whether the buyer’s right to withhold performance may be
granted even if there is a defect in the document provided by the seller. In this regard, it should
be noted that Article 91(e) of the CESL requires that the seller has to tender the documents
representing or relating to the goods which may be required by the contract. This means that
a document to be procured by the seller must be in accordance with the type, form, and
manner in which it must be handed over as required by the contract. So, if the seller fails to
procure the documents as required by the contract, the buyer can exercise the right to
withhold performance pursuant to article 113 of the CESL. Here the question is whether such
right to withhold performance can be applied to the withholding of payment. As to this
question, it is necessary to examine article 126 of the CESL which provides that the time of
the payment comes due upon the delivery of goods. Here, the mode of delivery agreed
between the parties may decide the time when payment must be made (CESL Art. 94). Insofar
as the seller is required to procure documents enabling the buyer to take over the goods from
the carrier or representing the goods in order to carry out his duty to deliver the goods, his
111
failure to procure such documents as required by the contract may allow the buyer to refuse
to pay the price (CESL Art. 113). One thing to be noted here is that the documents would not
include documents other than those for the disposal of goods unless agreed otherwise by the
parties.
c) Comparison and Assessment

As discussed above, the CISG does not provide explicit provisions on whether the buyer
can exercise the right to withhold performance in case of non-conformity of documents.
However, it was argued that the buyer’s right to withhold performance could be deduced
through article 58(1) of the CISG. According to this article, the buyer has a right to refuse to
take delivery of documents and to pay the price only if the seller fails to deliver the documents
controlling the disposal of goods. Insofar as the documents to be tendered are those not
controlling the disposal of goods, the buyer’s right to refuse to pay the price could be derived
by analogy from the bilateral relationship between articles 30 and 53 of the CISG, and this
view was justified by usages and practices broadly accepted in documentary sales contracts.
However, this view may open to criticism in that the bilateral relationship does not refer to
the buyer’s obligation to take over the documents tendered by the seller. This is because,
contrary to article 30 of the CISG, article 53 of the CISG does not expressively provide the
obligation to take over the documents as one of the buyer’s main obligations.
The CESL does not have an expressive provision for the buyer’s right to withhold
performance in case where the seller tenders defective documents. However, compared to the
position of the CISG, it was not difficult to infer that the buyer’s right to withhold
performance was able to be accepted through article 113 of the CESL which was provided for
the buyer’s general right to withhold performance. In addition, the bilateral contractual
relation between articles 91 and 123 of the CESL justifies the buyer to refuse to take over
documents that do not conform to the contract. Here contrary to the position of the CISG,
those articles deal with the seller’s obligation to tender documents (CESL Art. 91(e)) and the
buyer’s obligation to take delivery of documents (CESL Art. 123(1)(c)).
It was maintained both under the CISG and the CESL that the buyer might withhold the
payment if the seller tenders the documents which are not in accordance with the contract.
But this is applied only in the where the documents are those used for the disposal of the
goods. In this regard, it is regrettable to say that they do not properly reflect practices and
usages established in the area of documentary sale of goods. This may cause
misunderstanding and disagreement on document requirements by the contracting parties
in practice. Therefore, the parties should pay special attention to this and it would be desirable
for them to explicitly agree on this to prevent unnecessary disputes in advance.
3.2. Severity of Non-Conformity

3.2.1. The CISG
The CISG does not provide how serious non-conformity would allow the buyer to withhold
performance. It could be argued that it should reach the same degree of severity of non-
conformity as amounting to being sufficiently serious or material to terminate the contract.
However, it is generally understood that it does not require non-conformity to amount to a
fundamental breach of contract which entitles the buyer to terminate the contract (CISG Art.
25; Lee Byung-Mun, 2002, 266).
Although not clearly stated under the CISG, the degree of severity of non-conformity which
allows the buyer to withhold performance may be inferred from Article 71 of CISG which
provides the right to withhold performance for anticipatory breach of contract (Lee Byung-
112
Mun, 2002, 266). This is because this article deals with the same right to withhold
performance as in the case of an actual breach of contract. According to article 71 of CISG,
the buyer may withhold his performance if it becomes apparent that the seller will not
perform a substantial part of his obligations. Then, here the question is what “a substantial
part of performance” in this article means. Although not clear, it is submitted that the
‘substantial part’ of the seller’s obligations does not belong to a fundamental breach which is
specified in article 72 of the CISG (Bianca and Bonell, 1987, 519; Enderlein and Maskow,
1992, 286; Karton, 2009, 883; Kröll, Mistelis and Viscasillas, 2018, 888; Lee Byung-Mun, 2002,
266). Otherwise, article 71 of the CISG should have taken the same approach as article 72
does in terms of severity of breach (Lee Byung-Mun, 2002, 266; Schwenzer, 2016, 1008). As
the right to withhold performance is a self-help remedy which is intended to withhold the
buyer’s performance and to coerce the seller’s performance without avoidance of the contract,
it seems that the buyer’s right to withhold performance does not need to be based on the
seller’s fundamental breach (Jafarzadeh, 1998, 125; Lee Byung-Mun, 2002, 266).
Given that a fundamental breach of contract is not required for the buyer’s right to
withhold performance, the next question is how serious non-conformity should be for the
buyer to exercise the right. It is submitted that it should be decided by virtue of the principle
of proportionality that the buyer’s withholding of performance should be consummated to
the seller’s breach since the right to withhold performance is based on the principle of bilateral
contractual relations (Nyer, 2006, 36-40). Therefore, as long as the seller’s breach allows the
buyer to terminate the contract (CISG Art. 49(1)(a)) or claim replacement (CISG Art. 46(2)),
the buyer should be vested the right to refuse to take delivery and pay the price (Lee Byung-
Mun, 2002, 267). In addition, even if the seller’s breach amounts to a breach which does not
allow the buyer to terminate the contract but only to claim damages (CISG Arts. 45(1)(b), 74,
77), price reduction (CISG Art. 50) or repair (CISG Art. 46(3)), the buyer may be still entitled
to withhold performance until the seller cures non-conformity of goods or documents (Lee
Byung-Mun, 2002, 267). However, in case of minor non-performance, the buyer’s right to
refuse to take delivery of goods may not be granted pursuant to the principle of good faith
(Bianca and Bonell, 1987, 392; Lee Byung-Mun, 2002, 267). In the same vein, one must also
take into account the contract as a whole to decide what constitutes a substantial part of the
obligations and find out a fair balance between the parties to be achieved after the exercise of
the buyer’s right to withhold performance in a way to reduce the possibility of the buyer
abusing the right (Kröll, Mistelis and Viscasillas, 2018, 889; Schwenzer, 2016, 1009).
3.2.2. The CESL

As reviewed earlier, the CESL grants the buyer the right to withhold performance where
there is any non-conformity in the goods or documents (Schulze, 2012, 516). However, it
does not clearly provide the degree of severity of non-conformity which allows the buyer to
exercise the right to withhold performance. What is clear is that the degree of severity of non-
conformity for exercising the right to withhold performance is not the same as a fundamental
breach of a contract which is required for the buyer’s right to terminate the contract (CESL
Art. 114). This can be inferred from the fact that the provisions for the withholding of
performance and termination of contract under the CESL are separated (CESL Arts. 113-
114). If there is any non-conformity in the goods or documents, the extent to which the buyer
will be able to withhold performance may be decided by virtue of the severity of non-
conformity (Schulze, 2012, 516). Therefore, one must assess the severity of non-conformity
as well as the remedies available to the buyer due to non-conformity (Schulze, 2012, 516). If
the buyer can claim the replacement of goods, the buyer may be allowed to reject the goods
or refuse to take over the documents and to withhold the entire payment (Schulze, 2012, 516).
113
The same can be applied if non-conformity entitles the buyer to claim repair by the seller
(Schulze, 2012, 516). Where non-conformity of goods is to be repaired by the buyer, the
payment may be withheld until the repair of goods is completed. At this time, the right to
withhold performance may depend on the type of non-conformity and its impact on the
functioning of goods (Schulze, 2012, 516).
3.2.3. Comparison and Assessment

As reviewed above, both the CISG and the CESL do not express clearly for the degree of
severity of non-conformity which may allow the buyer to exercise the right to withhold
performance. Although not clear enough, both the CISG and the CESL has shown us the same
direction on this matter. First, the point is that non-conformity does not necessarily have to
amount to a fundamental breach of contract for the buyer to exercise the right to withhold
performance. Second, it seems that the right to withhold performance is depended upon the
degree of severity of non-conformity as well as what remedies are available to the buyer
pursuant to the principle of proportion. This position seems plausible particularly when one
takes into account the right to withhold performance as a coercive measure which may hold
the seller strict adherence to whatever agreed in the contract (Nyer, 2006, 36-40).
The only possible difference can be found in the question of how both jurisdictions deals
with trivial non-conformity. The CESL looks not clear and no commentary on this matter,
whereas the position of the CISG says the buyer should not be granted the right to withhold
performance for trivial non-conformity based on the principle of good faith. However, this
may not be true because the same principle is clearly provided in article 2 of the CESL and it
can be interpreted without any doubt to prevent the buyer from abusing the right to withhold
performance.
As examined above, those rules do not seem to give us a clear cut. However, the issue of
conformity often arises in practice and causes disputes between the contractual parties about
the severity of non-conformity for the right to withhold performance. Therefore, the
contracting parties shall keep this in mind and be better off to specify the clear standard rule
in the contract.
3.3. Non-Performance of Additional Obligations

3.3.1. The CISG
Additional obligations may be different from the nature of the bilateral obligations under
the contract. Therefore, it has been debated whether the breached party is granted the right
to withhold performance when such additional obligations are not performed (Schwenzer,
2016, 1006; Schwenzer, Hachem and Kee, 2012, 553). 7 The CISG does not also express
whether the buyer can exercise the right to withhold performance in relation to non-
performance of the seller’s additional obligations.8 However, it is submitted that the buyer
7
It is doubtful whether the right to withhold performance is properly applied in practice if this right is
granted for the breach of additional obligations. For example, if the buyer’s obligation to take delivery
is classified as the additional obligation under the governing law, it does not seem reasonable for the
seller to exercise the right to withhold performance to coerce the buyer into taking delivery (Schwenzer,
Hachem and Kee, 2012, 553).
8
The CISG does not differentiate between main and additional obligations, or bilateral and non-bilateral
obligations (Kröll, Mistelis and Viscasillas, 2018, 350; Schwenzer, 2016, 429-459). Thus, any breach of
additional obligations that are not regarded as typical obligations of a contract for the sale of goods may
still amount to a fundamental breach of contract (Kröll, Mistelis and Viscasillas, 2018, 350; Schwenzer,
2016, 429-459). If non-performance of additional obligations belongs to a fundamental breach of the
114
should be granted a general right to withhold performance at least in certain circumstances
(Schwenzer, 2016, 1006; Schwenzer, Hachem and Kee, 2012, 553). They argue that the buyer’s
obligation which will be withheld and the seller’s obligation which is not performed are not
necessarily the equivalent of each other (Schwenzer, 2016, 1006). This view is derived from
the principle underlying article 71 of the CISG, the importance of self-help remedy without
judicial intervention and the general principle of reasonability (Schwenzer, 2016, 1006 et
seq.). Therefore, for example, if the seller fails to perform the obligation to give prior notice
about a delivery place as required by the contract, the buyer may withhold the payment
(Schwenzer, 2016, 883). And the buyer is also allowed to withhold to pay the price where the
seller fails to tender a receipt of payment as agreed in the contract (Schwenzer, 2016, 883). In
addition, the buyer may be entitled to withhold payment if the seller has an additional
installation duty and his installation has caused the buyer’s working place unsafe (Schwenzer,
Hachem and Kee, 2012, 553). Then the question remains whether the buyer should be allowed
to withhold performance for the seller’s failure to perform any kind of additional obligations.
Although not clear, it may depend upon the actual circumstances of the case and the
interpretation of the contract pursuant to article 8 of the CISG.
3.3.2 The CESL

Article 113 of the CESL which provides the right to withhold performance is, as examined
above, based upon the reciprocal nature of contract (Schulze, 2012, 514). This might be
interpreted that insofar as the seller’s obligations are not reciprocal by their nature, the buyer’s
right to withhold performance cannot be applied to the breach of additional obligations.
However, one must note that article 113 of the CESL explicitly provides that the buyer has the
right to withhold performance where the seller has failed to perform or tender performance.
And thus, it is necessary to examine the concept of the seller’s non-performance under article
113 of the CESL to determine whether the buyer can exercise the right to withhold
performance for the breach of additional obligations. Article 87 of the CESL defines the
concept of the seller’s non-performance such as non-performance, delay of delivery and
delivery of defective goods (CESL Art. 87). However, the kinds of non-performance stated in
Article 87 of the CESL are just the most common cases of non-performance. The list of these
cases of non-performance is not exhaustive and mostly provides only illustrations to give
guidelines on the concept of non-performance under the CESL (Schulze, 2012, 400). The
concept of non-performance stated in Article 87 of the CESL is defined as the failure of the
contracting party to fulfill his obligations and it also refers to the discrepancies between the
party’s obligations and actual performance (Schulze, 2012, 399). In other words, it is very a
comprehensive concept that includes the failure to perform all kinds of obligations (Schulze,
2012, 399). Therefore, it can be inferred that any kind of failure to perform additional
obligations also belongs to the concept of non-performance that is provided in the CESL.
Based on this, it can be assumed that the buyer can withhold his performance if there is any
seller’s breach of additional obligations under article 113 of the CESL.

Both the CISG and the CESL do not explicitly provide whether the buyer has the right to
withhold performance in relation to non-performance of the seller’s additional obligations.
In this regard, there has been a controversy over whether the right to withhold performance
may be granted for the breach of additional obligations. This is because the right to withhold
contract, the buyer may exercise the right to withhold performance (Schwenzer, 2016, 884).
115
performance is based on the bilateral contractual relation but additional obligations cannot
be defined as a bilateral obligation between the contract parties due to its nature (Schwenzer,
Hachem and Kee, 2012, 553). Despite such controversy, it seems that we can infer that the
buyer may withhold performance for the breach of the seller’s additional obligations.
However, it is doubtful whether it is possible to exercise the right to withhold performance
for all cases of the breach of additional obligations. As regards this matter, it should be
interpreted that the buyer’s right to withhold performance is allowed for the seller’s failure to
perform any kind of additional obligations. This is justified when one takes into account the
primary playing role of the right to withhold performance as a coercive measure to force the
seller to comply with any of his additional obligations (Nyer, 2006, 36-40). However, as
examined above, there would be misunderstandings or disputes between the parties as to the
buyer’s right to withhold performance for the seller’s breach of additional obligations since
there is no explicit provision and compromised view on this. Therefore, practitioners should
establish clear and objective standards and clearly specify this in their own contract.
4. Buyer’s Readiness for Performance and Requirement of Notice

4.1. Buyer’s Readiness for Performance
4.1.1. The CISG and the CESL
The buyer’s readiness and ability to perform his obligations can be considered as another
requirement to exercise the right to withhold performance (Schwenzer, Hachem and Kee,
2012, 553; Schulze, 2012, 514). The CISG and the CESL do not provide explicitly for this
requirement. However, it seems that it may be inferred through the bilateral relations between
the buyer’s and the seller’s main obligations (CESL 91, 123; CISG Art. 30, 53; Lee Byung-Mun
and Kim Dong-Young, 2020, 809). The relations explain, as noted before, that a contractual
party undertakes to perform his part of the contract in the expectation that the other party
will also likewise perform (Schwenzer, Hachem and Kee, 2012, 548). On the basis of such
relations, it may be asserted that the seller’s obligation to deliver the goods or documents may
arise only when the buyer is ready to take delivery of the goods and pay the price, has the
intention to do so, and actually performs in accordance with the contract (Jafarzadeh, 1998,
123). Namely, if the buyer is not ready to do his obligation or has no intention to so, as there
is no obligation of the seller to perform, the right to withhold performance may not be applied
to this case (Jafarzadeh, 1998, 123). This is because it may be unreasonable if the right to
withhold performance is granted to the buyer who seeks to avoid performing his obligation
even where he is not ready to perform or has no intention to do so (Schwenzer, Hachem and
Kee, 2012, 553). This requirement is also justified when one takes into account the fact that
the buyer’s obligation to take delivery of the goods includes not only taking over the goods
but also all the acts which are necessary for the seller to deliver the goods.

Both the CISG and the CESL do not provide the buyer’s readiness for performance as a
requirement for the buyer to exercise the right to withhold performance. But the requirement
could be derived through the bilateral relations between both parties’ obligations. If this
requirement is not accepted, the right to withhold performance can be used for the buyer as
a means of avoiding his obligation even if he is not ready to perform or not intended to
perform (Schwenzer, Hachem and Kee, 2012, 553). In other words, it may be meaningless to
grant the buyer the right to withhold performance if the buyer already breaches one of his
116
obligations (Schwenzer, Hachem and Kee, 2012, 553).
As regards this requirement, in practice, the buyer seeking to withhold performance often
faces the seller’s claim for the buyer’s breach of a contract based on the lack of readiness to
perform his part of the contract. In order to avoid this situation, the buyer is advised to take
a precautionary measure before he exercises his right to withhold performance, for example,
the deposit of payment by the buyer or the performance (payment) guarantee by a third party.
Furthermore, it would be reasonable to eliminate the room for controversy by explicitly
providing in the contract what requirements must be fulfilled to show the buyer’s readiness
for performance before he exercises the right to withhold performance.
4.2. Requirement of Notice

4.2.1. The CISG and the CESL
The CISG has an expressive provision for a notice of withholding performance in an
anticipated breach of contract (CISG Art. 71(3)). There has been controversy as to the
consequence of failure to provide a notice of withholding performance in case of the other
party’s anticipated breach of contract. On the one hand, it is argued that the notice is required
for the right to withhold performance in case of the other party’s anticipated breach of
contract (Kröll, Mistelis and Viscasillas, 2018, 904). 9 On the other hand, the others contend
that the notice is not necessarily regarded as a requirement for the right to withhold
performance, but the failure to provide notice simply leads to a claim for damages by the other
party (Huber and Mullis, 2014, 343; Schwenzer, 2016, 1015). In contrast, there is no explicit
provision on the requirement of notice for the buyer to exercise the right to withhold
performance for the seller’s actual breach of contract under the CISG. However, in the case
of the CISG, although arguable, the requirement of notice may be inferred from Article 71(3)
of the CISG. This is based upon the idea that there is no reason to differentiate the notice
requirement for the right to withhold performance between anticipated and actual breach of
contract. In addition, it can be by analogy drawn from article of 26 of the CISG which provides
a notice requirement for the right to avoid the contract (Schwenzer, Hachem and Kee, 2012,
554). It is argued that this view is consistent with a general principle underlying the CISG
under Article 7(2) of the CISG that the contractual parties can exercise their rights only by
providing notice (Schwenzer, Hachem and Kee, 2012, 554). Once the buyer is determined to
exercise his right to withhold performance, he is required to immediately notify the seller of
his intention to do so (CISG Art. 71(3)). The meaning of such immediate notice is that the
buyer has to provide a notice without any avoidable delay before or as soon as his withholding
of performance (Kröll, Mistelis and Viscasillas, 2018, 904; Schwenzer, 2016, 1014). As regards
the contents of the notice, it is submitted that it should indicate on which grounds the buyer
exercises his right to withhold performance (Kröll, Mistelis and Viscasillas, 2018, 905;
Schwenzer, 2016, 1014).
Unlike the position of the CISG, the CESL does not provide for the requirement of notice
to exercise the right to withhold performance for anticipated and actual breach of contract
(CESL Art. 113, 118). According to the commentary book of the CESL, the buyer can exercise
the right to withhold performance without a separate notice if all the requirements are
fulfilled (Schulze, 2012, 515).

As examined above, the CISG seems to, although no consensus reached among academics
9
This is a majority view taken by case holdings.
117
and courts, require the buyer to provide a notice of withholding performance as a
precondition for the exercise of the right to withhold performance. On the other hand, the
CESL does not provide any notice requirement for the right to withhold performance. The
position of the CESL may be criticized as follows. First, it may not achieve the purpose of the
right to withhold performance which encourages the parties to perfect the contract by an
exchange of communication and cooperation and induces the other party to restate his
commitment to the contract (Kröll, Mistelis and Viscasillas, 2018, 904). That is, if the buyer
does not give any notice on the withholding of performance, the seller may lose the
opportunity to take a timely cure on his non-performance (Schulze, 2012, 584). This result
may not help at all to save the contract by enhancing communication and cooperation
between parties. Second, the seller should investigate the reason by himself why the buyer
refused to perform and the buyer is not even obliged to answer the seller’s question about the
buyer’s refusal to perform (Schulze, 2012, 584). This is directly opposed to the provision of
article 3 of the CESL that imposes the obligation to co-operate with each other to the extent
that it can be expected for the performance of obligations under the contract (Schulze, 2012,
584). Third, where there is no notice of the buyer to withhold his performance, the seller may
misunderstand the buyer’s withholding of performance as an anticipatory breach that may
induce the seller to terminate the contract pursuant to article 136 of the CESL. However,
insofar as there is still controversy as to the existence of notice requirement, although the
buyer may often provide a notice of withholding performance in his interest in practice, it is
desirable for the buyer to do so without avoidable delay in order to remove any risk of losing
the right to withhold performance and putting himself in breach of contract (Kröll, Mistelis
and Viscasillas, 2018, 904). In addition, it is necessary for the parties to negotiate and agree to
the requirement of notice in the contract to make this more clearly in practice.
5. Conclusion
This paper examined the requirements of the right to withhold performance through the
comparative study based on the CISG and the CESL for the buyer’s right to withhold
performance by the seller’s actual breach of obligation. In this conclusion, we will present the
following considerations and implications required when practitioners designate and use the
CISG and the CESL as the governing laws by focusing on the differences found through the
analysis and review by the comparative study.
First, the CISG explicitly recognizes the buyer’s right to withhold performance for a certain
case such as the early delivery or excess quantity, but it does not explicitly provide whether
the right to withhold performance is recognized where the seller delivers the goods with the
defects in quality, packing, quantity, the documents or the right. However, many scholars
have tried to draw the buyer’s general right to withhold performance from relevant provisions
of the CISG. In the case of the non-conformity of goods, we could infer that the buyer may
withhold his performance under the CISG by the bilateral relations between articles 30 and
53 of the CISG which provide the main obligations of the parties. And, on the basis of a
general principle drawn from article 71 of the CISG, it could be assumed that the buyer’s right
to refuse the payment under article 58 of the CISG may be applied to the cases of the seller’s
failure to perform other obligations than the delivery of goods and handing over of
documents controlling their disposition. Furthermore, we could see that the buyer’s right to
refuse to take delivery can be recognized by analogy from articles 58, 46(1), 47(1) of the CISG
on the premise that the buyer’s right to reject the goods under article 86 of the CISG is
equivalent to his right to refuse to take delivery under article 52 of the CISG. In case that the
118
seller delivers the goods, which are not free from a third party’s right or claim, it can be also
inferred in the same way like as non-conformity of goods based on the bilateral relations
between articles 30 and 53 of the CISG. Lastly, where the seller delivers the defective
documents, we could assume that the buyer may withhold performance through article 58(1),
articles 30 and 34 of the CISG.
On the other hand, the CESL explicitly provides that the buyer has a right to withhold
performance where the seller fails to tender performance or perform the contract. It also
provides that the buyer may withhold his performance in case of early delivery, lack of
quantity, and the delivery of excess quantity. But it is not clear whether it can be applied if
there is any defect of goods or documents or non-conformity of the right. However, in case
of the non-conformity of the goods, we could assume that the buyer may withhold
performance to the extent that is justifiable based on the analogical interpretation of the
relevant provisions of the CESL (Article 91, Article 99, Article 123, Article 126 and Article
113). In the case where the seller tenders the defective documents, we can also infer that the
buyer may have the right to withhold performance through Article 113 and the bilateral
contractual relation between both parties’ obligations under articles 91 and 123 of the CESL.
As discussed so far, the CESL provides the right to withhold performance more specifically
than the CISG, but the issue of defective goods or documents or non-conformity of the right
is still not clear. This may result in disputes or uncertainties between the contract parties.
Therefore, it would be appropriate for the contract parties to explicitly provide this part in the
contract.
Second, both the CISG and the CESL do not explicitly provide whether the performance
must be due as the requirement for the buyer’s withholding of performance. However, we
found that it was not difficult to accept it as the requirement of the right to withhold
performance by inferring from the relevant provisions. This is because whether the buyer
exercises the right to withhold performance for the seller’s breach is, of course, applicable only
when the time of performance comes due. In general, the time of performance of one party is
the time when goods are delivered and payment is made unless agreed otherwise by the
parties, and the risk of the parties is to be allocated when fixing the order of performance.
Therefore, it is necessary for the parties to negotiate and specify the necessary conditions in
relation to the exercise of the right to withhold performance in concluding the contract after
the parties fully review the risk by the difference in the order of performance.
Third, both the CISG and the CESL do not clearly provide what degree of non-conformity
entitles the buyer to exercise the right to withhold performance. As to this severity of non-
conformity, there are some arguments that it should be the same level of seriousness to
terminate the contract. However, we could infer that the degree of severity of defects is not
required to be a fundamental breach of contract under the CISG and the CESL, and the
defects should be significant enough to make one of the remedies be exercised to recognize
the buyer’s right to withhold performance. In addition, the right to withhold performance
may be recognized proportionally depending on the degree of its severity, and it seems that it
may not be accepted for very minor defects based on the principle of good faith. However,
the ambiguity of the conformity standard may cause disputes between the parties about the
severity of non-conformity for the right to withhold performance in practice. Therefore, it is
necessary for the contract parties to keep this in mind and specify clear standards in the
contract.
Fourth, it is not clearly expressed whether the buyer may withhold the performance in
relation to the seller’s non-performance of additional obligations in the CISG and the CESL.
However, we could see that the right to withhold the performance may be granted to the
buyer in specific circumstances under CISG. And in the case of the CESL, based on the
119
analogical interpretation that any kind of failure to perform additional obligations also
belongs to the concept of non-performance that is provided in the Article 87 of the CESL, it
can be also inferred that the buyer may withhold his performance if there is any seller’s breach
of additional obligations under article 113 of the CESL. However, this also may cause
differences in interpretation and controversy. Therefore, it would be desirable for
practitioners to establish clear and objective standards and specify this in the contract.
Fifth, both the CISG and the CESL do not provide the buyer’s readiness for performance
as the premise requirement to withhold performance. However, this principle could be
assumed through the bilateral relations between the obligations of the buyer and the seller. If
this principle is not accepted, the buyer may use this to simply avoid its obligations where he
is not ready or not willing to perform its obligation under the contract. In addition, if the
buyer tries to withhold performance without readiness to perform his obligations of a
contract, the buyer may face the seller’s claim for the buyer’s breach of a contract. In order to
avoid this situation, the buyer needs to take a precautionary measure such as the deposit of
payment by the buyer or the performance (payment) guarantee by a third party before
exercising his right to withhold performance. In conclusion, it would be desirable to eliminate
the room for controversy by explicitly providing the requirement of the buyer’s readiness for
performance in the contract to prevent this issue in advance.
Lastly, it seems that the buyer can exercise the right to withhold its obligation through a
notice under the CISG in case of being the seller’s breach of the contract. However, there is
no obligation for prior notice in the CESL. This may be controversial, but Article 113 of the
CESL does not require the buyer to give a notice for the refusal of performance (Schulze, 2012,
516). However, as examined above, this may lead to disputes between the parties. Therefore,
it would be advisable for the buyer to give a prior notice on the withholding of performance
based on the principle of good faith. In addition, the parties are advised to specify the
contents, method and time of notice in the contract in practice.
References
Bianca, C. M. and M. J. Bonell (1987), Commentary on the International Sales Law, The 1980 Vienna
Sales Convention, Milan, Italy: Giuffre.
Enderlein, F. and D. Maskow (1992), International Sales Law, New York, NY: Oceana Publications.
Ha Kyung-Hyo, Se-Il Ko, Kyu-Wan Kim, Sang-Joong Kim, Kyoung-Il Paek, Byung-Jun Lee, Joon-
Hyun and Jin-Ki Lee (2014), Common European Sales Law, Seoul: Se-Chang Press.
Huber, P. and A. Mullis (2014), The CISG - A New Textbook for Students and Ractitioners (2nd ed.),
München, Germany: European Law Publisher.
Jafarzadeh, M. (1998), The Buyer’s Remedies for Seller’s Non-Conforming Delivery: A Comparative
Study under English Law, the UN Convention on Contracts for the International Sale of Goods
and Shiah Law (Doctoral Dissertation), Sheffield, UK: Sheffield University.
Karton, J. (2009), “Contract Law in International Commercial Arbitration: The Case of Suspension
of Performance”, International and Comparative Law Quarterly, 58(4), 863-896.
Kröll, S., L. Mistelis and P. P. Viscasillas (eds.), (2018), UN Convention on Contracts for the
International Sale of Goods (the CISG) (2nd ed.), München, Germany: C. H. Beck.
Lee, Byung-Mun (2002), “A Comparative Study on the Buyer’s Right to Withhold Performance for
the Seller’s Delivery of Defective Goods and Documents in International Sales within the CISG,
English Law and Korean Law”, The International Commerce and Law Review, 17, 251-293.
Lee, Byung-Mun and Dong-Young Kim (2020), “A Study on the Buyer’s Right to Withhold
Performance for the Seller’s Actual Non-performance under the CISG”, Industrial Economic
120
Research Review, 33(3), 795-815.
Nyer, D. (2006) “Withholding Performance for Breach in International Transactions: An Exercise in
Equations, Proportions or Coercion”, Pace International Law Review, 18(1), 29-82.
Oh, Won-Suk and Kang-Hun Ha (2013), Understanding the UN Convention on the International Sale
of Goods, Seoul: Sam-Young-Sa.
Paek, Kyoung-Il (2013), “Der Vertragsschluss im Gemeinsamen Europäischen Kaufrecht”, Journal
of Comparative Private Law, 20(2), 433-476.
Schulze, R. (2012), Common European Sales Law (CESL), Baden-Baden, Germany: Nomos
Verlagsgesellschaft.
Schwenzer, I. (ed.), (2016), Commentary on the UN Convention on the International Sale of Goods
(the CISG) (4th ed.), New York, NY: Oxford University Press.
Schwenzer, I., P. Hachem and C. Kee (2012), Global Sales and Contract Law, New York, NY: Oxford
University Press, 548-559.
Shim, Gap-young and Chong-Seok Shim (2015), “A Study on the Seller’s Obligations and the Buyer’s
Remedies under Common European Sales Law: Focusing on Comparison with the CISG”,
Korea International Commerce Review, 30(4), 51-71.
Suk, Kwang-Hyun (2010), Legal Theory on Contracts for International Sale of Goods: Commentary
on the United Nations Convention on Contracts for the International Sale of Good, Seoul: Pak-
Young-Sa.
UNCITRAL (n.d.), Status: United Nations Convention on Contracts for the International Sale of
Goods (Vienna,1980) (the CISG). Available from https://uncitral.un.org/en/texts/salegoods/
conventions/sale_of_goods/theISG/status
We, Kye-Chan (2015), “Obligations of the Buyer under the Contract for International Sale of Goods:
focused on the duty of the buyer to pay the price and to take delivery of the goods under Proposal
for Regulation of Common European Sales Law”, Study on Law and Politics, 15(4), 1645-1678.
https://doi.org/10.35611/jkt.2020.24.8.121 121
Effects of Market Diversity on Performance

of Exporting Companies:
An Inverted U-shaped Relationship* JKT 24(8)
Jungeun Lee
Department of Business Administration, The Catholic University of Korea, Korea
Chang-Bong Kim†
Department of Trade and Logistics, Chung-Ang University, Korea
Dong-Jun Lee
Department of Trade and Logistics, Chung-Ang University, Korea
Abstract
Purpose – The principle aim of this study is to further investigate the relationship between market
diversity and export performance. We examine the benefits and costs of geographic market diversity
regarding the number of countries exported to by firms on their export performance. Based on the
financial risk reduction model and the entry costs model, we propose a way to incorporate the costs
and benefits aspects of market diversity.
Design/methodology – To empirically investigate our research question, the curvilinear relationship
between market diversity and export performance, we built a secondary panel data set between 2015
and 2019, containing 17,863 observations of Korean exporting companies. A generalized least squares
panel estimator with fixed effects was employed to test the hypothesis, and the statistical package, Stata
14, was used.
Findings – Our main findings are as follows: As market diversity increases, export performance
increases because exporters can diversify and reduce financial risks in export markets. However, the
relationship between the two does not grow. As it peaks, the entry costs increase due to the high market
diversity, thereby outweighing the benefits, leading, eventually to decrease in the export performance.
Consequently, there is an inverted U-shaped relationship between market diversity and export
performance.
Originality/value – In the export and trade literature, the impact of market diversity on export
performance has not been addressed yet, despite the importance of this subject. Many scholars have
assumed a positive linear relationship between the two, considering only the decrease in market risks
as the number of overseas markets increases, without examining the increase in the entry and
management costs. Therefore, our study contributes by providing a new perspective for analyzing the
characteristics and outcomes of market diversity.
Keywords: Curvilinear Relationship, Export Performance, Geographical Market Diversity

JEL Classifications: F10, F18, M10
1. Introduction
When a company does not secure proper growth and profits in the domestics market due
to the limited customer demand, it is likely to enter foreign markets in order to overcome this
* This work was supported by the Ministry of Education of the Republic of Korea and the National
Research Foundation of Korea (NRF-2019S1A5A2A03034788).
†
Corresponding author: kimchangbong@cau.ac.kr
122
situation (Hanson et al., 2016). There are several ways to enter foreign markets, however,
companies consider exporting the most because it provides a mechanism to expand the
market base and profitability from the fierce competition in the domestic market (Belich and
Dubinsky, 1995). Companies lay their foundations in foreign markets by exporting, thereby
expanding without significant investment in those markets (Mahoney et al., 2001). Numerous
firms argue profitability through export activities. According to the “Trade Statistics by
Business Characteristics in 2019” published by KOSTAT (Statistics Korea), the number of
companies engaged in export activities in South Korea in 2019 was approximately 97,000, and
their total export value was 54.1 billion dollars, achieving the highest record ever (KOSTAT,
2019).
Even though firms participating in export activities are steadily increasing, not all
companies can achieve a “sustainable” competitive advantage through them. A survey on the
survival rate of exporters in South Korea, found out the one-and five-year survival rate were
49.8% and 17.4%, respectively, indicating that 8 out of 10 export companies eventually shut
down the trade within five years (KCS, 2018). One of the reasons why exporting companies
struggle to achieve a sustainable competitive advantage is the uncertainty around
international markets. The fluctuation of the annual export amount by country and
geographic region (e.g., Southeast Asia, Middle East, Latin American. E.U., etc.) is quite large
(see KOSTAT, 2019).
The environment surrounding businesses and markets is continuously changing, with
uncertainty being the distinct feature of the global economy. Some country macro-level
changes significantly affect the economic conditions. Unpredictable changes in market
demands are caused by macroeconomic shifts (Lee Seung-Hyun and Chung, 2007), and
exporting companies need to deal with this uncertainty to secure constant profitability. For
exporters, the best way to manage the potential risks of an unstable foreign market is to
diversify and expand their portfolio. When a company exports to a single market, its export
performance is mainly influenced by the conditions of that sole market. When the single
market suffers a recession caused by an external collapse, an exporting company relying solely
on that market would be severely threatened. However, even if there is a crash in one market,
the company can manage and reduce risks by securing other foreign markets affected
differently from the economic situations of the crashed market. Therefore, exporting
companies need to manage how many countries they export to, in order to gain sustainable
competitive advantages.
In this study, we mainly focus on the impact of the number of exporting countries
represented by market diversity on the performance of exporting companies. However, in the
export and trade literature, this issue has not yet been addressed despite its importance. There
may be various reasons for this, however, we believe the major issue is the difficulty of
obtaining reliable and credible data. Most empirical studies on exporting activities have
mainly obtained data on their research variables through surveys, thus acquiring subjective
and limited information because companies are extremely reluctant to provide objective data
to researchers (Francis and Collins-Dodd, 2000; Leonidou, Katsikeas and Samiee, 2002).
Moreover, objective data are rarely publicly available (Robertson and Chetty, 2000).
Therefore, in the trade literature, the level of analysis is usually confined to a few exporters
and their self-reported capabilities. Hence, to answer the research question, firm-level
objective secondary data are collected and then organized as a panel data set.
As mentioned earlier, there are many ways for a firm to enter a foreign market, however,
diversification and export would both similarly result in expanding the geographical market
base of the firm. Therefore, the theoretical backgrounds and frameworks explaining the
outcome of the geographical market expansion in the diversification literature can be
Effects of Market Diversity on Performance of Exporting Companies: An Inverted U-shaped Relationship
123
appropriately applied to the corporate export activities research.
Two competing arguments with opposing results about the effect of geographical market
expansion appear in the diversification and international business literature (Tallman and Li,
1996). Various researchers insist that a high level of international market diversity could
positively affect corporate performance by reducing market risks. This line of work starts with
Agmon and Lessard’s (1977) argument based on a financial risk reduction model. In the
international market, under a high level of uncertainty, the company expands its market base
to reduce risk by minimizing performance variance (Mansi and Reeb, 2002). Conversely, the
other research stream argues that a high level of international market diversity could instead
negatively affect corporate outcomes. This line of arguments mainly focuses on the cost side.
There are costs of entering and managing various markets. Prior studies have proved the
presence of a sunk entry costs in exporting (Das, Roberts and Tybout, 2007). Entry costs refer
to the total costs of collecting the necessary information for successful exports (Moxnes,
2010). Through quality information, exporting companies need to establish country-specific
marketing, distribution, and sales methods. Since these are tailored differently to the
situations and conditions of each overseas market, more costs are incurred if companies
export to various markets.
Which of these two arguments best describes the export performance? Further, are they in
any way reconcilable? Until now, most scholars have assumed a positive linear relationship
between market diversity and export performance, without investigating the nonlinear
relationship between them considering the cost aspects of managing foreign markets. To
answer the research question, we summarize these two perspectives, generate a hypothesis,
and then test them in an empirical setting consisting of: Korean exporting companies listed
on the database of the Korea Trade-Investment Promotion Agency (KOTRA) between 2015
and 2019. Using corporate-level panel data on exporters, we investigate the relationship
between the level of market diversity and export performance.
2. Theory and Hypothesis

Some studies have tried to identify factors associated with exporters’ geographical market
expansion, mainly focusing on the antecedents of geographical expansion, such as the level
of innovation, market-related knowledge, and experiences (e.g., Belich and Dubinsky, 1995;
Johansson and Karlsson, 2007). However, studies on the result of geographical expansion
have not received enough attention yet. Therefore, this study focuses on the consequence of
geographical market expansion, which is market diversity in the form of increasing the
number of countries that a firm exports to.
2.1. Benefits and Costs of Geographical Market Diversity

The limited size of internal regional markets provides a natural incentive for companies to
export their products overseas (Lee Seung-Hyun and Chung, 2007). In this way, companies
could overcome stagnated growth and gain profits. However, the overseas markets’
conditions and situations are unstable, and unexpected shifts in demands are caused by
macroeconomic changes (Lee Seung-Hyun and Chung, 2007), creating uncertainties.
Exporting companies could manage those market uncertainties by expanding their foreign
market base. According to the financial risk reduction model, exporting companies can
reduce the risk of market uncertainties by not having all their commitments in one country
(Agmon and Lessard, 1977; Hisey and Caves, 1985). When companies are dispersed,
conducting exporting activities across various countries, increased risks in one market could
124
be offset by potentially reduced risks in another, thereby minimizing risks and performance
variance (Mansi and Reeb, 2002). Specifically, when a firm exports to a single market, its
export performance is largely dependent on the conditions of that market. When they are
stable, there is no change; when the conditions are unstable, however, the exporting company
can experience immense losses. The company can manage uncertainties and reduce risks by
investing in other markets that are influenced differently from the economic situations of the
crashed market. Thereby, exporting companies could appropriately respond to unpredicted
changes in the macro-environment. Moreover, exporting companies with high market
diversity could benefit from more stable earnings compared to competitors with low market
diversity without efficient means of lowering risks (Hisey and Caves, 1985). As a result,
exporting companies with high market diversity enjoy higher risk reduction benefits than
those with low market diversity, thus improving their export performance.
Under this circumstance, international business scholars insist that multinational
companies need to manage the overall risks they are exposed to by constantly reconfiguring
their foreign market portfolio to environmental changes (Benito and Welch, 1997; Wilson,
1980). Likewise, exporting companies can also have a different export performance
depending on how the company manages and controls the exporting portfolio. For example,
if there are some exporting countries, and most of the export performance results from only
a few major markets, it is expected that the effectiveness of geographical market diversity
would be inferior to the costs of managing several markets.
Recent exporting literature, such as the trade models of Chaney (2008), and Eaton, Kortum,
and Kramarz (2004), considers country-specific costs. According to Moxnes (2010), country-
specific costs are specific foreign market entry costs for exporting companies. When
exporting companies enter a new foreign market, they need to collect the required
information, such as customer requirements, their lifestyle, laws and regulations, and
accessible distribution channels specific to that market for successful exporting. By obtaining
and utilizing quality information, exporting companies can acquire an in-depth
understanding of each foreign market and its conditions and situations. In addition, there are
other possible costs, such as adjusting some product characteristics to follow country-specific
laws, regulations, and product standards; developing functional-level strategies, including
marketing and distribution to reach and attract the foreign customers; finding adequate
foreign clients; and negotiating contracts with them on favorable terms. Whereas some of
these costs are common to every export market, such as acquiring global-level information,
most of them are generated specifically per country (Moxnes, 2010). Previous exporting
research has proved the presence of sunk entry costs (Das, Roberts and Tybout, 2007). Based
on this perspective, exporting companies with high market diversity spend more on entry
costs than firms with low market diversity, eventually decreasing export performance.
The two opposing arguments on the relationship between market diversity and export
performance could be summarized as follows: Exporting companies with more dispersed
markets enjoy more comprehensive risk reduction benefits than those with fewer
concentrated markets, securing enhanced export performance. However, exporting
companies with more dispersed markets tend to have higher costs compared to those with
fewer concentrated markets. Therefore, country-specific entry costs can alter the benefits of
the geographic market diversity of exporting companies. There could be a trade-off
relationship between the entry costs and benefits of the financial risk reduction for exporting
companies on export performance. Accordingly, in this study, we propose a curvilinear
research model that incorporates both the costs and benefits of exporting companies on
export performance. Thus, we propose:
125
H1: There exists an inverted U-shaped relationship between market diversity and export
performance.
3. Methodology
3.1. Data and Sample
We identify all exporting companies that are listed on the database of the Korea Trade-
Investment Promotion Agency (KOTRA) between 2015 and 2019 in South Korea. The
agency established in 1962, is a government-invested institution under the Ministry of Trade,
Industry and Energy. The agency has collected significant data on trade and investment,
trying to facilitate trade and investment between South Korea and other countries by
providing information on markets and countries through the Trade Big Data Platform. It
contains information about 76,000 overseas markets from 129 trade centers in 84 countries
around the world, obtained from over 500 export conferences and exhibitions every year, and
1.8 billion global export statistics data. Then, we collect data regarding corporate export
activities from the agency and the Korea Customs Service. Other corporate-level data are
obtained from the Korea Information Service Database (KISLINE), which is similar to
COMPUSTAT from Standard & Poor’s (Chang and Hong, 2000), providing credible and
relevant corporate and financial information. In total, 6,151 firms met all the criteria resulting
in a longitudinal cross-sectional (panel) data set of 17,863 observations.
3.2. Measures
3.2.1. Dependent Variable
Export performance. Export performance could be measured by both objective and
subjective methods. In the exporting and trade literature, more studies have used the
subjective questionnaire-based method over objective methods based on secondary data (e.g.,
Robertson and Chetty, 2000; Katsikeas, Piercy and Ioannidis, 1996), despite the importance
of objective indicators. Moreover, since most of the studies were based on single-collection
surveys, the findings could be limited. There are some issues acquiring objective data in the
export setting. Primarily, exporting companies tend to be reluctant to provide objective
secondary data to researchers (Francis and Collins-Dodd, 2000; Leonidou, Katsikeas, and
Samiee, 2002). Moreover, objective data are rarely publicly available (Robertson and Chetty,
2000). However, if objective data is accessible, we believe that there are more benefits of using
it over subjective methods to measure export performance by avoiding social desirability bias
in the self-reported survey. Therefore, for this study, we adopt an objective proxy to measure
export performance.
Much of the empirical literature, where export performance is used as a dependent variable,
adopts “export intensity,” that is, the ratio of export sales to total sales, as a proxy of export
performance. However, there has been some criticism regarding the use of this type of
indicator (Cooper and Kleinschmidt, 1985; Sousa, Martínez‐López and Coelho, 2008;
Zucchella, Palamara and Denicolai, 2007). For example, a firm conducting insufficient export
activities with a large overseas market may perform better than a company with a large market
share of a relatively small overseas market (McGuinness and Little, 1981). Therefore, to
capture the export performance of companies, we employ a measure based on the
recommendation of Sousa (2004) and Katsikeas, Leonidou and Morgan (2000). In particular,
we apply the relative export performance measure suggested by Cadogan, Cui and Li (2003).
126
Export performance is calculated by dividing the absolute value of total export sales by the
number of employees in a given year, controlling for the impact of a company’s size on export
sales. The absolute values of total export sales of each exporting company between 2015 and
2019 are obtained from the publication by the Korea Customs Service, and data regarding the
number of employees are from KISLINE.
3.2.2. Independent Variable

Market Diversity. Following the research conducted by Johansson and Karlsson (2007)
and Wheeler, Ibeh and Dimitratos (2008) market diversity is measured by the number of
export countries (destinations) of each exporting company using the data from the KOTRA
and the Korea Customs Service. This is the most common measure to capture geographical
market diversity (Wheeler, Ibeh and Dimitratos, 2008). In addition, a variable that squared
the market diversity (Market Diversity Squared) was created for testing curvilinear
relationships between market diversity and export performance.
3.2.3. Control Variables

Based on a thorough review of previous research on exporting performance, we found
control variables designed to eliminate potential confounding factors that might affect export
performance. We added four control variables to our final research model: company size,
prior export performance (t-1), product diversity, and year dummies.
Since larger exporting companies tend to have more resources and capabilities for better
export performance, company size is measured by the number of employees. Prior export
performance (t-1) is also controlled for in the model and is calculated by dividing the absolute
value of total export sales by the number of employees a year before the given year (t). The
market level, as well as product diversity, could affect the export performance. Therefore, we
control for product diversity measured by the number of export products of exporting
companies. Finally, there is a potential variation regarding export performance across the
years of this study (2015-2019). To control for this time effect in the panel data, year dummy
variables were included in the analysis.
3.3. Estimation
To test the hypotheses, a generalized least squares (GLS) estimator was used employing the
Stata 14 statistical package. The reason for adopting the GLS estimator is as follows: Since we
have panel data, a pooled-ordinary least squares estimator may be inefficient. There can be
firm-specific differences in the error term capturing unobserved factors, which could affect
export performance. The GLS method takes this heteroskedasticity into account, and
therefore, it is the most efficient estimator. Moreover, to determine the most appropriate
panel data analysis model between the random-effect and fixed-effect models, we used the
Hausman test. Based on the result of the test, the best model used is the fixed effect. Thus, in
this study, the fixed effect panel regression was employed.
After obtaining the results of the regression analysis, additional analysis was performed to
confirm whether the nonlinear relationship between the market diversity and the export
performance is statistically significant. Therefore, Utest was conducted based on the Lind and
Mehlum (2010)’s suggestion. The result of the test supported the significance of the inverted
U-shape between the independent variable and dependent variable, which supports our
hypothesis.
127
Table 1 presents the descriptive statistics and correlations matrix for all the research
variables we used to test the hypothesis. Market Diversity has a mean of 12.56, ranging from
1 to 126, while Export Performance has a mean of 14.76.
Table 1. Descriptive Statistics and Correlation Matrix

No Variables N Mean SD 1 2 3 4 5 6
1 Export Performance 17,863 14.76 63.73 1.00
2 Prior Performance (t-1) 17,863 14.28 61.28 0.80* 1.00
3 Market Diversity 17,863 12.56 14.93 0.12* 0.10* 1.00
4 Market Diversity Squared 17,863 380.78 1019.79 0.08* 0.07* 0.90* 1.00
5 Product Diversity 17,863 10.87 19.98 0.08* 0.07* 0.29* 0.21* 1.00
6 Company Size 17,863 112.29 318.46 0.13* 0.08* 0.20* 0.19* 0.43* 1.00
Note: *p <0.05.
Table 2 presents the fixed effect panel regression estimates of the relationship between the
level of geographical market diversity and export performance.
Model 1 is the baseline, including all the control variables and fixed effects for the events.
Prior export performance, product diversity, and company size have significant positive
impacts on the export performance of companies (β =0.116, β =0.038, β =0.186,
respectively, with p <0.001). In Models 2 and 3, we enter market diversity and market
diversity squared, respectively, with control variables and fixed effects to investigate
whether the two explanations (reduction of risks and increase of costs) might be reconciled,
forming an inverted U-shaped (curvilinear) relationship between the market diversity and
the export performance. In Model 2, the coefficient of market diversity on export
performance is positive and significant (β =0.099, p <0.01). Namely, when the level of
market diversity increases, holding other variables constant, the export performance of the
company would increase. In Model 3, with the market diversity squared, we finally test our
hypothesis, the inverted U-shaped relationship between market diversity and export
performance. The coefficient of market diversity squared is negative and significant (β =-
0.040, p <0.05), meaning that market diversity has a curvilinear effect on export
performance. Thus, Hypothesis 1 is supported.
The export performance of companies rises before peaking at the highest market diversity
level, after which it starts to decline. More precisely, export performance decreases after a level
of market diversity of 79 in our research setting.
Finally, the effects of interest from our full model are illustrated in Figure 1. Using the
regression coefficients and intercepts from Model 3, it is evident that the effect magnitudes of
market diversity are substantively significant.
128
Table 2. Estimates for Fixed Effects Models of Export Performance
1 2 3
Export Export
Variables Export Performance
Performance Performance
Market Diversity 0.421*** 0.576**
[0.099] [0.135]
Market Diversity Squared -0.003**
[-0.040]
Prior export performance (t-1) 0.121*** 0.120*** 0.120***
[0.116] [0.115] [0.115]
Number of Exporting Products 0.121*** 0.094*** 0.091***
[0.038] [0.030] [0.029]
Company Size 0.000*** 0.000*** 0.000***
[0.186] [0.182] [0.183]
Year Dummies 2016 -1.444** -1.567*** -1.582***
[-0.008] [-0.009] [-0.009]
2017 0.012 -0.191 -0.212
[0.000] [-0.001] [-0.001]
2018 0.796 0.542 0.508
[0.005] [0.003] [0.003]
2019 0.216 -0.046 -0.079
[0.002] [-0.000] [-0.001]
Constant 8.997*** 4.259*** 3.315***
Observations 17,863 17,863 17,863
Number of firms 6,151 6,151 6,151
R2 0.20 0.23 0.23
F 136.8 126.2 112.5
Notes: 1. Normalized beta coefficients are in brackets.
2. *p <0.1, **p <0.05, ***p <0.001.
Fig. 1. Predicted Export Performance as a Function of Market Diversity

129
5. Discussion and Implications
This study investigates the curvilinear (inverted U-shaped) relationship between market
diversity and export performance of exporting companies. To test our hypothesis, we analyze
panel data of 17,863 observations between 2015 and 2019 in South Korea. Our analysis found
that as market diversity increases, the export performance also increases because exporters
can diversify and reduce financial risks in export markets. However, the relationship between
the two changes as it moves upward. As it peaks, the entry costs increase due to the high
market diversity, therefore, the costs outweigh the benefits, leading to decrease in export
performance. Therefore, we can conclude that there is a curvilinear inverted U-shaped
relationship between market diversity and export performance.
This study provides theoretical and practical contributions and implications by analyzing
the curvilinear relationship between market diversity and export performance, which has
been relatively overlooked in the field of trade.
First, this study expands the export and trade literature by focusing on the effect of market
diversity in discussing the export performance. When conducting research to identify
antecedents of export performance, most studies so far have mainly focused on the effect of
the subjective and self-reported internal characteristics of an exporting company because data
have been collected through questionnaires. For that reason, the effect of corporate-level
objective variables, like the export market portfolio, or market diversity on export
performance, has not received enough attention yet. In fact, the majority of exporting
companies export to more than one country. According to our panel data of 17,863, only
2,274 exporting companies, accounting for about 12.73% of the total, export to a single
country. The remaining 87% of firms export to more than one market, with an average of 12-
13 exporting countries. In this paper, we highlighted market diversity, which has not been
sufficiently addressed in the trade literature, as a main variable explaining export
performance. By doing so, we can secure a deeper understanding of the impact of objective
and observable characteristics of exporters on performance.
Second, this study provides insight to the trade literature, by examining the curvilinear
relationship between market diversity and export performance. Many scholars, especially in
the international business field, have mainly assumed a positive linear relationship between
market (international) diversity and the corporate financial performance, such as Return on
Asset (ROA) and Tobin’s Q, considering only the decrease in market risks as the number of
overseas markets increases (e.g., Agmon and Lessard, 1977; Hisey and Caves, 1985). On the
other hand, some scholars in the field of trade recently found that market diversity increases
the cost of companies, specifically the entry and management costs (e.g., Chaney, 2008; Eaton,
Kortum and Kramarz, 2004; Moxnes, 2010). However, both lines of research have limitations
that assume a linear relationship between market diversity and corporate performance. The
mixed impact of market diversity on corporate performance might indicate that it might not
be appropriate to assume a linear relationship between market diversity and corporate
performance. In this paper, we have found that this relationship is non-linear, breaking away
from the existing assumption of a linear relationship. Thus, by proposing a way to incorporate
the costs and benefits aspects of market diversity, this study analyzes further the relationship
between market diversity and performance, especially regarding export.
Third, this study adopts more favorable data to distinguish the relationship between market
diversity and export performance. We performed an empirical analysis by using objective
secondary data at the corporate level, which is distinct from other studies that have conducted
empirical analyses with data obtained through questionnaire surveys. Since most of the
survey data for hypothesis testing in the trade literature are obtained through a one-time
130
survey, it is difficult to assume a causal relationship between independent and dependent
variables and avoid the issues of endogeneity because the independent and dependent
variables reflect the same point in time. In addition, it could be exposed to a risk of sample
selection bias. Panel data used in this study were established for all domestic exporters from
2015 to 2019. Through panel data analysis, we can observe how firms change their behaviors
at various points in time and deal with the firm-specific differences and time-fixed effects in
the error term capturing unobserved factors. Therefore, some of the limitations of survey data
can be overcome. The number of observations in the panel data we used for hypothesis testing
is about 20,000, which is enough to reflect real business situations and generalize the results
of this study.
Forth, this study provides practical contributions and implications to exporting companies
and their managers. Based on our finding, continuous increase of the export market base to
reduce the financial risks of markets could have a negative impact on export performance due
to the costs of managing and entering numerous markets. Accordingly, managers of
exporting companies consider the costs that must be incurred to export to a specific country
and the benefits involved therein, and they should adjust and evaluate the geographical
market diversity to an optimal level. Then, the companies will be able to achieve a sustainable
competitive advantage. Our findings show that the way exporting companies manage their
export countries affects the company’s export performance, thus providing valuable practical
implications.
Despite the various contributions and implications of this work, there are some limitations
and avenues for future research.
First, we chose product diversity of exporting companies as one of the control variables.
However, there could be some possible interaction effects between market and product
diversity, affecting export performance (e.g., Tallman and Li, 1996). This study did not
address the contingent factors in order to focus on the main effect on the curvilinear
relationship between market diversity and export performance. However, if a follow-up study
considering this issue is conducted, it will be useful in providing more insight into the existing
literature.
Second, market diversity in this study is defined as the number of countries to which a
company exports. However, it would be essential to define market diversity differently in the
context of global economic integration. For example, it would be of great significance to study
the geographic market diversity by dividing it into regions, such as Southeast Asia, Middle
East, Latin America, European Union, etc., rather than by countries. Countries tied to similar
regions are likely to share similar characteristics. So, even if the number of exporting countries
of an exporter increases within the same region, the benefits of risk reduction might be limited
and costs for entering and managing the market might not increase significantly. Future
research that reflects this issue could provide more contributions and implications.
References
Agmon, T. and D. R. Lessard (1977), “Investor Recognition of Corporate International
Diversification”, The Journal of Finance, 32(4), 1049-1055.
Belich, T. J. and A. J. Dubinsky (1995), “Factors Related to Information Acquisition in Exporting
Organizations”, Journal of Business Research, 33(1), 1-11.
Benito, G. R. and L. S. Welch (1997), “De-Internationalization”, Management International Review,
1997/2, 7-25.
Cadogan, J. W., C. C. Cui and E. K. Y. Li (2003), “Export Market-Oriented Behavior and Export
131
Performance”, International Marketing Review, 20(5), 493-513.
Chaney, T. (2008), “Distorted Gravity: The Intensive and Extensive Margins of International Trade”,
American Economic Review, 98(4), 1707-1721.
Chang, Sea-Jin and Jae-Bum Hong (2000), “Economic Performance of Group-Affiliated Companies
in Korea: Intragroup Resource Sharing and Internal Business Transactions”, Academy of
Management Journal, 43(3), 429-448.
Cooper, R. G. and E. J. Kleinschmidt (1985), “The Impact of Export Strategy on Export Sales
Performance”, Journal of International Business Studies, 16(1), 37-55.
Das, S., M. J. Roberts and J. R. Tybout (2007), “Market Entry Costs, Producer Heterogeneity and
Export Dynamics”, Econometrica, 75(3), 837-873.
Eaton, J., S. Kortum and F. Kramarz (2004), “Dissecting Trade: Firms, Industries, and Export
Destinations”, American Economic Review, 94(2), 150-154.
Francis, J. and C. Collins-Dodd (2000), “The Impact of Firms’ Export Orientation on the Export
Performance of High-Tech Small and Medium-Sized Enterprises”, Journal of International
Marketing, 8(3), 84-103.
Hanson, D., M. A. Hitt, R. D. Ireland and R. E. Hoskisson (2016), Strategic Management:
Competitiveness and Globalisation, South Melbourne, Australia: Cengage.
Hisey, K. B. and R. E. Caves (1985), “Diversification Strategy and Choice of Country: Diversifying
Acquisitions abroad by US Multinationals, 1978–1980”, Journal of International Business
Studies, 16(2), 51-64.
Johansson, S. and C. Karlsson (2007), “R&D Accessibility and Regional Export Diversity”, The
Annals of Regional Science, 41(3), 501-523.
Katsikeas, C. S., L. C. Leonidou and N. A. Morgan (2000), “Firm-Level Export Performance
Assessment: Review, Evaluation, and Development”, Journal of The Academy of Marketing
Science, 28(4), 493-511.
Katsikeas, C. S., N. F. Piercy and C. Ioannidis (1996), “Determinants of Export Performance in a
European Context”, European Journal of Marketing, 30(6), 6-35.
Korea Customs Service (KCS) (2018), Statistics for Business Trade Activities in 2018, Daejeon,
Korea: Author.
KOSTAT (2019), Trade Statistics by Business Characteristics in 2019, Daejeon, Korea: Author.
Lee, Seung-Hyun and Chris Chang-Wha Chung (2007), “Geographic Diversification: Risk Reduction
or Operational Flexibility”. In S. B. Tallman (Ed.), A New Generation in International Strategic
Management, Chaltomham, UK: Edward Elgar Publishing.
Leonidou, L. C., C. S. Katsikeas and S. Samiee (2002), “Marketing Strategy Determinants of Export
Performance: A Meta-Analysis”, Journal of Business Research, 55(1), 51-67.
Lind, T. J. and H. Mehlum (2010), “With or Without U? The Appropriate Test for a U-Shaped
Relationship”, Oxford Bulletin of Economics and Statistics, 72(1), 109-118.
Mahoney, D., M. Trigg, R. Griffin and M. Pustay (2001), International Business: A Managerial
Perspective, Melbourne, Australia: Pearson Education Australia.
Mansi, S. A. and D. M. Reeb (2002), “Corporate Diversification: What gets Discounted?”, The
Journal of Finance, 57(5), 2167-2183.
McGuinness, N. W. and B. Little (1981), “The Influence of Product Characteristics on the Export
Performance of New Industrial Products”, Journal of Marketing, 45(2), 110-122.
Moxnes, A. (2010), “Are Sunk Costs in Exporting Country Specific?”, Canadian Journal of
Economics/Revue Canadienne D'économique, 43(2), 467-493.
Robertson, C. and S. K. Chetty (2000), “A Contingency-Based Approach to Understanding Export
Performance”, International Business Review, 9(2), 211-235.
132
Sousa, C. M. (2004), “Export Performance Measurement: An Evaluation of the Empirical Research
in the Literature”, Academy of Marketing Science Review, 2004(9), 1-22.
Sousa, C. M., F. J. Martínez-López and F. Coelho (2008), “The Determinants of Export Performance:
A Review of the Research in the Literature between 1998 and 2005”, International Journal of
Management Reviews, 10(4), 343-374.
Tallman, S. and J. Li (1996), “Effects of International Diversity and Product Diversity on the
Performance of Multinational Firms”, Academy of Management Journal, 39(1), 179-196.
Wheeler, C., K. Ibeh and P. Dimitratos (2008), “UK Export Performance Research: Review and
Implications”, International Small Business Journal, 26(2), 207-239.
Wilson, B. D. (1980), “The Propensity of Multinational Companies to Expand through Acquisitions”,
Journal of International Business Studies, 11(1), 59-64.
Zucchella, A., G. Palamara and S. Denicolai (2007), “The Drivers of the early Internationalization of
the Firm”, Journal of World Business, 42(3), 268-280.
https://doi.org/10.35611/jkt.2020.24.8.133 133
Factors of Korea-China Product Trade

According to GVC Changes: Focused on FTA
Su-Young Kwak JKT 24(8)
Department of Global Trade, Dongguk University, Seoul, South Korea
Mun-Seong Choi
Department of International Trade, Dongguk University, Seoul, South Korea
Yong-Hwan Kim
Department of Media and Communication, Dongguk University, Seoul, South Korea
Do-Hyung Lee†
Department of Global Trade, Dongguk University, Seoul, South Korea
Abstract
Purpose – The purpose of this study is to analyze the determinants of commodity trade in Korea and
China and to examine the implications of China’s GVC shift from export to domestic market on its
impact on Korea’s trade.
Design/methodology – This study selected 30 major trading partner countries. The dependent variable
is the trade volume, and the independent variables are general economic factors such as gross domestic
product (GDP), GDP per capita, distance, and FTA.
Findings – The trade pattern of Korea’s commodities shows that GDP has a positive relationship with
trade, import, and export. Distance has a significant negative relationship with total trade, import, and
export. FTA is significant for import but it is not significant for total trade and export. The trade
pattern of China’s commodities shows that GDP has a significant positive relationship with total trade,
import, and export. Distance has a negative relationship with trade, import, and export. GDP per
capita is not significant for total trade and import, but it is significant for export. FTA is significant
for total trade and export, but it is not significant for import.
Originality/value – Existing papers were studied mainly in certain industrial sectors such as
agriculture, manufacturing, automobile industry and steel industry. This paper attempts to collects
vast amounts of data about the 30 countries of Korea and China respectively and analyzes by Random
Effect Model dividing the goods (0 to 9) in units of STIC (Rev. 4). The major contribution is that the
decision factors affecting commodity trade can be analyzed in SITC units (0-9) to obtain analysis
results that are subdivided by product group and organized by product.
Keywords: China FTA, GVC (Global Value Chain), Korea-China FTA, Korea FTA
JEL Classifications: F01, F10, F14
1. Introduction
The two countries, Korea and China, established diplomatic relations since 1992 and have
developed rapidly through multilateral cooperation such as trade, investment, and human
exchange for the past 27 years. Korea-China commodity trade increased 38 times from $6.4
billion in 1992 to $243.4 billion in 2019, achieving explosive growth exceeding the world trade
rate (by 4.2 times) during the same period. In other words, China, Korea’s fifth-largest trading
†
Corresponding author: dohyunglee@dongguk.edu
134
partner in 1992, has continued to maintain its largest trading partner since rising to the top
in 2004.
Korea-China trade volume and proportion continued to grow, and Korea’s trade surplus
with China continued to increase. With the change of “Global Value Chain” (hereinafter
referred to as GVC) in the protectionist mood, China has been transformed into a domestic-
oriented economic structure. In this regard, it has been argued that in the case of Korea, the
free trade zone could promote Korean companies to better access the Chinese domestic
market. The increase in trade and investment between Korea and China has been centered
on processing trade according to China’s investment-led growth policy, but China’s industrial
structure has recently begun to change. As the economic growth trend is rapidly shifting
toward consumption and services, changes in the trade structure between the two countries
are showing signs of transformation.
In addition, South Korea’s investment in China and vice versa fell 46.3 percent and 32.3
percent, respectively, during the first half of 2017 due to political conflicts, including China’s
opposition to the deployment of Terminal High Altitude Area Defense (THAAD). However,
to enter China’s fast-growing domestic market, increasing the proportion of exports of
consumer goods and development of products based on field surveys is essential. Therefore,
Korean companies need to come up with a new strategy to advance into China.
This study aims to analyze the factors affecting the Korea-China FTA and the amount of
trade that has been made so far to create a new strategy following China’s GVC changes.
Therefore, in order to analyze the factors influencing the trade value of Korea and China, 30
countries were selected for each of the major trading partners of Korea and China by gravity
model using dependent variable(trade amount), independent variables(GDP, per capita
income, distance, FTA dummy) for 2007-2015.
The contribution of this study collects vast amounts of data based on 30 countries of Korea
and China(Korea’s main partner: China, USA, Hong Kong SAR, Viet Nam, Japan, Singapore,
India, Mexico, Malaysia, Australia, Philippines, United Kingdom, Indonesia, Thailand,
Germany, Turkey, Canada, Russian Federation, Brazil, Netherlands, Norway, Italy, Poland,
France, Belgium, Spain, Chile, Peru, New Zealand, Colombia; China’s main partner: USA,
Hong Kong SAR, Japan, Rep. of Korea, Germany, Viet Nam, India, Netherlands, United
Kingdom, Singapore, Malaysia, Russian Federation, Australia, Thailand, Mexico, Indonesia,
United Arab Emirates, Philippines, Canada, Italy, France, Brazil, Pakistan, Turkey, Chile,
Saudi Arabia, New Zealand, Peru, Iceland, Switzerland) respectively and analyzes through
Random Effect Model dividing the goods (0 to 9) in units of STIC (Rev. 4). The major
contribution is that the decision factors affecting commodity trade can be analyzed in SITC
units (0-9) to obtain analysis results that are subdivided by product group and organized into
product.
In addition, this study also aims to present implications for the impact of China’s GVC
fluctuation, which has been converted from export to domestic market, on Korea’s trade.
2.1. Previous Studies of the Related Korea-China FTA
Kang Jun-Gu, Kim Tae-Jin and Shim Seung-Jin (2017) focused on service trade using the
WIOT proposed by WIOD, and proposed the policy implications by analyzing the effects of
the transition as well as the creation of the value-added trade. Kim Gyu-Jeong and Kim Min-
Ho (2016) suggested that Korea-China FTA seems to have impact on explosive increase in
135
trade volume between Korea and China and is necessary to set up a detailed export strategy
for each of the five strategic industries of Jeollabuk-do. In this regard, we analyzed the
competitiveness and linkage among the five major industries of Jeonbuk using the Trade
Specialization Index (TSI) and Trade Association Index (GL). Kong Su-Jin (2015) considered
the relationship between bilateral and trilateral agreements, taking into account the possibility
that complex bilateral agreements would coexist in the context of the Korea-China-Japan
trilateral agreement. To this end, the contents of each agreement were examined and was
reviewed the basic structure of the bilateral and trilateral investment agreements of the three
countries in Northeast Asia.
Baik Il (2015) combined the Marxist warnings on the expansion of non-discriminatory
trade with the competition and risk of trade expansion during the FTA period. He also
concentrated on the characteristics of international trade between the two countries and
achieving mutually beneficial growth through the number of cases is impossible. Seo Chang-
Bae (2016) thought this may be an opportunity factor for opening up a new consumer market
of 1.4 billion population and $10 trillion in the Korea-China FTA, but it can be a threat factor
considering China’s competitiveness is rising sharply.
Lee Hack-Chun and Ko Zoon-Ki (2013) reviewed the status of trade remedies, and the
status and environment of trade between Korea, China and Japan, and lastly the FTA policies
among the three countries. Based on this, Korea-China-Japan trade liberalization, the FTA
benefit, and the ultimate promotion plan of the three countries will be developed in the future.
Zhou Kun (2009) analyzed the anticipated effects carefully to promote long-term win-win
strategies and to open up vulnerable industries in stages. Duan Pei-Pei (2009) presented the
short-term effect and the long-term economic impact of the Korea-China FTA. Also, political
and diplomatic influences are presented. Yeo Taek-Dong and Choi Eui-Hyun (2012) suggest
that the effects of the Korea-China FTA on the region are different depending on the detailed
industrial characteristics of the region. Kim Dae-Jeong (2015) argued that the expansion of
trade with China had a ripple effect of economic growth (1.5% ~ 2.1%). An Ni (2019)
investigated the export effect of China’s agricultural products on the Korea - China FTA, and
in order to strengthen the bilateral agricultural and international competitiveness of both
agricultural products, both countries will need to cooperate with each other. Chang Eun-Gap
(2019) empirically analyzed the case study method and found the financial characteristics of
individual companies’ relation to the stock price. Yu Min-Gyan (2019) investigated and
utilized primary data such as basic statistical data of China and Korea and analyzed secondary
data such as research reports and monographs of related research institutes. Based on related
articles and data analysis, this paper analyzed the current status of start-ups by Chinese college
student at the present stage and the effect Korea-China FTA has on the start-up. Liang Yu-Jie
(2019) studied a comparative analysis of the online trading industry and related trade between
the two countries in terms of industry scale, payment methods, logistics, customs clearance,
and legal & policy aspects. Combined with the signing of the Korea-China FTA agreement,
the change of trade dependence after the signing of the agreement was calculated as well as
the impact of the agreement on online transactions between the two countries. Li Yue-Xin
(2019) analyzed quantitative analysis of the effects of the Korea-China FTA on Chinese
agricultural exports by conducting the trade gravity model based on qualitative analysis. Lim
Byung-Ho (2019) aimed to analyze the economic effects of FTA by introducing the concept
of value-added export and evaluate the influence toward GVC countries. The outcome of
FTA was analyzed concerning the increase of exports and value-added exports in regions
where GVC is deepened, such as East Asia, and the possibility of its ripple effects that will
spread across multiple countries.
This paper intends to concentrate on the effects of the Korea-China FTA on commodity
136
trading. In particular, we will select 30 major countries that have already concluded FTAs
between Korea and China, and analyze the effect of the FTA with the gravity model. The
empirical analysis results draw implications for Korea and China’s product trade.
2.2. Previous Studies of the Related GVC

GVC is a division of labor between countries, and international division of labor is carried
out in stages of production and sales of goods beyond the production of final goods within
one country. GVC is divided into seven stages such as R&D, design, procurement, manu-
facturing, logistics, marketing, and services, which are necessary to complete production and
sales, asthe unit fortrade transitioned from goods to tasks (Oh Dong-Yoon, 2018). Previous
studies related to GVC are as follows.
KOTRA (2015) drew implications for Korean companies’ participation in GVC through
an understanding of the recent mega FTA, which is the flow of the recent trade agreement, as
well as expanding the production network with GVC.
Park Jung-Soo et al. (2019) identified major domestic service sectors that can cooperate
with each other based on the analysis of the trade structure between Korea and China, and
develops industrial cooperation plans between the two countries for their overseas expansion
in various fields through a qualitative method. KOTRA (2017) confirmed the expansion of
Southeast Asia while the quantitative expansion of China’s trade continued. Considering that
many Korean companies have established local production systems in China, it is necessary
to use the expansion of China’s exports to Southeast Asia as an opportunity to expand exports
of Chinese-made Korean companies’ products to Southeast Asia. KDI (2018) attempted to
explore ways to develop cooperation between Korea and developing countries using GVC,
which has been attracting attention as a major means to support sustainable development in
developing countries.
KITA (2020) explained that emerging countries’ trade dependence on advanced countries
fell from 66% in 2000 to 52% in 2017. As the industrial structure of emerging countries has
advanced, dependence on imports and exports to advanced countries has declined in almost
all stages of processing, including consumer goods, capital goods, and intermediate goods,
and transactions between emerging countries have become more active.
Kang Nae-young and Kang Seong-Eun (2020) said that the global value chain has slowed
since 2011. This is because the vertical division of labor between advanced and emerging
countries has weakened as emerging countries, which used to function as global production
plants, grew rapidly and increased the self-sufficiency rate of intermediate goods. GVC
participation rates of advanced and new countries have remained at 59% and 48%,
respectively, since 2011. However, the service industry’s position in the global value chain is
growing. They explained that it is necessary to enhance the competitiveness of export
products not only to increase the value of the service industry, but also to promote the
convergence between service and manufacturing industries, focusing on the service industry
related to the manufacturing industry. Cho Jae-Han and Kim In-Cheol (2020) is concerned
about the global economic downturn and domestic economic shock due to Covid-19, and
believes it is necessary to understand the changes in the global value chain and to take effective
mid to long term countermeasures in order to minimize threats and find opportunities. In
response to the reorganization of the global supply chain, expanding the attraction of
domestic and foreign corporate investment is suggested, as well as focusing on structural
innovation and quality improvement of foreign economic policy goals and capabilities, and
linking export companies’ digital capabilities and investment policies.
Existing papers have been mainly studied in specific industries such as agriculture,
137
manufacturing, automotive industry and steel industry. This paper collects a vast amount of
data on 30 countries in Korea and China and analyzed products (0 ~ 9) in a Random Effect
Model divided by STIC (Rev. 4) units. In addition, this study attempted to establish a new
strategy along with the change of GVC in China by deriving factors that affect the Korea-
China.
3. Methodology
3.1. Research Method
In order to achieve the purpose of the study, we analyzed the effect of the Korea-China FTA
on the future trade effects of goods export and imports in Korea and China. Related data were
collected from sites such as KITA, Uncomtrade and UNCTAD. Data from the Uncomtrade
site was used to collect data from 2007 to 2015 of both countries and used panel data analysis.
The econometric analysis using panel data means that both the time series analysis and the
cross-sectional area analysis are carried out simultaneously.
Since both time series data and cross-sectional data can be used, there is an advantage that
the estimation error occurring in the time series process and the estimation error occurring
in the data of the regional unit can be controlled. In addition, the panel model has the greatest
significance to overcome the limitations of missing variables that are not included as
independent variables even though they have a significant effect on dependent variables.
The major advantages and disadvantages of panel analysis are as follows.
First, dynamic relationships can be estimated because objects are repeatedly observed.
Second, the unobserved heterogeneity of groups can be considered in the model. Third, it can
mitigate efficient estimator and multicollinearity problems by providing various information
and variable volatility. On the other hand, the disadvantages of panel analysis are as follows.
First, there are difficulties in collecting data. Second, there may be a correlation between panel
groups. Third, there is a disadvantage that the length of the time variable is short when the
individual is a panel group.
In addition, STIC (REV. 4) was divided into 10 categories (0 ~ 9) to analyze the trade,
export and import.
3.2. Research Model

In this study, data collected during the nine years (2007-2015) were used for 30 countries
respectively among the major trading countries from Korea and China. Dependent variable
uses the trade volume of imports and exports between Korea and China. As the independent
variables uses the GDP, population, distance, and FTA dummy variables. The model used in
this study is (1).
ln௜௝௧ ଴ ଵ ln
௜௧
௝௧ ଶ ln
௜௧
௝௧ ଷ ௜௝ ସ ௧
௜௝௧ 1
௜௝௧ !""#$ %& ' &()* # ")* +&,&& -*&!. #
*#/ )*&*0 "&
௜௧
௝௧
1)2 '
+&,&& -*& !. # *#/ )*&*0 "&

௜௧
௝௧ 1)2 '
)&* 2) +&,&& -*&!. #
*#/ )*&*0 "&
௜௝ 32& +&,&& -*&!. # *#/ )*&*0
138
௧

Korea’s or China’s import and export patterns can be analyzed if commodity value of
import (Equation (2)) and commodity value of export (Equation (3)) are substituted instead
of commodity value of export and import as dependent variables.
ln !௜௝௧ "଴ # "ଵ ln$%&'௜௧ ( %&'௝௧ ) # "ଶ ln$'%&'௜௧ ( '%&'௝௧ ) # "ଷ & *௜௝ # "ସ ௧
# +௜௝௧ ,- 2
!௜௝௧
/ 0 1

ln,(௜௝௧ "଴ # "ଵ ln$%&'௜௧ ( %&'௝௧ ) # "ଶ ln$'%&'௜௧ ( '%&'௝௧ ) # "ଷ & *௜௝ # "ସ ௧
# +௜௝௧ ,- 3
3௜௝௧
/ 0 1 4

The independent variables used in the study are the variables used in many previous studies
as shown in Table 1. The previous study using gravity models included cultural and other
variables such as language and race, but this study was excluded because it was aimed at
analyzing product trade patterns and understanding the effects of FTAs.
Table 1. Independent Variables of Antecedent Research

Variables Researcher
GDP Choi Bong-Ho (2005), Ham Shee-Chang (1997), Nam Ki-Chan, Nam Hyung-
Sik, and Kang Dal-Won (2013)
Park Ho, Jang Hyun-Mi, Kim Sang-Youl (2016), Park Jae-Jin (2005),
GDP per capita Choi Bong-Ho (2005), Kang Bong-Kyung (2009), Nam Ki-Chan, Nam Hyung-
Sik, and Kang Dal-Won (2013), Park Ho, Jang Hyun-Mi, Kim Sang-Youl (2016),
Park Young-Gl, Yi Chae-Deug (2010)
Distant Choi Bong-Ho (2005), Ham Shee-Chang (1997), Kang Bong-Kyung (2009),
Nam Ki-Chan, Nam Hyung-Sik, and Kang Dal-Won (2013), Park Ho, Jang
Hyun-Mi, Kim Sang-Youl (2016),
Park Jae-Jin (2005),
FTA Kang Bong-Kyung (2009), Park Ho, Jang Hyun-Mi, Kim Sang-Youl (2016), Park
Young-Gl, Yi Chae-Deug (2010)
Table 2 shows the definitions of variables used in the study and the data provided in the
UN statistics on commodity value in export and import as sources. As an independent
variable, the gross domestic product of both countries used the UN statistics (www.
uncomtrade). Distance used the data provided at http://www.cepii.fr/CEPII/en/bdd_modele/
bdd_modele.asp, and population used UNCTAD. In addition, the dummy variable, FTA, was
assigned 1 for signed and 0 for non-signed.
139
Table 2. Definitions and Sources of Variables
Variables Description Source
Commodity value of trade (export, UNCOMTRADE
TR(EX, IM) import)
GDP GDP of both countries UNCTAD
PGDP GDP per capita of both countries UNCTAD
DIST Distance CEPII
FTA Whether FTA is signed Ministry of Foreign Affairs
and Trade
As for the independent variables used in the study, GDP indicates the size of the economy
and the market size of the country. The increased GDP in both countries means that
economies of scale have comparative advantage due to increased productivity. GDP per
capita is the total annual production of real GDP divided by the total population, which is the
size of the overall production and expenditure of a country. It is an indicator of individual
income and expenditure and is a variable that can determine whether trade size is influenced
by income level. Therefore, as the GDP per capita of both countries increases, it is expected
to have a positive effect on the trade volume of commodities between two countries. Distance
is a measure based on the physical distance between the administrative capitals of two
countries. Increased distance can be regarded as a typical trade cost, which means an increase
in transport cost and time required to quantitatively measure. Therefore, as the distance
between two countries increases, it is expected to have a negative impact on the amount of
trade in commodities between them.
If the coefficients of the FTA dummy variables and economic integration variables show
positive values, this implies expansion of trade through the effect of trade creation. When
shown negative values, this implies reduction of trade through the effect of trade conversion.
The effect of trade creation is to move the production factors such as capital and labor and to
allocate resources efficiently by eliminating tariffs and non-tariff barriers through FTAs or
customs unions. This signify the expansion effect of trade as economies of scale occur. On the
other hand, the trade-off effect is the negative effect of the removal of trade barriers, which
means that commodities imported from foreign countries with low cost of ownership are
imported from countries with higher production costs. Therefore, the conclusion of an FTA
is expected to have a positive effect on trade volume.
4. Analysis
4.1. Korea-China Commodity Trade Status
Trade in goods between Korea and China increased 38 times from $6.4 billion in 1992 to
$243.4 billion in 2019, achieving explosive growth exceeding the world trade increase (4.2
times) during the same period. Over the past 27 years, as the industrial relations between
Korea and China have developed cooperatively, the dependence on commodity trade
between the two countries has sharply increased. As China’s share of trade in Korea exploded
from 4% in 1992 to 23.4% in 2016, China’s trade rankings remained fifth place in 1992 and
maintained its first place from 2004 to 2019. In China trade, Korea’s share also increased from
4.2% in 1993 to 6.8% in 2016, and as of 2019, Korea became China’s fourth largest exporter
and one largest importer. In particular, the ranking of Korea in imports from China rose from
140
4th in 1998 to 1st in 2013, surpassing Japan, and remains first place until 2019 as shown in
Table 3.
Table 3. Changes in Trade in Commodity between Korea and China

Division 1992 2004 2016 2019
China’s Trade 5th 1st 1st 1st
position in Export 6th 1st 1st 1st
Korea’s trade Import 5th 2nd 1st 1st
Korea’s 1998 2004 2016 2019
position in Trade 4th 4th 4th 3rd
China’s trade Export 5th 4th 4th 4th
Import 4th 3rd 1st 1st
Source: Korea International Trade Association (2020).
In the early days of diplomatic relations, we can see recent quick shift from simple light
industry and heavy chemical-oriented items to high value-added items related to information
and communication technology (ICT). In 1992, Korea exported steel products and textile
products to China and imported vegetable materials (feeds) and crude oil. As shown in Table
4, “inter-industry” trade which was originally the main focus recently changed to “intra-
industry” trade forms such as semiconductors and displays as the two countries’ industries
advanced.
The following are the top 10 export items that have led Korea’s exports for 10 years. As of
2019, the nation’s major export items were semiconductors, automobiles, petroleum pro-
ducts, displays, synthetic resins, ships, steel plates, wireless communication devices and
plastic products. There has been little change in the nation’s top 10 export items over the past
10 years, and only slight change in the ranking. The top 10 export items accounted for 56.1%
of the total exports (542.33 billion dollars), and the top five export items accounted for 40.7%.
A previous study on trade between Korea and China, Gyunggi Future Development
Institute (2008) divided trade structure into China export structure and import structure and
analyzed the characteristics of recent changes in the Korean economy in a comprehensive
manner. Jeon Kwang-Myung and Noh Won-Jung (2008) examined the patterns and causes
of change in the trade structure between Korea and China in terms of the development of the
industries of both countries, analyzed the impact of these structural changes on Korea’s future
imports and exports, and seeked policy implications and countermeasures. The Korea
International Trade Association (2011) analyzed the status of linkage between Korean-
Chinese cities and regions in the industry and logistics sector, and analyzed the difficulties of
major regional companies in both Korea and China. Han Sang-Wan and Cho Gyu-Rim
(2014) analyzed the structural causes of sluggish exports to China and found ways to respond
to the risk of China, which is threatened by the stable growth of the Korean economy, through
future export channels. Yang Pyung-Seop (2014) assessed the 2014 economic cooperation
between Korea and China as a focus on trade and investment in Korea-China economic
relations and presented the direction of cooperation between Korea and China. Lee Bu-
Hyeong and Chun Yong-Chan (2016) tried to analyze the phenomenon of mutual economic
dependence between Korea and China and present policy directions on the future direction
of economic exchanges between the two countries. Yang Pyung-Seop and Park Min-Sook
(2017) analyzed the changes and causes of Korea’s trade balance with China and presented
countermeasures. Nam Soo-Jung et al. (2018) presented detailed cooperation plans for major
areas in full consideration of changes in internal and external conditions surrounding the two
Table 4. Changes in the Proportion of Korea’s Top Ten Import and Export Items (Unit: %)
Export Import
Ranking
1992 2000 2019 1992 2000 2019
1 semiconductor 17.3 semiconductor 15.1 semiconductor 8.9 crude oil 14.0 crude oil 15.7 crude oil 11.7
2 car 7.9 computer 8.5 clothing 8.5 semiconductor 9.3 semiconductor 12.4 semiconductor 6.6
3 Petroleum 7.5 car 7.7 Marine 5.4 Natural gas 4.1 computer 4.9 Aircraft and 3.2
products Offshore parts
Structures and
Components
4 Automotive 4.2 Petroleum 5.3 Video 4.8 Petroleum 3.5 Petroleum 3.1 Petroleum 2.7
Parts products equipment products products products
5 Flat panel 3.8 Marine 4.9 Artificial long 4.4 Coal 2.8 Natural gas 2.4 Prime mover 2.4
display and Offshore fiber fabric and pump
sensor Structures and
Components
6 Synthetic resin 3.7 Wireless 4.6 furniture 4.2 Wireless 2.7 Semiconductor 2.3 Wood 2.2
communication communication manufacturing
device device equipment
7 Marine 3.7 Synthetic resin 2.9 Steel sheet 4.1 car 2.4 Gold, silver and 1.7 Textile and 2.1
Offshore platinum chemical
Structures and machinery
Components
8 Steel sheet 3.4 Steel sheet 2.8 computer 3.9 computer 2.3 Wired 1.6 computer 2.0
communication
equipment
9 Wireless 2.6 clothing 2.7 car 3.7 Fine Chemical 2.3 Steel sheet 1.5 Measurement 2.0
communication Raw Materials Control
device Analyzer
10 Plastic products 1.9 Video 2.1 Sound 3.2 clothing 2.2 Fine Chemical 1.4 Coal 2.0
equipment equipment Raw Materials
Note: MTI 3 digits.
Source: Korea International Trade Association (2020).
141
142
economies. Lee Chan‑Woo (2018) analyzed the changes of the trade in value added between
Korea and China, using the World input and output Database (WIOD) of 2005, 2010 and
2014. After joining the $1 trillion trade club for the first time in 2011, Korea achieved $1
trillion in trade for seven years, excluding years from 2015 to 2016. Certain items such as
semiconductors and automobiles contributed to the country’s rise to become an export
powerhouse.
4.2. Basic Statistics of Variables

In order to identify the characteristics of the variables used in the analysis before the
empirical analysis, the basic statistics of the variables in the model Table 5 and Table 6 shows
that there is a correlation between independent variables and confirmed on the matter of
multiple collinearity between independent variables in Table 7. In addition, the dispersion
expansion index of each independent variable was smaller than 10, indicating that there was
no multi-collinearity problem.
Table 5. Basic Statistics of Variables

Type N Average S.D Min. Value Max. Value
‫ܺܲܦܩ‬௄ ‫ܲܦܩ‬ 270 2.268 3.843 8.691 2.499
‫ܲܦܩ‬௜௧ ܺ௣௖௞ ‫ܲܦܩ‬ 270 649.306 556.097 19.749 2728.371
‫ܶܵܫܦ‬௜ 270 8142.973 4.516 955.651 18375.180
Table 6. Correlations
Div. C GDP pcGDP DIST FTA
C 1.000
GDP 0.523*** 1.000
pcGDP 0.032 0.341*** 1.000
DIST - 0.658*** - 0.079 0.243*** 1.000
FTA 0.112* - 0.081 - 0.090 - 0.140** 1.000
Note: *p<0.1, **p<0.05, ***p<0.001.
Table 7. VIF between Independence Variables

Div. VIF Tolerance
GDP 1.0484 0.9538
PcGDP 1.0475 0.9547
DIST 1.0317 0.9693
FTA 1.0298 0.9710
4.3. Results
Table 8 summarizes the results of the fixed effects and random effects models with
dependent variables for Korea’s trade, import, and export products. The trade patterns of
Korean commodities show that trade, import, and export between Korea and its trade
partners is in proportion to the market size of Korea and its trade partners.
The trade pattern of Korean commodities shows that GDP has a positive relationship with
total trade, import, and export which is significant at the 1% level. GDP per capita of Korea
143
and its trade partners has a negative relationship with trade, import, and export which is not
significant. Distance has a negative relationship with trade, import, and export which is
significant at the 1% level. The FTA was derived as a significant variable in import, but as an
insignificant variable in total trade and export.
Table 8. Korean Commodity Analysis Results

Trade Import Export
Division Random Random Random
Fixed Effect Fixed Effect Fixed Effect
Effect Effect Effect
–9.886 14.165 –8.083 11.146 –26.889 13.946
C
(–1.254) (4.774) (–0.683) (2.809) (–2.302) (4.256)
1.197*** 0.471*** 1.067* 0.485*** 1.977 0.504***
GDP
(3.198) (5.023) (1.899) (3.865) (3.564) (4.845)
–0.857** –0.047 –0.729 –0.064 –1.731 –0.099
PcGDP
(–2.061) (–0.455) (–1.168) (–0.457) (–2.809) (–0.862)
–1.001*** -0.809*** –1.119***
DIST – - -
(–5.154) (–3.141) (–5.330)
0.008*** 0.027 0.160*** 0.172*** –0.070 –0.030
FTA
(0.263) (0.902) (3.334) (3.809) (–1.473) (–0.676)
Obs. 270 270 270 270 270 270
Adj. R2 0.984 0.456 0.973 0.337 0.970 0.301

F-stat. 521.503*** 57.323*** 298.890*** 35.201*** 272.058 29.973***
H-M 4.559 4.025 7.356*
Notes: 1. *p<0.1, **p<0.05, ***p<0.001.
2. Figures in parenthesis are t-statistics.
Table 9 showed the results of Korea’s trade by item. Item 6 (Manufactured goods classified
chiefly by material) is affected by GDP, GDP per capita, distance, and FTA. Item 5 (Chemicals
and related products) and item 7 (Machinery and transport equipment) are affected by GDP,
distance and FTA, except GDP per capita. All items were affected by GDP excluding number
3 (Mineral fuels, lubricants and related materials) and number 9 (Commodities and
transactions not classified elsewhere in the SITC). GDP per capita affects only item 5
(chemicals and related products), while the remaining items have no effect. The distance
affected all items except item 0 (food and live animals), item 4 (Animal and vegetable oils, fats
and waxes). FTA affected only the items 0 (food and live animals), 3 (mineral fuels, lubricants
and related materials), 5 (chemicals and related products), 6 (Manufactured goods classified
chiefly by material) and 7 (machinery and transportation equipment).
Looking at the Korean import by item in Table 10, import volume was influenced by GDP,
excluding item 9 (Commodities and transactions not classified elsewhere in the SITC). GDP
per capita affects only item 7 (machinery and transport equipment) and not all items. The
distances affected only items 3, 5, 6, 7, 8, and 9, and FTA affected only items 0, 5, 6, 7 and 8.
Item 7 (machinery and transport equipment) was wholly influenced by four variables (GDP,
GDP per person, distance, FTA). Items 5 (chemicals and related products), 6 (Manufactured
goods classified chiefly by material), and 8 (Miscellaneous manufactured articles) were
affected by three variables except GDP per capita.
144
Table 9. Korea’s Trade Analysis Result by Item (Random Effect Model)
Item. 0 1 2 3 4 5 6 7 8 9
Mineral
Crude Animal and Commodities
fuels, Chemicals Manufactured Machinery
Food and Beverages materials, vegetable Miscellaneous and transactions
lubricants and related goods classified and transport
Commodity live and tobacco inedible, oils, fats manufactured not classified
and related products chiefly by equipment
animals (0) (1) except fuels and waxes articles (8) elsewhere in the
materials (5) material (6) (7)
(2) (4) SITC(9)
(3)
0.704*** 0.982*** 0.680*** 0.403 0.537* 0.592*** 0.426*** 0.426*** 0.702*** -0.079
GDP
(4.672) (4.360) (3.603) (1.318) (1.918) (6.502) (4.428) (4.049) (5.278) (-0.259)
-0.179 0.157 -0.275 0.095 -0.138 -0.159 -0.214** 0.032 -0.178 0.316
PcGDP
(-1.070) (0.631) (-1.311) (0.282) (-0.444) (-1.575) (-2.006) (0.271) (-1.205) (0.938)
-0.556* -1.132** -0.146 -1.723*** -0.768 -1.170*** -1.059*** -1.148*** -1.405*** -2.833***
DIST
(-1.782) (-2.558) (-0.370) (-2.965) (-1.393) (-6.337) (-5.373) (-5.389) (-5.247) (-4.838)
0.254*** 0.256 0.157*** 0.406* 0.105 0.292*** 0.122*** -0.030*** 0.049 0.071
FTA
(5.210) (1.981) (2.824) (1.748) (0.654) (7.842) (3.465) (-0.676) (0.828) (0.331)
Notes: 1. *p<0.1, **p<0.05, ***p<0.001.
Table 10. Korea’s Import Analysis Result by Item (Random Effect Model)
Div. 0 1 2 3 4 5 6 7 8 9
Mineral
Crude Animal and Commodities
fuels, Chemicals Manufactured Machinery
Food and Beverages materials, vegetable Miscellaneous and transactions
lubricants and related goods classified and transport
Commodity live and tobacco inedible, oils, fats manufactured not classified
and related products chiefly by equipment
animals (0) (1) except fuels and waxes articles (8) elsewhere in the
materials (5) material (6) (7)
(2) (4) SITC(9)
(3)
0.710*** 1.143*** 0.671*** 0.826** 0.530* 0.643*** 0.449*** 0.409** 0.560*** 0.275
GDP
(4.487) (3.786) (2.885) (2.124) (1.783) (4.486) (3.218) (2.218) (3.434) (0.586)
-0.255 0.145 -0.304 -0.320 -0.246 0.117 -0.287* 0.213* -0.062 0.743
PcGDP
(-1.454) (0.433) (-1.180) (-0.750) (-0.749) (0.741) (-1.857) (1.045) (-0.342) (1.443)
-0.183 -0.552 0.045 -1.547** -0.420 -1.065*** -0.818*** -2.360*** -1.827*** -2.763***
DIST
(-0.564) (-0.932) (0.093) (-2.150) (-0.728) (-3.789) (-2.923) (-6.410) (-5.560) (-3.218)
0.314*** 0.335* 0.079 0.372 0.125 0.247*** 0.140** 0.253*** 0.229*** 1.153**
FTA
(5.630) (1.950) (1.110) (1.042) (0.660) (3.002) (2.199) (2.865) (3.199) (2.302)
Notes: 1. *p<0.1, **p<0.05, ***p<0.001.
145
Looking at the results of Korea’s export by item in Table 11, 4 variables (GDP, GDP per
capita, distance, FTA) influenced item 5 (chemical products and related products) and item
6 (Manufactured goods classified chiefly by material). All items were affected by all three
variables except GDP per capita excluding item 2 (Crude materials, inedible, except fuels) and
item 7 (machinery and transport equipment). Item 8 (Miscellaneous manufactured articles)
were affected by all three variables except the FTA dummy variable. In case of Korea’s exports,
all items were affected by distance.
Table 12 summarizes the results of the random effects model with dependent variables for
China’s trade, import, and export products. The trade pattern of Chinese commodities shows
that GDP has a positive relationship with trade, import, and export which is significant at the
1% level. The trade pattern of China’s commodities shows that total trade, import, and export
are in proportion to the market size rather than people’s income patterns. Distance has a
negative relationship with trade, import, and export. GDP per capita is significant for export
but is not significant for total trade, indicating that trade patterns based on market size are
more visible than trade patterns based on income levels. FTA was a significant variable for
trade and export, but was not significant for import.
As for the results of China’s trade products by item in Table 13, in case of item
6(Manufactured goods classified chiefly by material) for trade was affected 4 variables (GDP,
GDP per capita, distance, FTA). Item 5 (chemicals and related products) was affected GDP,
GDP per capita, distance. Item 3 (Mineral fuels, lubricants and related materials) was affected
excluding GDP per capita. GDP has positive effect on trade for all items except item 4 (Animal
and vegetable oils, fats and waxes) and item 9 (Commodities and transactions not classified
elsewhere in the SITC).
Looking at China’s imports by item in Table 14, item 0 (food and live animals) was affected
by three variables (GDP per person, distance, and FTA), excluding GDP. Item 3 (Mineral
fuels, lubricants and related materials) was affected by three variables (GDP, distance, FTA),
excluding GDP per capita. Item 6 (Manufactured goods classified chiefly by material) was
affected by three variables (GDP, GDP per capita, FTA) except for distance.
Looking at China’s exports by item in Table 15, item 5 (chemicals and related products)
was affected by all variables (GDP, GDP per capita, distance, FTA). Item 2 (Crude materials,
inedible, except fuels), No. 3 (Mineral fuels, lubricants and related materials), No. 6
(Manufactured goods classified chiefly by material), No. 7 (machinery and transportation
equipment) were affected by 3 Variables (GDP, GDP per capita, distance) except for FTA. In
the case of distance, all items were affected.
5. Conclusion and Implications

This study analyzed the FTA and the trade status between Korea and China, and also, the
trade volume changes resulting from it. We examined the implications of China’s GVC shift
from export to domestic market and its impact on Korea’s trade.
5.1. Conclusion
The main findings of this study are as follows. First, as a result of the analysis of Korea’s
trade (import and export commodity), the dependent variables are classified as the random
effect model. As the GDP is influenced by trade, import and export and its trade patterns of
Korean commodities are in proportion to the market size rather than the national income
pattern. In the case of export, the distance is proportional to the assumption of the gravity
146
Table 11. Korea’s Export Analysis Result by Item (Random Effect Model)
Div. 0 1 2 3 4 5 6 7 8 9
Crude Mineral fuels, Chemicals Machinery Commodities
Animal and Manufactured
Food and Beverages materials, lubricants and and and Miscellaneous and transactions
vegetable goods classified
Commodity live animals and tobacco inedible, related related transport manufactured not classified
oils, fats and chiefly by
(0) (1) except materials products equipment articles (8) elsewhere in the
waxes (4) material (6)
fuels (2) (3) (5) (7) SITC(9)
GDP 0.690*** 1.190*** 0.386 0.690*** 1.190*** 0.386 0.690*** 1.190*** 0.386 0.690***
(7.989) (7.133) (1.560) (7.989) (7.133) (1.560) (7.989) (7.133) (1.560) (7.989)
PcGDP - - - - - - - - - -
DIST -2.008*** -2.133*** -2.462*** -2.008*** -2.133*** -2.462*** -2.008*** -2.133*** -2.462*** -2.008***
(-5.276) (-5.070) (-3.159) (-5.276) (-5.070) (-3.159) (-5.276) (-5.070) (-3.159) (-5.276)
FTA 0.136* 0.326* 0.248 0.136* 0.326* 0.248 0.136* 0.326* 0.248 0.136*
(1.722) (1.764) (1.015) (1.722) (1.764) (1.015) (1.722) (1.764) (1.015) (1.722)
Notes: 1. *p<0.1, **p<0.05, ***p<0.001.
Table 12. China’s Commodity Analysis Results

Division Trade Import Export
Fixed Effect Random Effect Fixed Effect Random Effect Fixed Effect Random Effect
C –53.712 10.850 12.005 10.235 –109.088 5.607
(–3.469) (2.391) (0.527) (1.933) (–7.600) (1.212)
GDP 2.870 0.537*** 0.183 0.450** 5.073*** 0.779***
(4.602) (3.589) (0.199) (2.571) (8.774) (5.115)
–2.678*** –0.227 0.149 –0.132 –4.989*** –0.477***
PcGDP
(–4.087) (–1.447) (0.154) (–0.717) (-8.212) (-2.984)
-0.761*** -0.557* -0.928***
DIST – - -
(–2.970) (–1.882) (–3.533)
0.088 0.122** 0.143 0.136 0.038 0.103*
FTA
(1.412) (2.005) (1.556) (1.521) (0.662) (1.825)
Obs. 216 216 214 214 216 216
Adj. R2 0.984 0.740 0.967 0.596 0.990 0.737
F-stat. 507.718*** 153.622*** 242.904*** 79.571*** 802.791*** 151.602**
H-M 20.850*** 0.130 70.190***
Notes: 1. *p<0.1, **p<0.05, ***p<0.001.
Table 13. China’s Trade Analysis Result by Item (Random Effect Model)
Div. 0 1 2 3 4 5 6 7 8 9
Crude Mineral fuels, Commodities
Animal and Chemicals Manufactured
Food and Beverages materials, lubricants and Machinery Miscellaneous and transactions
vegetable and related goods classified
Commodity live and tobacco inedible, related and transport manufactured not classified
oils, fats and products chiefly by
animals (0) (1) except fuels materials equipment (7) articles (8) elsewhere in the
waxes (4) (5) material (6)
(2) (3) SITC(9)
GDP 0.320* 0.289 0.451** 0.473* 0.102 0.702*** 0.611*** 0.581*** 0.666*** 0.391
(1.917) (1.037) (2.493) (1.673) (0.315) (4.777) (4.328) (3.043) (3.528) (0.760)
PcGDP 0.138 0.255 -0.203 -0.187 0.259 -0.388** -0.338** -0.321 -0.222 -0.252
(0.790) (0.872) (-1.069) (-0.632) (0.764) (-2.510) (-2.281) (-1.601) (-1.117) (-0.472)
DIST -0.753*** -0.462 0.291 -1.617*** -0.262 -1.029*** -0.752*** -1.183*** -0.883*** -2.080**
(-2.687) (-0.991) (0.963) (-3.454) (-0.487) (-4.118) (-3.158) (-3.602) (-2.753) (-2.512)
FTA 0.173 0.172 0.124 -0.591** -0.247 0.109 0.227*** 0.121 0.047 -0.136
(1.644) (0.892) (0.962) (-2.321) (-1.001) (1.568) (2.888) (1.622) (0.522) (-0.185)
Notes: 1. *p<0.1, **p<0.05, ***p<0.001.
Table 14. China’s Import Analysis Result by Item (Random Effect Model)
Div. 0 1 2 3 4 5 6 7 8 9
Commodities
Food and Crude Mineral fuels, Animal and Chemicals Manufactured
Beverages Machinery Miscellaneous and transactions
live materials, lubricants and vegetable and related goods classified
Commodity and and transport manufactured not classified
animals inedible, except related materials oils, fats and products chiefly by
tobacco (1) equipment (7) articles (8) elsewhere in the
(0) fuels (2) (3) waxes (4) (5) material (6)
SITC(9)
GDP 0.505** 0.444 0.477** 0.717* 0.384 0.406* 0.567*** 0.582 0.900*** 0.596
(2.501) (1.130) (2.519) (1.672) (0.988) (1.764) (2.665) (1.538) (3.164) (0.891)
PcGDP 0.165 0.487 -0.212 -0.093 0.067 -0.135 -0.390* -0.413 -0.456 0.126
(0.778) (1.181) (-1.065) (-0.210) (0.165) (-0.560) (-1.744) (-1.039) (-1.525) (0.182)
DIST 0.043 0.357 0.439 -1.970*** 0.352 -1.125*** -0.524 -1.983*** -2.102*** -3.097***
(0.127) (0.545) (1.383) (-2.870) (0.549) (-2.867) (-1.473) (-3.090) (-4.326) (-2.765)
FTA 0.052 0.056 0.100 -1.328** -0.166 0.080 0.610*** 0.324 -0.127 -0.636
(0.348) (0.179) (0.770) (-2.055) (-0.435) (0.776) (4.087) (1.728) (-1.025) (-0.623)
Notes: 1. *p<0.1, **p<0.05, ***p<0.001.
147
148
Table 15. China’s Export Analysis Result by Item (Random Effect Model)
Div. 0 1 2 3 4 5 6 7 8 9
Mineral Commodities
Crude Animal and Machinery
Food and fuels, Chemicals Manufactured and
Beverages materials, vegetable and Miscellaneous
live lubricants and related goods classified transactions not
Commodity and tobacco inedible, oils, fats transport manufactured
animals and related products chiefly by classified
(1) except fuels and waxes equipment articles (8)
(0) materials (5) material (6) elsewhere in the
(2) (4) (7)
(3) SITC(9)
GDP 0.565*** 0.082 0.558*** 0.380** 0.120 0.812*** 0.625*** 0.645*** 0.660*** -0.352
2.833 (0.355) (3.733) (2.103) (0.509) (6.912) (5.054) (3.832) (3.535) (-1.037)
PcGDP -0.226 0.257 -0.352** -0.387** 0.465* -0.474*** -0.317** -0.377** -0.219 -0.530
-1.079 (1.068) (-2.244) (-2.050) (1.898) (-3.839) (-2.441) (-2.134) (-1.116) (-1.556)
DIST -1.460*** -1.650*** -1.175*** -1.782*** -1.458*** -0.977*** -0.951*** -0.958*** -0.755** -1.753***
-4.298 (-4.369) (-4.691) (-6.027) (-3.832) (-4.902) (-4.585) (-3.317) (-2.385) (-3.402)
FTA 0.155* 0.172 0.138 0.273 -1.123*** 0.079** 0.069 0.100 0.074 -0.094
1.743 (0.704) (1.415) (1.343) (-3.450) (1.388) (0.871) (1.489) (0.804) (-0.146)
Notes: 1. *p<0.1, **p<0.05, ***p<0.001.
149
model, but the distance is not significant for import. The reason for this is that most of the
imported commodities are essential iron ore, crude oil, etc., which are irrelevant to the
distance as raw materials. The FTA was a significant variable in import, but was not a
significant variable in total trades and export. The reason for this result is the imbalance
between imports and exports. This shows that import is increasing after the conclusion of the
FTA, and needs countermeasures to solve the international imbalance.
Second, as a result of the analysis, GDP is highly influenced by China’s trade, import, and
export products. The trade pattern of Chinese commodity shows that trade, imports, and
exports is in proportion to the market size rather than the national income pattern. The
distance has a negative effect in trade, import and export of China’s commodities like the
assumption of gravity models. GDP per capita is not significant for trade indicating that trade
patterns based on market size are more visible than trade patterns based on income levels, but
the case of exports is significant. The FTA was a significant variable in trades, exports, and
the elimination of tariffs and non-tariff barriers through FTA will enable the efficient transfer
of resources and production factors such as capital and labor.
5.2. Implications and limitation

The increase in trade and investment between Korea and China has been centered on
processing trade according to China’s investment-led growth policy, but China’s industrial
structure has recently begun to change. As the trend of economic growth moved rapidly to
consumption and services, changes in the trade structure between the two countries are
visible. In this regard, the meaning of the Korea-China FTA and the change in GVC in the
process of transition from export to domestic demand in China have the following
implications for Korean companies.
First, the FTA between Korea and China could promote greater access to China’s domestic
market. In China, the successful conclusion of the Korea-China FTA has a positive impact on
China’s “Belt and Road Initiative (BRI)” policy. In addition, China’s current economic
development face a number of problems, including regional economic development
imbalances, slowing economic growth, shrinking export markets and environmental issues.
Korea-China FTA can provide China with favorable factors such as Korea’s technology,
funds, market and development experience, which helps speed up China’s economic changes
and industrial upgrades.
Second, a strategy is needed to target the Chinese domestic market of Korean exporters.
With the government’s support for contactless marketing using measures to boost con-
sumption in China, not only consumer goods but also intermediate goods companies such as
materials, parts and equipment need strategic methods to communicate with local buyers
online.
Third, expansion of its global partnership business is evident from automobile and
shipbuilding equipment to aviation, home appliances, machinery and semiconductors in
preparation for the reorganization of GVC. It is also necessary to expand the support for
export of intermediate goods by identifying alternative demands following the transition to
the Chinese supply chain.
Fourth, Korea can expand the opening of follow-up investment service trade in China’s
already signed FTA. Looking at the mutual relationship between liberalization of product
trade and service trade, it showed the characteristics of “product priority” before 2006, but
the characteristics of the expansion and development of products and services since 2008. The
major conclusion of the Korea-China FTA service sector was that China promised to
liberalize services and investment in a negative list manner for the first time.
150
This study has the following limitations although it derives meaningful implications above.
Unit value is only available until 2015 and was not able to analyze after the conclusion of
Korea-China FTA. For this reason, panel data of this paper was limited to nine years from
2007 to 2015. Because of the lack of data, the explanatory variables in the gravity model only
considered general macroeconomic factors such as GDP, per capita income, etc. For the
analysis of the Korea-China FTA, analyzing by long-term data collection is necessary for
which this is an issue that needs to be studied in the future.
References
An, Ni (2019), A Study on the Export Effect of Chinese Agricultural Products to Korea in Korea-
China FTA, (Master’s Thesis), Daejeon, Korea: Pai Chai University.
Baik, Il (2015), “A Korea-China-Japan FTA: A Win-Win View of Growth”, MARXISM, 12(1), 193-
227.
Chang, Eun-Gap (2019), A Study on the Effect of FTA on Trade Structure and Firm Value Focusing
on the Changes in the Trading Structure and Stock Price Response of Korean Stock Market
Caused by Korea vs. US FTA and Korea vs. China FTA (Doctoral Dissertation), Seoul: Hankuk
University of Foreign Studies.
Cho, Jae-Han and In-Cheol Kim (2020), Mid- to Long-term Countermeasures Following Corona 19
and Global Value Chain Reorganization, Seoul: Korea Institute for Industrial Economics &
Trade.
Choi, Bong Ho (2005), “A Study on Analysis of Trade Pattern of Korea and Effect of Regional
Economic Integration by Gravity Model; Using Panel Data Analysis”, Journal of Industrial
Economics and Business, 18(3), 1035-1052.
Duan, Pei-Pei (2009), The Influence and the Corresponding Plans of China-Korea FTA on China’s
Manufacture Industry (Master’s Thesis), Daejeon, Korea: Pai Chai University.
Gyeonggi Future Development Institute (2008), Changes in Trade Structure between Korea and
China and Its Implications, Suwon, Korea: Author.
Ham, Shee-Chang (1997), “Analysis of Global Economic Integration Trend through Gravity Model”,
The Korean Journal of Economic Studies, 43(4), 151-181.
Han, Sang-Wan and Gyu-Rim Cho (2014) “Changes in the Trade Structure between Korea and China
and Implications: Response to Changes in China’s Industrial Structure Is Urgent”, Weekly
Economic Review, 608.
Jeon, Kwang-Myung and Won-Jung Noh (2008), Changes in Korea-China Trade Structure and
Countermeasures, Seoul: Bank of Korea.
Kang, Jun-Gu, Tae-Jin Kim and Seung-Jin Shim (2017), “Trade Creation and Diversion Effect of
Korea-China FTA on Service Trade in Value Added”, Kukje Kyungje Yongu, 23(1), 1-26.
Kang, Nae-Young and Seong-Eun Kang (2020), “Changes in the Status and Implications of the
Service Industry within the Global Value Chain Structure”, Trade Focus, April.
KDI (2018), Development Cooperation Plan between Korea and Developing Countries Using Global
Value Chain, Seoul: Author.
Kang, Bo-Kyung (2009), “The Analysis Using the Gravity Model of Trade Creation Effect of
Economic Integration in East Asia”, Journal of Industrial Economics and Business, 22(4), 1961-
1974.
Kim, Dae-Jeong (2015), “A Study on the Effects and Countermeasures of the Korea-China FTA on
Domestic SMEs”, Korean Academic Society of Business Administration Integrated
Conference, 2275-2290.
Kim, Gyu-Jeong and Min-Ho Kim (2016), “Korea’s export Strategy to China after Korea-China FTA:
151
Focusing on the Jeonlabukdo”, Asia-Pacific Journal of Business & Commerce, 8(1), 35-64.
Kong, Su-Jin (2015), “Relationship between Bilateral and Trilateral Investment Agreements by
Korea, China and Japan”, The Korean Journal of International Law, 60(2), 35-66.
KITA (2011), Strategy to Revitalize China’s Economic Cooperation and Trade Investment in China:
The Korea-China Trade Structure and China's Policy Change and Strategies for Link
Cooperation in China Industry and Logistics, Seoul: Author.
KITA (2020), Paradigm Shift in Global Value Chain and Future of Korean Trade, Seoul: Korea
International Trade Association.
KOTRA (2015), Establishing a Strategy to Expand Participation in GVC using the FTA Network
(KOCHI Material, No. 15-63), Seoul: Author.
KOTRA (2017), East Asian Value Chain Changes and Implications from Changes in Trade Structure
(KOCHI Material, No. 17-014), Seoul: Author.
Lee, Bu-Hyeong and Yong-Chan Chun (2016), “The Importance and Direction of Economic
Relations between Korea and China”, Hyundai Research Institute, 16(8).
Lee, Chan‑Woo (2018), The Changes of Trade in Value-Added Between Korea and China, Kongju,
Korea: Chungnam Research Institute.
Lee, Hack-Chun and Zoon-Ki Ko (2013), “Review on the Trade Remedy According to Korea-China-
Japan FTA in the Future”, Kangwon Law Review, 38(1), 273-304.
Li, Yuexin (2019), Analyzing the Effect of the Korea-China FTA Concluded by Using the Gravity
Model on the Export of Chinese Agricultural Products and Drawing Out the Strategic
Utilization Plan (Master’s Thesis), Daejeon, Korea: Woosong University.
Liang, Yu-Jie (2019), The Impacts of Korea-China FTA on On-line Trade between Two Countries
(Master’s Thesis), Seoul: Sungkyunkwan University.
Lim, Byung-Ho (2019), A Study on the Economic Effects of FTAs as Evaluated by Value-added
Exports (Doctoral Dissertation), Incheon, Korea: In-Ha University.
Nam, Ki-Chan, Hyung-Sik Nam and Dal-Won Kang (2013), “An Analysis on the Impact of Korea-
Chile FTA on Busan Port: Focusing on the Inbound Container from Chile”, Journal of Korean
Navigation and Port Research, 37(6), 681-687.
Nam, Soo-Jung, Yun-Jong Wang, Soon-Chan Park, Seung-Chan Park, Jung-Won Cho, Jong-In Kim
et al. (2018) Policies and Vision for a New Leap Forward in Korea-China Economic
Cooperation (KIEP Research Report, No.18-30), Sejong, Korea: KIEP.
Oh, Dong-Yoon (2018), “SMEs Globalization Using Global Value Chain”, Korean Journal of
Business Administration, 31(1), 181-200.
Park, Ho, Hyun-Mi Jang, and Sang Youl Kim (2016), “Analysis of Changes in Maritime Trade
Patterns according to the Changing International Trade Environment: The Case of Major
Container Ports in Korea”, Journal of Shipping and Logistics, 32(2), 239-254.
Park, Jae-Jin (2005), “An Evaluation on Trade Impact of Open Regionalism in APEC Region”, Korea
Trade Review, 30(5), 43-67.
Park, Jung-Soo, Sang-Hyun Lee, Soon-Hak Lee, Eun-Hee Choi and Hong-Sik Lee (2019), Korea-
China Industrial Cooperation Plan for Overseas Expansion of Service Industry: Focused on
Global Value Chain Approach (Policy Material, No. 2019-364), Sejong, Korea: Korea Institute
for Industrial Economics & Trade.
Park, Young-Gil and Chae-Deug Yi (2010), “Empirical Analysis of EU Economic Integration on
Korean Trade”, The Korean Society of Contemporary European Studies, 28(3), 261-282.
Seo, Chang-Bae (2016), “Significance of Korea-China FTA and How to Use the Business in China”,
Korea-Chinese Social Science Studies, 14(1), 243-278.
Yang, Pyung-Seop (2014, December 28-29), “Foreign Economic Policy (Korea-China Economic
Relations)”, National Institute of Foreign Affairs, China Research Center Expert Workshop,
Seoul, Korea.
152
Yang, Pyung-Seop and Min-Sook Park (2017), “Changes in Korea’s Trade Balance with China and
Its Implications”, World Economy Today, 17(24), 1-20.
Yeo, Taek-Dong and Eui-Hyun Choi (2012), “Economic Effects of Korea-China FTA for the Korean
Regional Industry: Focused on the Firm-level Survey Results in Daegu Province”, Sino-Soviet
Affairs, 36(4).
Yu, Ming-Yan (2019), The Influence of the Korea-China FTA on the Business start-up of College
Students (Master’s Thesis), Busan, Korea: Dong-A University.
Zhou, Kun (2009), A Research on Economic Effects and Promotion of a Korea-China FTA (Master’s
Thesis), Seoul: Sejong University.
https://doi.org/10.35611/jkt.2020.24.8.153 153
Global Productivity and Market Structure

Implications of the US-China Trade War:
A CGE Modeling Approach JKT 24(8)
Jaewon Jung†
Department of International Trade, Dankook University, South Korea
Abstract
Purpose – As the US-China trade war intensifies and lasts long time, there is growing concern about
its potential effects on the global economy. In particular, for the countries like Korea that have a large
economic dependence on the economy of the two countries, the US-China trade war may have a great
repercussion in many ways. The aim of this paper is to investigate the global productivity and market
structure implications of the US-China trade war for Korea, as well as for other surrounding countries
and regions.
Design/methodology – In this paper, we develop a full multi-country/region multi-sector computable
general equilibrium (CGE) model of global trade incorporating heterogeneous workers and firms in
individual skill levels and used technologies. We then calibrate the model using a global Social
Accounting Matrix (SAM) dataset extracted from the recently released GTAP 10 Database, and assess
the potential effects of the US-China trade war on the aggregate real productivity and the market
structure for Korea, as well as for other surrounding countries and regions.
Findings – We show that the US-China trade war may largely affect the aggregate productivity in each
sector in each country/region, as well as the global market structure through entry and exit of firms,
which results finally in considerable changes in the industrial comparative advantage of each
country/region. Though the effects are diverse sector by sector, the results show that Korea may also
be affected significantly: concerning the real productivity implications, it is shown that the machinery
industry may be affected the most negatively; on the other hand, it is shown that the number of
exporting firms may decrease the most in the other transports industry.
Originality/value – As the US-China trade war intensifies, many studies have tried to estimate the
possible implications, and for this usually the CGE models have largely been used as the standard tool
for evaluating the impacts of changes in trade policies. Standard CGE models, however, cannot be
used to assess the global productivity and market structure implications due to the symmetric and
simplified base assumptions. This paper is the first to analyze and quantify the possible impacts of the
US-China trade war on the aggregate productivity and global market structure using a CGE model
incorporating endogenous skill-technology assignment of heterogeneous workers and firms.
Keywords: Computable General Equilibrium (CGE) Modeling, Firm/Worker Heterogeneity and

International Trade, Market Structure, Productivity, US-China Trade War
JEL Classifications: F15, F62
1. Introduction
The US-China trade war which started in 2018 is ongoing and even intensifying. Though
there have been several attempts and deals to resolve the trade disputes between the US and
China, the majority opinion seems to be that the US-China trade war will last since it is finally
a combat to win the economic supremacy in the twenty-first century. As the US-China trade
†
First and Corresponding author: jung@dankook.ac.kr
154
war continues and intensifies, there is growing concern about its potential effects on the
global economy, considering the economic importance of the two countries in the world
economy.
Since the beginning of the US-China trade war, many studies have been trying to estimate
the possible implications. For example, recently using a standard CGE model, Li, Balistreri
and Zhang (2020) find that the US-China trade war can significantly decrease the welfare of
the two countries, while other countries’ welfare may increase due to trade diversion effect.
Itakura (2020) also evaluates the impact of the US-China trade war using a recursive dynamic
CGE model of global trade, and argues that the US-China trade war can significantly decrease
the real GDP of the two countries, though the trade diversion effect may contribute to a small
positive impact on the real GDP for other countries.
On the other hand, focusing on the indirect impact on third countries through links in
global supply chains, Mao and Gorg (2020) find that the US-China trade war may hit the
most heavily other third countries, such as the EU, Canada, and Mexico, because of their close
trading relationship with the US. This result implies that a country like Korea may also be
heavily affected by the US-China trade war given its close trading relationship with China and
the US. Table 1 shows Korea’s main trading partners. China and the US have been the first
and second country both in export and import. In particular, for an export-driven, small, and
open economy such as Korea, the US-China trade war might cause fundamental changes in
the industrial comparative advantage, and thus in the long-run growth path. Related to the
topic, recent work by Jung Jae-Won and Kim Tae-Hwang (2019) suggest that mega-FTA
formation and/or participation may be a good strategic trade policy option to mitigate any
possible negative effects. However, given the large dependence of Korean economy on the
two countries, if the US-china trade war would last long time, it might have a great
repercussion on Korean economy in many ways.
China gradually opened up its economy to the rest of the world and has experienced rapid
economic growth over the last decades. Among the China’s market-oriented economic
reforms to open its door to foreign trade and investment, two monumental events would be
the normalization of diplomatic relations with the US in 1979 (after that, the US granted Most
Favored Nations status (MFN) to China in early 1980) and the China’s accession to WTO,
which made China the world’s largest manufacturing and exporting nation. At the time of
these events, most foreign firms were eager to enter into trade relations with China due to its
large market potential. A survey research by Tung (1982) reports that most US firms
perceived trade relations with China to be more important to their image as a leader in the
industry than to their overall profitability. Many researchers also highlighted the importance
of institutional quality to enhance overall trade and FDI flows (see, e.g., Angkinand and Chiu,
2011; Wei, 2000). There is now a large consensus that institutional progress in many
developing countries, such as in China, has been one of the key factors behind the rapidly
globalizing world over the last decades.
Table 1. Korea’s Main Trading Partners (2018, Rank Order of Export)

(Unit: US $ Million)
Export Import Trade Balance
1. China 162,125 106,489 55,636
2. USA 72,720 58,868 13,852
3. Vietnam 48,622 19,643 28,979
4. Hong Kong 45,996 1,997 43,999
5. Japan 30,529 54,604 -24,075
Global Productivity and Market Structure Implications of the US-China Trade War: A CGE Modeling Approach
155
Export Import Trade Balance
6. Taiwan 20,784 16,738 4,046
7. India 15,606 5,885 9,721
8. Philippines 12,037 3,569 8,468
9. Singapore 11,782 7,974 3,808
10. Mexico 11,458 5,090 6,368
11. Australia 9,610 20,719 -11,109
12. Germany 9,373 20,854 -11,481
13. Malaysia 8,994 10,206 -1,212
14. Indonesia 8,833 11,161 -2,328
15. Thailand 8,505 5,582 2,923
Source: KITA (2019).
The US and China have been the two largest nations that lead the international trade and
global economy. Given the market power and size of these two countries in the world
economy today, it is inevitable that the US-China trade war will affect considerably the global
value chains (GVCs) and trade, as well as the two countries. Concerning the current US-
China trade war, economists claim to distinguish between the strategic competition and the
economic competition since while the strategic competition might normally be a zero-sum
game, a fair economic competition could generally create a win-win outcome in the long run
(see, e.g., Liu and Woo, 2018). Economists have traditionally thought of market structure and
competition as the fundamentals for economic efficiency and productivity (see, e.g., Syverson,
2004, and references therein). For this reason, one of the main concerns of trade economists
has been how a trade reform or changes in trade environment would affect the overall market
structure and productivity by entry and exit of firms and by restructuring of trade systems.
The aim of this paper is to investigate the global productivity and market structure
implications of the US-China trade war for Korea, as well as for other surrounding countries
and regions. We assess the potential effects of the US-China trade war on the aggregate real
productivity of surrounding countries and regions, as well as on that of each sector in each
country and region. We also study how the US-China trade war may affect the global market
structure by investigating the potential impacts on the number of domestic and exporting
firms in each sector in each country/region, which is closely related to the global industrial
comparative advantage.
For this, a new CGE model is developed. Though market structure, competition and
productivity have been the central issues for policy makers, conventional CGE models have
mostly assumed a perfectly competitive world composed of homogeneous agents and
technologies, which made it impossible to study the global productivity and market structure
implications of any policy changes. As Jung Jae-Won (2020) shows, considering the close
interplay between heterogeneous agents and technologies, as well as the competition among
heterogeneous firms, should be important for policy assessment not only quantitatively but
also even qualitatively. In this paper, we first develop a new CGE model incorporating
heterogeneous workers and firms. The firm heterogeneity literature in international trade
highlights that firms are different (heterogeneous) in many aspects. Many systematic links
between the characteristics of firms and their degree of internationalization have been
156
uncovered. In particular, in terms of the productivity, there is now ample evidence that
exporting firms are more productive and use higher technologies than domestic firms (see,
e.g., Aw, Chung and Roberts, 2000; Bernard and Jensen, 1999; Clerides, Lach and Tybout,
1998; Girma, Gorg and Strobl, 2004; Helpman, Melitz and Yeaple, 2004). Based on evidence,
and differently from most previous CGE models, we model two types of firms (domestic vs.
exporting) using different technologies (low-tech vs. high-tech). Domestic firms use a low-
fixed-cost high-marginal-cost technology, while exporting firms use a high-fixed-cost low-
marginal-cost technology. Workers are differentiated in their skill levels and endogenously
sorted between technologies according to their respective comparative advantage, so that the
aggregate productivity (sectoral and country/region-wide) is endogenously determined by
skill-technology assignment in equilibrium. This modeling approach is also closely related to
the recent theoretical development in international trade highlighting the globalization-
induced real productivity gains (see, e.g., Antras, Garicano and Rossi-Hansberg, 2006;
Blanchard and Willmann, 2016; Costinot and Vogel, 2010; Helpman, Itskhoki and Redding,
2010; Grossman and Maggi, 2000; Grossman, 2004; Jung Jae-Won, 2017/2019/2020; Melitz,
2003; Yeaple, 2005). Among others, Jung Jae-Won (2019) highlights that if technology would
exhibit any increasing returns to skill, the equilibrium skill-technology matching itself could
have considerable implications for economy-wide aggregate productivity.
The developed multi-country/region multi-sector CGE model with heterogeneous workers
and firms was calibrated using a global Social Accounting Matrix (SAM) dataset extracted
from the recently released GTAP 10 Database. As will be shown, the simulated results show
that the US-China trade war may generate pervasive negative productivity effects for all
countries and regions. It is also shown that the US-China trade war may largely affect the
global market structure through entry and exit of firms, and considerably change the
industrial comparative advantage of each country/region. Though relatively less pronounced
than other countries and regions, it is shown that Korea also experiences a negative aggregate
productivity effect of a fall of 0.007%. In terms of the effects on Korean exporting firms, it is
shown that exporting firms decrease in the petroleum and chemicals, machinery, and other
transports industries, while there is a more entry of exporting firms in the steel and metal,
electronics and electrics, motor vehicles and parts, and other manufacturing industries, with
a largest decrease of 11.547% in the other transports industry and a highest increase of 8.100%
in the other manufacturing industry.
The rest of the paper is organized as follows. In Section 2, we describe the basic model and
its CGE application. In Section 3, we study the effects of US-China trade war on the country
and sector-wide productivity and market structure. Section 4 concludes with some remarks.
2. The Model and Application

2.1. Model Description
2.1.1. Technologies and Production
We model two types of firms using different technologies in each sector. When entering
the market, firms choose whether to serve only domestic market or export. Exporting to
foreign markets requires higher set-up fixed costs. Also, there is now ample evidence that
exporting firms are more productive than domestic firms. Based on evidence, we assume two
different technologies in each sector: low-tech (‫ )ܮ‬and high-tech (‫ ;)ܪ‬we associate domestic
firms with low-tech-low-fixed-cost technology, and exporting firms with high-tech-high-
fixed-cost technology. In what follows, if no confusion arises, country/region and sector
157
indexes are omitted.
Production of final goods requires primary factors and intermediate goods. Following the
standard CGE models, we assume that capital is homogeneous and the Armington
assumption applies to the intermediate demands. On the other hand, existence of different
technologies implies worker allocation to different technologies. We assume that workers are
differentiated by their individual skill level . The skill distribution in the population in each
country/region is given by with density on support [ , ].
We denote by , ∈ , , the technology-augmented productivity of a worker with
skill level . is increasing in and continuous. Also, at a given skill level , using high
technology is more productive than using low technology: . Furthermore, we
assume a comparative advantage aspect of skill in technologies. Specifically, we assume that:
1 1

0 , (1)

with . Our technology specification implies that workers with higher
skill level are relatively more productive when they are matched with higher technologies.
Consequently, in our framework, workers with relatively low skill levels have relative
comparative advantage in low technology (working in domestic firms using low-tech),
while workers with relatively high skill levels have relative comparative advantage in high
technology (working in exporting firms using high-tech). There should, then, be a skill
threshold in each sector. Let ∗ denote the equilibrium skill thresholds between and
: ∗ . Finally, in equilibrium, workers with ∈ , ∗ are matched
with low-tech domestic firms and worker with ∈ ∗ , are matched with high-tech
exporting firms. The equilibrium skill threshold ∗ is, of course, different sector by sector
and endogenously determined. Fig. 1 illustrates the equilibrium technology/skill
assignment.
In a perfectly competitive labor market, the threshold worker’s wage should be the same
whether he is assigned to domestic firms or exporting firms, which leads to the following
no-arbitrage condition:
∗ ∗ , (2)
where and are technology-specific efficiency wage rates.
Fig. 1. Equilibrium Skill Allocation to Different Technologies

158
2.1.2. Competition and Equilibrium
We assume Dixit-Stiglitz preferences over a continuum of varieties, and monopolistic
competition to prevail in each sector. Firms charge a constant mark-up over marginal
production costs:
ఙ ఙ
௅ ௅ and ு ு . (3)
ఙିଵ ఙିଵ
Zero-profit conditions from free entry satisfy that mark-up revenues exactly cover the fixed
costs of each firm type:
ଵ ଵ
௅ ௅ ௅ ௅ and ு ு ு ு , (4)
ఙ ఙ
where ௅ and ு are demands for individual varieties, and ௅ and ு are technology-specific
fixed costs for each firm type. These free entry conditions determine the equilibrium number
of each firm type: ௅ and ு .
For the derivation of demands for individual varieties, usual constant elasticity of
substitution (CES) formulation is used. Also, for the supply decision of exporting firms
between domestic and foreign market, usual constant elasticity of transformation (CET)
formulation is used. Note that in each country/region there are three groups of goods:
varieties of domestic low-tech firms, varieties of high-tech exporting firms supplied for
domestic market, and varieties of foreign high-tech exporting firms imported for domestic
market.
One important feature of this model is that the technology-augmented total efficiency units
of labor are endogenously determined by the skill-technology assignment in equilibrium,
though the labor supply itself is fixed in each country/region as in conventional CGE models.
In each sector-s in each country/region-i, the technology-augmented total efficiency units of
labor are given by:
௦௨௣ ௦௨௣

௜௦்௢௧ ௜௦௅ + ௜௦ு , (5)
௦௨௣ ௭∗
௜௦௅ ௜௦ ௭ ௜௦௅ ௦௛
௜௦ , (6)

௦௨௣ ௭
௜௦ு ௜௦ ௭ ∗ ௜௦ு ௦௛
௜௦ , (7)

where ௜௦ is a scale parameter and ௦௛

௜௦ is a sectoral labor employment share variable with
∑௦ ௦௛
௜௦ =1 in each country/region.
In a perfectly competitive labor market, the sectoral employment share variable ௦௛௜௦ is
determined by the no-arbitrage condition of the average worker in each sector:
௦௨௣ ௦௨௣ ௦௨௣ ௦௨௣
௅ ௜௦௅ ு ௜௦ு ௅ ௜௧௅ ு ௜௧ு
௦௛ ଴௜௦௧ , (8)
௜௦ ௦௛
௜௧
where ଴௜௦௧ is a parameter of the initial average wage difference between sector s and t.
Equilibrium wages are determined by labor supply and demand in efficiency units:
159
௦௨௣
‫ܮ‬௜௦௅ ൌ ሺ‫ܮ‬ௗ௘௠
௜௦௅ ൅ ݂௜௦௅ ሻܰ௜௦௅ , (9)
௦௨௣
‫ܮ‬௜௦ு ൌ ሺ‫ܮ‬ௗ௘௠
௜௦ு ൅ ݂௜௦ு ሻܰ௜௦ு . (10)
All the other basic settings and equations follow the conventional full multi-country/region
multi-sector global CGE models with monopolistic competition.
2.2. CGE Application

The developed model is calibrated using a global Social Accounting Matrix (SAM) dataset
extracted from the recently released GTAP 10 Database. The database was aggregated to 11
countries/regions and 10 sectors.
The 11 countries and regions are Korea, China, Japan, USA, EU, India, ASEAN 4, ASEAN
6, Oceania 2, America 4, and Rest of World (RoW). Country groups are divided depending
on their mega-FTA participation. ASEAN 4 countries include Vietnam, Malaysia, Singapore,
and Brunei which participate in both the Regional Comprehensive Economic Partnership
(RCEP) and the Comprehensive & Progressive Agreement for Trans-Pacific Partnership
(CPTPP), while ASEAN 6 countries include Thailand, Indonesia, Philippines, Myanmar,
Cambodia, and Laos which participate only in RCEP. Oceania 2 countries include Australia
and New Zealand which participate in both RCEP and CPTPP. America 4 countries include
Canada, Mexico, Peru, and Chile which participate only in CPTPP.
The 10 sectors include primary, petroleum & chemicals, steel & metal, electronics &
electrics, machinery, motor vehicles & parts, other transports, other manufactures, utility &
construction, and services. We apply our developed model to the manufacturing sectors. On
the other hand, for the primary, utility & construction, and services sectors, conventional
perfect competition settings with homogeneous goods are used. Following Table 2
summarizes the classification of countries/regions and industries.
Table 2. Classification of Countries/Regions and Industries

Country / Region Industry
Korea Primary
China Petroleum & Chemicals
Japan Steel & Metal
USA Electronics & Electrics
EU Machinery
India Motor Vehicles & Parts
ASEAN 4 Other Transports
ASEAN 6 Other Manufactures
Oceania 2 Utility & Construction
America 4 Services
Rest of World
Notes: 1. ASEAN 4: Vietnam, Malaysia, Singapore, Brunei (RCEP & CPTPP participation).
2. ASEAN 6: Thailand, Indonesia, Philippines, Myanmar, Cambodia, Laos (RCEP participation).
3. Oceania 2: Australia, New Zealand (RCEP & CPTPP participation).
4. America 4: Canada, Mexico, Peru, Chile (CPTPP participation).
160
For the model calibration, we need to specify the functional form of , ∈ , . We
assume linear technologies as shown in Fig. 1:
1 , ∈ , , (11)
and following Bernard and Jensen (2004), we set / =1.085.

For the skill distribution, we assume uniform distribution and normalize the skill levels so
that 0 and 1.
Fixed costs , ∈ , are calibrated so that Equation (4) is satisfied for all monopolistic
competition sectors in all countries and regions. All the other parameter values are also
calibrated so that we exactly reproduce the initial Social Accounting Matrix (SAM) extracted
from the recently released GTAP 10 Database.
Finally, in this paper we also fully model the Rest of World (RoW) and choose the aggregate
consumption price index of the Rest of World (RoW) as our numeraire. In Jung Jae-Won
(2020), the Rest of World was treated rather exogenously.
3. Effects of the US-China Trade War

Given our model construction and calibration, in this section we investigate the effects of
the US-China trade war. Specifically, we will consider a case where the US and China increase
reciprocally the tariff rates by 1.5 times for all imported goods from each other, and analyze
how such bilateral tariff increases between the G2 affect the global productivity and market
structure.
For the global productivity implications, we will investigate how the bilateral tariff increases
between the G2 affect the skill/technology thresholds and the technology-augmented
efficiency units of labor in each industry and in each country/region. For the global market
structure implications, we will investigate the induced effects on the number of each firm type
(low-tech domestic vs. high-tech exporting) in each sector and in each country/region.
In the Appendix, the results of the unilateral tariff increases are also reported.
3.1. Implications on the Productivity

Table 3 first shows the effects on the skill/technology thresholds ∗ in manufacturing
industries. As described before, a decrease (leftward shift) of ∗ implies that now more firms
and workers are matched with high technology in that sector. Note that Table 3 shows the
induced percentage changes of the skill/technology thresholds ∗ , and that a negative number
in the cell indicates a leftward shift of ∗ (skill/technology-upgrading), while a positive
number indicates a rightward shift of ∗ (skill/technology-downgrading).
We can see that the sectoral effects as well as the effects for each country/region are largely
diverse. For the US and China, it is shown that overall negative effects prevail. For the US, the
motor vehicle and parts industry may be affected the most negatively with an increase of the
threshold by 32.924%, while, for the China, the other manufacturing industry was affected
the most negatively with an increase of the threshold by 96.698%. On the other hand, some
industries may experience some positive effects resulted from the indirect general equilibrium
effects. For example, it is shown that the machinery industry in China may have a
skill/technology upgrading effect with a leftward shift of the threshold by 5.156%. In the US,
it is shown that the steel and metal industry may experience a slightly positive effect with a
161
leftward shift of the threshold by 0.091%.
Not surprisingly, the US-China trade war influences not only the countries directly
concerned, but also the other countries. Looking at the effects on Korea, it is shown that the
petroleum and chemicals, electronics and electrics, machinery, and other transports
industries are negatively affected with an increase of the skill/technology thresholds, while the
steel and metal, motor vehicles and parts, and other manufacturing industries are positively
affected with a decrease of the thresholds. In terms of the effects on the skill/technology
thresholds, Korea may have the most negative effect in the machinery industry with a
rightward shift of the threshold by 12.753%.
The effects for the other countries and regions are also largely diverse. Japan is affected the
most negatively in the steel and metal industry with a rightward shift of the threshold by
24.866%, while the EU is affected the most negatively in the petroleum and chemicals industry
with a rightward shift of the threshold by 32.413%. Otherwise, it is shown that the other
transports industry in the America 4 countries is affected the most negatively with a rightward
shift of the threshold by 50.988%.
Table 3. Effects of US-China Trade War on Thresholds

(Unit: % Changes)
Petrol. Steel Motor Other Other
Electr. Mach.
Chem. Metal Parts Transp. Manuf.
Korea 3.933 -6.670 4.120 12.753 -4.562 8.396 -7.594
China 20.906 2.281 17.617 -5.156 -2.587 2.512 96.698
Japan -48.306 24.866 5.219 24.747 -1.889 14.795 -8.083
USA 2.006 -0.091 2.388 3.245 32.924 11.748 1.255
EU 32.413 -1.242 -6.172 4.676 0.298 9.017 7.450
India 18.001 -8.664 -6.985 -4.011 -11.904 -8.683 1.768
ASE4 13.650 23.356 13.522 3.367 7.119 11.126 29.522
ASE6 31.567 6.584 24.237 33.708 0.062 10.353 33.523
Ocea2 3.745 7.957 2.487 4.694 1.900 1.960 -0.810
Amer4 11.628 -13.505 3.572 0.466 19.066 50.988 4.208
RoW -19.514 11.834 17.323 4.033 5.107 -3.494 -5.325
Table 4 shows the induced effects on the sectoral real productivity for each country/region.
We measure the sectoral real productivity level as the average efficiency units of labor in each
௭∗ ௭
sector, i.e., ‫׬‬௭ ߮௅ ሺ‫ݖ‬ሻ ݃ሺ‫ݖ‬ሻ݀‫ ݖ‬൅ ‫׬‬௭ ∗ ߮ு ሺ‫ݖ‬ሻ ݃ሺ‫ݖ‬ሻ݀‫ݖ‬. Following the shifts of the skill/technology
thresholds, Table 4 shows the induced percentage changes of the average real productivity
level in each sector and for each country/region.
We can see again that the sectoral effects as well as the effects for each country/region are
largely diverse. We see that for the sectors where the skill/technology threshold moved
rightwards there is a fall in the average real productivity, while for the sectors where the
threshold moved leftwards there is a rise in the average real productivity.
For the US and China, it is shown that there is an overall negative real productivity effect.
For the China, we can see that the other manufacturing industry experiences the most
negative real productivity effect with a decrease of the average real productivity by 0.567%.
162
For the US, it is shown that the motor vehicles and parts industry may have the most
negative real productivity effect with a fall of the average real productivity by 0.152%. In
total, the aggregate real productivity decreases by 0.103% and 0.034% in China and the US,
respectively. Thus, the results show that in terms of the real productivity China may be
affected more negatively than the US following the reciprocal tariff increases between them.
Concerning Korea, it is shown that the average real productivity decreases in the petroleum
and chemicals, electronics and electrics, machinery, and other transports industries by
0.016%, 0.017%, 0.054% and 0.035%, respectively, which leads to an overall decrease of
0.007%. On the other hand, it is shown that Korea’s real productivity loss may be relatively
small compared to other countries and regions, except for the India. India has a positive
aggregate productivity effect with an increase of 0.010%, while all the other countries and
regions experience negative aggregate productivity effects. In our classification of regions, the
ASEAN countries are shown to be affected the most negatively after China. The aggregate
real productivity decreases by 0.090% and 0.063% in ASEAN 6 and ASEAN 4 countries,
respectively, which are larger than even for the US.
Table 4. Effects of US-China Trade War on Real Productivity

(Unit: % Changes)
Electr. Mach. Total
Korea -0.016 0.026 -0.017 -0.054 0.018 -0.035 0.029 -0.007
China -0.091 -0.009 -0.076 0.020 0.010 -0.010 -0.567 -0.103
Japan 0.145 -0.111 -0.021 -0.110 0.007 -0.063 0.031 -0.017
USA -0.008 0.000 -0.010 -0.013 -0.152 -0.049 -0.005 -0.034
EU -0.149 0.005 0.024 -0.019 -0.001 -0.037 -0.031 -0.030
India -0.078 0.033 0.027 0.016 0.044 0.033 -0.007 0.010
ASE4 -0.058 -0.103 -0.057 -0.014 -0.029 -0.046 -0.134 -0.063
ASE6 -0.145 -0.027 -0.107 -0.156 0.000 -0.043 -0.155 -0.090
Ocea2 -0.015 -0.033 -0.010 -0.019 -0.008 -0.008 0.003 -0.013
Amer4 -0.049 0.050 -0.014 -0.002 -0.083 -0.253 -0.017 -0.053
RoW 0.070 -0.050 -0.074 -0.016 -0.021 0.014 0.021 -0.008
3.2. Implications on the Market Structure

Now we investigate the effects on the sectoral number of firms in each country/region. Note
that the changes in the number of each firm type (domestic or exporting) do not necessarily
coincide with the changes in the skill/technology thresholds. A leftward shift of the threshold
‫ ∗ ݖ‬implies that a worker who was previously matched with low-tech domestic firms is now
matched with high-tech exporting firms. However, the total variation of number of firms is
not only induced by the allocation of workers within each sector, but also by the allocation of
workers between sectors in equilibrium.
Table 5 first shows the percentage changes of domestic firms in each sector and for each
country/region. It is shown that following the reciprocal tariff increases between the US and
China, more domestic firms prevail in both countries. Such effects are more pronounced in
China. In the other manufacturing industry in China, it is shown that the domestic firms
163
increase by 93.991%, which is followed by the petroleum and chemicals industry with an
increase of 22.738%. In the US, we see also an overall increase of domestic firms though less
pronounced compared to China. It is shown that in the US the motor vehicles and parts
industry is more dominated by domestic firms.
Concerning Korea, the effects on the sectoral domestic firms are quite diverse. It is shown
that the number of domestic firms decreases in the steel and metal, motor vehicles and parts,
other transports, and other manufacturing industries by 5.665%, 3.706%, 0.852% and 3.027%,
respectively, while increases in the petroleum and chemicals, electronics and electrics, and
machinery industries by 0.564%, 7.805% and 9.219%, respectively.
Otherwise, it is shown that following the reciprocal tariff increases between the US and
China, the ASEAN countries are highly dominated by low-tech domestic firms. Among
others, it is shown that the low-tech domestic firms increase the most in the other
manufacturing industry of the ASEAN countries, with an increase of 38.952% and 33.042%
in ASEAN 4 and ASEAN 6 countries, respectively.
Table 5. Effects of US-China Trade War on Domestic Firms

(Unit: % Changes)
Electr. Mach.
Korea 0.564 -5.665 7.805 9.219 -3.706 -0.852 -3.027
China 22.738 6.943 21.017 1.405 5.259 12.698 93.991
Japan -46.877 19.425 1.970 15.916 -2.505 11.791 -2.625
USA 4.952 -0.573 3.070 0.553 29.511 12.460 7.266
EU 26.621 1.461 -0.303 5.429 1.913 9.732 8.886
India 18.102 -4.299 -5.657 -4.821 -12.979 -7.450 2.768
ASE4 5.853 20.684 18.578 14.967 4.935 12.784 38.952
ASE6 28.440 16.784 32.402 41.488 4.588 15.119 33.042
Ocea2 5.033 5.586 0.481 0.299 -1.538 0.862 3.611
Amer4 9.918 -6.229 17.666 -1.246 18.896 1.944 6.232
RoW -16.282 4.410 2.939 -0.128 1.131 -2.494 -1.623
Finally, we investigate the effects on the high-tech exporting firms. As mentioned before,
note that an increase(a decrease) of domestic firms does not necessarily lead to a
decrease(an increase) of exporting firms. The total variation of number of firms is not only
induced by the allocation of workers within each sector, but also by the allocation of
workers between sectors in equilibrium. If more workers move to a sector due to an
expansion of that sector in equilibrium, both domestic and exporting firms may increase.
Likewise, if less workers are allocated to a sector due to a contraction of that sector, both
domestic and exporting firms may decrease in that sector. The final outcomes would
depend not only on the competition within a country (competition between firms and
sectors within a country), but also on the competition between countries and regions
(competition with foreign firms and sectors).
Table 6 shows the percentage changes of exporting firms in each sector and for each
country/region. As before, it is shown that the sectoral effects are largely diverse. In China, it
is shown that exporting firms decrease in the petroleum and chemicals, electronics and
164
electrics, and other manufacturing industries, while increase in the steel and metal,
machinery, motor vehicles and parts, and other transports industries; exporting firms
decrease the most in the other manufacturing industry with a fall of 37.880%, while increase
the most in the motor vehicles and parts industry with a rise of 9.162%.
In the US, it is shown that exporting firms decrease in the steel and metal, electronics and
electrics, machinery, motor vehicles and parts, and other transports industries, while increase
in the petroleum and chemicals, and other manufacturing industries; exporting firms
decrease the most in the motor vehicles and parts industry with a fall of 15.012%, while
increase the most in the other manufacturing industry with a rise of 5.415%.
In Korea, it is shown that exporting firms decrease in the petroleum and chemicals,
machinery, and other transports industries, while increase in the steel and metal, electronics
and electrics, motor vehicles and parts, and other manufacturing industries; exporting firms
decrease the most in the other transports industry with a fall of 11.547%, while increase the
most in the other manufacturing industry with a rise of 8.100%.
It would also be of our great interest to look at the effects on the Japanese exporting firms
given their high competition with the Korean exporting firms in the international market. In
Japan, it is shown that exporting firms decrease in the steel and metal, electronics and
electrics, machinery, and other transports industries, while increase in the petroleum and
chemicals, motor vehicles and parts, and other manufacturing industries; exporting firms
decrease the most in the machinery industry with a fall of 16.119%, while increase the most
in the petroleum and chemicals industry with a rise of 22.921%.
Table 6. Effects of US-China Trade War on Exporting Firms

(Unit: % Changes)
Electr. Mach.
Korea -4.754 3.759 1.846 -7.996 2.724 -11.547 8.100
China -6.837 3.612 -4.228 9.098 9.162 8.850 -37.880
Japan 22.921 -13.732 -5.087 -16.119 0.116 -8.294 9.343
USA 2.074 -0.447 -0.275 -3.849 -15.012 -3.981 5.415
EU -16.549 3.242 8.860 -1.141 1.490 -2.928 -1.647
India -7.035 8.397 4.244 0.732 3.452 4.840 0.279
ASE4 -11.896 -11.154 -1.131 9.737 -4.792 -2.959 -5.132
ASE6 -14.462 6.713 -3.614 -8.179 4.497 0.049 -13.433
Ocea2 -0.255 -5.262 -2.918 -5.970 -4.097 -1.840 4.792
Amer4 -6.059 14.247 12.001 -1.885 -7.658 -45.853 0.245
RoW 12.152 -11.005 -18.238 -5.530 -5.726 2.430 6.104
Finally, the results of Table 6 also indicate that the US-China trade war may induce
considerable changes in the industrial comparative advantage of each country/region. Other
things being equal, an increase of exporting firms in a certain industry of a country implies
that that country has more comparative advantage for that industry, and exports more those
goods to the global market.
In Table 6, we can see that the effects are more pronounced in certain industries. For
example, for the petroleum and chemical goods, Japan’s comparative advantage may increase
significantly, while that of the EU and the ASEAN decreases significantly. On the other hand,
165
Japan may lose considerably its comparative advantage in the steel and metal goods and in
the machinery goods. Concerning Korea, it is shown that Korea may lose considerably its
comparative advantage in the other transports industry, while gain comparative advantage in
the other manufacturing industry. It is also worthy of notice that the US may lose largely its
comparative advantage in the motor vehicles and parts industry.
4. Conclusion
Though there have been several attempts and deals to resolve the trade disputes between
the US and China, the ongoing US-China trade war is even intensifying and it is widely
believed that the US-China trade war will last long time since it is finally a combat to win the
economic supremacy in the twenty-first century. If it would be the case, not only countries
like Korea that have a large economic dependence on the economy of the two countries but
also other outside countries might have great repercussions through links in global supply
chains.
In this paper, we developed a full multi-country/region multi-sector computable general
equilibrium (CGE) model of global trade incorporating endogenous skill-technology
assignment of heterogeneous workers and firms, and investigated possible impacts of the US-
China trade war on the aggregate productivity and global market structure.
The simulated results show that the US-China trade war may largely affect the aggregate
productivity in each sector in each country/region, as well as the global market structure
through entry and exit of firms using different technologies, which may finally result in
considerable changes in the industrial comparative advantage of each country/region.
Though the effects may be diverse sector by sector and country by county, the results show
that all countries may be affected significantly given the importance of the two economies in
the world economy as well as the close economic interdependence in today’s globalized world.
Above all, it was shown that the US-China trade war might decrease the aggregate
productivity of all countries and regions. One exception of an aggregate productivity gain was
found in the case of the India. In a dynamic setting, however, it may be very likely that the
pervasive aggregate productivity losses all over the world would finally affect negatively the
India too, and such dynamic negative effects might be much larger for export-driven, small,
and open economies such as Korea. The results of this paper may therefore suggest that the
US-China trade war is not just about the US and China, and thus that other third countries
have also high incentives to resolve the trade disputes together.
Finally, the results of the paper also imply that if the US and China would increase bilateral
tariffs for some targeted industries, all the productivity and market structure implications
could change again not only quantitively but also qualitatively. Also, extending the basic
model to dynamic settings and analyzing dynamic implications may provide other interesting
dynamic insights. Though market structure, competition and productivity have been the
central issues for policy makers, conventional CGE models have mostly assumed a perfectly
competitive world composed of homogeneous agents and technologies. The heterogeneous-
agent/technology framework of this paper may widely be used to evaluate or re-evaluate the
impacts of any policy changes when the competition and productivity issues are important
(or at least not negligible). I leave them for future research.
166
References
Angkinand, A. P. and E. M.P. Chiu (2011), “Will Institutional Reform Enhance Bilateral Trade Flows?
Analyses from Different Reform Aspects”, Journal of Economic Policy Reform, 14(3), 243-258.
Antras, P., L. Garicano and E. Rossi-Hansberg (2006), “Offshoring in a Knowledge Economy”,
Quarterly Journal of Economics, 121, 31-77.
Aw, B., S. Chung and M. Roberts (2000), “Productivity and Turnover in the Export Market: Micro-
Level Evidence from the Republic of Korea and Taiwan (China)”, World Bank Economic
Review, 14, 65-90.
Bernard, A. B. and J. B. Jensen (1999), “Exceptional Exporter Performance: Cause, Effect, or Both?”,
Journal of International Economics, 47, 1-25.
Bernard, A. B. and J. B. Jensen (2004), “Exporting and Productivity in the USA”, Oxford Review of
Economic Policy, 20(3), 343-357.
Blanchard, E. and G. Willmann (2016), “Trade, Education, and the Shrinking Middle Class”, Journal
of International Economics, 99, 263-278.
Clerides, S., S. Lach and J. Tybout (1998), “Is Learning by Exporting Important? Micro-Dynamic
Evidence from Colombia, Mexico, and Morocco”, Quarterly Journal of Economics, 113, 903-
947.
Costinot, A. and J. Vogel (2010), “Matching and Inequality in the World Economy”, Journal of
Political Economy, 118, 747-786.
Girma, S., H. Gorg and E. Strobl (2004), “Exports, International Investment, and Plant Performance:
Evidence form a Non-Parametric Test”, Economic Letters, 83, 317-324.
Grossman, G. M. and G. Maggi (2000), “Diversity and Trade”, American Economic Review, 90,
1255-1275.
Grossman, G. M. (2004) “The Distribution of Talent and the Pattern and Consequences of
International Trade”, Journal of Political Economy, 112, 209-239.
Helpman, E., M. J. Melitz and S. R. Yeaple (2004), “Export versus FDI with Heterogeneous Firms”,
American Economic Review, 94 300-316.
Helpman, E., O. Itskhoki and S. Redding (2010), “Inequality and Unemployment in a Global
Economy”, Econometrica, 78, 1239-1283.
Itakura, K. (2020), “Evaluating the Impact of the US-China Trade War”, Asian Economic Policy
Review, 15, 77-93.
Jung, Jae-Won (2017), “Openness, Productivity and Welfare with Heterogenous Firms”, Journal of
Korea Trade, 21(4), 349-365.
Jung, Jae-Won (2019), “Technology, Skill, and Growth in a Global Economy”, Economic Theory, 68,
609-641.
Jung, Jae-Won and Tae-Hwang Kim (2019), “Macroeconomic Buffer Effects of Mega-FTA
Formation: A CGE Analysis for Korea”, Journal of Korea Trade, 23(3), 118-137.
Jung, Jae-Won (2020), “Introducing Roy-like Worker Assignment into Computable General
Equilibrium Models”, Applied Economics Letters, 27, 503-510.
Li, M., E. J. Balistreri and W. Zhang (2020), “The US-China Trade War: Tariff Data and General
Equilibrium Analysis”, Journal of Asian Economics, 69, 1-13.
Liu, T. and W. T. Woo (2018), “Understanding the U.S.-China Trade War”, China Economic Journal,
11(3), 319-340.
Mao, H. and H. Gorg (2019), “Friends Like This: The Impact of the US-China Trade War on Global
167
Value Chains”, World Economy, 43, 1776-1791.
Melitz, M. J. (2003), “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry
Productivity”, Econometrica, 71, 1695-1725.
Syverson, C. (2004), “Market Structure and Productivity: A Concrete Example”, Journal of Political
Economy, 112(6), 1181-1222.
Tung, R. L. (1982), “U.S.-China Trade Negotiations: Practices, Procedures and Outcomes”, Journal
of International Business Studies, 13(2), 25-37.
Wei, S. J. (2000) “How Taxing is Corruption on International Investors?”, Review of Economics and
Statistics, 82(1), 1-11.
Yeaple, S. R. (2005), “A Simple Model of Firm Heterogeneity, International Trade, and Wages”,
Journal of International Economics, 65, 1-20.
168
Appendix
Table A. Effects on Thresholds (Unilateral Tariff Increases by the US)

(Unit: % Changes)
Electr. Mach.
Korea 1.839 3.854 15.010 28.869 -3.998 5.293 -6.310
China 10.726 -5.541 39.001 -17.289 -7.026 -3.166 77.673
Japan -25.046 55.612 24.145 63.578 -3.658 33.744 -4.819
USA -11.302 -11.090 -19.363 -12.312 -6.790 -13.830 -3.315
EU 10.641 -1.065 -7.198 5.197 55.632 -24.429 1.231
India 25.075 -3.371 -3.658 -0.791 -4.139 -30.002 7.067
ASE4 30.692 12.950 23.983 28.270 8.981 -1.352 20.249
ASE6 33.566 -1.790 43.231 19.431 9.042 13.085 38.971
Ocea2 2.083 -5.566 -0.318 0.575 0.753 2.786 -1.482
Amer4 -14.176 48.687 17.715 33.839 7.004 -2.031 -10.173
RoW 7.711 18.852 20.070 9.950 2.816 21.390 1.693
Table B. Effects on Real Productivity (Unilateral Tariff Increases by the US)

(Unit: % Changes)
Electr. Mach. Total
Korea -0.007 -0.016 -0.064 -0.131 0.016 -0.022 0.024 -0.028
China -0.045 0.021 -0.184 0.062 0.027 0.012 -0.427 -0.076
Japan 0.087 -0.281 -0.107 -0.332 0.014 -0.156 0.019 -0.108
USA 0.042 0.041 0.069 0.046 0.026 0.051 0.013 0.041
EU -0.044 0.004 0.027 -0.021 -0.281 0.085 -0.005 -0.034
India -0.112 0.013 0.014 0.003 0.016 0.101 -0.029 0.001
ASE4 -0.140 -0.055 -0.106 -0.128 -0.037 0.005 -0.088 -0.078
ASE6 -0.155 0.007 -0.208 -0.084 -0.037 -0.055 -0.184 -0.102
Ocea2 -0.008 0.021 0.001 -0.002 -0.003 -0.011 0.006 0.001
Amer4 0.052 -0.240 -0.076 -0.157 -0.029 0.008 0.038 -0.058
RoW -0.032 -0.082 -0.087 -0.041 -0.011 -0.094 -0.007 -0.051
169
Table C. Effects on Domestic Firms (Unilateral Tariff Increases by the US)
(Unit: % Changes)
Electr. Mach.
Korea 1.107 4.054 17.304 23.374 2.268 -9.472 -2.781
China 11.939 -1.973 30.559 -10.279 0.925 1.634 75.027
Japan -23.400 46.862 26.736 47.002 0.915 32.204 -0.261
USA -8.418 -4.820 -14.866 -14.873 -12.743 -16.235 1.528
EU 9.649 -0.200 -1.182 4.860 41.522 -13.415 2.713
India 25.765 1.404 4.941 -1.471 -4.191 -27.753 7.756
ASE4 18.580 13.985 29.519 22.016 11.300 0.828 31.904
ASE6 34.243 9.388 53.611 46.661 9.615 7.200 36.942
Ocea2 3.717 -4.252 13.951 -5.160 2.046 -6.941 1.855
Amer4 -14.012 11.162 51.825 -5.533 15.822 -4.374 -6.870
RoW 8.287 15.880 20.094 7.566 14.070 -0.238 4.174
Table D. Effects on Exporting Firms (Unilateral Tariff Increases by the US)

(Unit: % Changes)
Electr. Mach.
Korea -1.444 -1.338 -4.060 -15.080 8.216 -15.812 6.360
China -3.187 6.065 -20.357 15.924 11.578 6.267 -30.948
Japan 12.487 -26.081 -7.584 -32.296 6.267 -14.207 6.795
USA 7.872 11.724 13.635 1.796 -3.891 2.502 6.379
EU -5.073 1.300 9.527 -2.366 -28.748 25.763 0.969
India -9.387 6.348 10.507 -0.374 1.585 15.589 -2.155
ASE4 -20.226 -4.243 -5.402 -15.446 -1.490 2.758 0.958
ASE6 -12.715 12.174 -10.906 13.387 -3.075 -10.103 -16.462
Ocea2 0.766 3.626 14.459 -5.916 0.983 -10.455 3.993
Amer4 5.863 -39.422 19.973 -38.746 5.234 -1.608 7.873
RoW -2.539 -9.746 -7.868 -6.006 9.710 -24.701 1.754
Table E. Effects on Thresholds (Unilateral Tariff Increases by China)

(Unit: % Changes)
Electr. Mach.
Korea -5.998 -4.438 -2.534 0.267 -8.324 -11.474 0.091
China 2.327 1.761 -1.631 3.158 -0.068 -0.998 3.686
Japan -4.317 -1.415 2.550 -5.428 0.874 -8.957 -0.632
USA 8.272 5.264 11.802 9.042 37.479 18.404 3.107
EU 18.046 1.990 4.402 3.520 -21.486 40.309 4.920
India 1.501 -2.876 -2.436 -2.067 -7.255 8.390 -0.806
ASE4 -7.937 9.263 -1.709 -21.628 -0.217 11.712 2.449
ASE6 1.977 -1.491 0.556 2.773 -15.168 -5.425 1.774
Ocea2 1.820 10.557 2.606 3.841 1.345 0.019 3.004
Amer4 15.771 -20.969 1.291 -7.762 16.359 43.012 9.179
RoW -13.349 -2.336 -5.118 -2.472 1.832 -15.260 -2.994
170
Table F. Effects on Real Productivity (Unilateral Tariff Increases by China)
(Unit: % Changes)
Electr. Mach. Total
Korea 0.023 0.017 0.010 -0.001 0.032 0.043 0.000 0.018
China -0.009 -0.007 0.006 -0.013 0.000 0.004 -0.015 -0.005
Japan 0.017 0.006 -0.010 0.021 -0.003 0.034 0.002 0.009
USA -0.034 -0.021 -0.049 -0.037 -0.176 -0.080 -0.012 -0.059
EU -0.078 -0.008 -0.018 -0.014 0.076 -0.192 -0.020 -0.036
India -0.006 0.011 0.010 0.008 0.028 -0.035 0.003 0.003
ASE4 0.030 -0.038 0.007 0.076 0.001 -0.049 -0.010 0.002
ASE6 -0.008 0.006 -0.002 -0.011 0.055 0.021 -0.007 0.008
Ocea2 -0.007 -0.044 -0.010 -0.015 -0.005 0.000 -0.012 -0.014
Amer4 -0.067 0.074 -0.005 0.030 -0.070 -0.207 -0.038 -0.040
RoW 0.049 0.009 0.020 0.010 -0.007 0.056 0.012 0.021
Table G. Effects on Domestic Firms (Unilateral Tariff Increases by China)

(Unit: % Changes)
Electr. Mach.
Korea -5.282 -3.509 -0.190 -2.226 -9.098 -0.198 0.837
China 2.685 1.997 0.041 2.996 1.195 4.343 4.024
Japan -4.342 -1.484 -0.158 -4.146 -0.868 -7.338 -0.100
USA 8.681 1.552 10.495 9.369 39.597 22.281 3.852
EU 15.256 2.875 5.621 4.295 -16.835 22.132 4.774
India 1.124 -2.081 -2.827 -2.119 -7.978 8.561 -0.712
ASE4 -6.755 6.387 -1.868 -7.737 -4.888 7.417 1.618
ASE6 0.318 0.297 -0.364 -4.331 -11.836 2.974 1.759
Ocea2 1.429 8.059 0.302 3.378 -3.373 6.491 3.047
Amer4 14.782 -8.517 5.049 1.794 11.228 5.903 7.889
RoW -11.376 -4.163 -6.950 -4.070 -7.080 -5.683 -2.408
Table H. Effects on Exporting Firms (Unilateral Tariff Increases by China)

(Unit: % Changes)
Electr. Mach.
Korea 3.172 2.753 3.439 -2.589 2.436 17.851 0.710
China -0.575 -0.467 2.356 -1.401 1.290 5.809 -1.134
Japan 1.695 0.492 -3.622 3.543 -2.068 5.402 0.787
USA -2.889 -5.515 -5.723 -3.254 -13.204 -4.136 -0.506
EU -9.321 0.075 -0.593 -0.653 15.032 -26.655 -2.083
India -0.967 1.971 0.562 0.765 2.076 -3.154 0.414
ASE4 4.496 -6.198 0.516 27.925 -4.599 -8.284 -1.772
ASE6 -2.397 2.417 -1.134 -7.937 10.231 11.231 -0.717
Ocea2 -1.100 -6.320 -3.249 -1.953 -5.162 6.463 -1.144
Amer4 -7.027 25.458 3.178 13.754 -10.589 -38.352 -4.764
RoW 7.737 -0.960 0.060 -0.676 -9.414 18.037 1.798
https://doi.org/10.35611/jkt.2020.24.8.171 171
Is Text Mining on Trade Claim Studies

Applicable? Focused on Chinese Cases of
Arbitration and Litigation Applying JKT 24(8)
the CISG*
Cheon Yu
Department of International Trade, Mokpo National University, South Korea
DongOh Choi
Department of International Trade, Mokpo National University, South Korea
Yun-Seop Hwang†
Department of International Trade, KyungHee University, South Korea
Abstract
Purpose – This is an exploratory study that aims to apply text mining techniques, which
computationally extracts words from the large-scale text data, to legal documents to quantify trade
claim contents and enables statistical analysis.
Design/methodology – This is designed to verify the validity of the application of text mining
techniques as a quantitative methodology for trade claim studies, that have relied mainly on a
qualitative approach. The subjects are 81 cases of arbitration and court judgments from China
published on the website of the UNCITRAL where the CISG was applied. Validation is performed by
comparing the manually analyzed result with the automatically analyzed result. The manual analysis
result is the cluster analysis wherein the researcher reads and codes the case. The automatic analysis
result is an analysis applying text mining techniques to the result of the cluster analysis. Topic
modeling and semantic network analysis are applied for the statistical approach.
Findings – Results show that the results of cluster analysis and text mining results are consistent with
each other and the internal validity is confirmed. And the degree centrality of words that play a key
role in the topic is high as the between centrality of words that are useful for grasping the topic and
the eigenvector centrality of the important words in the topic is high. This indicates that text mining
techniques can be applied to research on content analysis of trade claims for statistical analysis.
Originality/value – Firstly, the validity of the text mining technique in the study of trade claim cases is
confirmed. Prior studies on trade claims have relied on traditional approach. Secondly, this study has
an originality in that it is an attempt to quantitatively study the trade claim cases, whereas prior trade
claim cases were mainly studied via qualitative methods. Lastly, this study shows that the use of the text
mining can lower the barrier for acquiring information from a large amount of digitalized text.
Keywords: Semantic Network Analysis, Text mining, Topic Modeling, Trade Claim
JEL Classifications: F14, K12, M16
1. Introduction
Changes in the international trading environment, such as new trade protectionism,
bilateralism, the Fourth Industrial Revolution, and climate change, increase not only
* This work is supported by Research Funds of Mokpo National University in 2020 and is based on a part
of the first author’s doctoral dissertation.
†
Corresponding author: rusiahys@khu.ac.kr
172
uncertainty in international commerce but also the possibility of trade claims being caused.
In international commercial practice, preventing and managing trade claims is critical not
only for the contracting parties but also for the national economy. For the contracting parties,
trade claims increase the transaction costs and a cancellation of contracts results in direct
losses to the company’s profits and reputation as well as hurting the company’s competitive
advantages due to potential loss of future contracts. At the macroeconomic level, trade claims
can negatively affect the balance of international payments and foreign debts and may induce
international trade conflicts between countries (Shin Koon-Jae, 2007). Therefore, with such
high amounts of uncertainty, trade claim management is crucial for companies to secure
economic performance.
One of the ways to effectively manage trade claims is to identify and respond to the frequent
trade claims at the national level. The trading parties interpret and apply the contents of the
contract based on their country’s legal system, economic environments, commercial
practices, cultural values, and beliefs (La Porta et al., 1998; Zhou and Peng, 2010). This implies
that trading parties of the same nationality will have comparable commercial behavior
patterns and trade claims may appear similar. If a country’s trade claims appear in a specific
pattern, a trading party planning or carrying out exports or imports for the country may have
useful implications for claim management. However, since previous research on trade claims
is mainly conducted on individual cases, research on trade claim patterns at the national level
is insufficient.
The following academic and practical contributions are expected to be gained by
understanding trade claims patterns at the national level and analyzing their characteristics
through statistical techniques. First, it is possible to secure objective and reproducible
research results. A study on the interpretation of individual cases has the advantage that it
informs the interpretation of the theory of law applied to the case. However, it has a limitation
from being derived from single cases, which places considerable restrictions on
generalization. Therefore, there is a need for a methodology that identifies the overall pattern,
connects internal concepts, and organizes them into a form that can be interpreted. It is
expected that this study will be able to substantially satisfy this goal.
Second, the statistical approach is highly scalable compared to studies focused on
individual cases and legal interpretation. This is because the statistical approach allows
analysis of the importance of content as well as interpretation of the regulations of the law.
Furthermore, if specific patterns can be analyzed by country, it is possible to check the
differences between national patterns due to various macroscopic influences such as culture
and economic systems. In practice, this provides a basis for judging which regulations and
trade claims should be observed by individual companies in accessing foreign markets. Also,
identifying the differences between countries can help to set the market expansion strategies
for different markets.
The following two elements are needed to clarify the characteristics of trade claims at the
national level. One is to secure data on trade claims across the country. Legal cases are mainly
used as the source of the subject of research on claims between trading parties in international
business transaction. Claims arising during business transaction are a kind of business secret.
Because it is not publicly disclosed, it is more dependent on legal cases. This study aims to
secure trade claims data for the entire country through the trade claims legal cases applied to
the United Nations Convention on the Contract for the International Sale of Goods of each
country published on the website of the United Nations Commission on International Trade
Law. The UNCITRAL discloses trade claims legal cases of member countries for the purpose
of raising the awareness of CISG as an international trade norm and for the purpose of unified
interpretation and application.
Is Text Mining on Trade Claim Studies Applicable?
Focused on Chinese Cases of Arbitration and Litigation Applying the CISG
173
The other is the analysis method of the collected large-scale text data. As the content of
trade claims data across the country is vast, traditional methods not only take a lot of time to
analyze, but have a high probability of error during the analysis process. This study intends
to perform analysis by applying a data mining technique that has been actively used for
unstructured text data. Text mining is a methodology for structuring input texts and finding
patterns in unstructured data, and has the advantage of automatically processing large
amounts of text (Ponweiser, 2012). Validity needs to be confirmed to adopt the results of
trade claims analysis using text mining (Newman et al., 2010). In this study, the internal
validity is confirmed by comparing the analysis results derived through text mining with the
analysis results derived through the traditional method.
This is an exploratory study to apply text mining techniques to legal cases of trade claims
as a quantitative approach. The content of the legal cases is identified using word frequency
and centralities generated by semantic network analysis on the corpus because the text
contains information and includes words and sentences to convey that information (Laver,
Benoit and Garry, 2003). The occurrence and frequency of a particular word is representative
of the characteristics of the text (Evans et al., 2007). And centralities generated by semantic
network analysis shows the meaning of the content by deriving the role of each word and the
relationship between words by representing words as nodes and links (Freeman, 2005).
The subjects of analysis are the Chinese cases published on the UNCITRAL website
wherein CISG is applied. The CISG is a universal governing law that determines imputation
for claims to international business transaction. This is an international treaty signed by 94
countries and applies to the international sale of goods in preference to domestic law
(UNCITRAL website). This study focuses on the fact that the cases are textual documents
that contain the contents of claims between the trading parties.
The remainder of the paper is organized as follows: Section 2 reviews the theoretical
background and related works of text mining techniques, Section 3 develops the framework
for applying text mining on the trade claim analysis, Section 4 presents and explains the
results of text mining, and Section 5 concludes the paper and discusses the implications not
only theoretically but also practically.
2. Theoretical Background and Literature Review

2.1. Extracting Information from the Text
Text contains information (Laver, Benoit and Garry, 2003) and includes words and
sentences to convey that information. Information can be obtained by reading and
understanding the words and sentences that make up the text. A well-known method of
acquiring information from text is content analysis. Content analysis is a research method
that uses a series of procedures to derive valid inferences from text. It is useful for
quantitatively transforming the unstructured text to understand its characteristics (Berelson,
1952; Holsi, 1969; Weber, 1990). Content analysis can quantitatively classify and verify the
contents of the document using a structured system when research data exists as
documentation. In this regard, it is useful in studies of communication, tourism, marketing,
and law. That is, if you use the same method with the same data, anyone can replicate the
same results (Hall and Wright, 2008).
Despite the usefulness of content analysis, it is time-consuming and expensive for massive
text analysis, which has limited its use. However, due to the advancement of information
technology, the time and cost of quantifying information contained in digital texts has
174
drastically lowered, increasing its use in various fields. For example, in the financial sector, it
is used to predict changes in asset prices or to estimate the effect of new information through
extracted text from news, social media, and available data published by a company. In the
macroeconomic sector, text analysis is used to predict inflation and unemployment, or to
estimate the effects of government policies (Gentzkow, Kelly and Taddy, 2019). Text mining,
content analysis using data mining techniques, is a representative methodology for acquiring
information from dizitalized text (Evans et al., 2017).
The meaning in a text is not understood by a single word. Since the meaning of a word is
determined by other words distributed together with it, several words must be considered
simultaneously to interpret and accept meaning. Thus, the meaning of a word is determined
by the context, which is the environment for the word, and this environment is structured by
other words (Harris, 1954). Therefore, interpreting the context allows understanding of the
meaning of the word and measuring the similarity between words (McDonald and Ramscar,
2001). According to the distributional hypothesis, words used in similar contexts have similar
meanings (Deerwester et al. 1990; Harris, 1954; Weaver, 1955).
Vector space models (hereafter VSMs) enable statistical processing of digital texts by
positioning words in a geometric space. This is a method of substituting the dimensions of
the semantic space with words and then placing them in the space according to their degree
of appearance together. Two words with similar linguistic contexts are placed in a closer
semantic space, allowing the similarity of a corpus of text to be analyzed (McDonald and
Ramscar, 2001). That is, VSMs derive co-occurrence frequency information and statistically
estimate the similarity of meaning between words, phrases, and documents.
Trade claims are a type of business secret and are not publicized often, so judicial decisions
are the main source of information on them. Legal documents are in text form and reading
and understanding them can help obtain information about the claims. For example, if the
word ‘non-conformity’ appears frequently along with ‘quality’ and ‘goods’, it can be seen that
the ruling is related to product quality non-conformity. This implies that content analysis of
legal documents is possible through the application of text mining.
2.2. Related Works

Research on applying text mining to digital texts is being carried out in many fields for
descriptive and causal analysis. These studies are largely divided into three categories. The
first are studies that interpret the meaning through frequency analysis of the words that
appear. Evans et al. (2007) regarded the occurrence and frequency of a particular word as
representing the characteristics of the text to present a methodology for automatically
classifying legal documents. Lee Ju-Yeon, Han Seung-Hwan and Kwon Ki-Seok (2015)
analyzed the keywords of 24,019 legal papers from 2004 to 2013 to divide contents of the study
into tangible issues such as consumer protection and default on debt, and ideological issues
such as freedom of expression and human rights. Iselin and Siliverstovs (2013) attempted to
verify to what extent the R-word index proposed by The Economist can predict GDP growth.
The R-word index was the frequency of the word ‘recession’ that appeared in articles in the
Financial Times and the Wall Street Journal.
The second are studies that compare the similarity of words that appear and interpret their
meaning. Allee, Elsig and Lugg (2017) demonstrated through text mining that the
incompatible preferential trade agreements (hereafter PTAs) and the World Trade
Organization (WTO) were strongly linked to each other. To this end, they analyzed how often
individual country PTAs refer to the WTO. The result showed that 90% of new PTAs refer to
the WTO, confirming that the WTO still remains a focal point. Ghani et al. (2006) proposed
175
the use of text mining as a method to estimate the expected sales of new products by extracting
products with similar properties and values through it and predicting sales by analyzing those
products data. This is a study based on the assumption that products with similar occurrence
words will have similar properties. Laver, Benoit and Garry (2003) conducted a study to
identify the policy position by analyzing the text generated by politicians. Laver, Benoit and
Garry (2003) began their study from the assumption that text contains information. They
demonstrated a method of comparing and analyzing texts from politicians based on reference
texts regarding past policy positions.
Lastly, there are studies using semantic network analysis. This is a method of analyzing the
meaning of the content by deriving the role of each word and the relationship between words
by representing words as nodes and links. This is the use of co-occurrence in which a specific
word is mentioned along with other words shared with it, and concepts with a similar context
(Freeman, 2005). Lee and Jung (2019) applied semantic network analysis to current social
sustainability literature over time. Ruiz and Barnett (2015) conducted a semantic network
analysis to determine how HPV vaccine information is presented online and what concepts
co-occur. Zarei, Sharifi and Chaghouee (2017) conducted a semantic network analysis to
examine the delay causes of complex construction project in the Oil-Gas-Petrochemical
sector. Jiang, Barnett and Taylor (2016) used a semantic network analysis to understand the
national political environment affecting the formation of news frames on international
political issues.
2.3. Prior Studies on the Application of CISG in China

China has introduced and utilized the CISG since 1988. The CISG not only had a major
influence on the formation of domestic contract law in China but is also evaluated as taking
an important position in China’s international transactions and legal practices. This is due to
the fact that the CISG was used as a reference to enact the Contract Law of the People’s
Republic of China on October 1, 1991 to unify the existing economic contract law, foreign
economic contract law, and technology contract law (Yongping and Weidi, 2008).
Accordingly, the Supreme People’s Court and the China International Economic and Trade
Arbitration Commission (CIETAC) pay close attention to the CISG, and there are studies
being actively conducted on the Chinese trade claims in which the CISG is applied. Related
studies mainly include Chinese CISG reservations (Wang and Andersen, 2004), their impact
on domestic law (Zeller, 1999), the comparison between the Contract Law of the People’s
Republic of China and the CISG (Shen, 1997; Wang, 1989a/1989b/1989c; Yang, 2004), and
the application of CISG in arbitration practice (Jacobs and Huang, 2005; Mohs and Zeller,
2006; Wu, 2005).
A representative study on cases of CISG application in China was done by Yongping and
Weidi (2008). This study was conducted on the scope of the CISG, the CISG and party
autonomy, the CISG and trade customs, formal validity of the contract, product
nonconformity and notification obligation, buyer inspection obligation, etc. based on China’s
CISG legal cases published by the Pace Law School. Shulman and Singh (2010) argued that
China’s CIETAC judged foreign companies without prejudice through an analysis of the
Chinese arbitration cases applying the CISG. Ramaswamy (2017) conducted a study through
legal case analysis on the applicability and limitations of the CISG in the sale of goods in China
and Brazil. Yu Cheon (2018) studied the application of the relational contract theory by using
the mutuality-flexibility frame targeting the decisions of applying CISG in China.
The preceding studies above have adopted the method of deriving issues and presenting
implications based on the legal interpretation of provisions on the application of the CISG’s
176
governing laws, obligations of the trading parties, contract violations and compensation for
damages, etc. and related legal cases. This is effective for application to individual cases, but
has limitations in the analysis of overall trade claims trends. Therefore, the goal of this study
is to analyze the contents of trade claims by applying text mining for the all cases in China in
which the CISG was applied.
3. A Text Analytic Framework

3.1. Text
The subjects of this study are 81 cases of arbitration and court judgments from China
published on the website of the United Nations Commission on International Trade Law
where the CISG was applied. Individual legal cases can be found on the website of the United
Nations Commission on International Trade Law. The total number of words is 36,848. The
document structure for legal cases where the CISG was applied consists of case identification,
UNCITRAL abstract, classification of issues present, citation, and case text. The subject of
analysis in this study is the content of the claim and the content of the judgment on the claim.
Therefore, data was collected excluding the general information of the case.
The contents of the case text first introduce the contract and the process of the transaction,
and the items to be traded, the nationality of the trading party, the payment method, whether
or not it was a split transaction, and incoterms are presented. Subsequently, the process and
cause leading to the claim are presented, and the arguments of the trading parties are
described. In the second half, a case text is presented which describes the results of the dispute
between trading parties and the provisions on which the judgment is based. Lastly are the
contents of the decision of the responsible party and compensation for damages.
3.2. Analytic Procedure

The goal of this study is to analyze the content characteristics of 81 trade claims cases at the
national level and verify validity by applying a text mining technique. Validation is performed
by comparing the manually analyzed result with the automatically analyzed result (Evans et
al., 2007; Laver, Benoit and Garry, 2003). The manual analysis result is the cluster analysis
wherein the researcher reads and codes the case. The automatic analysis result is an analysis
applying the text mining technique to the result of the cluster analysis.
The cluster analysis result is used as a standard for comparison of the internal validity of
content analysis through text mining. Internal validity compares the similarity between the
content identified through manual work and the content estimated through topic modeling.
In other words, the content validity of text mining results can be verified by comparing the
result of topic modeling derived through text mining with the result of cluster analysis derived
manually (Newman et al., 2010). The degree of agreement between the manual analysis result
by the author and the automatic analysis result by computer means the degree of applicability
of the text mining technique.
The detailed analysis procedure is as follows. First, 81 cases collected are individually
documented. For cluster analysis, data is converted by coding based on the contents of the
claims in individual cases. The trade claim content is identified based on the information
contained therein after reading the case. For example, the keywords such as ‘seller, buyer,
goods, L/C, failure, breach of contract, damage’ indicate that the content of a trade claim in
the case is related to payment. In this way, they are coded as ‘Refusal of Product
Acceptance=1’, ‘Failure of Product Acceptance=2’, ‘Refusal of Payment=3’, ‘No Issuance of
177
L/C= 4’, ‘Nonpayment=5’, ‘Delay of Delivery=6’, ‘Failure of Delivery=7’, ‘Refusal of
Delivery=8’, ‘Product Nonconformity=9’. At this time, the refusal and failure of receiving
goods are classified based on the words used in the judgment while filing a claim, and delay,
failure, and refusal of delivering goods are also classified based on the words used in the
judgment. The party at fault is coded by classifying it into ‘seller=1’ and ‘buyer=2’.
After developing the coding manual, two cross-checks are conducted on each piece of
research with one researcher who majored in international commerce to secure the validity.
Cluster analysis is conducted based on the trade claims and party at fault to classify cases.
To determine the number of clusters, a hierarchical cluster analysis is performed using the
squared Euclidean distance, and the appropriate number of clusters is determined based on
the coefficient growth rate according to the average linkage method. K-means cluster analysis
is performed using the number of clusters determined through hierarchical cluster analysis.
Fig. 1. Analytic Procedure
Prior to text mining for each cluster, data pre-processing is performed for all cases. A
thesaurus list is created by searching for words used in the text, and a defined words list is
created so that proper nouns composed of several words can be recognized as a single word.
After that, words that are not applied to the topic, such as specific items and units of the
product, and words that are judged not to be directly related to the content of trade claims
such as CISG, case, court, arbitration, etc., are included in the exception list. After that, all of
the cases go through a process of review via word frequency analysis to determine whether
words that match the research purpose appear.
After conducting topic modeling to derive detailed topics for each cluster, the internal
validity of the topics is verified through comparison with the cluster analysis results by
reviewing the high ranking words. After that, semantic network analysis is conducted to
analyze the contents of each topic. At this time, the length of the word is set to 2 or more, the
window size indicating the formation of links of adjacent words is set to 2, and a semantic
network is formed using the top 50 words. The learning method uses Markov Chain Monte
Carlo, and the number of iterations is set to 1,000. Based on the semantic network analysis
results, the network centrality index for each topic is calculated and analyzed.
3.3. Analytic Strategy

3.3.1. Cluster Analysis
This study categorizes trade claims cases by the cluster analysis. Individuals within
classified clusters are classified so that they have as similar characteristics as possible and are
178
different between clusters (Hair et al., 2006). In particular, cluster analysis is often used in the
case of grouping in exploratory situations like this study because grouping is based on the
characteristics revealed without prior information about the group (Hardiman et al., 1990).
For the data composed of the entire sample, the number of clusters ‘K’ is determined, and the
central point is calculated for the corresponding variable in each cluster so that all subjects
are assigned to the cluster for analysis. Here, K is the number of clusters derived via
hierarchical cluster analysis. The goal of the cluster analysis result is to confirm the internal
validity of the text mining result.
3.3.2 Text Mining Techniques

This study is designed to explore the applicability of text mining techniques to digitalized
trade claims cases using topic modeling and semantic network analysis.
a) Topic Modeling
Topic modeling is a probability modeling algorithm to extract topics that are hidden latent
variables through text analysis. This is possible because words are used to describe a particular
topic. The topic modeling technique can derive meaningful topics by mechanically analyzing
the contents of the entirety of the text data according to a certain algorithm, so its use in
content analysis research is increasing rapidly. Topic modeling is a text mining technique
based on a probabilistic framework, and has the advantage of automatically constructing and
understanding large amounts of text data, enabling search and summarize (Hornik and Grün,
2011). A topic is a latent variable in the text, meaning the context revealed through the
connection of words appearing in the document (Tong and Zhang, 2016). In other words, it
is a method that allows the topic to be estimated via words classified through machine
learning and natural language processing. Topic modeling assumes that words in the
document are not randomly selected, but they are generated by combining the word
distribution of the topic and the topic distribution of the document. Accordingly, it is possible
to estimate the distribution of topics and the distribution of topics within the document.
Topic modeling does not require pre-determining the category of code or meaning, and it is
automatically classified by specifying the number of topics. In this regard, it is considered
more inductive than traditional text mining methods (Nahm Choon-Ho, 2016). This is
because the process of deriving the topic is clear, so it is highly replicable by other researchers
and does not depend much on prior knowledge.
In order to verify the validity of the topic extracted through topic modeling, internal or
external validity can be used. The purpose of internal validity is to compare the similarity
between the content identified through manual work and the content estimated through topic
modeling. In this study, after the researcher clusters the manually coded data, topic modeling
is performed for each cluster to review the words constituting each topic to verify the internal
validity of each topic. When the topic derived through text mining and the contents of cluster
analysis coincide with each other, it is considered that the validity is secured.
b) Semantic Network Analysis

Semantic Network Analysis is an application of Social Network Analysis to text
(Wasserman and Faust, 1994). SNA uses an object as a connection point while semantic
network analysis sets the word as a node and analyzes the structural relationship of the words
that make up the message in the text. Semantic Network Analysis is a research method used
179
to identify potential topics by identifying the relationship between words in the corpus. In
particular, each word in the corpus is set as one node, and the form and strength of the
relationship between each node, as well as the location of the node on the network, are
presented using the centrality index. By doing so, it reveals the context of the words more
clearly (Jang and Barnett, 1994).
SNA uses indices such as density, centrality, and centralization in order to understand
structural characteristics (Hansen, Shneiderman and Smith, 2009). Density refers to the
overall connection strength between nodes in the network. This is measured by the number
of connected lines, and a high density means a network with a strong relationship. Centrality
is an index that explains what position a node occupies in the overall network structure, and
refers to the relative importance of a node. Representative centrality indices include degree
centrality, between centrality, closeness centrality, and eigenvector centrality (Freeman,
2005).
In this study, degree centrality, between centrality, and eigenvector centrality are used to
understand the characteristics of word networks within a cluster of cases applying the CISG.
This is because degree centrality can identify words that play a key role in the cluster of cases
and between centrality is useful for grasping the topic of the network by connecting key
words. Eigenvector centrality reveals the content that is important to the topic.
4. Results
4.1. Descriptive Statistics
The results of the analysis of the descriptive statistics of the Chinese cases applying the CISG
are as follows. First, there were 42 cases (51.9%) of sellers and 39 cases (48.1%) of buyers as
claimant. For the contents of claims, 25 cases (30.9%) were Product Nonconformity, 14 cases
(17.3%) were Nonpayment, 9 cases (11.1%) were Refusal of Product Acceptance and Refusal
of Payment respectively, 8 cases (9.9%) were No Issuance of L/C. For party at fault, 35 cases
(43.2%) were attributable to the seller and 46 cases (56.8%) were attributable to the buyer.
4.2. Result of Cluster Analysis

In order to determine the number of clusters, a hierarchical cluster analysis was conducted
based on the trade claim content and party at fault. K-means cluster analysis was performed
by applying four, which is the number of clusters determined. The party at fault and claim
contents by cluster are shown in Table 1 below. Cluster 1 is the Buyer at fault, Product
Acceptance group, and 10 cases are clustered. Cluster 2 is the Seller at fault, Delivery group,
and 14 cases are clustered. Cluster 3 is Buyer at fault, Payment and L/C group, and 31 cases
are clustered. Cluster 4 is Seller at fault, Product Nonconformity group, and 26 cases are
clustered. The differences between groups in each cluster for party at fault were verified using
Duncan when homogeneity of variance was secured according to Levene’s test and Dunnett
T3 was used when homogeneity of variance was rejected. The difference between clusters’
claim content was verified through Fisher’s Exact Test. It was clustered so that differences
between clusters occur at 99%.
180
Table 1. Result of Cluster Analysis
Post hoc
Item Cluster 1 Cluster 2 Cluster 3 Cluster 4
(F Value)
Party at Seller 1 11 0 23 Cluster 1,3
Fault Buyer 9 3 31 3 > Cluster 2,4
(F=60.641***)
Claim Refusal of Product 9 0 0 0 Fisher’s exact
Content Acceptance Test
Failure of Product 1 0 0 0 =
Acceptance 164.068***
Refusal of Payment 0 0 9 0
Issuing L/C 0 0 8 0
Nonpayment 0 0 14 0
Delay of Delivery 0 2 0 0
Failure of Delivery 0 6 0 0
Refusal of Delivery 0 6 0 0
Product 0 0 0 26
Nonconformity
No. of Cases 10 14 31 26
Notes: 1. Fisher’s Exact Test conducted due to the existence of low expected frequency.
2. *** P<0.01.
4.3. Result of Text Mining on Each Cluster of Trade Claim Collections1

4.3.1. Result of Word Frequency
Prior to topic modeling analysis, word frequency analysis was performed for the entire text.
Among the 36,848 words, a total of 15,855 words were extracted by excluding similar words,
designated words, and exception words. The most frequently used word in all judgments was
buyer, 1286 times (8.11%), followed by seller (1078 times, 6.80%), good (684 times, 4.31%),
payment (319 times, 2.01%), order (265 times, 1.67%), deliver (261 times, 1.65%), claim (240
times, 1.51%), price (210 times, 1.32%), breach of contract (192 times, 1.21%), L/C (157 times,
0.99%). The top 20 words are presented in Table 2.
Table 2. Result of Word Frequency

Rank Word Freq. Ratio Rank Word Freq. Ratio
1 buyer 1286 8.11 11 damage 154 0.97
2 seller 1078 6.80 12 interest 132 0.83
3 goods 684 4.31 13 loss 128 0.81
4 payment 319 2.01 14 request 122 0.77
5 order 265 1.67 15 issue 111 0.70
6 deliver 261 1.65 16 quality 109 0.69
7 claim 240 1.51 17 time 105 0.66
8 price 210 1.32 18 inspect 104 0.66
9 breach of contract 192 1.21 19 shipping 102 0.64
10 L/C 157 0.99 20 obligation 101 0.64
1
As a result of conducting topic modeling for each of the four clusters, it was found that all of the four
clusters were composed of two topics, and topics related to ‘damage compensation’ were common
within all clusters. This study aims to derive the contents of trade claims through text mining, and the
presentation of the results on the topic of ‘damage compensation’ is omitted.
181
This result shows that a corpus properly reflects the characteristics of trade claims because
trading partners, goods, the price of the goods, payment conditions, orders, and payments are
in the top rank. Also, words such as delivery time, product quality and nonconformity, as well
as inspection, damage and loss, cost, and interest are ranked next.
4.3.2. Results of Text Mining on Cluster 1: ‘Buyer at fault - Product Acceptance’

Cluster
According to cluster analysis based on the manual work, Cluster 1 appeared as a trade claim
related to ‘Buyer at fault - Product Acceptance’. Ten of the 81 cases fall into this. And topics
that were latent in the 10 cases in Cluster 1 were extracted using topic modeling techniques.
As a result, Cluster 1 was divided into two topics. The words mainly used in the target topic
were in the order of ‘buyer, seller, goods, shipping, payment, deliver, loss, claim, cost,
obligation’ showing that it is related to the shipment and delivery of goods by buyers and
sellers.
Semantic network analysis was conducted to more clearly understand the content of the
target topic of Cluster 1. As a result, the network showed high degree centrality in ‘buyer,
goods, seller, payment, shipping, cost, deliver, obligation, data, defect’. On the other hand,
between centrality was high in ‘B/L, shipping, inspect, payment, error, reasonable, data’. At
this time, words such as ‘refuse, request, loading, notify’ did not appear in degree centrality
or between centrality, but appeared in eigenvector centrality. The text mining analysis result
shows that Cluster 1 is about ‘Product Acceptance’, which is consistent with the cluster
analysis result. The cases where the buyer refuses to accept the goods are generally document
nonconformity, quality nonconformity, and noncompliance with the delivery date. This can
be inferred from words such as ‘refuse, error, date, request, notify’.
Table 3. Results of Text Mining on Cluster 1

Result of Topic Result of Semantic Network Analysis
Modeling
Rank (Words Degree Centrality Between Centrality Eigenvector Centrality
Ratio=52%)
Word Freq. Word Index Word Index Word Index
1 buyer 138 buyer 0.688 buyer 0.323 buyer 0.488
2 seller 129 good 0.583 seller 0.185 good 0.467
3 good 85 seller 0.563 good 0.168 seller 0.454
4 shipping 34 payment 0.313 BL 0.089 deliver 0.289
5 payment 30 shipping 0.292 shipping 0.060 payment 0.248
6 deliver 25 cost 0.271 inspect 0.048 shipping 0.215
7 loss 24 deliver 0.229 payment 0.045 refuse 0.153
8 claim 18 obligation 0.188 error 0.042 request 0.136
9 cost 17 date 0.167 reasonable 0.037 loading 0.106
10 obligation 14 defect 0.167 date 0.023 notify 0.096
4.3.3. Results of Text Mining on Cluster 2: ‘Seller at fault – Delivery’ Cluster

related to ‘Seller at fault – Delivery’. 14 of the 81 cases fall into this. Topics that were latent in
the 14 cases in Cluster 2 were extracted using topic modeling techniques. As a result, Cluster
182
2 was divided into two topics. The words mainly used in the target topic were ‘buyer, seller,
deliver, goods, L/C, order, breach of contract, payment’ showing that it is related to the
delivery and payment.
target topic of Cluster 2. As a result, the network showed high degree centrality in the order
of ‘seller, buyer, goods, deliver, payment, L/C, claim, breach of contract, agreement, time’.
Between centrality was in the order of ‘buyer, seller, goods, deliver, L/C, payment, claim, date,
breach of contract, cost’, and eigenvector centrality was in the order of ‘seller, buyer, deliver,
goods, claim, payment, order, breach of contract, L/C, date’. In the semantic network of the
target topic of cluster 2, words with high degree centrality have high between centrality and
high eigenvector centrality. At this time, the high degree centrality and eigenvector centrality
of ‘seller’ and ‘deliver’ indicates that the network is mainly composed of contents related to
seller delivery. The text mining analysis result shows that Cluster 2 is about ‘Deliver &
Payment’. This is consistent with the cluster analysis result. The main causes of claims related
to Deliver include seller’s refusal of delivery or non-compliance with deliver data by seller due
to buyer’s non-payment. This can be inferred from the co-occurrence words ‘payment, L/C,
date’.

Modeling
Ratio=67%)
1 buyer 171 seller 0.837 buyer 0.263 seller 0.513
2 seller 161 buyer 0.796 seller 0.258 buyer 0.491
3 deliver 64 good 0.612 good 0.110 deliver 0.376
4 goods 62 deliver 0.551 deliver 0.060 good 0.330
5 L/C 46 payment 0.408 L/C 0.040 claim 0.172
6 order 39 L/C 0.347 payment 0.036 payment 0.168
7 breach of 35 claim 0.306 claim 0.020 order 0.148
contract
8 payment 32 breach of 0.286 date 0.020 breach of 0.136
contract contract
9 claim 30 agreement 0.265 breach of 0.020 L/C 0.132
contract
10 damage 21 time 0.265 cost 0.010 date 0.111
4.3.4 Results of Text Mining on Cluster 3: ‘Buyer at fault – Payment and L/C’ Cluster
related to ‘Buyer at fault – Payment and L/C’. 31 of the 81 cases fall into this. Topics that were
latent in the 31 cases in Cluster 3 were extracted using topic modeling techniques. As a result,
Cluster 3 was divided into two topics. The words mainly used in the target topic were in order
of ‘buyer, seller, goods, payment, order, delivery, price, obligation, fail, resell’ showing that it
is related to payment between trading parties.
183
of ‘buyer, seller, goods, payment, deliver, price, order, obligation, agreement, data’. Between
centrality of ‘buyer, seller, goods, payment, deliver, price’ was in the same order as degree
centrality but ‘agreement, order, fail, present’ demonstrated a difference. Eigenvector
centrality was the highest in ‘payment’ followed by ‘buyer, goods, seller, deliver, price, order,
obligation, fail, agreement’ in that order. In the semantic network of the target topic of cluster
3, there was not much difference between words in terms of degree centrality and between
centrality. The fact that payment is high in eigenvector centrality indicates that the network
is composed of contents related to payment. The text mining analysis result shows that
Cluster 3 is about ‘Payment’. This is consistent with the cluster analysis result. The main
causes of claims related to payment include breach or failure of the buyer’s obligation to pay
payment at the time of delivery on agreement to supply of goods. This can be inferred from
the co-occurrence words ‘fail, price, obligation, agreement’.

Modeling
Rank (Words Ratio=46%) Degree Centrality Between Centrality Eigenvector Centrality
1 buyer 347 buyer 0.660 buyer 0.222 payment 0.485
2 seller 292 seller 0.638 seller 0.183 buyer 0.475
3 good 185 good 0.617 good 0.177 good 0.448
4 payment 159 payment 0.553 payment 0.131 seller 0.386
5 order 76 deliver 0.468 deliver 0.063 deliver 0.271
6 deliver 72 price 0.362 price 0.047 price 0.205
7 price 59 order 0.298 agreement 0.024 order 0.140
8 obligation 35 obligation 0.255 order 0.021 obligation 0.127
9 fail 32 agreement 0.213 fail 0.013 fail 0.093
10 resell 21 date 0.213 present 0.012 agreement 0.061
4.3.5 Results of Text Mining on Cluster 4: ‘Seller at fault – Product Nonconformity’

Cluster
related to ‘Seller at fault – Product Nonconformity’. 26 of the 81 cases fall into this. Topics
that were latent in the 26 cases in Cluster 4 were extracted using topic modeling techniques.
As a result, Cluster 4 was divided into two topics. The words mainly used in the target topic
were ‘buyer, seller, goods, claim, damage, inspect, quality, breach of contract, defect’ showing
that it is related to product nonconformity.
of ‘buyer, seller, goods, inspect, claim, time, quality, damage, defect, period’. Between
centrality was in the order of ‘buyer, seller, goods, inspect, quality, time, period, claim, defect,
contractual’ and eigenvector centrality was in the order of ‘buyer, seller, goods, claim, damage,
inspect, breach of contract, defect, quality’. The text mining analysis result shows that Cluster
4 is about ‘Product Nonconformity’. This is also consistent with the existing cluster analysis.
These are cases where trade claims are caused by product nonconformity or defect.
184
Modeling
Ratio=51%)
1 buyer 630 buyer 0.792 buyer 0.345 buyer 0.531
2 seller 496 seller 0.583 seller 0.157 seller 0.450
3 good 352 good 0.521 good 0.107 good 0.384
4 claim 142 inspect 0.417 inspect 0.067 claim 0.382
5 damage 88 claim 0.333 quality 0.059 damage 0.229
6 inspect 86 time 0.313 time 0.044 Inspection 0.142
7 quality 78 quality 0.292 period 0.028 breach of 0.137
contract
8 breach of 76 damage 0.271 claim 0.027 defect 0.118
contract
9 defect 66 defect 0.271 defect 0.022 quality 0.114
10 time 60 period 0.250 contractual 0.020 compensation 0.109
According to the above findings, the results derived through topic modeling and semantic
network analysis are consistent with the characteristics of the group derived through cluster
analysis based on the manual work. Thus, it can be seen that the content of trade claims of
each cluster by the manual method and by the automatic method are very similar.
Table 7. Comparison of Two Approaches in Each Cluster

Result of Cluster Analysis Results of Text Miming
Cluster 1 Buyer at fault – Product Acceptance Product Acceptance
Cluster 2 Seller at fault – Delivery Delivery & Payment
Cluster 3 Buyer at fault – Payment & L/C Payment
Cluster 4 Seller at fault – Product Nonconformity Product Nonconformity
Each cluster is named based on the characteristics of the cases included in the cluster: the
trade claim content classified by the researchers & the party at fault. Two criteria are named
based on information obtained by researchers after reading legal cases. For example, in the
legal case, the following sentence means Seller at fault: The tribunal stated that the seller
would be liable for latent defects where the buyer had raised a claim within two years after the
acceptance of the flanges. And the trade claim content is identified based on the information
contained therein after reading the text such as ‘seller, buyer, goods, damage, inspect, quality,
defect, etc.’. Those keywords indicate that the content of a trade claim is related to Product
Nonconformity 2 . Therefore, this case belongs to cluster 4(Seller at fault - Product
Nonconformity) according to k-means cluster analysis. Cluster 4 is a group consisting of
Seller at fault - Product Nonconformity cases. The word frequency and centrality index of the
text mining on Cluster 4 clearly reveal the contents of Cluster 4. The keywords that the
researcher obtained manually information to classify the trade claim content appear the same
in the text mining results.
Therefore, it is confirmed that text mining technique can be used as a statistical method to
typify cases and interpret their meaning. Furthermore, when topic modeling and semantic
2
The sentence and words are part of case 770 used in the study.
185
network analysis are applied, it is possible to grasp which concepts have important meanings
in the process of calculating the centrality index. There is also an advantage in that the results
can be interpreted in a richer manner because the connectivity with other concepts can be
considered at the same time.
5. Conclusion
This study is designed to verify the validity of the application of text mining for trade claims
research. To this end, 81 cases applying the CISG in China published on the UNCITRAL
website were collected and used. First, 81 cases were converted into data by applying the
existing method and cluster analysis was then conducted. Finally, text mining techniques
were applied to each cluster afterwards.
The summary of the analysis results is as follows. First, the cases applying the CISG were
divided into four clusters by the claim content and the party at fault, ‘Buyer at fault – Product
Acceptance’, ‘Seller at fault – Delivery’, ‘Buyer at fault - Payment & L/C’, ‘Seller at fault -
Product Nonconformity’. According to the text mining results for each cluster, Cluster 1,
‘Buyer at fault – Product Acceptance’, showed high frequency and centrality in ‘shipping,
deliver, defect, B/L, inspect, notify, etc.’ and appeared to be related to trade claims that
occurred during the product acceptance process. Cluster 2, ‘Seller at fault – Delivery’, showed
high frequency and centrality in ‘deliver, time, date, L/C, payment, etc.’ and appeared to be
trade claims in connection with product delivery and payment. Cluster 3, ‘Buyer at fault -
Payment & L/C’, showed a high frequency and centrality in ‘payment, fail, resell, date, price,
etc.’ indicating trade claims related to payment. Cluster 4, ‘Seller at fault - Product
Nonconformity’, showed a high frequency and centrality in ‘damage, inspect, quality, defect,
period, etc.’ indicating trade claims related to product nonconformity.
The implications of this study are as follows. First, the validity of the text mining technique
in the study of trade claim cases is confirmed. Table 7 shows the comparison of the cluster
analysis results and text mining results. Specifically, it was found that the characteristics of
the cluster and the details derived through text mining coincide. Also, the degree centrality of
words that play a key role in the topic was high as was the between centrality of words that
are useful for grasping the topic and the eigenvector centrality of the important words is high
in the topic. Therefore, the results of cluster analysis and text mining results are similar to
each other and confirm that the results of trade claim analysis through text mining has
internal significance. This indicates that text mining techniques can be applied to research on
content analysis of trade claims.
Secondly, this study has implications in that it is an attempt to quantitatively study the trade
claim cases, whereas prior trade claim cases were mainly studied via qualitative methods. This
is because qualitative research is suitable for interpretation of legal texts and case analysis
accordingly (Ramaswamy, 2017; Wang and Adnserson, 2004; Yongping and Weidi, 2008; Yu,
2019). However, this approach has a limitation in that it is difficult to identify the common
characteristics of multiple cases. In this study, text mining was conducted to quantitatively
analyze the contents of trade claim cases. As a result of the analysis, it was possible to secure
a result with internal validity. Recently, there have been an increasing number of research
cases applying text mining techniques to unstructured text data. Text mining refers to the
process of extracting high-quality information from a large amount of text (Ponweiser, 2012).
The application of this methodology is expected to provide new implications to the
methodologies used for research on trade cases, which have previously remained centered on
individual case analysis.
186
Lastly, the usefulness of text mining techniques for content analysis of trade claims is
confirmed. There are 10 trade claim cases in Cluster 1. Reading all 10 cases, grasping and
organizing the contents by classifying them requires considerable effort and time. However,
if you examine the words with high centrality shown in Table 7, you can easily see that it is
made up of claims related to the acceptance of products. The same method can be applied to
the remaining the clusters. The traditional method of content analysis is not only time-
consuming and expensive, but also limited in its use as it requires skilled labor (Laver, Benoit
and Garry, 2003; Evans et al., 2007). This study shows that the use of the text mining can
lower the barrier for acquiring information from a large amount of text.
However, this study has the following limitations. First, the application of text mining is
useful for understanding the types of trade claims but it appears that additional analysis is
needed regarding the legal principles and the presence or absence of the liability of the claim.
Second, comparative analysis is necessary to examine the differences in the characteristics of
Chinese trade claims and other countries. Since this study was conducted only on Chinese
trade claim cases, it is somewhat unreasonable to interpret the results of the study as unique
characteristics of Chinese trade claims. Lastly, this study was conducted on only 81 cases
published on the website of the United Nations Commission on International Trade Law
among cases applying the CISG in China. In order to derive more generalized results, it is
necessary to secure a larger amount of data. This limitation can be overcome by research in
the future.
References
Allee, T., M. Elsig and A. Lugg (2017), “The Ties between the World Trade Organization and
Preferential Trade Agreements: A Textual Analysis”, Journal of International Economic
Law, 20(2), 333-363.
Berelson, B. (1952), Content Analysis in Communication Research, Glencoe, Ill: Free Press.
Deerwester, S., S. T. Dumais, G. W. Furnas, T. K. Landauer, T. K. and R. Harshman (1990), “Indexing
by Latent Semantic Analysis”, Journal of the American Society for Information Science, 41(6),
391-407.
Evans, M., W. McIntosh, J. Lin and C. Cates (2007), “Recounting the Courts? Applying Automated
Content Analysis to Enhance Empirical Legal Research”, Journal of Empirical Legal Studies,
4(4), 1007-1039.
Freeman, L. C. (2005), “Graphic Techniques for Exploring Social Network Data”. In P. J. Carrington,
J. Scott and S. Wasserman (Eds.), Models and Methods in Social Network Analysis, New York,
NY: Cambridge University Press, 248-269.
Gentzkow, M., B. Kelly and M. Taddy (2019), “Text as Data”, Journal of Economic Literature, 57(3),
535-74.
Ghani, R. and A. E. Fano (2002, December 9-12), “Using Text Mining to Infer Semantic Attributes
for Retail Data Mining”, 2002 IEEE International Conference on Data Mining, Maebashi City,
Japan.
Hair, J. F., W. C. Black, B. J. Babin, R. E. Anderson and R. L. Tatham (2006), Multivariate Data
Analysis (6th ed.), Upper Saddle River, NJ: Pearson Prentice Hall.
Hall, M. A. and R. F. Wright (2008), “Systematic Content Analysis of Judicial Opinions”, Calif. L.
Rev., 96, 63-122.
Hansen, D., B. Shneiderman and M. Smith (2009), “Analyzing Social Media Networks: Learning by
Doing with NodeXL”, Computing, 28(4), 1-47.
Hardiman, R. T., R. Lacey and Y. M. Yi (1990), “Use of Cluster Analysis for Identification and
187
Classification of Farming Systems in Qingyang County, Central North China”, Agricultural
systems, 33(2), 115-125.
Harris, Z. S. (1954), “Distributional Structure”, Word, 10(2-3), 146-162.
Holsti, O. R. (1969), Content Analysis for the Social Sciences and Humanities, Reading, MA:
Addison-Wesley Publishing Company.
Hornik, K. and B. Grün (2011), “An R Package for Fitting Topic Models”, Journal of Statistical
Software, 40(13), 1-30.
Iselin, D. and B. Siliverstovs (2013), “The R-word index for Switzerland”, Applied Economics
Letters, 20(11), 1032-1035.
Jacobs, M. S. and Y. Huang (2005), “An Arbitrator’s Power and Duties Under Art. 114 of Chinese
Contract Law in Awarding Damages in China in Respect of a Dispute under a Contract
Governed by CISG”, Mealey’s International Arbitration Report, 20, 39-43.
Jang, H. Y. and G. A. Barnett (1994), “Cultural Differences in Organizational Communication: A
Semantic Network Analysis 1”, Bulletin of Sociological Methodology/Bulletin de Méthodologie
Sociologique, 44(1), 31-59.
Jang Heung-Hoon (2002), “Study on the Chinese International Commercial Culture and Negotiation
Strategy with the Chinese”, Journal of Korea Culture Industry, 2(1), 173-188.
Jiang, K., G. A. Barnett and L. D. Taylor (2016), “Dynamics of Culture Frames in International News
Coverage: A Semantic Network Analysis”, International Journal of Communication, 10, 3170-
3736.
Kumar, A. and A. Paul (2016), Mastering Text Mining with R, Birmingham, UK: Packt Publishing
Ltd.
Laver, M., K. Benoit and J. Garry (2003), “Extracting Policy Positions from Political Texts Using
Words as Data”, American Political Science Review, 97(2), 311-331.
Lee Ju-Yeon, Seung-Hwan Han and Ki-Seok Kwon (2015), “A Study on Korean Legal Research
Trend During the Last 10 Years based on Keyword Network Analysis”, Ajou Law, 8(4), 519-
539.
Lee, K. and H. Jung (2019), “Dynamic Semantic Network Analysis for Identifying the Concept and
Scope of Social Sustainability”, Journal of Cleaner Production, 233, 1510-1524.
Lin, D. and P. Pantel (2001), “Discovery of Inference Rules for Question-Answering”, Natural
Language Engineering, 7(4), 343-360.
Manning, C., P. Raghavan and H. Schütze (2010), “Introduction to Information Retrieval”, Natural
Language Engineering, 16(1), 100-103.
McDonald, S. and M. Ramscar (2001), “Testing the Distributional Hypothesis: The Influence of
Context on Judgements of Semantic Similarity”, Proceedings of the Annual Meeting of the
Cognitive Science Society, 23 (23), 611-616.
Mohs, F. and B. Zeller (2006), “Commentary: Penalty and Liquidated Damages Clauses in CISG
Contracts Revisited”, Mealey’s International Arbitration Report, 21(6), 1-5.
Nahm Choon-Ho (2016), “An Illustrative Application of Topic Modeling Method to a Farmer’s
Diary”, Cross-Cultural Studies, 22(1), 89-135.
Newman, D., J. H. Lau, K. Grieser and T. Baldwin (2010, June 2-4), “Automatic Evaluation of Topic
Coherence”, Human Language Technologies: The 2010 Annual Conference of the North
American Chapter of the Association for Computational Linguistics, Los Angeles, California,
USA.
Ponweiser, M. (2012), Latent Dirichlet Allocation in R (Diploma Thesis), Vienna: WU Vienna
University of Economics and Business.
Porta, R. L., F. Lopez-de-Silanes, A. Shleifer and R. W. Vishny (1998), “Law and Finance”, Journal
of Political Economy, 106(6), 1113-1155.
Ramaswamy, M. P. (2017), “The Scope of CISG in China and Brazil and Facilitation of Trade with
188
Portuguese Speaking Countries”, International Journal of Social Sciences, 3(2), 78-92.
Ruiz, J. B. and G. A. Barnett (2015), “Exploring the Presentation of HPV Information Online: A
Semantic Network Analysis of Websites”, Vaccine, 33(29), 3354-3359.
Salton, G., A. Wong and C. S. Yang (1975), “A Vector Space Model for Automatic Indexing”,
Communications of the ACM, 18(11), 613-620.
Shen, J. (1997), “Declaring the Contract Avoided: The UN Sales Convention in the Chinese Context”,
New York International Law Review, 10 (1), 7-57.
Shin, Koon-Jae (2007), “An Empirical Study on the Relationship between the Characteristics of Firms
and Contracts and the Experience of Claims”, Korea Trade Review, 32(5), 155-178.
Shulman, M. R. and L. Singh (2010), “China’s Implementation of the UN Sales Convention through
Arbitral Tribunals”, Columbia Journal of Transnational Law, 48, 242-286.
Tong, Z. and H. Zhang (2016, May 21-22), “A Text Mining Research based on LDA Topic
Modelling”, The Sixth International Conference on Computer Science, Engineering and
Information Technology, Vienna, Austria.
United Nations Commission on International Trade Law (UNCITRAL) (2016), UNCITRAL Digest
of Case Law on the United Nations Convention on Contracts for the International Sale of Goods,
Vienna: UNCITRAL.
Wang, C. (1989a), “Comparison Between the CISG and the FECL (Part I)”, Chinese Legal Science,
3, 106-116.
Wang, C. (1989b), “Comparison Between the CISG and the FECL (Part II)”, Chinese Legal Science,
4, 114-116.
Wang, C. (1989c), “Comparison Between the CISG and the FECL (Part III)”, Chinese Legal Science,
5, 115-119.
Wang, X. and C. B. Andersen (2004), “The Chinese Declaration Against Oral Contracts Under the
CISG”, Vindobona Journal of International Commercial Law and Arbitration, 8, 145-165.
Weaver, W. (1955), “Translation”, Machine Translation of Languages, 14(15-23), 10.
Weber, R. P. (1990), Basic Content Analysis, Newburry Park, CA: Sage Publications.
Wu, D. (2005), “CIETAC’s Practice on the CISG”, Nordic Journal of Commercial Law, 2, 1-46.
Yang, Y. (2004), “Suspension Rules under Chinese Contract Law, the UCC and the CISG: Some
Comparative Perspectives”, China Law & Practice, 18(7), 23-27.
Yongping, X. and L. Weidi (2008), “Selected Topics on the Application of the CISG in China”, Pace
International Law Review, 20, 61-103.
Yu, Cheon (2018), “Analysis of CISG Clauses and China’s Decision Cases using Relational Contract
Theory: Focusing on the Mutuality-Flexibility Frame”, International Commerce and
Information Review, 20(4), 289-308.
Yu, C. (2019), “An Empirical Study on Trade Claim Management from a Relational Perspective”,
Journal of Korea Trade, 23(6), 14-32.
Zarei, B., H. Sharifi and Y. Chaghouee (2018), “Delay Causes Analysis in Complex Construction
Projects: A Semantic Network Analysis Approach”, Production Planning & Control, 29(1), 29-
40.
Zeller, B. (1999), “CISG and China”. In R. W. Michael (Ed.), CISG and China: Dialog Deutschland-
Schweiz Ⅶ, Geneva, Switzerland: University of Geneva, 7-22.
Zhou, J. Q. and M. W. Peng (2010), “Relational Exchanges versus Arm’s-Length Transactions during
Institutional Transitions”, Asia Pacific Journal of Management, 27(3), 355-370.
https://doi.org/10.35611/jkt.2020.24.8.189 189
MNC Subsidiary’s Entrepreneurship and

Knowledge Transfer: Evidence from MNC
Subsidiaries in South Korea* JKT 24(8)
Kangmun Lee
Department of Business Administration, Chonnam National University, South Korea
Ji Yeon Yang
Department of Global Business, The University of Suwon, South Korea
Taewoo Roh†
Department of International Trade and Commerce, Soonchunhyang University, South Korea
Abstract
Purpose – This paper attempted to verify the process by which a multinational corporation (MNC)’s
subsidiary practices entrepreneurship to create effective knowledge (KC) in the local market. We have
looked at whether subsidiary entrepreneurship (SENT) has a moderation effect in creating knowledge
for the local market when a subsidiary has been given autonomy (AUT) from the headquarters (HQ).
We also argue that when a subsidiary creates meaningful knowledge, the effect of the increased status
by the HQ within the MNC network position (NP) has an indirect effect on whether knowledge is
transferred to other overseas subsidiaries (KTO).
Design/methodology – This paper used a structural equation model (SEM) of 282 effective foreign
companies invested in Korea. To test the hypothesis about the process of SENT on KTO, descriptive
statistics, confirmatory factor analysis, reliability, convergent and discriminant validities, and
common method bias were analyzed using STATA. In addition, the moderation effect was verified
along with SEM. The moderation effect of AUT on SENT and KC was presented graphically by
confirming \mathrm{\pm1} standard deviation of AUT for the main effect.
Findings – Our findings are as follows. First, while the hypothesis about the direct effect of SENT and
KC on KTO was not supported, all other hypotheses were supported. Second, both the AUT and
moderating effect and the indirect effect of NP were significant. In the conclusion, these findings are
discussed in relation to its various theoretical and practical implications.
Originality/value – This study attempted to contribute to the knowledge creation theory of MNC by
contemplating how subsidiaries can move away from HQ and grow in the local market. Although
there is still a shortage of foreign investment in the Korean market, our practical implications offer
guidance for how current subsidiaries can develop more than other overseas subsidiaries.
Keywords: Knowledge Transfer to Overseas Subsidiaries, Multinational Corporation, Subsidiary

Autonomy, Subsidiary Entrepreneurship, Subsidiary Knowledge Creation, Subsidiary
Network Position
JEL Classifications: F23, L23, L62
1. Introduction
Since discussions on the relationship between headquarters and subsidiaries first began, the
role of subsidiaries has been recognized as an important factor in enhancing the capabilities
of multinational corporations (MNCs). No matter how excellent the headquarter (HQ) is, it
* This work was supported by the Soonchunhyang University Research Fund.
†
Corresponding author: troh@sch.ac.kr
190
is also located in one country, meaning it is limited in its ability to secure the same level of
information on various distant countries as it has on local companies. In fact, in the early
stages, foreign subsidiaries of MNCs were given the resources and knowledge of the HQ to
virtually enter the local market, accept the strategic direction of the HQ, and carry out this
passive role (Adenfelt, 2010). However, in recent years, there have been many cases of
subsidiaries helping enhance the capabilities of MNCs by creating and transferring
knowledge in the local market (Paterson and Brock, 2002; Yang, Mudambi and Mayer, 2008).
Then, more specifically, what differentiated factors did subsidiaries that accumulate and
create knowledge as active subjects have compared to other subsidiaries? In the existing
research on knowledge creation (KC) and transfer, knowledge characteristics such as tacit or
formal knowledge (Nonaka, Katushiro and Dai, 1996), the complexity of knowledge (Inkpen
and Tsang, 2005), the relationship between the knowledge created by the subsidiary and the
knowledge of HQs (Yang et al., 2008), and the knowledge transfer process (Goh, 2002; Zhang
et al., 2010) as important factors influencing successful KC and transfer. However, this
presupposes that all overseas subsidiaries act as active agents (Gupta and Govindarajan,
2000). In other words, there had been insufficient consideration of the fundamental factors
for the overseas subsidiary to act as an active entity (Minbaeva, 2007). KC and knowledge
transfer to overseas subsidiaries (KTO) can improve the competitive advantages of HQs and
other subsidiaries, but not all overseas subsidiaries can play this role. In fact, there are
differences in roles depending on the knowledge gap between overseas subsidiaries (Yang et
al., 2008), so it is important to consider the preceding factors that caused these differences.
Further, for effective reverse knowledge transfer, it is important to understand what factors
the foreign subsidiary should consider as a transferring entity.
Which subsidiaries can create knowledge more effectively? Further, which subsidiaries can
more effectively transfer this created knowledge? In this study, we examine the above two
research questions. Research on KC of MNCs has also been conducted from two perspectives.
One is the transfer of knowledge from HQ to an overseas subsidiary, and the other is when
knowledge is transferred from a specific subsidiary to the HQ or to other subsidiaries (Gupta
and Govindarajan, 2000). Initially, the foreign subsidiary of an MNC was recognized as the
entity executing the knowledge transferred from the HQ, so the research was conducted with
a narrow focus on the transfer of knowledge from the HQ to the subsidiary (Minvaeva et al.,
2003). Another early study on subsidiaries considered them as active entities that seek and
acquire new resources and capabilities rather than settling for the current situation (Ambos,
Ambos and Schlegelmilch, 2006; Birkinshaw, Hood and Young, 2005; Lee, 2011; Sumelius
and Sarala, 2008). The subsidiary is an independent organization and makes various strategic
choices in a process that is influenced by the entrepreneurship and autonomy of the
subsidiary (O’Brien et al., 2019). In this regard, existing research on KC has inexplicably
focused on the absorptive capacity of a company. However, there has been insufficient
research on how the creation of knowledge is activated or when the absorptive capacity is
achieved through knowledge sharing within an MNC’s network (Song, 2014). However, the
entrepreneurship and autonomy act (AUT) is a critical motivating factor for the
organization’s KC (Dess et al., 2003; Zahra, Nielsen and Bogner, 1999). The entrepreneurship
of a subsidiary is formed by a subsidiary that conducts business activities locally, and it can
ultimately be seen as an essential factor in expanding the contribution of subsidiaries to the
global competitiveness of MNCs (Birkinshaw, 1999).
The autonomy of subsidiaries under entrepreneurship is also vital for the transfer of created
knowledge. Yuhan-Kimberly, a joint venture subsidiary between Yuhan from Korea and
Kimberly-Clark from the United States (U.S.), sold products manufactured in Korea to the
local market. However, Yuhan Kimberly’s performance in the Korean market was
MNC Subsidiary’s Entrepreneurship and Knowledge Transfer: Evidence from MNC Subsidiaries in South Korea
191
remarkable. While Kimberly-Clark was inferior in the global market compared to its main
rival, Procter & Gamble (P&G), it showed an overwhelming market share in the Korean
market. Based on these achievements, Yuhan-Kimberly was given the authority to oversee six
Northeast Asian countries, including Korea, Japan, China, Taiwan, Hong Kong, and
Mongolia (Cho Dong-Sung, 2014). This example demonstrates the importance of subsidiary
autonomy and entrepreneurship in the transfer of subsidiary knowledge.
The extant studies have mentioned various obstacles to the transfer of knowledge of foreign
subsidiaries, that is, the transfer of reverse knowledge. This is because various situational
factors must be met to transfer and utilize knowledge from overseas subsidiaries to HQs and
other subsidiaries. For example, the transfer of knowledge from HQs to subsidiaries is natural.
However, the transfer of knowledge from a foreign subsidiary to a HQ requires the HQ to
recognize the value of the knowledge generated by the subsidiary, and even if the value is
recognized, it must undergo various verifications of the possibility of transferring knowledge
(Yang et al., 2008). This indicates that the relationship between HQs and subsidiaries and
knowledge integration mechanisms (Gupta and Govindarajan, 2000; Hakanson and Nobel,
2001) or the absorptive capacity of HQs and subsidiaries (Ambos et al., 2006), which have
been discussed primarily in previous studies on KTO, need to be further discussed and
supplemented. Accordingly, this study aims to contribute to the existing research and provide
practical implications for MNCs by presenting discussions on (1) effective KC and (2)
effective transfer of the knowledge created by active subsidiaries.
The composition of this study is as follows. In the literature review and hypotheses part,
entrepreneurship, KC, and KTO are discussed, and based on these discussions, six hypotheses
are presented. In the methodology part, samples, data collection, estimation models, and
operational definitions for each variable are described. Next, the research results, descriptive
statistics, reliability analysis, factor analysis, correlation, and structural equation model (SEM)
results are presented. Finally, in the conclusions and implications section, the study is
summarized, theoretical and practical implications are discussed, and the limitations and
future research directions are discussed.
2. Literature Review and Hypothesis Development

2.1. MNC Subsidiary’s Entrepreneurship
Entrepreneurship is a key element for a company’s survival, change, and growth (Dess et
al., 2003; Zahra et al., 1999). Entrepreneurship has also been addressed in a variety of contexts
in research on MNCs. Strategic entrepreneurship is used to innovate products, technologies,
and processes, establish new organizations to foster entrance to new sectors and markets, and
secure organizational competitive advantage. Moreover, the subsidiary initiative refers to the
active, autonomous, and risk-taking behaviors of MNCs’ overseas subsidiaries, and it is
included in the main topic of entrepreneurship research (Ahsan and Fernhaber, 2019).
Therefore, this study understands the entrepreneurship of subsidiaries as the leading acts of
subsidiaries, and our discussion proceeds according to this.
Subsidiary entrepreneurship (SENT) is formed by a unit organization that performs
management activities locally, and it can ultimately be regarded as a focal factor that expands
the subsidiary’s contribution to the global competitiveness of MNCs (Birkinshaw, 1999; Scott
et al., 2010). Existing studies of MNCs have generally discussed the roles of subsidiaries
through two approaches. The first approach recognizes the subsidiary as a passive entity that
simply follows the orders of the HQ. The HQ is at the center of an MNC’s decision-making
192
structure, and its subsidiaries are sub-organizations within the MNC, and they are subject to
coordination and integration (Gates and Egelhoff, 1986; Picard, 1980).
However, a new research approach for SENT has begun to appear. Studies using this
approach have attempted to understand HQ in the context of subsidiary management, which
differs from the conventional method of understanding subsidiaries in the context of HQ
management. As independent organizations, subsidiaries are active entities that develop
unique resources and capabilities for their survival and growth (Young et al., 1994), and in
this development process, subsidiaries make various strategic choices (Birkinshaw, 1997).
The various actions taken by a subsidiary in an attempt to survive and grow can be seen as
actions influenced by SENT (O’Brien et al., 2019; Scott, Gibbons and Coughlan, 2010). In the
stream of literature on MNCs, SENT has been discussed in various forms. Among them, as
mentioned earlier, Birkinshaw (1997) named the subsidiary’s entrepreneurship-reflected
behaviors as subsidiary initiatives and defined subsidiary initiatives as active, proactive
actions that lead to new approaches attempting to leverage and expand the resources held by
subsidiaries. Recent studies on SENT have mainly focused on the influence of
entrepreneurship on the generation of unique resources and the capabilities of subsidiaries,
as well as the relationship between entrepreneurship and various environmental factors for
subsidiaries. The level of the relationship between the HQs and subsidiaries, which is affected
by various activities (i.e., sales, evaluation, and acceptance of subsidiary ideas), is closely
related to the entrepreneurship of subsidiaries (Schmid et al., 2014).
2.2. Subsidiary Entrepreneurship (SENT) and Knowledge Creation (KC)

In general, subsidiary entrepreneurship contributes to the evolution of subsidiary roles. In
other words, the entrepreneurial subsidiary acquires a more critical role and corresponding
obligation within the MNC network (Birkinshaw and Hood, 1998; Birkinshaw, 1998). The
entrepreneurship level of the subsidiary contributes to its responsiveness to the local market
as well as its global learning and integration capabilities (Birkinshaw, 1997). Further, the
entrepreneurship of subsidiaries can help them seize various opportunities in the local
market, the internal market of MNCs, and the global market, while also contributing to the
development of new products and markets, or enabling the more efficient operation of MNCs
across the enterprise (Clark and Ramachandran, 2019). However, subsidiaries that lack
entrepreneurship tend to be less sustainable (Reilly et al., 2012). In addition, low levels of
entrepreneurship lead to inefficient operations, loss of goals, reduced motivation from the
intervention of the HQ, and lack of trust in subsidiaries (Foss et al., 2012). In particular, roles
and responsibilities corresponding to the evolution of subsidiaries include activities
iinvolving the transferring or sharing of knowledge (Lee et al., 2020). In this regard, if a
specific subsidiary aims to play a more influential role within the hierarchical network of
MNCs, a subsidiary with a higher level of entrepreneurship can achieve higher performance
(Williams and Lee, 2011). Some of the actions taken by a subsidiary to achieve
entrepreneurship may include the transfer of knowledge by the subsidiary to its HQs or to
peer subsidiaries within the multinational network (Ambos and Birkinshaw, 2010;
Dörrenbächer and Gammelgaard, 2016). Based on this, this study can derive the following
logic based on a study indicating that the entrepreneurship of subsidiaries and active activities
in the network have a positive relationship with each other. Subsidiaries cannot increase their
influence through the initiative of their subsidiaries unless they draw the attention of the HQ
(Ambos et al., 2010). Therefore, subsidiaries will focus on actively developing their
capabilities and publicizing their achievements to attract favorable attention from HQs. In
this process, entrepreneurship can make the efforts of subsidiaries more active to foster a
positive interest in HQs (Conroy and Collings, 2016). Moreover, this effort can be
193
implemented as an action of expanding knowledge transfer to peers of the subsidiary
(Bouquet and Birkinshaw, 2008; Frost et al., 2002).
In transferring a subsidiary to the HQ and peers, the first step is to create knowledge
compatible with the MNC’s global strategy. In our study, we suppose that a subsidiary’s
entrepreneurship can contribute to the creation of such knowledge. In general,
entrepreneurial-based organizations are proactively and voluntarily oriented toward risk-
taking, learning, and change (Kostova, 1999; Zahara and Covin, 1995). In addition,
entrepreneurial organizations (corporations) value active communication among members,
emphasize the development and sharing of new ideas, and provide corresponding rewards
(Mudambi, Piscitello and Rabbiosi, 2014).
Therefore, in general, the higher the level of entrepreneurship of a subsidiary, the more the
subsidiary can accumulate its knowledge and know-how in the process of grasping various
opportunities related to local business (Clark and Ramachandran, 2019) and actively
overcome them (Scott et al., 2010). The subsidiary achieves KC in the process of securing the
capability to respond to the local market, and through this process it ultimately evolves itself.
The main driving force behind this evolutionary process is the subsidiary initiative
(Birkinshaw and Hood, 1998), which this study regards to be the same as the subsidiary’s
entrepreneurship.
Birkinshaw and Hood (1998) noted that MNC’s HQs could induce innovation in
subsidiaries by providing them with new management opportunities and projects that they
have not experienced while also ensuring dynamic local activities. In other words, by making
efforts to promote the entrepreneurship of subsidiaries (such as capturing new market
opportunities), HQs ultimately induce the subsidiary to create new knowledge. Mudambi et
al. (2014) presupposed that entrepreneurship contributes to the creation of new knowledge,
and also mentioned that it plays a positive role in acquiring local knowledge of subsidiaries.
In particular, Mudambi et al. (2014) suggested that the direction of the entrepreneurship of
the subsidiary should be compatible with that of the HQ, and that the knowledge acquired or
created by the subsidiary should also have an appropriate scope compared to the knowledge
of the HQ. The KC of subsidiaries can be realized by subsidiaries accepting the
entrepreneurship of MNCs within the subsidiary organization (Lee and Williams, 2007;
Wang and Wang, 2020). In other words, entrepreneurship can contribute to the
organization’s KC activities. Scott et al. (2010) noted that entrepreneurship could help
companies establish creative strategies in response to various local environmental changes.
Further, the establishment of a creative response strategy is refined as knowledge inside the
organization is accumulated, and this can ultimately serve as the basis for the creation of new
knowledge.
H1: SENT of an MNC has a positive effect on KC for the local market.
2.3. Moderating effect of Subsidiary Autonomy (AUT) between SENT

and KC
Clark and Ramachandran (2019) found that the higher the level of entrepreneurship of a
subsidiary, the more that subsidiary can grasp various opportunities related to local business
in general and accumulate unique knowledge and know-how in the process of actively
overcoming any obstacles (Scott et al., 2010). In the process of overcoming these obstacles,
the identification of various opportunities and proactive responses can be considered to be
closely related to the AUT of subsidiaries. Lumpkin et al. (2009) mentioned that the AUT
enjoyed by an organization is ultimately closely related to the level of entrepreneurship of that
organization. Their research presented AUT as a tool for measuring the level of
194
entrepreneurship. However, in studies on MNCs, the entrepreneurship of subsidiaries can be
discussed separately from AUT. This is because the subsidiary’s AUT cannot be secured by
itself, but the scope of the subsidiary’s activity, that is, the AUT, is regulated under the
approval of the HQ (Gammelgaard et al., 2012). Therefore, the higher the level of AUT
enjoyed by the subsidiary, the broader the scope of the subsidiary to identify and respond to
market opportunities in the local market, which will also contribute to an increased possibility
that the subsidiary will accumulate unique knowledge and know-how (Geleilate et al., 2019).
It can also be inferred that the AUT of the subsidiary will have a positive moderating effect
independently of the relationship between the entrepreneurship and KC of the subsidiary.
Therefore, we propose the following hypothesis regarding the moderating effect of AUT.
H2: AUT in the local market will positively moderate the relationship between SENT and KC.
2.4. Subsidiary Network Position (NP) between KC and Knowledge

Transfer to Overseas Subsidiaries (KTO)
Subsidiaries of MNCs evolve through the accumulation of capabilities. Evolution simply
means advancements in the management activities of subsidiaries, but the accumulation of
capabilities leads to a change in both the network position (NP) and the activities of the
subsidiary within the MNC (Birkinshaw and Hood, 1998; Gupta and Govindarajan, 1994).
The act of developing and accumulating subsidiary-specific knowledge capabilities is the first
step toward changing the properties of subsidiaries as centers of excellence (CoE).
Subsidiaries with unique knowledge capabilities outperform other subsidiaries, which can in
turn lead to additional investments in HQ (Frost et al., 2002). Egeraat and Breathnach (2012)
summarized the change in capability as an improvement in the R&D process and noted that
this could be a foremost driver for the evolution of subsidiaries of MNC. In other words, the
existence of an appropriate system for developing new knowledge and accumulating it is
closely related to the evolution of subsidiaries.
However, it is important to note that the enhancement of NP within an MNC’s network is
based on the strategic decision of the HQ. in the organizational structure of MNCs,
subsidiaries are obliged to follow the orders of HQs. AUT discussed in subsidiary studies
mostly refers to the AUT of local management activities (Kawai and Strange, 2014), and the
evolution of NP within the MNC network is not an issue that subsidiaries can decide for
themselves. The HQ of an MNC set criteria for the evaluation of the activities of its
subsidiaries. These criteria are used to analyze and evaluate the contribution of subsidiaries
(Corry and Cormican, 2019), and in turn change the NP of certain subsidiaries in the network
of MNC (Andersson et al., 2005; Sumelius and Sarala, 2008).
In summary, subsidiaries of MNCs are evaluated by HQs in the process of self-knowledge-
generating actions. If the created knowledge is recognized as having value, the HQ is more
likely to enhance the status of its subsidiary. Therefore, the following hypothesis can be
proposed.
H3: KC of an MNC will has a positive effect on NP within the MNC network.
If the subsidiary shares the unique knowledge it has acquired locally within the MNC
network, the subsidiary has to risk the disappearance of the scarcity of its knowledge. In other
words, the competitive advantage of the subsidiary based on KC may be lost. Therefore, the
HQ must reward subsidiaries for their KTO and attempt to serve as a catalyst to enrich
knowledge-sharing behavior (Björkman et al., 2004). In addition, support from HQs is also
required for expenses such as securing infrastructure for knowledge sharing (Forsgren et al.,
195
2000; Szulanski, 1996). Therefore, when the HQ has a system in place to promote KTO, the
HQ may recognize the contributing activities of certain subsidiaries and suggest elevated NP,
additional investment, and extended scope of activities compared to other subsidiaries in the
multinational network. This will allow HQs to appropriately motivate the subsidiary’s KTO
(Dellestrand, 2020; Frost et al., 2002; Want et al., 2019).
Additionally, enhancing the NP of subsidiaries can create a positive circulation structure
between knowledge sharing and NP enhancement, in addition to the meaning of HQs’
compensation for knowledge sharing activities. Subsidiaries that have once experienced NP
enhancement will continue to focus on KTO to maintain enhancement. Enhancing and
maintaining the influence of subsidiaries within the network of MNCs can be a sufficient
motivating factor for knowledge sharing (Najafi-Tavani et al., 2015).
H4: NP of an MNC will have a positive effect on KTO within the MNC network.
The order of the discussions mentioned above is illustrated in Fig. 1 below.
Fig. 1. Research Model
3. Methodology
3.1. Samples
Recently, in Korea, an innovative economy has emerged that creates high added value. In
achieving the creation of new growth engines, the development of in-house technology
capabilities is also important, but it is possible to achieve these goals more quickly through
acquiring skills by attracting foreign leading companies using foreign direct investment.
Notably, in the case of industries where the technology level is not high but where
development is required, such a strategy using foreign direct investment can be very useful.
In consideration of these points, this study attempts to examine the functions and roles of
MNC subsidiaries invested in Korea from the perspective of subsidiary entrepreneurship.
The sample of this study consisted of companies in Gyeonggi and Seoul; specifically, the
population was the 17,938 foreign-invested companies registered in the Ministry of Trade,
Industry and Energy (MOTIE). The sample was extracted by simple random sampling using
the address book of companies registered in MOTIE. The survey was conducted over about
5 months, from August to December 2019. Among companies with subsidiaries in Gyeonggi
and Seoul, 800 subsidiaries were randomly selected, 302 responded, and 282 valid companies
were ultimately selected as the final sample (response rate: 37.6%). A pilot test was conducted
for the companies in question through questionnaires, and as a result of the reliability
measurement, Cronbach alpha was estimated to be 0.7 or higher, thereby ensuring reliability,
as listed in Table 1. Then, Haman’s single factor test (Podsakoff et al., 2003) and the variation
inflation factor test (VIF) (Kock and Lynn, 2012) were conducted to confirm the common
196
method bias. The results confirmed that more than half of the explanatory power of the
variance extracted through the total factors was explained, and that the coefficients of the
correlation coefficient were all 0.8 or higher, thus showing a high correlation (Harman, 1976).
The independent variable in this study is the subsidiary entrepreneurship (SENT), the
dependent variable is knowledge transfer to overseas subsidiaries (KTO). In addition, the
moderation variable is the subsidiary autonomy (AUT), and the subsidiary knowledge
creation (KC) of the subsidiary and the subsidiary network position within MNCs (NP) serve
as the indirect bridge to KTO. This study verified the proposed research model with the
maximum likelihood of the structural equation (SEM) using STATA 16.
3.2. Variable Measurements and Validity

To measure each variable in this study, we use the 7-point Likert scale (1: not very much,
7: very agree). The Likert scale is a tool that can best measure whether or not a subject agrees
or disagrees with the variable. Table 1 summarizes the measurements and data in terms of the
variables considered in this study. In this study, Kostova (1999)’s tool was applied to measure
ENT, and the KC of subsidiaries was measured using four questionnaires (Sumelius and
Sarala, 2008). In addition, Li (2005) and Lee et al. (2008) were applied for the measurement
of KTO while Najafi-Tavani et al. (2015) and Gammelgaard et al. (2012) were applied for
AUT. Finally, NP was measured using four questionnaires (Andersson, Forsgren and Holm,
2002; Najafi-Tavani, Giroud and Andersson, 2014). After analyzing the reliability and
convergent validity for all constructs, it was found that both exceeded the reference value.
According to Fornell and Larcker (1981), the threshold of AVE should be higher than 0.5 to
show that the measurement size of the proposed model converges, as presented in Table 1.
Table 1. Reliability and Convergent Validity

Construct Item Est. Std. Est. S.E. p t AVE C.R. α Ref.
SENT SENT1 1.00 0.87 0.02 0.00 44.66 0.70 0.90 0.90 Kostova (1999)
SENT2 0.86 0.83 0.02 0.00 35.86
SENT3 0.91 0.81 0.02 0.00 32.48
SENT4 0.89 0.82 0.02 0.00 34.60
KC KC1 1.00 0.92 0.01 0.00 86.38 0.76 0.93 0.93 Sumelius and
KC2 1.02 0.97 0.01 0.00 126.1 Sarala (2008)
KC3 0.84 0.78 0.03 0.00 30.54
KC4 0.90 0.81 0.02 0.00 36.56
KTO KTO1 1.00 0.75 0.03 0.00 27.12 0.68 0.89 0.89 Li (2005) and
KTO2 1.45 0.98 0.01 0.00 82.06 Lee et al. (2008)
KTO3 1.13 0.69 0.03 0.00 20.79
KTO4 1.29 0.84 0.02 0.00 42.20
AUT AUT1 1.00 0.77 0.03 0.00 27.96 0.71 0.91 0.90 Najafi-Tavani et al.
AUT2 1.08 0.80 0.03 0.00 31.20 (2015) and
AUT3 1.42 0.93 0.01 0.00 63.33 Gammelgaard et al.
AUT4 1.28 0.86 0.02 0.00 46.28 (2012)
NP NP1 1.00 0.84 0.02 0.00 39.10 0.74 0.92 0.92 Andersson et al.
NP2 0.94 0.81 0.02 0.00 34.50 (2002) and Najafi-
NP3 1.06 0.93 0.01 0.00 69.56 Tavani et al. (2014)
NP4 0.91 0.85 0.02 0.00 44.93
Note: SENT=Subsidiary entrepreneurship, KC=Subsidiary knowledge creation, KTO=Knowledge
transfer to overseas subsidiaries, AUT=Subsidiary autonomy, NP=Subsidiary network position
within MNCs, Est.=estimate, Std. Est.=standardized estimate, S.E.=standard error,
AVE=average variation extracted, C.R.=composite reliability, ߙ=Cronbach’s alpha.
197
Table 2. Fit Indexes for CFA
Fit Indices Value
Comparative fit index (CFI) 0.844
Tucker–Lewis index (TLI) 0.815
Standardized root mean squared residual (SRMR) 0.069
Root mean squared error of approximation (RMSEA) 0.073
In Table 2, CFA is mainly used for certain types of research for two reasons: First, CFA is
used to determine the dimensions of the scale at which all concepts are measured. Secondly,
it is used to test whether the measurements of the structure are consistent with the
researcher’s understanding of the nature of the proposed model (߯ ଶ =509.784, df=202).
According to Hu and Bentler (1999), CFI and TLI must be 0.8 or higher, and SRMR and
RMSEA must be less than 0.08 to be considered adequate. CFI and TLI were both found to
be in the acceptable range of 0.844 and 0.815, respectively, while the SRMR was 0.069 and
RMSEA was 0.073, meaning the results were within the valid ranges of all four indices. In the
case of CFI, 0.036 was higher in CFA than in EFA, and 0.041 in CFA was also higher in TLI.
It was confirmed that the SRMR decreased by 0.004, and that the RMSEA decreased by 0.012.
Therefore, in this study, the factor was found to be better for CFA than for EFA.
Table 3. Intra-construct Correlations and Discriminant Validity

Constructs 1 2 3 4 5
1 SENT 0.87
2 KC 0.54* 0.84
3 KTO 0.16* 0.26* 0.83
4 AUT 0.19* 0.28* 0.08* 0.86
5 NP 0.11* 0.37* 0.37* 0.01* 0.82
Notes: 1. Correlations in diagonal are the square root of AVE.
2. *p<0.05.
As listed in Table 3, the square roots of AVE were ENT (0.87), KC (0.84), KTO (0.83), AUT
(0.86), and NP (0.82), respectively; therefore, it shows a decent convergence validity. To
determine the discriminant validity for the proposed model, the square root of AVE and the
correlation coefficient were compared. Only KC was significant at 0.84, and all AVE square
root values were higher than the correlation coefficient. In particular, the squared value of
ENT was 0.87, thus showing the largest value, and the difference between AUT (0.86) and NP
(0.01) was 0.85, with the largest difference between the AVE square root and the correlation
coefficient, and with the smallest difference (0.33) between ENT (0.87) and KC (0.54).
Therefore, the discriminant validity of this study was verified.
3.3. Common Method Bias Test

The risk of common method bias (CMB) is always inherent in structural equation models
using surveys. Podsakoff et al. (2003) recommended Harman’s single factor test using
primary factor analysis and suggested that there is no problem with CMB when the variance-
based explanatory power of one factor is lower than 50%. The result of the single-factor
analysis in this study showed that the explanatory power was 25.7%; thus, there is no CMB
issue. Next, in the recently emerged multivariate methodology, the latent variable (LV) is used
more often than it had been in the past, and the methodology focuses on the multiple
collinearities that can occur between the latent variables rather than Harman’s single factor.
198
In this context, Kock and Lynn (2012) suggested that the risk for CMB is small when the
lateral collinearity Variation inflation factor (VIF) between latent variables is lower than 3.3.
This study adopted the method proposed by Podsakoff et al. (2003) and Kock and Lynn
(2012) confirmed that the variance expansion factor between variables was lower than 2.
Therefore, it was ultimately estimated that there was no risk of CMB. Lai et al. (2016) also
verified CMB using this same method.
4.1. Descriptive Analysis
Table 4 shows the results of descriptive statistics for the subsidiaries of MNC in Gyeonggi
and Seoul in South Korea that have been randomly extracted for the analysis in this study. Of
the 282 companies that submitted valid responses, U.S. companies accounted for the largest
share with 105 (37.23%), followed by Japanese companies with 96 (34.04%), the second
largest. The combined portion of U.S. and Japanese companies was 71.27%, which accounted
for a high percentage of the total sample. Regarding the number of employees, 84 companies
(29.79%) accounted for the largest share with 51 to 150 employees, while 72 companies
(25.53%) with less than 25 employees accounted for the second-largest share. In terms of the
year of the establishment of subsidiaries in Korea, the largest number was 102 (36.17%)
between 1996 and 2000, and 30 (10.64%) since 2005 was the smallest number of companies
established during the sample period. For total sales, 108 (38.3%) companies exceeded 950
billion won, and 93 (32.98%) companies accounted for between 11 billion won and 300 billion
won. In terms of exports, 153 companies (39.36%) accounted for the highest proportion, with
less than 10 billion won. Exports, which accounted for the next largest share, exceeded KRW
35 billion, and 96 companies (34.04%) accounted for this.

Item Category Frequency (n) Proportion (%)
Country of Origin U.S. 105 37.23
Japan 96 34.04
Germany 21 7.45
U.K. 18 6.38
Sweden 12 4.26
Others 30 10.64
Number of Employees ≤ 25 72 25.53
26-50 66 23.4
51-150 84 29.79
> 150 60 21.28
Year Established in Korea ≤ 1995 78 27.66
1996-2000 102 36.17
2001-2005 72 25.53
> 2005 30 10.64
Total Sales (Bil. Won) ≤ 10 24 8.51
11-300 93 32.98
301-950 57 20.21
> 950 108 38.3
Export Sales (Bil. Won) ≤ 10 153 39.36
11-35 33 11.7
> 35 96 34.04
Total 282 100
199
4.2. SEM Results
Table 5 lists the results of testing six hypotheses through structural equations. The
investigation showed that the fit of the model in this study was found to be acceptable (χ2 =
502.604, df = 196, χ2 /df = 2.564, RMSEA = 0.075, CFI = 0.856, TLI = 0.832, SRMR = 0.077)
(Hu and Bentler, 1999). Standardization coefficients were used for all variables. Among paths,
the coefficient of H1 representing a path from SENT to KC was found to be the largest at
0.399 (p < 0.001). It was verified that AUT has a moderating effect on the relationship between
SENT and KC, and it was confirmed to be significantly supported (β = 0.171, p < 0.01); thus,
H2 was supported. The coefficient of KC on NP was positively significant (β = 0.366, p < 0.01).
Therefore, H3 was also supported. Finally, regarding H4, the variable with the second greatest
influence on the dependent variable KTO was NP (β = 0.374, p <0.001). The explanatory
power of the total effect on KC was 44.4% affected by SENT, and this was found to be
significant at the 0.1% level.
Table 5. SEM Results

Hypothesized paths Coefficient t-value Testing
***
H1 SENT → KC 0.399 6.33 Supported
**
H2 SENT×AUT → KC 0.171 2.69 Supported
***
H3 KC → NP 0.366 6.67 Supported
***
H4 NP → KTO 0.374 6.62 Supported
Total effect Indirect effect
KC NP KTO NP KTO
λௌாே் =0.444*** λ௄஼ =0.395*** λ௄஼ =0.089** λௌாே்ൈ஺௎் =0.009* λ௄஼ =0.089***
λௌாே்ൈ஺௎் =0.024** λௌாே்ൈ஺௎் =0.009* λே௉ =0.226*** λௌாே் =0.175*** λௌாே்ൈ஺௎் =0.002*
λௌாே் =0.175*** λௌாே்ൈ஺௎் =0.002* λௌாே் =0.040**
λௌாே் =0.040**
Notes: 1. Standardized beta coefficients.
2. *p<0.05, **p<0.01, ***p<0.001.
ଶ ଶ ଶ
3. R2 of Bentler-Raykov: ௄஼ = 0.263, ே௉ = 0.134 and ௄்ை = 0.140.
Fig. 2. Moderation Effect of AUT between ENT and KC

200
Meanwhile, it was found to have 2.4% explanatory power when SENT and AUT interacted
on KC (p < 0.01). The explanatory power of the total effect on NP was 39.5% (p < 0.001), it
was 17.5% for SENT (p < 0.001), and it was 1.0% for the SENT interacting with AUT (p <
0.05). For KTO, KC, NP, and SENT had explanatory powers of 8.9%, 22.6%, and 4.0%,
respectively, and they were found to be significant at the overall 5% level. When SENT
interacted with AUT, the explanatory power for KTO was estimated to have a significance of
0.2% (p < 0.05). The indirect effect on NP was found to have 17.5% explanatory power in the
case of SENT (p < 0.001), but the explanatory power decreased when SENT interacted with
AUT (p < 0.05). Finally, in terms of the indirect effect on KTO, KC and ENT had respective
explanatory powers of 8.9% (p < 0.001) and 4.0% (p < 0.01), while the explanatory power
decreased to 0.2% when SENT interacted with AUT. As a result, the significance level also
decreased to 5%, indicating that the explanatory power decreased when all variables
interacted with the AUT. However, since all of the interaction terms between SENT and AUT
were found to be significant, it can be seen that all of the moderating effects of AUT affect
KC. According to Bentler-Raykov’s explanatory power (R2), KC was 26.3%, NP was 13.4%,
and KTO was 14.0%.
4.3. Moderation Effect

In terms of the relationship between ENT and KC, Fig. 2 presents a graph showing the
result when AUT interacts with ENT. When the involvement of AUT was high, the
moderation effect on KT was found to be greater. In other words, the greater the autonomy
of subsidiaries, the greater the moderating effect on KT. This moderation effect had already
been confirmed by Lumpkin et al. (2009). However, this is a more differentiated finding
because we have targeted the context of MNC subsidiaries that have invested directly in new
countries.
4.4. Discussion
First, SENT was found to positively affect activities to learn new knowledge in the local
market. In this study, since SENT was related to the spirit of challenge, risk-taking for
innovation, and encouragement for creativity for new ideas, the adoption of H1 implies that
the subsidiary’s tendency contributes to actively motivating the organization’s new
knowledge learning activities. As subsidiaries belong to a hierarchical structure of HQ of
MNCs, creating a unique organizational orientation and influence on local activities may be
limited under the HQ control and management. However, the results of this study confirmed
that the subsidiary’s unique entrepreneurial propensity has a high significance on its local
knowledge creation. In this vein, it was possible to confirm the importance of forming a
unique entrepreneurial propensity. Second, it was possible to see how the focal subsidiary
provided knowledge to other subsidiaries after creating the knowledge itself, not the benefits
or support provided from the headquarters. Among the numerous subsidiaries connected to
the HQ, the focal subsidiary will naturally dominate the subsidiary network when it has
created excellent knowledge. This means that if the subsidiary is well equipped with SENT,
its knowledge reaches the NP, proves its excellence to other companies, and spreads it to other
subsidiaries. Third, we confirmed the empowerment effect of autonomy among the roles of
the subsidiary. Local knowledge creation was reinforced when the HQ granted autonomy
within the MNCs to which the subsidiary belongs. It is possible to recognize the importance
201
of establishing the direction that the subsidiary’s entrepreneurial orientation should aim for
in the HQ-subsidiary relationship. Thus, SENT-oriented organizational orientation (e.g.,
taking risks for challenge and innovation, encouraging creativity for new ideas) has a positive
effect on learning unique local knowledge, while guaranteeing the autonomy of subsidiaries
bolsters the knowledge network of MNCs.
5. Conclusion
This study aims to contribute to the existing research and provide practical implications to
MNCs by presenting discussions on (1) effective knowledge creation and (2) valuable
knowledge transfer within subsidiaries. Specifically, this study examined the effects of SENT
and AUT as a major factor in the creation of subsidiary knowledge of MNCs, as well as the
indirect effect of NP within MNCs between KC and KTO. The proposed model is verified
through SEM to determine whether our hypotheses have sound reasoning.
All four of our hypotheses were significantly supported. First, in H1, it was found that
enterprise and spirit had a positive effect on the KC of overseas subsidiaries. It can be
concluded that SENT has a positive effect by acting as a motivating factor for the creation of
new knowledge in the organization. Second, in H2, AUT confirmed the moderating effect on
the relationship between SENT and KC. This leads to the logical reasoning that autonomy
and delegation of authority are necessary for active KC because subsidiaries are clearly
affected by the control of HQs. Third, the adoption of H3 and H4 confirmed the importance
of NP as a knowledge transmitter from KC to KTO. NP within MNCs enables the logical
reasoning that it increases knowledge transfer to the HQ and other subsidiaries by giving
them a more favorable position from where they can appreciate the value of the knowledge
created.
For example, to jointly develop a process that makes it possible to shorten the process and
logistics, American company A attempted to contract an MOU with a Korean logistics
company that has been trading for a long time. This was enough to be reported to the HQ
office, and at the same time, attract attention and envy from other subsidiaries. However, as
new knowledge (e.g., the patent for the new logistic process) was created from the MOU, the
HQ shared these achievements with subsidiaries in the world, and as a result, the Korean
subsidiary naturally gained authority power compared to other overseas subsidiaries. In this
respect, the interviewee of the Korean subsidiary said that it had exhibited entrepreneurship
because the company intentionally desired to be recognized by the HQ. He pointed out that
if he had taken action to help other subsidiaries for shared assets, the outcome with a Korean
logistics company might not have been excellent. In summary, it can be said that MNC’s
subsidiaries are knowingly or unknowingly competing with subsidiaries in other countries
within their network, so they do not actively encourage knowledge transfer either directly or
indirectly. On the other hand, if entrepreneurial behavior creates new knowledge and gains
recognition and rewards from HQ to some extent, it can be understood that knowledge can
be shared with other subsidiaries after that.
Based on the results of this study, we can draw the following implications. First, in this
study, we confirmed that the knowledge-sharing activities of subsidiaries are directly affected
by the original SENT. A series of actions based on SENT can be summarized as a Subsidiary
Initiative (Ahsan and Fernhaber, 2019). Further, the Subsidiary Initiative refers to a
subsidiary’s actions that contribute to the enterprise level of MNC (Birkinshaw et al., 1998).
202
Existing studies dealing with the entrepreneurship of a subsidiary have viewed the Subsidiary
Initiative, which reflects the entrepreneurial spirit of the subsidiary, as a competency (e.g.,
knowledge) creation or innovation activity of the subsidiary, and they have mainly focused
on identifying the factors that influence this. By contrast, we attempted to deal with the
impact of SENT on KC included in the Subsidiary Initiative and knowledge sharing (or
reverse transfer) of subsidiaries (Birkinshaw and Hood, 1998). This allowed us to examine
the internal structure of the Subsidiary Initiative.
Second, we examined the relationship between entrepreneurship, knowledge creation, and
knowledge sharing activities of subsidiaries. Generally, entrepreneurship has a positive effect
on knowledge creation (Sarabi et al., 2020). However, there have been an insufficient number
of studies investigating the relationship with entrepreneurship in the case of knowledge
sharing (or reverse transfer). We argued that there would be a process that requires further
investigation between entrepreneurship and knowledge-sharing activities. Therefore, we
presented both the creations of new knowledge by subsidiaries and rewards by HQ for
acknowledging these contributing actions (e.g., improvement of the status in the network) as
a component of the process that connects the entrepreneurship of the subsidiary and its
knowledge-sharing activity. For example, it is difficult to attract the attention of the HQ
because existing knowledge is of low value, and it is also difficult to use it for knowledge
sharing activities of the subsidiary. Therefore, a subsidiary of an MNC can secure valuable
knowledge by creating new knowledge and then share the knowledge created later (Egeraat
and Breathnach, 2012). The newly created knowledge makes the HQ interested in the
subsidiary, and this leads the HQ to positively evaluate the subsidiary (Corry and Cormican,
2019). Based on the positive evaluation, the HQ is likely to allocate more slack resources to
the subsidiary (Dellestrand et al., 2020) and improve its subsidiary in its MNC network
(Baraldi and Ratajczak-Mrozek, 2019). Ultimately, the subsidiary can play as a CoE, leading
itself to have experienced an improved position to actively perform knowledge sharing
activities. Such an in-substitutable role will become more appropriate to their principal
position to maintain network power (Frost et al., 2002).
Third, this study reaffirms the importance of AUT in the knowledge creation of a
subsidiary (Kawai and Strange, 2014). For the first time, this study examined the effect of the
autonomy enjoyed by subsidiaries in the field of product production on the knowledge
creation of subsidiaries. The results of this study show that securing autonomy in the product
design, technology development, and manufacturing process can more positively strengthen
the impact of subsidiary entrepreneurship on knowledge creation.
Despite the above theoretical and practical implications, this study has the following
limitations. First, SENT in this study is considered as the entrepreneurial spirit at the
organizational level. Most existing studies have analyzed entrepreneurship as the degree of
recognition of a firm’s CEO or top management team (TMT). Since SENT in this study
targets organizational units, it can be said that this study made a new attempt in terms of
research analysis units for entrepreneurship of subsidiary. Future studies examining SENT
should consider how the CEO perceives the subsidiaries’ entrepreneurial statuses throughout
the entire unit of analysis (Zahra et al., 1999). Second, the sample of the questionnaire
conducted for analysis in this study was limited to South Korea. Although South Korea is a
member of the OECD and has some degree of local knowledge that is distinct from emerging
countries, it is expected that more meaningful implications can be drawn by comparing
developed and emerging countries in which the motivations for knowledge learning may
vary.
203
References
Adenfelt, M. (2010), “Exploring the Performance of Transnational Projects: Shared Knowledge,
Coordination and Communication”, International Journal of Project Management, 28(6), 529-
538.
Ahsan, M. and S. A. Fernhaber (2019), “Multinational Enterprises: Leveraging a Corporate
International Entrepreneurship Lens for New Insights into Subsidiary Initiatives”, Journal of
International Management, 25(1), 51-65.
Ambos, T. C. and J. Birkinshaw (2010), “Headquarters’ Attention and Its Effect on Subsidiary
Performance”, Management International Review, 50(4), 449-469.
Ambos, T. C., B. Ambos and B. B. Schlegelmilch (2006), “Learning from Foreign Subsidiaries: An
Empirical Investigation of Headquarters’ Benefit from Reverse Knowledge Transfers”,
International Business Review, 15(3), 294-312.
Ambos, T. C., U. Andersson and J. Birkinshaw (2010), “What Are the Consequences of Initiative-
Taking in Multinational Subsidiaries?”, Journal of International Business Studies, 41(7), 1099-
1118.
Andersson, U., I. Bjdersso and M. Forsgren (2005), “Managing Subsidiary Knowledge Creation: The
Effect of Control Mechanisms on Subsidiary Local Embeddedness”, International Business
Review, 14(5), 521-538.
Andersson, U., M. Forsgren and U. Holm (2002), “The Strategic Impact of External Networks:
Subsidiary Performance and Competence Development in The Multinational Corporation”,
Strategic Management Journal, 23(11), 979-996.
Baraldi, E. and M. Ratajczak-Mrozek (2019), “From Supplier to Center of Excellence and Beyond:
The Network Position Development of a Business Unit Within “IKEA Industry”, Journal of
Business Research, 100, 1-15.
Birkinshaw, J. (1999), “The Determinants and Consequences of Subsidiary Initiative in Multinational
Corporations”, Entrepreneurship Theory and Practice, 24(1), 9-36.
Birkinshaw, J. and N, Hood (1998), “Multinational Subsidiary Evolution: Capability and Charter
Change in Foreign-Owned Subsidiary Companies”, Academy of Management Review, 23(4),
773-795.
Birkinshaw, J., N. Hood and S. Jonsson (1998), “Building Firm-Specific Advantages in Multinational
Corporations: The Role of Subsidiary Initiative”, Strategic Management Journal, 19(3), 221-
242.
Birkinshaw, J., N. Hood and S. Young (2005), “Subsidiary Entrepreneurship, Internal and External
Competitive Forces, and Subsidiary Performance”, International Business Review, 14(2), 227-
248.
Björkman, I., W. Barner-Rasmussen and L. Li (2004), “Managing Knowledge Transfer in MNCs:
The Impact of Headquarters Control Mechanisms”, Journal of International Business Studies,
35(5), 443-455.
Bouquet, C. and J. Birkinshaw (2008), “Weight Versus Voice: How Foreign Subsidiaries Gain
Attention from Corporate Headquarters”, Academy of Management Journal, 51(3), 577-601.
Cho, D. S. (2014), Mechanism-based Perspective: A New Strategic Paradigm for Integrated
Management, Seoul: Seoul Economics and Management Publication.
Clark, K. and I. Ramachandran (2019), “Subsidiary Entrepreneurship and Entrepreneurial
Opportunity: An Institutional Perspective”, Journal of International Management, 25(1), 37-50.
Conroy, K. M. and D. G. Collings (2016), “The Legitimacy of Subsidiary Issue Selling: Balancing
Positive & Negative Attention from Corporate Headquarters”, Journal of World Business, 51(4),
204
612-627.
Corry, S. and K. Cormican (2019), “Towards Innovation in Multinational Corporation Subsidiaries:
Development of an Instrument to Select and Evaluate Value Driven Strategies”, Journal of
Technology Management & Innovation, 14(1), 25-33.
Dellestrand, H., P. Kappen and O. Lindahl (2020), “Headquarter Resource Allocation Strategies and
Subsidiary Competitive or Cooperative Behavior: Achieving a Fit for Value Creation”, Journal
of Organization Design, 9(1), 1-16.
Dess, G.G., R. D. Ireland, S. A. Zahra, S. W. Floyd, J. J. Jay and P. J. Lane (2003), “Emerging Issues
in Corporate Entrepreneurship”, Journal of Management, 29(3), 351-378.
Dörrenbächer, C. and J. Gammelgaard (2016), “Subsidiary Initiative Taking in Multinational
Corporations: The Relationship between Power and Issue Selling”, Organization Studies, 37(9),
1249-1270.
Fornell, C. and D. F. Larcker (1981), “Evaluating Structural Equation Models with Unobservable
Variables and Measurement Error”, Journal of Marketing Research, 18(1), 39-50.
Gammelgaard, J., F. Mcdonald, A. Stephan, H. Tüselmann and C. Dörrenbächer (2012), “The Impact
of Increases in Subsidiary Autonomy and Network Relationships on Performance”,
International Business Review, 21(6), 1158-1172.
Geleilate, J. M. G., D. S. Andrews and S. Fainshmidt (2019), “Subsidiary Autonomy and Subsidiary
Performance: A Meta-Analysis”, Journal of World Business, 55(4), 101049.
Goh, S. W. (2002), “Managing Effective Knowledge Transfer: An Integrative Framework and Some
Practice Implication”, Journal of Knowledge Management, 6(1), 23-30.
Gupta, A. K. and V. Govindarajan (2000), “Knowledge Flow within Multinational Corporations”,
Strategic Management Journal, 21(4), 473-496.
Hakanson, L. and R. Nobel (2001), “Organizational Characteristics and Reverse Knowledge
Transfer”, Management International Review, 40(Special Issue), 29-48.
Harman, H. H. (1976), Modern Factor Analysis, Chicago, IL: University of Chicago Press.
Hu, L. T. and P. M. Bentler (1999), “Cutoff Criteria for Fit Indexes in Covariance Structure Analysis:
Conventional Criteria Versus New Alternatives”, Structural Equation Modeling: A
Multidisciplinary Journal, 6(1), 1-55.
Inkpen, A. C. and E. W. Tsang (2005), “Social Capital, Networks, and Knowledge Transfer”,
Academy of Management Journal, 30(1), 146-165.
Kawai, N. and R. Strange (2014), “Subsidiary Autonomy and Performance in Japanese Multinationals
in Europe”, International Business Review, 23(3), 504-515.
Kim, J. O. and C. W. Mueller (1978), Factor Analysis: Statistical Methods and Practical Issues,
Newbury Park, CA: Sage.
Kock, N. and G. Lynn (2012), “Lateral Collinearity and Misleading Results in variance-based Sem:
An Illustration and Recommendations”, Journal of The Association for Information Systems,
13(7), 546-580.
Kostova, T. (1999), “Transnational Transfer of Strategic Organizational Practices: A Contextual
Perspective”, Academy of Management Review, 24(2), 308-324.
Lai, C., Q. Wang, X. Li and X. Hu (2016), “The Influence of Individual Espoused Cultural Values on
Self-Directed Use of Technology for Language Learning Beyond the Classroom”, Computers
in Human Behavior, 62, 676-688.
Lee, J. Y., A. Jiménez and K. R. Bhandari (2020), “Subsidiary Roles and Dual Knowledge Flows
between MNE Subsidiaries and Headquarters: The Moderating Effects of Organizational
Governance Types”, Journal of Business Research, 108, 188-200.
Lee, Kang-Moon (2011), “The Determinants of Subsidiary Reverse Knowledge Transfer: The Case
205
of Foreign MNC Subsidiaries in Korea”, Journal of CEO and Management Studies, 14(1), 1-
25.
Lee, R. P., Q. Chen, D. Kim and J. L. Johnson (2008), “Knowledge Transfer between Multinational
Corporations’ Headquarters and Their Subsidiaries: Influences on and Implications for New
Product Outcomes”, Journal of International Marketing, 16(2), 1-31.
Lee, S. H. and C. Williams (2007), “Dispersed Entrepreneurship within Multinational Corporations:
A Community Perspective”, Journal of World Business, 42(4), 505-519.
Li, J. and R. P. Lee (2015), “Can Knowledge Transfer within MNCs Hurt Subsidiary Performance?
The Role of Subsidiary Entrepreneurial Culture and Capabilities”, Journal of World Business,
50(4), 663-673.
Li, L. (2005), “The Effects of Trust and Shared Vision on Inward Knowledge Transfer in Subsidiaries’
Intra-and Inter-Organizational Relationships”, International Business Review, 14(1), 77-95.
Lumpkin, G. T., C. C. Cogliser and D. R. Schneider (2009), “Understanding and Measuring
Autonomy: An Entrepreneurial Orientation Perspective”, Entrepreneurship Theory and
Practice, 33(1), 47-69.
Minbaeva, D. T. (2007), “Knowledge Transfer in Multinational Corporations”, Management
International Review, 47(4), 567-593.
Minbaeva, D. T. P., I. Bjorkman, C. Fey and H. J. Park (2003), “MNC Knowledge Transfer, Subsidiary
Absorptive Capacity and HRM”, Journal of International Business Studies, 34(1), 586-599.
Mudambi, R., L. Piscitello and L. Rabbiosi (2014), “Reverse Knowledge Transfer in MNEs:
Subsidiary Innovativeness and Entry Modes”, Long Range Planning, 47(1-2), 49-63.
Najafi-Tavani, Z., A. Giroud and U. Andersson (2014), “The Interplay of Networking Activities and
Internal Knowledge Actions for Subsidiary Influence Within MNCs”, Journal of World
Business, 49(1), 122-131.
Najafi-Tavani, Z., G. Zaefarian, P. Naudé and A. Giroud (2015), “Reverse Knowledge Transfer and
Subsidiary Power”, Industrial Marketing Management, 48, 103-110.
Nonaka, I., U. Katushiro, and S. Dai (1996), “From Information Processing to Knowledge Creation:
A Paradigm Shift in Business Management”, Technology in Society, 18(2), 203-218.
O’Brien, D., S. P. Scott, U. Andersson, T. Ambos and N. Fu (2019), “The Microfoundations of
Subsidiary Initiatives: How Subsidiary Manager Activities Unlock Entrepreneurship”, Global
Strategy Journal, 9(1), 66-91.
Paterson, S. L. and D. M. Brock (2002), “The Development of Subsidiary Management Research:
Review and The Practical Analysis”, International Business Review, 11(1), 139-163.
Podsakoff, P. M., S. B. Mackenzie, J.-Y. Lee and N. P. Podsakoff (2003), “Common Method Biases
in Behavioral Research: A Critical Review of the Literature and Recommended Remedies”,
Journal of Applied Psychology, 88(5), 879-903.
Scott, P., P. Gibbons and J. Coughlan (2010), “Developing Subsidiary Contribution to the MNC—
Subsidiary Entrepreneurship and Strategy Creativity”, Journal of International Management,
16(4), 328-339.
Song, J. Y. (2014), “Subsidiary Absorptive Capacity and Knowledge Transfer within Multinational
Corporations”, Journal of International Business Studies, 45(1), 73-84.
Sumelius, J. and R. Sarala (2008), “Knowledge Development in MNC Subsidiaries: The Influence of
MNC Internal and External Knowledge and Control Mechanisms”, Thunderbird International
Business Review, 50(4), 245-258.
Van Egeraat, C. and P. Breathnach (2012), “The Drivers of Transnational Subsidiary Evolution: The
Upgrading of Process R&D in The Irish Pharmaceutical Industry”, Regional Studies, 46(9),
1153-1167.
206
Wang, N. and Y. Wang (2020), “Does Parenting Matter in Subsidiary Innovation in Emerging
Economies? Exploring the Role of Parent Superior Competitiveness in Affecting Subsidiary
Contextual Ambidexterity”, International Business Review, 101673 (forthcoming).
Wang, N., Y. Hua, G. Wu, C. Zhao and Y. Wang (2019), “Reverse Transfer of Innovation and
Subsidiary Power: A Moderated Mediation Model”, Journal of Business Research, 103, 328-
337.
Yang, Q., R. Mudambi and K. E. Mayer (2008), “Conventional and Reverse Knowledge Flows in
Multinational Corporation”, Journal of Management, 34(5), 882-902.
Yu, Y., Y. Liu and T. Bai (2019), “Does Attention from Headquarters Influence Subsidiary Behavior?
A Social Psychological Perspective”, Long Range Planning, 52(4), 101877.
Zahra, S. A., A. P. Nielsen and W. C. Bogner (1999), “Corporate Entrepreneurship, Knowledge, and
Competence Development”, Entrepreneurship Theory and Practice, 23(3), 169-189.
Zhang, H., C. Shu, X. Jiang and A. Malter (2010), “Managing Knowledge for Innovation: The Role
of Coorperation, Competition, and Alliance Nationality”, International Journal of Marketing,
18(4), 74-94.
Journal of Korea Trade Volume 24 Number 6 2020 The Korea Trade Research Association
The Korea Trade
www.newktra.org
JKT

Korea Trade and Halal Cosmetic

Uploaded by

Copyright:

Available Formats

You might also like

Korea Trade and Halal Cosmetic

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Korea Trade and Halal Cosmetic

Uploaded by

Copyright:

Available Formats

ISSN 1229-828X

Professor Tae-Hwang Kim

(until December 31, 2021)

MANAGING EDITOR (until December 31, 2020)

Guidelines for authors can be found at:

JKT Vol. 24, No. 8

Professor Eun-Jae Lee Professor Adriana Grigorescu

Professor Hyuk-Soo Cho Professor John C. Yi

Professor Nam-Suk Choi Professor Hongmu Lee

The Impact of Input and Output Tariffs on

Keywords: Comparative Advantage, Employment, Imported Intermediate Inputs, Tariff

2. Literature Review and Research Hypotheses

Fig. 1. Correlations between Import Input Coefficients and Industrial Characteristics

Sources: Author’s calculation using KOSIS, UN COMTRADE, and BOK.

where i and t refer to sectors and years, respectively.

3.2. Robustness Check6

3.3. Data Collection and Summary Statistics

௜௧  ∑௡௝ୀଵ ௝௜ ௝௧ (3)

Table 2 reports summary statistics. I checked correlation coefficients and found no

Table 2. Descriptive Statistics

Table 3. Main Empirical Results

4.2. Heterogeneous Effects across Industries

Table 4. Empirical Results of Interaction Terms: Labor Quality

4.3. Robustness Results

Table 6. Main Empirical Results in the Arellano-Bond Estimator

Table 7. Empirical Results of Interaction Terms in the Arellano-Bond Estimator

The Impact of COVID-19 and Korea’s New

2.1. GVC after the Covid-19 Outbreak

2.2. Korea’s New Southern Policy and COVID-19

2.3. Quantitative Studies on Korea’s GVC with NSP Countries

Table 1. Summary of Studies on Korea’s GVC with NSP Countries

Fig. 1. Structure of Research Methodology

whereX  : gross output vector of Country s,

Fig. 2. Structure of the GTAP CGE Model

National economic structure Income distribution

Structure for consumer’s demand Top structure for producer

Source: Various sources such as Hertel and Tsigas (1997).

Table 2. Aggregation Scheme for Country/Region and Sector

3.2. Evaluation of Korea’s Exports with GVC Decomposition

Fig. 3. DID Results on Korea’s Major Exporting Intermediate Goods (%)

Note: Korea’s monthly export data used.

Table 3. GVC Decomposition of Korea’s Exports to Major Countries

4. Simulation Results and Interpretation

4.1. Impact of COVID-19

Table 5. Impact of COVID-19 on Korea’s Total Export by Country/Region (Unit: %)

Table 6. Decomposition of COVID-19 Impacts on VA (Unit: %)

Table 7. Impacts of COVID-19 on Korea’s Exports to NSP Region by Sector (Unit: %)

4.2. VA Decomposition Including NSP Effects

Table 9. Impacts of COVID-19 and NSP on Vertical Specialization (VS) (Unit: %)

Table 10. Impacts of COVID-19 and NSP on RDV Index (Unit: %)

5. Conclusion and Policy Implications

Safeguarding Korean Export Trade through

2.1. eWom and Brand Crisis Management

2.2. Risk Management and Product Development Life Cycle

3.1. Amazon Product Reviews

Table 1. Distribution of Korean Product Reviews by Product Category

We employed established “smoke-term” mechanisms to identify the reviews that mentioned

3.2. YouTube Product Review Videos

4.1. Amazon Product Reviews

௜௧ ∑௡௝ୀଵ ௝௜ ௝௧ (3)

whereX : gross output vector of Country s,

Note: p<0.1, p<0.05, p<0.001.