Download as pdf or txt
Download as pdf or txt
You are on page 1of 19

World Development 138 (2021) 105205

Contents lists available at ScienceDirect

World Development
journal homepage: www.elsevier.com/locate/worlddev

Returns to education in Vietnam: A changing landscape


Seamus McGuinness a,b,c,⇑, Elish Kelly a,b,c, Pham Thi Thu Phuong d, Ha Thi Thu Thuy e,
Adele Whelan a,b
a
The Economic and Social Research Institute, Whitaker Square, Sir John Rogerson’s Quay, Dublin 2, Ireland
b
Department of Economics, Trinity College, Dublin, Ireland
c
IZA – Institute of Labor Economics, Bonn, Germany
d
Institute of Regional Sustainable Development, Vietnam Academy of Social Sciences, Viet Nam
e
National Centre for Socio-Economic Information and Forecast, Hanoi, Viet Nam

a r t i c l e i n f o a b s t r a c t

Article history: Using data from the Vietnam Household Living Standards Survey (VHLSS), this paper examines the
Accepted 11 September 2020 returns to education in Vietnam in 2002, 2008, 2010, 2012 and 2016, and how these returns have chan-
Available online 11 November 2020 ged over time. The returns to educations of men and women displayed a linear pattern in 2002, 2008 and
2010, with earnings rising with increased levels of education. The 2002 to 2010 period was one which
JEL Classification: saw particularly rapid increases in both the growth of educated labour supply and the returns to school-
I26 ing for both males and females. The results also suggest that there existed substantial excess demand for
J23
male and female graduate labour over the period. However, the situation changed radically during the
J24
O53
2010 to 2012 period, despite the economy continuing to grow, the supply of educated labour expanded
P20 at a much slower rate and the wage premiums to education declined. Consequently, during 2010 to 2012
any excess demand for educated labour was eliminated in both the male and female labour markets. The
Keywords: results show that the nature of labour demand altered rapidly during the 2010 to 2016 period, with the
Returns to education relative demand for educated labour slowing substantially, which suggests that the trajectory of the
Relative labour demand Vietnamese economy has not recently been driven by a growth in higher value-added firms demanding
Skill-biased technological change
ever higher quantities of educated labour. The periods 2012 to 2016 saw some stabilisation in the labour
Vietnam Household Living Standards Survey
market, with the returns to education remaining relatively unchanged, albeit at a much lower level com-
(VHLSS) data
Vietnam pared to 2010, and increases in the supply of educated workers to the labour market much lower than
before. The study raises important policy questions regarding the future direction of policy approaches
to both skills and FDI in Vietnam.
Ó 2020 Elsevier Ltd. All rights reserved.

1. Introduction grew in estimating the returns to education in countries transition-


ing from socialist systems to more liberal free market states.
The extensive research that has been undertaken on the returns The initial research in this area focused on Eastern European
to education has established that, in line with the predictions of countries, before and after the introduction of market reforms in
the Becker (1964) human capital model, earnings rise with educa- the late 1980 s (e.g., Rutkowski (1996) for Poland; Večerník
tional attainment and individuals earn a substantial premium from (1995) and Münich et al. (2005) for the Czech Republic; Orazem
gaining qualifications (see for example, Rumberger, 1980; Grubb, and Vodopivec (1995, 1997) for Slovenia; and Brainerd (1998) for
1993; Psacharopoulos and Patrinos, 2004; Strauss and De la Russia).1 In more recent times, increased attention has been paid
Maisonneuve, 2009; Heckman et al., 2008; McGuinness et al., to estimating the returns to education in Asian countries that have
2009; Jensen, 2010). Most of the research in this area relates to undergone significant economic reform. This includes China (see,
returns in developed countries. However, in the mid-1990s interest for example, Fang et al., 2012; and Gao and Smyth, 2015) and
Vietnam.2
⇑ Corresponding author at: The Economic and Social Research Institute, Whitaker
Square, Sir John Rogerson’s Quay, Dublin 2, Ireland.
1
E-mail addresses: seamus.mcguinness@esri.ie (S. McGuinness), elish.kelly@esri. See also Krueger and Pischke (1995) for Eastern Germany after unification with
ie (E. Kelly), ptp110@gmail.com (T.T.P. Pham), hathuy212@gmail.com (T.T.T. Ha), West Germany.
2
adele.whelan@esri.ie (A. Whelan). Vietnamese studies are discussed seperately below.

https://doi.org/10.1016/j.worlddev.2020.105205
0305-750X/Ó 2020 Elsevier Ltd. All rights reserved.
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

More recently, based on a review of 705 studies spanning 1950 to 2010. The second period included an adjustment following the
to 2014, it was found that every year of schooling undertaken global economic crises. However, with economic growth between
increases earnings by 8.8 per cent, and that the rate of return has 2010 and 2016, averaging at 6.4 per cent, only slowed marginally
remained more or less constant over time and is similar for low, (World Bank, 2019). Unlike previous studies, we map demand
middle and high income countries (Psacharopoulos and Patrinos, and supply changes against each other to assess the extent to
2018). Barro and Lee (2013) estimate educational attainment for which the labour market for workers with various levels of quali-
146 countries to show that the average number of years of school- fications was in a state of imbalance and the degree to which the
ing is almost eight years. However, this varies across countries, patterns of relative demand altered following the global recession.
with it being eleven years for advanced countries and seven years This study considers three aspects: (i) has the demand for labour
for developing countries. They also show that the proportion of the been continually skewed towards more educated workers over
population with tertiary education is 32 per cent in advanced time; (ii) the extent to which the global recession represents a
economies and 10 per cent in developing. This research by Barro structural break in the relationship between education and earn-
and Lee (2013) highlights that with development the average years ings; and (iii) the extent to which the pattern of results vary by
of schooling can be expected to increase, and also a higher propor- gender. Data from the Vietnamese Living Standards Survey (VLSS)
tion of the population receiving tertiary level education.3 is used to conduct the analysis.
In relation to Vietnamese studies, Stroup and Hargrove (1969) The remainder of the paper is structured as follows. Section 2
were one of the first to examine the returns to education. Focussing presents a summary of economic and labour market developments
specifically on rural South Vietnam, they found that an additional in Vietnam between 2002 and 2016, along with some of its labour
year of schooling increased earnings by 17 per cent. Doan et al. market institutions. We focus specifically on those factors that may
(2018) used Vietnamese Household Survey data to estimate the have had an impact on the returns to education in Vietnam, such as
returns to education between 1998 and 2014. They found that foreign direct investment (FDI) and minimum wage legislation. The
the return to a year of schooling increased from 3.8 per cent in data, sample and methodologies used in the paper are discussed in
1998 to 10.4 per cent in 2008 before falling back to 6 per cent in Section 3. The educational and earnings profile of Vietnamese
2014.4 Older studies have generated a range of estimates. For exam- employees is presented in Section 4, while the economic returns
ple, the estimated return to a year of education varied from 1.9 per to education and results from our relative demand for labour anal-
cent up to 7.5 per cent between 1993 and 1998 (Doan and Stevens, ysis are presented in Sections 5 and 6 respectively. Finally, Section 7
2011; Nguyen, 2004; Liu, 2006; Pham and Reilly, 2007; Phan and concludes the paper.
Coxhead, 2013).5 Moock et al. (2003), using earning data from the
1992–1993 Vietnam Living Standards Survey (VLSS), estimated the
average private rates of return to primary and university education 2. Background context
at 13 and 11 per cent, respectively, but only 4 to 5 per cent at sec-
ondary and vocational levels. Returns to higher education are 2.1. Economic and labour market developments
reported as higher for females: 12 per cent compared to 10 per cent
for males. Despite the negative effects of the 1997 to 1999 economic crisis,
In relation to over time estimates, Gallup (2002) found that as stated, in the succeeding years Vietnam has experienced strong
returns to education in Vietnam increased from 1993 to 1998; GDP growth rates averaging over 6 per cent annually (World Bank,
while Liu (2006), Pham and Reilly (2007), and Oostendrop and 2019). Specifically, over the main timepoints we examine, from
Doan (2013) also reported increases for the period from 1998 to 2002 to 2010 and 2010 to 2016, the average growth rates were
2006. However, Phan and Coxhead (2013) found that returns to 6.7 per cent and 6.4 per cent respectively. The World Bank
education in Vietnam declined between the periods 1993 to 2002 upgraded Vietnam to a middle-income country in 2010. However,
and 2002 to 2008. it still remains a relatively poor country with incomes for the vast
Vietnam is one of the only South East Asian emerging econo- majority, especially those in rural areas, being far below middle-
mies not to have gone into recession in the global economic down- income levels.
turn. The GDP growth rate has been remarkably steady at The implementation of the 2001–2010 National Socio-
approximately 6 per cent per annum over the past twenty years. Economic Development Plan contributed to the economy’s perfor-
Regarding returns to education in Vietnam, the mixed findings that mance during the 2000s, as has FDI, upgraded infrastructure,
have been derived to date are most likely due to the employment amended legal and policy systems, and increased participation in
of different estimation techniques across the various studies (e.g., the world economy. The latter grew considerably after Vietnam’s
Ordinary Least Squares versus Instrumental Variables) along with accession to the World Trade Organization (WTO) in 2007, as did
differences in the defined sample and control variables included FDI. Between 2007 and 2010, Vietnam attracted $123 billion in
in the estimated models. FDI: this surpassed its previous successful period for FDI, which
This paper proposes to add to the literature in this area by esti- was $20.8 billion between 2001 and 2005 (Quoc and Thi, 2018).
mating the returns to education in Vietnam over four differing Vietnam has continued to attract significant FDI since 2010 (ILO
timeframes from, 2002 to 2008, 2008 to 2010, 2010 to 2012, and & ILSSA, 2018).
2012 to 2016. For the bulk of the first two periods, the Vietnamese The presence of FDI in a country has been found to have a pos-
economy underwent significant changes and achieved high levels itive impact on human capital accumulation. In examining 16
of economic growth averaging at a rate of 6.7 per cent from 2002 Asian countries, for the period 1985 to 2010, Zhuang (2017) found
that an increase in FDI was associated with an increase in sec-
3 ondary education. FDI was found to have a negative impact on ter-
This paper updates their earlier estimates of educational attainment from 107
countries (Barro and Lee, 2001). tiary education in the East Asian countries examined, unless the
4
These figures relate to the results from a standard OLS. The authors get slightly FDI had came from OECD countries. While investment in education
different results after applying a Heckman selection correction. has been a national policy priority for Vietnam, and it has now
5
Glewwe and Jacoby (2004) using panel data for the same time period (1993– achieved universal pre-school and primary education (Patrinos
1998) find that school enrolment for children grew faster in households that
experienced greater increases in wealth. These findings were shown over and above
et al., 2018), still only 23.7 per cent of its labour force had a voca-
the effects for locality-specific factors such as changes in the education returns, tional or higher education qualification in 2017 (ILO & ILSSA, 2018).
supply and quality of schools and the opportunity costs associated with schooling. Specifically, 9.5 per cent had a university or higher qualification,
2
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

3.2 per cent a college qualification, 5.4 per cent a middle-level These variations in unemployment rates by educational attain-
vocational schools qualification, and 3.5 per cent a primary voca- ment are likely due to the shift in the composition of economic
tional schools qualification. Of the remaining labour force, 61.9 activity and employment away from agriculture towards services
per cent were unskilled, 14.4 per cent had no certificate/degree, and manufacturing that has taken place in Vietnam as the econ-
and 2 per cent had short-term vocational training. While there omy has become more open to trade and the inflow of foreign
has been rapid expansion in the number of third-level colleges investment. Table 3 shows the composition of employment by
and universities in Vietnam, and enrolments in both public and pri- industry in 2002, 2010 and 2016, separately for males and females.
vate higher education institutions, these institutions are not neces- Excluding agriculture, overall manufacturing, other services and
sarily providing the courses that line up with the needs of the construction are the three largest sectors of employment. There
labour market (Patrinos et al., 2018). As a result of this, skills mis- were increases in the percentage of both males and females
matches is a growing issue in Vietnam, with the percentage of employed in manufacturing industries and commerce between
workers with a university degree or higher qualification working 2002 and 2016, while the proportions employed fell in ‘other ser-
in secondary education level jobs growing from 15.4 per cent in vices’ and mining and quarrying.
2012 to 22.2 per cent in 2017 (ILO & ILSSA, 2018). This issue The change in the composition of employment over the period
may, in turn, impact the wage return obtained by college and uni- will have been influenced by Vietnam’s National Socio-Economic
versity graduates. Development Plan. One of the objectives of this strategy has been
Economic growth over the recent period in Vietnam, assisted by to increase industrialisation in Vietnam, and manufacturing and
international trade and FDI has stimulated the country’s labour other services are two of the priority industries in this regard
market. As can be seen from Table 1, labour force participation (Hung, 2004). Another factor influencing the large share of Viet-
increased between 2002 and 2016, from 72.5 per cent to 77.3 per namese employed in both manufacturing and other services is gov-
cent, with the male rate increasing at a slightly faster pace. Viet- ernment strategies aimed at encouraging the movement of labour
nam’s employment rate also accelerated over this period, increas- from low-to-high productivity sectors, such as away from agricul-
ing from 71 per cent in 2002 to over 75 per cent in 2010, and it ture, forestry and fishing and mining and quarrying to manufactur-
remained stable at that rate in 2016. Again, male employment ing and other services. This compositional change in employment,
increased at a slightly faster pace to females. and consequential shift in the skill levels required by firms operat-
In general, unemployment tends to be low in Vietnam, fluctuat- ing in the high productivity sectors, is likely to have had an impact
ing between 2 and 3 per cent over the 2002 to 2016 time period on the returns to education in Vietnam over the period. Based on
(Table 1). There is, however, variation in the unemployment rate the education specific unemployment rates presented in Table 2,
by educational attainment. As can be seen from Table 2, which pre- it would appear that the industrial shift that has taken place in
sents education specific unemployment rates between 2010 and Vietnam since 2010 has resulted in greater demand for individuals
2016, unemployment tends to be highest among those with a pri- with vocational qualifications, and also those with high school or
mary, secondary or high school qualification, and lowest among lower, than for individuals with a college, university or higher
those with a vocational or college qualification and the unskilled. qualification.
Between 2010 and 2012, those with a university or higher qualifi- Given the role that FDI has played in Vietnam’s economic devel-
cation also recorded a relatively low unemployment rate. The opment, Table 4 shows the composition of employment by owner-
unemployment rate for the most educated workers, with univer- ship type in 2002, 2010 and 2016. The proportion of males and
sity or college qualifications, increased between 2010 and 2016. females employed in the Foreign sector has grown since 2002,
In contrast, the unemployment rates among some less qualified especially among females: 17.9 per cent were employed in this
workers (educated to primary, secondary and vocational level), sector in 2016, which was more than double the male percentage.
has been falling over the period. The largest share of male and female employment in 2016 was in

Table 1
Principal Changes in Vietnam’s Labour Market: 2002, 2010 and 2016 (%).

2002 2010 2016


Overall Male Female Overall Male Female Overall Male Female
Labour force participation rate 72.5 76.2 69.0 77.4 82.0 73.0 77.3 82.4 72.5
Employment rate 70.9 74.8 67.4 75.3 80.1 70.8 75.6 80.6 71.0
Unemployment rate 2.1 1.9 2.3 2.9 2.7 3.0 2.3 2.4 2.2

Source: Vietnam General Statistic Office and Institute of Labour Science and Social Affairs.

Table 2
Unemployment Rates by Educational Attainment: 2010–2016 (%).

2010 2011 2012 2013 2014 2015 2016


Unskilled 3.8 2.5 2.6 2.5 1.6 2.4 2.2
Primary 21.0 20.7 19.0 15.5 14.0 17.1 15.4
Secondary 27.4 25.8 24.2 21.9 21.1 20.0 19.5
High school 21.0 19.0 18.9 17.5 17.6 15.0 17.4
Vocational primary 4.4 5.8 6.8 8.4 3.1 3.3 3.0
Vocational secondary 5.3 5.6 6.8 6.7 3.1 2.0 1.4
Vocational college1 – – – – 1.5 1.0 0.9
College 2.7 4.4 5.4 6.6 7.7 8.8 9.7
University and above 6.1 7.6 10.1 14.0 16.5 16.3 17.5

Source: Labor Force Survey (LFS), (Vietnam General Statistic Office).


Note: 1 Vocational college information is not available in the official LFS reports until 2014.

3
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table 3
Composition of non-farm employment by persons employed full-time by industries: 2002, 2010 and 2016 (%).

Male Male Male Female Female Female


2002 2010 2016 2002 2010 2016
Mining and Quarrying 3.9 1.9 1.7 1.6 1.0 0.5
Manufacturing 26.2 25.2 27.6 38.4 36.4 41.1
Electricity 1.7 1.8 1.9 0.5 0.9 0.8
Construction 21.6 22.0 21.4 3.8 3.2 2.6
Commerce 9.8 13.5 14.2 13.7 16.2 16.8
Transport 10.6 10.4 10.0 2.9 2.7 2.6
Finance 1.2 1.5 1.5 1.7 2.3 2.2
Other Services 25.0 23.7 21.7 37.4 37.3 33.4
100.0 100.0 100.0 100.0 100.0 100.0

Source: Calculated from Vietnam Household Living Standards Survey (VHLSS) 2002, 2010 and 2016.
Note: The Agricultural sector is excluded (see Section 4).

Table 4
Composition of Non-farm Employment by Persons Employed Full-time by Employment Sector: 2002, 2010 and 2016 (%).

Male Male Male Female Female Female


2002 2010 2016 2002 2010 2016
State 41.1 31.2 24.2 49.9 38.5 29.8
Private 54.6 62.8 67.9 42.1 48.6 51.9
Collective 1.4 0.8 0.7 1.3 0.8 0.4
Foreign 2.9 5.2 7.2 6.7 12.1 17.9

Source: Calculated from Vietnam Household Living Standards Survey (VHLSS) 2002, 2010 and 2016.

the private sector, and these shares have increased steadily since shares of its workforce are still employed in agriculture, forestry and
2002. On the other hand, state and collective sector employment fishing, which are low productivity sectors. Even though there has
have fallen between 2002 and 2016.6 The movement of employ- been a growth in employment in manufacturing industries in recent
ment out of the state and collective sectors into both the private years, and labour productivity is higher here than in agriculture, it
and foreign sectors between 2002 and 2016 reflects the macro shift has not improved much in recent years as many of these industries
of Vietnam from a centrally planned to market economy. are concentrated in subsectors of the economy with relatively low
Vietnam’s increased openness to international trade and indus- skilled labour that is used for processing and assembly work (ILO
trialisation has given rise towards rural to urban migration. Even & ILSSA, 2018).
though the majority of the population still live in rural areas, the Furthermore, even though three-quarters of the total number of
population growth rate in urban areas is three times that of rural waged workers are in the non-state sector, these workers receive
areas because of this internal migration. Consequently, the annual lower wages and have lower productivity levels than those in the
growth rate of the urban labour force is 2.5 per cent per year com- state and FDI sectors. This is because many of these individuals
pared to only 0.4 per cent for the rural labour force. As of 2017, the are in informal employment. Thus, even though most workers
urban labour force was 17.5 million and accounted for 32 per cent can obtain a job easily, obtaining a quality job in Vietnam remains
of the total labour force (ILO & ILSSA, 2018). Many internal challenging. This, in turn, will have implications for wages and eco-
migrants, however, find it hard to obtain jobs in the formal econ- nomic returns to education.
omy when they migrate to urban areas as such job opportunities
are limited (ILO & ILSSA, 2018). Consequently, many of them either 2.2. Vietnamese labour market institutions
have to work in the informal economy or are unemployed. Their
unemployment rate is much higher than the national rate: it stood Developments in Vietnam’s labour market is subject to both
at 9.3 per cent in 2016, which compares with 2.3 per cent for the Labour Codes and related institutions. Within the scope of this
total population. paper, only institutions that directly impact the relationship
Due to increased international competition, and greater integra- between wages and educational level are discussed. In particular,
tion provided by free trade agreements, Vietnams’ productivity we focus on their Labour Codes and minimum wage legislation.
levels have grown in recent years. However, they are still relatively The period 2002 to 2010 saw several emendations to Vietnam’s
low. For example, using the constant 2011 Purchasing Power Pari- Labor Codes8 in order to implement international regulations and
ety (PPP) values, Vietnam’s labour productivity in 2016 gained US to better protect the rights of both employees and employers as Viet-
$9,894. When comparing this to its neighboring Asian countries it nam became a more open economy. Before this time period, the
equalled approximately only 7 per cent of Singapore’s, 18 per cent 1994 Labour Code saw the introduction of the minimum wage.
of Malaysia’s, 37 per cent of Thailand’s, 42 per cent of Indonesia’s, According to this Labour Code, a worker’s wage is determined by
57 per cent of the Philippines’, and 87 per cent of Lao’s. Strikingly, negotiations between employer and employee representatives.
the difference in labour productivity between Vietnam and its However, this negotiated wage is not allowed to be lower than the
neighboring countries continues to increase.7 This is because large minimum wage that is regulated on by the State, which is known
as the general minimum wage. According to the 1994 Labour Code,
6
The collective sector is a typical sector under a centrally planned economy, which the general minimum wage is the lowest payment to unskilled/un-
was the case in Vietnam before ‘renovation’ (i.e., the move towards a more open trained workers, fixed on the basis of cost of living to ensure that
economy). The inclusion of this ownership type information in the analysis allows us employees performing the most elementary work in normal working
to explore how the shift from a centrally planned to a market economy affected
Vietnam’s labour market.
7 8
Data from the World Development Indicators (World Bank, 2017). Specifically, in the years of 2002, 2006, and 2007.

4
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

conditions claim their basic work capacity. The National Wage Coun- face methodology is employed to carry out the survey and, conse-
cil (NWC) of Vietnam was established in 2013 to replace the Ministry quently, has a very high response rate (more than 90 per cent). The
of Labour as Vietnam’s minimum wage oversight body.9 The NWC is micro data used for this study comprised of approximately 10,000
tasked with addressing the concerns of minimum wage earners. households in 2008, 2010, and 2012, and 30,000 households in
Each year they present their findings to the government, accompa- 2002 and 2016. The data were weighted to ensure that the sample
nied by any recommended changes to the country’s minimum wage was representative of the population.
rates. In this paper, we focus on the working-age population, which
Prior to 2006, the State had one general minimum wage (base relates to those aged 15 to 65. Given that the focus of the study
salary) that applied throughout the country. However, since 2006 is on identifying the returns to education and the demand for
(until 2012) there were FDI firm and non-FDI firm minimum wage labour, the sample consists of employees who were working more
rates, which varied according to the region that the firm operated than 14 hours per week.
in (Schmillen and Packard, 2016). The growth rate of the general Vietnam is primarily a rural country with a large share of all
and regional minimum wages has been relatively high.10 For exam- workers in agriculture (57 per cent) and the majority of workers
ple, over the past decade the general minimum wage grew at an (65 per cent) are self-employed. Given this, estimates here of the
average annual rate of approximately 15 per cent. Furthermore, returns to education are limited to the waged earners, thus exclud-
the growth rate of minimum wages has been considerably higher ing the self-employed sector, for which the VHLSS does not collect
than the growth rate of productivity. complete earnings information, and also agricultural workers.
While it is regulated by law that the minimum wage be applied Those employed in the agricultural sector were excluded from
to the whole labour market, in fact, it is mostly applied to the for- the analysis as their working arrangements tend to be more pre-
mal sector including the state and private firms (Nguyen Viet carious and, consequently, labour demand and educational returns
Cuong, 2010). However, even within the formal sector some are less clear for this group of workers. After these data restric-
employees receive payments that are below the minimum wage. tions, the final unweighted employee samples include 11,090 indi-
Under weak enforcement measures, employers may choose to hire viduals in 2002, 4509 in 2008, 4939 in 2010, 5162 in 2012, and
a varied workforce, including some workers receiving the man- 5598 in 2016.12 With these sample restrictions, the analysis relates
dated minimum wage rate and others receiving less, either through to approximately 28 per cent of all the workers: this must be borne
contracting or informally. Thus, this may impact on the returns to in mind when going through the results presented later in the
education in Vietnam, depending on how employers implement paper.13
the minimum and general wage legislation, and also compliance For the educational returns analysis, our dependent variable is
with the legislation. Between 2012 and 2017, Vietnam’s Labor the natural logarithm of hourly wages.14 The main independent
Codes have been further amended, supplemented or new codes variables used were educational attainment, work experience, indus-
issued. The measures that have been introduced have the aim of try,15 region,16 ownership type,17 marital status, and place of resi-
improving the legal environment for the labour market, and also dence.18 Educational attainment is defined according to the
to assist the socio-economic development of the country. For educational system in Vietnam, which is: i) no qualifications, ii) pri-
example, measures have been introduced to protect domestic mary, iii) secondary, iv) high-school, v) vocational training, vi) col-
workers, to extend coverage of mandatory social insurance lege, vii) university, and viii) advanced degree. In this paper, due
schemes to cover workers with labour contracts of minimum one to the declining numbers with basic qualifications, we merge the
full month, etc. (ILO & ILSSA, 2018). lowest three levels into a category defined as ‘Secondary or below’.
In general, the structure of the education system in Vietnam,
which consists of public and non-public schools, colleges and uni-
3. Data, sample and measures

Data from the Vietnam Household Living Standards Survey


12
(VHLSS) are used for the analyses conducted in this paper. The In the years 2002 and 2016, data consisted of a boosted income household
VHLSS is a nationally representative survey which was first imple- samples. Specifically, the data sample in 2016 (2002) include 9,000 household
samples for income-expenditure and 36,000 (30,000) household samples for income.
mented in 1992. A second survey was carried out in 1998, and
In other years (2006–2012) there are 9000 household samples for income-
since 2000 the survey has been conducted biennially. The VHLSS, expenditure only. That is why the sample size in 2002 is higher than in other survey
which is implemented by the General Statistics Office (GSO) of years.The household size on average is about 3.8 – 4.
13
Vietnam, is part of the living standards measurement study that This 28 per cent figure is an average of all the waves of VHLSS data used in this
is carried out in more than 100 countries.11 paper: 2002, 2008, 2010, 2012 and 2016. For the individual waves of data, the rate
varied between 21 and 33 per cent. This figure is slightly lower than that published in
VHLSS collects information on various aspects of the house- the Labor Force Survey (LFS). For example, based on the LFS data, the percentage of
holds’ living standards and individuals’ socio-economic status employees/waged earners has ranged between 30 and 41 per cent between 2007 and
including income, employment, education, and so on. A face-to- 2016. The reason for the discrepancy in the figures between the two sources is
because: (i) waged earners in LFS may include agricultural workers while our VHLSS
sample is refined to non-agricultural workers only; and (ii) the LFS includes all people
who are working regardless of their age and working hours, while for the analysis
9 conducted in this paper the VHLSS sample is restricted to include only those from 15
The NWC is made up of representatives from the Vietnam General Confederation
of Labor, the Vietnam Chamber of Commerce and Industry, the Vietnam Associations to 65.
14
of Small and Medium Enterprises and the Vietnam Cooperative Alliance. Each Hourly wage information is not available in the VHLSS, but it can be calculated
committee member serves a five-year term. using the annual earnings and hours worked information that is captured in the
10 dataset. In this paper, the hourly wage formula that was used is: annual earnings / 12
The general minimum wage applies to the government officials and public
servants, people who receive salaries, allowances, and employees working in the months / (number of hours worked*22 days).
15
government bodies, public service organisations of the Communist Party and the Classification of sector is based on the Vietnamese Standard Industrial Classifi-
State, state-budget funded social-political organisations and agencies at the central, cation 2007, which follows the International Standard Industrial Classification
provincial, district, and commune levels, and the armed forces. The regional minimum (Version 4).
16
wage applies to employees working in enterprises, cooperatives, farms, households, Vietnam is divided into 6 areas in the VHLSS data.
17
individuals and institutions and organisations that use employees under labor Ownership type consists of four categories: state, private, collective and foreign.
18
contract in accordance with the provisions of the Labor Code (Nguyen et al., 2017). As a sensitivity check, we re-ran our analysis without the economic and
11
Assistance was received from the World Bank for the first few waves of the ownership type variables and there was no significant change in the estimates
survey. associated with educational attainment.

5
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

versities, is broadly consistent with that found in other countries The share of vocationally qualified females increased from 17 to
(London, 2011). More specifically, the country’s education system 22 per cent between 2002 and 2008, before declining to 12 per cent
consists of two distinct components: (i) the academic system and in 2016, with much of this reduction occuring between 2012 and
(ii) the technical and vocational system. The academic system con- 2016. It is not clear what has driven the decline in the share of
sists of 6 education levels: (1) pre-primary education;19 (2) primary vocationally qualified employees over the period, but possible
education;20 (3) lower-secondary education;21 (4) upper-secondary explanations include a decline in the dominance of vocationally
education;22 (5) university;23 and (6) post graduate education.24 orientated sectors, demographic factors or continued educational
The technical and vocational system includes: (1) vocational train- advancement among workers holding such qualifications. The
ing;25 (2) college-level technical, technological and specialist skills corollary of these developments is that the proportion of Viet-
training;26 (3) university-level technical, technological and specialist namese employees with low levels of education (i.e., no qualifica-
skills training;27 and (4) post graduate training.28 tions, primary, secondary, and high school) has been falling. Within
Compared to other Asian economies, the Vocational Education both the male and female labour markets the share of employees
and Training (VET) sector in Vietnam is relatively new and is grow- with such low levels of educational attainment, declined dramati-
ing gradually over time. In 2006, there were a total of 861 VET cally between 2002 and 2016. For males, the share holding such
institutes in Vietnam: 262 VET schools and 599 VET centres.29 As basic qualifications fell from 55 per cent in 2002 to 41 per cent
of 2017, there were a total of 1976 institutes: 395 VET colleges (20 in 2012, with no notable decline observed between 2012 and
per cent), 541 secondary VET schools (27 per cent), and 1040 VET 2016. With respect to females, the share of employees with the
centres (52.6 per cent) (Viet, 2017). This is something that needs lowest levels of schooling fell from 50 to 39 per cent between
to be kept in mind when the returns to vocational training results 2002 and 2010 before increasing to 40 per cent in 2012 and declin-
are presented later in the paper. ing to 37 per cent by 2016. The share of vocationally qualified
males increased rapidy between 2002 and 2010 from 14 to 24
per cent, before falling back to 19 per cent in 2016.
4. Education and earnings profile of Vietnamese employees Table 6 shows hourly wages by educational attainment in 2002,
2008, 2010, 2012 and 2016, separately for males and females.
The composition of employment by educational attainment for Hourly wages are expressed in USD and have been deflated.30
males and females is reported in Table 5 for 2002, 2008, 2010, Table 6 also shows how these wage levels have changed over time.31
2012 and 2016. We make use of a larger number of waves of data As is seen in most countries, Vietnamese hourly earnings increase
to allow us to more easily pinpoint any major turning points in either with educational attainment. For example, in 2002 males (females)
the educational composition of employment or the returns to with no qualifications earned 0.59 (0.49) USD per hour, rising to
schooling. 1.90 (1.65) USD per hour for those with an advanced degree
In 2002, less than 15 per cent of male and female employees in education.
Vietnam had a university or advanced degree education. Instead, The pattern of change in hourly earnings growth could be gen-
the majority of the full-time employees in the workforce possessed erally characterised as increasing for more educated categories of
a high school or less qualification (69 per cent of males and 66 per workers between 2002 and 2008, with the most substantial growth
cent of females). However, the education profile of those in for workers with a university qualification or an advanced degree.
employment has changed as the economy expanded over the While the hourly earnings of highly educated males continued to
2002 to 2016 period. Specifically, the proportion of males with a grow strongly between 2008 and 2010, the data suggests that
third-level qualification (i.e., university or higher) increased by the earnings of educated females and less educated males began
over 6 percentage points to approximately 20 per cent over the to fall back during that period. It would appear that earnings
period, with the bulk of the expansion taking place at a relatively growth among educated females began to decline somewhat
even pace between 2002 and 2012. There was little growth in before that of males.
the proportion of male employees with third-level qualifications Within the male labour market, the hourly wages of workers
between 2012 and 2016. There was much more rapid growth in holding a university qualification increased by 16 per cent
the share of females with university level qualifications which between 2002 and 2008 and by 4 pecent between 2008 and
increased from 11 per cent in 2002 to 25 per cent in 2016. In con- 2010; the comparable figures for males with an advanced degree
trast to the male labour market, the proportion of females with were 10 and 48 per cent respectively. Thus while the hourly earn-
third-level education increased rapidly between 2012 and 2016. ings growth of males with a university qualification slowed
rapidly between 2008 and 2010, it accelarated rapidly for males
19
This education is for children below 6 years of age and includes two types of
with an advanced degree. Between 2010 and 2012, the hourly
education; namely, preschool for children below 3 years of age and kindergarten for earnings of more educated males declined in a linear fashion, with
children aged from 3 to 5. the earnings of males holding university and advanced qualifica-
20
The duration of this education level is 5 years (grade 1–5), and it is for pupils aged tions falling by 12 and 28 per cent respectively. Between 2012
from 6 to 10 years.
21
and 2016 the hourly earnings of all employees began to recover
This education level lasts for 4 years (grade 6–9), and it is for pupils aged from 11
to 15 years. somewhat, however, the pattern of growth appears to have
22
This education level is for 3 years (grade 10–12), and is for pupils aged from 15 to reversed relative to earlier periods. Specifically, the hourly earn-
18 years. ings growth of males with the three lowest levels of schooling
23
This education level can take from 4 to 5 years, and is undertaken by those who generally exceeded that of males holding university or advanced
obtained an upper-secondary education certificate.
24
qualifications.
This education level is pursued by those who obtained a university degree.
25
This training lasts for 1 to 2 years (level 2 skilled workers).
26
The duration of this type of training depends on an individual’s previous
30
education and training qualifications. GDP deflators from the World Bank were used: 41.7 for 2002; 84.0 for 2008; 100.0
27
The length of this type of training depends on an individual’s previous education for 2010; 134.5 for 2012; and 147.4 for 2016. The following exchange rates were
and training qualifications. used: 2002: 1 USD = 15,280 VND; 2008: 1 USD = 16,302 VND; 2010: 1 USD = 18,613
28
In addition, adults with a lower-secondary education certificate who desire to VND; 2012: 1 USD = 20,828 VND; 2016: 1 USD = 21,935 VND.
31
learn professional or new skills can attend short-term vocational training, which lasts The findings of Phan and Coxhead (2013) shows that while average real wages
less than a year. rose from 1993 to 2008, most of this increase took place during the 1990s and an
29
Information obtained from the General Department of Vocational Training. obvious slowdown in wage growth was observed in the 2000s.

6
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table 5
Composition of Non-farm Employment by Persons Employed Full-time by Education Level: 2002–2008, 2008–2010, 2010–2012, 2012–2016.

2002 2008 2010 2012 2016 2002–2008 2008–2010 2010–2012 2012–2016


% % % % % Percentage Point D Percentage Point D Percentage Point D Percentage Point D
Males:
Secondary or below 55.12 40.95 43.34 42.08 41.78 14.17 2.39 1.26 0.30
High school 14.21 13.59 10.74 10.94 12.8 0.62 2.85 0.20 1.86
Vocational training 14.44 24.76 23.76 22.33 19.45 10.32 1.00 1.43 2.88
College 2.46 4.22 4.07 4.93 5.84 1.76 0.15 0.86 0.91
University 13.16 15.77 16.87 18.37 18.64 2.61 1.10 1.50 0.27
Advanced degree 0.61 0.71 1.22 1.35 1.49 0.10 0.51 0.13 0.14
Females:
Secondary or below 49.9 35.85 38.8 40.86 36.79 14.05 2.95 2.06 4.07
High school 15.61 16.04 13.83 11.74 15.13 0.43 2.21 2.09 3.39
Vocational training 17.35 22.28 20.45 18.19 12.49 4.93 1.83 2.26 5.70
College 6.11 9.17 8.33 8.85 9.54 3.06 0.84 0.52 0.69
University 10.74 16.21 17.76 19.01 24.78 5.47 1.55 1.25 5.77
Advanced degree 0.29 0.45 0.83 1.35 1.27 0.16 0.38 0.52 0.08

Source: Calculations based on VHLSS data.

The pattern of growth for female earnings differed somewhat to 5. Returns to education in Vietnam 2002, 2008, 2010, 2012 and
males. Hourly earnings growth was almost universal and positively 2016
related to educational attainment between 2002 and 2008, how-
ever, the earnings of females holding both univeristy and advanced In this section of the paper, we apply the standard Mincer equa-
degrees fell between 2008 and 2010. While the earnings of males tion to explore the economic returns to education in Vietnam in
with advanced degrees grew rapidly between 2008 and 2010, 2002, 2008, 2010, 2012 and 2016. This equation can be specified
those of females holding these qualifications declined by over 30 as follows:
per cent during the same period. As was the case for males, 2010
to 2012 saw declines in the hourly wages of the most educated J
X
females, with the 2012 to 2016 period characterised by wage ln W ij ¼ Q ij aj þ dxi þ cx2i þ k þ ei ð1Þ
j¼1
growth across all levels of education. In line with the male pattern,
2012 and 2016 saw more rapid wage growth among workers with
the lowest levels of schooling; however, in contrast to males, where W is hourly earnings in Vietnamese dongs (thousands);32 Q is
females with advanced degrees saw their hourly earnings increase a set of dummy variables taking the value of 1 if a person holds a cer-
by 21 per cent over the period, helping to make up the the substan- tain education qualification (e.g. vocational, college level, etc.) and 0
tial losses incurred during the 2008 to 2012. otherwise; X is a vector of personal characteristics (e.g. marital sta-
Therefore, in summary, the pattern of increasing returns to edu- tus, industry, employment sector, region, urban–rural area, etc.); xi is
cation observed during most of the 2002 to 2010 period seems to a proxy for potential work experience;33 k denotes a series of selec-
have almost reversed in Vietnam, with the turning point appearing tion terms and ei is the error term. Separate male and female earn-
to occur in 2008 for females and 2010 for males. While 2002 to ings equations were estimated for both periods, the results for
2008 can be easily characterised as one of increasing wage growth which are presented in Tables 7 and 8 respectively. It should be
for workers with higher levels of schooling, the hourly wages of noted that our overall pattern of results hold when we include wide
females (males) with univerity and advanced degrees fell between ranging controls for sector, region, marital status etc.34
2008 and 2012 (2010 and 2012) respectively. Between 2012 and The male and female equations were augmented with a sample
2016 while wages growth was positive for workers of all education selection control to account for the truncated nature of the labour
levels, the rate of growth was generally much higher among force data. The first reason relates to the fact that labour market
employees with lower levels of schooling. non-participants can be non-randomly distributed across charac-
One factor likely to have helped preserve Vietnamese workers’ teristics that are also correlated with earnings.35 We also account
wages over the period, particularly for those with lower qualifica- for the possibility that our results are potentially influenced by a
tions, is the increases that took place in the Vietnamese Minimum
Wage. The particularly strong wage growth for those with a college
and higher education qualification during the 2002 to 2010 period
points to an increasing demand for such workers as the economy
has grown between 2002 and 2010. The introduction of the FDI 32
1 USD = 21,570 VND (2015).
33
and Non-FDI minimum wages in 2007 is likely to have played a Our work experience variable is estimated (in years) as age minus 15 for those
role in the observed wage increases over this time period as well. with no qualifications, primary and secondary educational attainment; age minus 18
for those with high school; age minus 19 for those with vocational; age minus 20 for
This could also be potentially related to the relaxation of the com-
those with college; age minus 22 for those with university and age minus 23 for those
mand economy wage grid, which is discussed in more detail by with an advanced degree. The age of 15 is the minimum working age in Vietnam.
Phan and Coxhead (2013). However, the relative change in wages 34
See Appendix Tables A1 and A2 for details.
35
during the later period of 2010 to 2012 is somewhat suprising, par- We ran a two-step Heckman selection model for both male and female samples.
ticularly given that the Vietnamese economy continued to grow The first stage is a probit model predicting the likelihood of working among people
aged 15–65 of all employment statuses. The sample selection control is derived from
during those years in spite of low global growth. A fall in the a stage one probit model of labour force participation (number of children in the
returns to education during periods of growth points to a number household were included as the additional variables in this stage one probit model).
of possibilities including an increase in employer power, whereby Specifically, the number of children less than age 6 and the number of children aged 6
firms are more fully able to capture rents from workers, or, alterna- to 17. This control is then included in the earnings equation at the second stage to
account for the fact that not all people in the labour force participate in the labour
tively, a decline in productivity levels. We will discuss these issues
market, particularly those with lower levels of education. Thus, the omission of this
in more detail in the conclusion section (Section 7). sample selection control would result in biased education return estimates.

7
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table 6
Hourly Non-farm Wages of Males and Females Employed Full-time by Education Level in Vietnam: 2002, 2008, 2010, 2012 and 2016 (USD).

2002 2008 2010 2012 2016 Change Change Change Change


2002–2008 (%) 2008–2010 (%) 2010–2012 (%) 2012–2016 (%)
Males:
Secondary or below 0.59 0.62 0.62 0.62 0.76 6.73 1.06 0.24 22.78
High school 0.82 0.79 0.79 0.82 0.91 2.06 1.07 3.44 10.66
Vocational training 0.86 0.96 0.96 0.91 1.04 11.58 0.03 5.37 14.61
College 0.92 1.13 1.09 1.05 1.09 23.21 3.67 3.85 3.62
University 1.38 1.61 1.68 1.48 1.65 16.42 4.42 11.69 11.03
Advanced degree 1.90 2.10 3.10 2.23 2.36 10.14 47.97 28.00 5.88
Females:
Secondary or below 0.49 0.50 0.50 0.50 0.67 3.11 0.45 0.33 32.75
High school 0.68 0.67 0.61 0.64 0.80 9.52 7.74 4.44 24.57
Vocational training 0.79 0.76 0.80 0.76 0.92 0.69 5.77 5.17 21.10
College 0.93 1.14 0.94 0.97 1.01 1.70 17.54 2.72 4.71
University 1.18 1.41 1.37 1.22 1.24 15.63 3.18 11.20 2.19
Advanced degree 1.65 2.91 2.02 1.53 1.86 22.44 30.56 24.26 21.41

Source: Calculations based on VHLSS data (figures deflated).


Notes: GDP deflators from the World Bank were used: 41.7 for 2002; 84.0 for 2008; 100.0 for 2010; 134.5 for 2012; and 147.4 for 2016. The following exchange rates were
used: 2002: 1 USD = 15,280 VND; 2008: 1 USD = 16,302 VND; 2010: 1 USD = 18,613 VND; 2012: 1 USD = 20,828 VND; 2016: 1 USD = 21,935 VND.

second form truncation which relates to our decision to exclude agri- 2016. By 2016, the return to a degree was some 9 percentage
cultural workers and those employed on a part-time basis from our points below the 2002 estimate. The rate of change in the return
sample; we consequently estimate a second selection equation that to postgraduate qualifications was also remarkable; the premium
accounts for wage determining differences in observables between increased from 99 to 122 per cent between 2002 and 2010 before
our sample and this excluded component.36 Consequently, our mod- falling back to around 95 per cent in both 2012 and 2016. Com-
els are augmented by two sample Heckman selection terms, Lambda pared to the reference category of secondary level education and
1 and Lambda 2. Lambda 1 accounts for differences in observables below, the wage premium to high school and vocational qualifica-
relative to the inactive and Lambda 2 acccounts for differences in tions were 23 and 34 per cent respectively in 2010, however, by
the waged workers not included in our sample. 2016 both returns had fallen by over 10 percentage points for both
One factor that we were not able to control for in estimating our levels of attainment. Relative to the reference category, the return
earnings models, which is often included in female wage equa- to a college level education increased from 55 to 61 per cent
tions, is ‘time out of the labour market’, which can be particularly between 2002 and 2010 before falling back to 36 per cent in
important in explaining female wages (McGuinness et al., 2011). 2016, with the majority of the decline in the premium occuring
However, in the case of Vietnam, female labor force participation between 2012 and 2016. The results emphasis a very dramatic
is less sensitive to child bearing and rearing, which results in reversal in the returns to education between 2010 and 2012 for
females spending very little time out of the labour market males, a period during which the Vietnam economy continued to
(World Bank, 2012); thus, this factor is not as important when esti- grow strongly. With respect to the other covariates in the model,
mating a Vietnamese female wage equation as it is for other there was no evidence of any substantial and sustained changes
countries. in the returns to potential labour market experience.
The male education return results (Table 7) display a linear pat- Results from the female returns to education models for 2002,
tern, which is observed in most countries. Specifically, the returns 2008, 2010, 2012 and 2016 are shown in Table 8. The results from
to education increase with higher levels of educational attainment. the female model reflect those found for males with respect to the
This finding is consistent with previous studies on Vietnam (Liu, returns to education. Education premiums again follow a linear
2006; Pham and Reilly, 2007; Phan and Coxhead, 2013; and pattern. The pattern with respect to university qualifications is
Oostendrop and Doan, 2013). In 2002, males with a high school similar, if somewhat more sporadic, to that observed for males.
education earned 24 per cent more compared to those with sec- The returns to a degree did not change markedly between 2002
ondary level education or below. Relative to the reference category, and 2008, however, it increased rapidly from 74 to 104 per cent
the 2002 returns for vocational, college, university and post- between 2008 and 2010 before falling to 86 and 82 per cent in
graduate qualifications were 27, 55, 67 and 99 per cent, 2010 and 2016 respectively. With respect to post-graduate qualifi-
respectively.37 cations, the wage premium increased from 124 to 141 per cent
There were some interesting changes in the returns to educa- between 2008 and 2010, before declining by 29 percentage points
tion for males between 2002 and 2016. Consistent with the between 2010 and 2012. The return to post-graduate degree sta-
descriptive evidence, between 2002 and 2008 the return to a uni- bilised between 2012 and 2016. The returns to lower levels of
versity degree increased by 12 percentage points to 79 per cent, schooling also followed a pattern of substantial increases during
before increasing slightly between 2008 and 2010. The return to the 2002 to 2010 period, which were particularly rapid during
a university degree fell by approximately 25 percentage points 2008 to 2010, and substantial declines over the 2010 to 2016 per-
between 2010 and 2012 before stabilising between 2012 and iod, with most of the adjustment occurring during 2010 to 2012.
The return to labour market experience fluctuated somewhat over
36
We use the same specification for the first stage equation as that used for the the period within the female labour market with no evidence of
labour market participation models. any consistent trend.
37
Please note, as Table 6 reports a log-level regression, the estimates included in the Therefore, in summary, the evidence suggests that while the
text are in line with the table coefficients. However, when the outcome variable is log returns to almost all levels of schooling increased rapidly for both
transformed, it is natural to interpret the exponentiated regression coefficients. These
males and females between 2002 and 2010, the rates of return fell
values included in the text are therefore estimates of the exponentiated regression
coefficients which can be calculated by using the following formula: exp(coefficient)- back dramatically over the 2010 to 2012 period, with 2010 mark-
1. ing a clear turning point for most qualification levels across both
8
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table 7
Returns to Education – Mincer Equations for Men: 2002, 2008, 2010, 2012 and 2016.

(1) (2) (3) (4) (5)


Men 2002 Men 2008 Men 2010 Men 2012 Men 2016
Education and experience controls:
Years worked 0.050*** 0.052*** 0.046*** 0.024*** 0.048***
(0.006) (0.009) (0.009) (0.009) (0.006)
Years worked squared 0.001*** 0.001*** 0.001*** 0.000** 0.001***
(0.000) (0.000) (0.000) (0.000) (0.000)
Education attainment (Ref: Secondary level or below):
High school 0.237*** 0.163*** 0.229*** 0.182*** 0.115***
(0.027) (0.053) (0.057) (0.044) (0.041)
Vocational training 0.272*** 0.282*** 0.336*** 0.161*** 0.198***
(0.030) (0.069) (0.053) (0.050) (0.053)
College 0.548*** 0.570*** 0.606*** 0.535*** 0.364***
(0.045) (0.082) (0.065) (0.049) (0.047)
University 0.669*** 0.794*** 0.812*** 0.564*** 0.580***
(0.031) (0.082) (0.071) (0.059) (0.058)
Graduate 0.989*** 1.084*** 1.223*** 0.955*** 0.948***
(0.093) (0.146) (0.141) (0.107) (0.097)
Selection Terms:
Lambda 1 0.058 0.035 0.205** 0.191** 0.026
(0.045) (0.077) (0.082) (0.082) (0.058)
Lambda 2 0.416*** 0.335 0.286 0.586*** 0.451**
(0.069) (0.245) (0.202) (0.157) (0.181)
Constant 1.756*** 1.870*** 2.048*** 2.487*** 2.394***
(0.074) (0.151) (0.119) (0.123) (0.100)
Observations 6,664 2,349 2,736 2,838 3,035
R-squared 0.259 0.319 0.321 0.316 0.296

Notes: Robust standard errors are reported in parentheses;


As this is a log-level regression, the precise estimates are calculated as exp(coefficient)-1.
*Significant at 10%; **Significant at 5%; ***Significant at 1%. Sample of non-farm and full-time employees.

labour markets. There is nothing to suggest that returns to educa- 2008. The declines in returns in the Mekong River Delta were more
tion recovered between 2012 and 2016, in fact, the wage premium signficant than those observed nationally or in the Red River Delta.
for some qualifications continued to decline over this period. For males, educational wage premiums increased sharply between
We have included a number of robustness tests within the 2002 and 2012 before falling off sharply thereafter to the point
Appendix A. Tables A1 and A2 estimate the returns to education where no wage premium was observed for high school or voca-
using an expanded model specifcation that controls for a wide tional qualification by 2016. A similar, somewhat more extreme,
range of factors including marital status, sector, region and owner- picture emerged for females with the returns to almost all qualifi-
ship type. While there were some marginal differences in the esti- cations falling to zero by 2016.
mated premiums associated with various levels of schooling
compared to those of our basic mincer equations, the trends in
terms of the educational wage premiums, and specifically the 6. Relative demand for labour in Vietnam 2002–2010
declines that occurred from 2010 onwards, are completely
consistent. It is obvious from our earlier analysis that a number of complex,
Furthermore, in order to ensure that the trends in the returns to and sometimes seemingly contradictory developments have been
education are universal in nature, and not driven by particular occurring within the Vietnamese labour market over recent years.
areas, we have estimated separate models for the Mekong River Between 2002 and 2016, the economy in Vietnam has continued to
Delta and Red River Delta regions. The regions were chosen on grow and the share of educated labour in employment has
the basis of having workable data samples that would allow for expanded with it. Consistent with the predictions of Skilled Bias
reliable estimation. Tables A2–A6 report the models for the Red Technological Change (SBTC) whereby the growth in labour
River Delta regions for females and males respectively. The results demand is skewed towards educated workers, the period 2002–
reflect the national picture, with the rates of return to higher levels 2010 saw significant increases in the returns to higher levels of
of educational attainment declining between 2010 and 2012 after education and training. However, despite continued economic
experiencing sharp increases. However, in contrast to the national growth, the pattern of rising returns to higher levels of education
picture, for some qualifications, the turning points differed from and training appeared to have largely reversed between 2010
2010. With respect to the Red River Delta for males, the falls in and 2016. Furthermore, the unemployment rate among more edu-
the premium to university and post-graduate qualifications were cated workers has risen in more recent years which is also points
more similar to those observed nationally, however, the declines towards a reversal of previous patterns.
in the returns to posgraduate qualifcations were more modest. Given such developments, our estimated wage equations, or an
There was more substantial falls in the returns to college level analysis of education specific unemployment rates, cannot tell us
qualifications which declined by 42 percentage points between anything about the extent to which the demand and supply of labour
2010 and 2012. For females, the declines were more substantial were in balance in any period and, more specifically, the degree to
across the three highest levels of educational attainment. Addition- which the fundamental labour demand and supply relationship has
ally, the return to postgraduate qualifications began falling after altered radically over time. Changes in rates of return point to move-
9
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table 8
Returns to Education – Mincer Equations for Women: 2002, 2008, 2010, 2012 and 2016.

(1) (2) (3) (4) (5)


Women 2002 Women 2008 Women 2010 Women 2012 Women 2016
Education and experience controls:
Years worked 0.068*** 0.046*** 0.103*** 0.076*** 0.092***
(0.008) (0.014) (0.015) (0.012) (0.009)
Years worked squared 0.001*** 0.001** 0.002*** 0.002*** 0.002***
(0.000) (0.000) (0.000) (0.000) (0.000)
Education attainment (Ref: Secondary level or below):
High school 0.151*** 0.017 0.228*** 0.116** 0.177***
(0.034) (0.075) (0.065) (0.049) (0.040)
Vocational training 0.307*** 0.195* 0.525*** 0.398*** 0.396***
(0.048) (0.104) (0.081) (0.105) (0.067)
College 0.859*** 0.534*** 0.817*** 0.723*** 0.634***
(0.043) (0.114) (0.064) (0.050) (0.044)
University 0.764*** 0.744*** 1.037*** 0.860*** 0.819***
(0.055) (0.124) (0.091) (0.121) (0.082)
Graduate 1.190*** 1.237*** 1.409*** 1.119*** 1.155***
(0.135) (0.280) (0.178) (0.142) (0.114)
Selection Term:
Lambda 1 0.193*** 0.099 0.268* 0.329** 0.471***
(0.073) (0.106) (0.149) (0.135) (0.098)
Lambda 2 0.790*** 1.378*** 0.592*** 0.726*** 0.459**
(0.083) (0.335) (0.205) (0.260) (0.227)
Constant 1.408*** 1.829*** 1.140*** 1.581*** 1.631***
(0.109) (0.212) (0.201) (0.216) (0.151)
Observations 4,417 1,738 2,013 2,176 2,403
R-squared 0.253 0.369 0.317 0.327 0.303

Notes: Robust standard errors are reported in parentheses.


As this is a log-level regression, the precise estimates are calculated as exp(coefficient)-1.
*Significant at 10%; **Significant at 5%; ***Significant at 1%. Sample of non-farm and full-time employees.

ments in demand at the intensive margin, while unemployment rates In terms of the results, a positive value indicates that the
indicate changes at the extensive margin. Neither indicator can suffi- relative demand for skilled labour is growing faster than relative
cient indentify changes in underlying equilibrium relationships over supply, a negative value that relative supply of skilled labour is
time. To address this question, we estimate the change in the growing more rapidly, and the greater the imbalance between rel-
demand for educated labour relative to the lowest qualifications ref- ative demand and relative supply the further the value will be from
erence category. This work is undertaken using the methodology zero.
developed by Katz and Autor (1999), which has been applied in a Results from the computation of Eq. (2) are presented in Table 9,
number of studies (e.g., Harkness and Machin, 1999; Liu, 2006; which reports not only relative wage and supply changes for the
McGuinness, et al., 2009) to identify the direction and strength of rel- periods 2002–2008, 2008–2010, 2010–2012 and 2012–2016 but
ative labour demand. The approach is based on a number of strong also the relative yearly demand for each education qualification.39
assumptions that are centered on a constant elasticity of demand; The results show that despite rapid increases in the relative supply
therefore, the approach is not flawless. However, it still provides a of educated workers, the demand for all categories of educated
good picture of the general nature of relative labour demand. labour grew faster than that of unskilled labour in the period from
The relative demand for labour between 2002, 2008, 2010, 2012 2002 to 2008. The level of relative demand was particularly high
and 2016 is estimated using Eq. (2): for mid-skilled labour (vocational and college) in the male labour
DW ij

DSij
 market and graduates in the female labour market. During the per-
DD W ij
þ r1 Sij iod 2008 to 2010 the pattern of higher relative demand for educated
¼ 1
ð2Þ labour continued in the female labour market, with relative demand
D r
particularly high for graduates and postgraduates. The 2008 and
where W is the wage return to a particular education level over the 2010 period the male labour market saw the relative demand for
lowest qualification base case, which come from Tables 7 and 8; i skilled and unskilled workers come more into balance, with the
and j denote the separate education groups used in the calculation; exception of the labour market for postgraduate workers for which
S represents the relative labour market shares; and r is the relative demand levels continued to be exceptionally high.
assumed rate at which employers can substitute between the differ- Consistent with our more descriptive analysis, the 2010 to 2012
ent types of labour. Following other authors, such as McGuinness period marked a substantial turning point in the relative demand
et al. (2009), we assume that r will take the value of 1.4. However, for more educated workers. During 2010 to 2012 the share of edu-
research shows that relative labour demand analyses are generally cated males in employment continued to grow more rapidly than
not responsive to different values of r (Harkness and Machin, that of unskilled workers, while in the female market supply
1999).38

38 39
As a robustness check we re-estimated our results, from Table 9, with differing For instance, take the reported figures for males 2002 to 2008 for High School \
values of r. Specifically, for 2002, 2010 and 2012 we used values between 0.7 and 2.8 Secondary level and below. Relative supply is calculated as the % change in the high
(half and double the value of 1.4 respectively) to confirm that the relative labour school to secondary or below employment share ratios in both years taken from
demand was not responsive to different values of r: our results, as reported, did not Table 4 i.e. ((13.59/40.95)/(14.21/55.12))*100=28.73. Relative demand is taken as the
change. high school wage premiums for both periods in Table 6 i.e. (23.7–16.3)= 7.4.

10
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

growth was limited to workers with postgraduate qualifications. At 2016 saw some stabilisation in the labour market, with the returns
the same time as the relative supply of educated labour continued to education remaining relatively unchanged, albeit at a much
to expand, albeit at a slower rate than previously, the wage pre- lower level compared to 2010, and increases in the supply of edu-
mium associated with various levels of schooling fell substantially cated workers to the labour market much lower than before. So
pointing to a shift away from unmet demand for educated workers. what are the potential explanations for this apparent shift in cir-
The one exception to the trend was for females with postgraduate cumstances and, in particular, why have we not observed any sub-
qualifications, whose relative demand continued to be positive stantial improvements in the returns to education during 2012 to
during 2010 to 2012 despite continued increases in relative supply. 2016?
Nevertheless, the 2010 to 2012 period appears to have marked a An obvious explanation for the lower relative demand for edu-
clear turning point in the demand for educated workers in cated labour is an increase in the dominance of lower value-added
Vietnam. firms within the economy. China is slowly losing its cost advantage
Finally, the 2012 to 2016 period is best characterised as one of and many companies are looking to diversify their operations by
stabilisation in the male labour market, whereby there has been a expanding into other Asian countries. Such developments may play
significant slowdown in the relative supply of more educated a favorable role allowing Vietnam to sustain its high growth rate.
workers and a stabilisation of returns. The wage premiums for edu- However, it does have implications for returns to education and
cated workers changed little over the period 2012 to 2016, despite the incentive for individuals to invest in their human capital. Many
having experienced large declines in the previous period. Within Chinese companies have already chosen to relocate their factories
the female labour market, the change in the wage premium for (in particular, labour-intensive manufacturing firms) across the
graduates was slightly positive from 2012 to 2016, despite signifi- border into Vietnam (Yang, 2016). The draw of low value-added
cant increases in relative supply, indicating a continued stronger production is evident, in terms of the cost and availability of
demand for graduates relative to unskilled workers. However, as labour. However, growth based on the accumulation of factor
was the case within the male labour market, the female graduate inputs will eventually experience diminishing returns and, as such,
wage premium remained substantially below its 2010 peak. does not constitute a sustainable strategy for continued growth.
Thus, we can summarise the situation as one whereby the 2002 Given Vietnam’s stage of development, a transition to a
to 2010 period saw a rapid shift in the relative demand away from productivity-driven growth trajectory is vital for past rates of eco-
unskilled towards skilled labour, during which the rapidly expand- nomic growth to be sustained into the future.
ing supply of workers with higher levels of schooling was unable to The decline in returns to schooling over the 2010 to 2016 period
meet labour market requirements. The pattern quickly reversed should be a particular concern for policy as the trend will tend to
itself during 2010 to 2012, which saw a movement of demand lower incentives among individuals to invest in education. There
away from educated labour despite a significant slowdown in rel- is a risk that the economy could move towards a low-skilled equi-
ative supply. For males, the period 2012 to 2016 is best charac- librium whereby lower returns create lower investment in human
terised as one of stabilisation during which the labour market for capital, which is more likely to attract inflows of lower skilled
graduate labour appears to have moved closer to equilibrium dur- employment, further lowering the returns to education and,
ing a period of low growth in both the education wage premia and thereby, giving rise to a ‘low skills trap’ situation. This, in turn,
relative supply. During 2012 to 2016, the female labour market has could give rise to a ‘middle income trap’, which is where a coun-
continued to absorb increased flows in the relative supply of grad- try’s economic growth is stalled and it is unable to move to the
uates without incuring any further large falls in the graduate wage next level of economic development partly because of inadeque-
premium. cies in high levels of human capital, either in terms of supply short-
ages, or what is being supplied not matching the labour market
7. Conclusion needs of more higher wage activities.
As mentioned, Vietnam has undertaken significant investment
Using data from the VHLSS, this paper examined the returns to in its education system, and there has been rapid expansion in
education in Vietnam over a number of different time frames from the number of public and private third-level institutions, along
2002 to 2016. The paper also assesses how the change in demand with enrollments. Nevertheless, less than a quarter (23.7 per cent)
for educated labour relative to those with secondary level qualifica- of its labour force had a vocational or higher education qualifica-
tions or below changed over time. In Vietnam, the GDP growth rate tion in 2017 (ILO & ILSSA, 2018). In addition, concern has been
has been remarkably steady at approximately 6 per cent per annum expressed that these institutions are not necessarily providing
over the past twenty years. For the bulk of the period from 2002 to courses that line up with the needs of the labour market (ILO &
2010 the Vietnamese economy underwent significant changes and ILSSA, 2018; Patrinos et al., 2018). As a result, skills mismatches
achieved high levels of economic growth averaging at a rate of 6.7 is a growing issue in Vietnam, with the percentage of workers with
per cent. The latter periods included an adjustment following the a university degree or higher qualification working in secondary
global economic crises. However, economic growth, averaging at education level jobs growing from 15.4 per cent in 2012 to 22.2
6.4 per cent, only slowed marginally (World Bank, 2019). per cent in 2017 (ILO & ILSSA, 2018). Thus, this could be part of
The male and female education returns displayed a linear pat- the reason for the decline in the returns to higher levels of educa-
tern in 2002, 2008 and 2010, with earnings rising with increased tion in Vietnam between 2010/2012 and 2016. If so, it is important
levels of education. The 2002 to 2010 period was one which saw that higher education institutions are incentivised to improve the
particularly rapid increases in both the growth of educated labour quality and the relevance of the course being provided to ensure
supply and the returns to schooling for both males and females. that they match with labour market needs and to enable Vietnam
The results also suggest that there existed substantial excess to move to the next level of economic development and focus on
demand for male and female graduate labour over the period. high-end economic activity.
However, the situation changed radically during the 2010 to It is important that policy makers do their utmost to ensure the
2012 period, despite the economy continuting to grow, the supply continued flow of high-quality FDI and investment so as to main-
of educated labour expanded at a much slower rate and the wage tain the growth of high-quality employment, thereby facilitating
premiums to education declined. Consequently, during 2010 to increasing labour market returns and incentives to invest in educa-
2012 any excess demand for educated labour was eliminated in tion, along with continued economic growth and development
both the male and female labour markets. The periods 2012 to within the country. Vietnam could also consider examining the
11
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table 9
Annual Average Relative Demand40

2002–2008 Wage change (%) Supply change (%) Relative demand per annum
Men:
High school/ Secondary or below 7.40 28.73 3.06
Vocational/ Secondary or below 1.00 130.80 22.03
College/ Secondary or below 2.20 130.90 22.33
University/ Secondary or below 12.50 61.30 13.13
Graduate/ Secondary or below 9.50 56.67 11.66
Women:
High school/ Secondary or below 13.40 43.03 4.04
Vocational/ Secondary or below 11.20 78.74 10.51
College/ Secondary or below –32.50 108.90 10.57
University/ Secondary or below 2.00 110.08 17.88
Graduate/ Secondary or below 4.70 115.99 20.43

2008–2010 Wage Change (%) Supply Change (%) Relative demand per annum
Men:
High school/ Secondary or below 6.60 25.33 8.04
Vocational/ Secondary or below 5.40 9.33 0.88
College/ Secondary or below 3.60 8.87 1.91
University/ Secondary or below 1.80 1.08 1.80
Graduate/ Secondary or below 13.90 62.36 40.91
Women:
High school/ Secondary or below 21.10 20.33 4.61
Vocational/ Secondary or below 33.00 15.19 15.51
College/ Secondary or below 28.30 16.07 11.78
University/ Secondary or below 29.30 1.23 21.13
Graduate/ Secondary or below 17.20 70.42 47.25

2010–2012 Wage change (%) Supply change (%) Relative demand per annum
Men:
High school/ Secondary or below 4.70 4.91 0.84
Vocational/ Secondary or below 17.50 3.20 13.85
College/ Secondary or below 7.10 24.76 7.41
University/ Secondary or below 24.80 12.15 11.29
Graduate/ Secondary or below 26.80 13.97 11.78
Women:
High school/ Secondary or below 11.20 19.39 17.54
Vocational/ Secondary or below 12.70 15.54 16.66
College/ Secondary or below 9.40 0.89 6.14
University/ Secondary or below 17.70 1.64 11.57
Graduate/ Secondary or below 29.00 54.45 6.92

2012–2016 Wage Change (%) Supply Change (%) Relative demand per annum
Men:
High school/ Secondary or below 6.70 17.84 2.11
Vocational/ Secondary or below 3.70 12.27 1.77
College/ Secondary or below 17.10 19.31 1.16
University/ Secondary or below 1.60 2.20 1.11
Graduate/ Secondary or below 0.70 11.16 2.54
Women:
High school/ Secondary or below 6.10 43.13 12.92
Vocational/ Secondary or below 0.20 –23.74 6.01
College/ Secondary or below 8.90 19.72 1.81
University/ Secondary or below 4.10 44.77 9.76
Graduate/ Secondary or below 3.60 4.48 2.38

Source: Own calculations based on the VHLSS 2002, 2008, 2010, 2012, and 2016.
Sample of non-farm and full-time employees.

human capital focussed economic growth strategy that Singapore income traps, and its GDP per capita grew from USD $5,041 when
pursed to see if such an approach could be adopted to enable it it transitioned to independence in 1965 to USD $60,742 in 2011
to continue its economic development. Specifically, Singapore (Saner and Yiu, 2014).41
achieved its structural transformation by focussing on upgrading Given the development experiences of other Asian countries, it
the skills of its workforce and emphasising higher value-added is reasonable to expect a shift in labour demand from predomi-
economic activities. As part of this strategy, it incentivised nantly manual jobs, towards more skill-intensive jobs that will
companies to invest in the human capital development of their require increased skills and competencies. For example, Vietnam’s
workers.40 This enabled Singapore to avoid low skills and middle Law on High Technology strategy (originally passed in 2009)
specifies a list of high-tech products that are given investment
40 priority and are entitled to benefit from corporate tax rates of
It should be noted that changes in the supply composition will incorporate a
number of factors, principally rising educational attainment and a reduction in the
agricultural sector, but also the impacts of older workers retiring out of the labour
41
market and changes in the labour force participation rate. 2011 PPP US dollars.

12
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

10 per cent for the first 15 years of operation (compared to the Data curation, Software, Formal analysis, Investigation. Ha Thi Thu
normal rate of 22 per cent). This has led to multinational high- Thuy: Data curation, Software, Formal analysis, Investigation. Adele
tech companies such as Apple, Bosch, Intel, LG Electronics, Micro- Whelan: Data curation, Software, Formal analysis, Investigation.
soft, Nokia and Samsung setting up operations in Vietnam. This,
in turn, is contributing to a positive shift in the country’s occupa-
tional structure. Specifically, the proportion of manual workers is Declaration of Competing Interest
declining from 40.8 per cent in 2013 to 37.5 per cent in 2017,
while there is an increase in the occupations requiring technical The authors declare that they have no known competing finan-
professions (ILO & ILSSA, 2018). In addition, expansion in infras- cial interests or personal relationships that could have appeared
tructure will need to be complemented by significant improve- to influence the work reported in this paper.
ments in efficiency and quality to close growing infrastructure
gaps that constrain competitiveness. It is important that policy
in Vietnam refocuses toward higher value-added activtities in Acknowledgements
order to sustain returns to human capital, which is a vital aspect
in ensuring that the country has sufficient capacity for continued This is a joint research between the Economic and Social
growth and development. Research Institute (ESRI), Ireland and the National Centre for
Socio-Economic Information and Forecast (NCIF), Ministry of Plan-
ning and Investment, Vietnam. We would like to thank Irish Aid,
CRediT authorship contribution statement Department of Foreign Affairs, Ireland for funding this joint
research (Grant ID: CIRL00761), and colleagues from both research
Seamus McGuinness: Funding acquisition, Conceptualization, institutes for their contributions to the research work. Sincere
Methodology, Supervision. Elish Kelly: Supervision, Data curation, thanks also to three anonymous referees for their insightful and
Software, Formal analysis, Investigation. Pham Thi Thu Phuong: helpful comments on an earlier draft of this paper.

13
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Appendix A

Table A1
Returns to Education – Mincer Equations with additional controls for Men: 2002, 2008, 2010, 2012 and 2016.

(1) (2) (3) (4) (5)


Men 2002 Men 2008 Men 2010 Men 2012 Men 2016
Education and experience controls:
Years worked 0.045*** 0.056*** 0.053*** 0.030*** 0.048***
(0.006) (0.009) (0.009) (0.009) (0.007)
Years worked squared 0.001*** 0.001*** 0.001*** 0.001*** 0.001***
(0.000) (0.000) (0.000) (0.000) (0.000)
Education attainment (Ref: Secondary level or below):
High school 0.158*** 0.149*** 0.163*** 0.154*** 0.114***
(0.026) (0.052) (0.058) (0.045) (0.041)
Vocational training 0.238*** 0.314*** 0.322*** 0.193*** 0.212***
(0.029) (0.071) (0.054) (0.054) (0.060)
College 0.591*** 0.683*** 0.667*** 0.581*** 0.368***
(0.049) (0.088) (0.068) (0.051) (0.048)
University 0.615*** 0.821*** 0.834*** 0.614*** 0.623***
(0.032) (0.086) (0.074) (0.065) (0.066)
Graduate 1.010*** 1.130*** 1.253*** 1.028*** 0.991***
(0.092) (0.149) (0.127) (0.105) (0.099)
Marital status (Ref: Single):
Married 0.105*** 0.054 0.094*** 0.065** 0.024
(0.025) (0.038) (0.036) (0.031) (0.034)
Widowed 0.131 0.135 0.191 0.085 0.145
(0.116) (0.092) (0.117) (0.116) (0.110)
Divorced 0.040 0.072 0.143 0.022 0.026
(0.087) (0.077) (0.168) (0.086) (0.053)
Separated 0.002 0.101 0.079 0.071 0.029
(0.147) (0.121) (0.157) (0.121) (0.076)
Economic industries (Ref: Manufacturing):
Mining and quarrying 0.155*** 0.432*** 0.277*** 0.168** 0.190***
(0.038) (0.082) (0.088) (0.068) (0.057)
Electricity 0.106** 0.157** 0.056 0.022 0.104*
(0.051) (0.077) (0.080) (0.058) (0.054)
Construction 0.011 0.161*** 0.069* 0.049* 0.033
(0.021) (0.030) (0.036) (0.025) (0.025)
Commerce 0.007 0.076* 0.037 0.028 0.041
(0.031) (0.043) (0.040) (0.030) (0.029)
Transport 0.123*** 0.140*** 0.070 0.086** 0.080**
(0.027) (0.041) (0.043) (0.040) (0.037)
Finance 0.053 0.188 0.147 0.400*** 0.212**
(0.073) (0.115) (0.114) (0.111) (0.085)
Other Services 0.180*** 0.078* 0.253*** 0.102*** 0.033
(0.024) (0.041) (0.046) (0.039) (0.033)
Location (Ref: Rural):
Urban 0.147*** 0.070*** 0.131*** 0.048** 0.034*
(0.016) (0.025) (0.024) (0.019) (0.019)
Sector types (Ref: Private):
State 0.186*** 0.066* 0.065 0.010 0.040
(0.022) (0.037) (0.040) (0.036) (0.030)
Collective 0.128* 0.087 0.178 0.454*** 0.052
(0.071) (0.102) (0.124) (0.116) (0.163)
Foreign 0.391*** 0.221*** 0.183*** 0.124*** 0.172***
(0.045) (0.047) (0.058) (0.040) (0.031)
Location (Ref: Red river Delta):
Northern Midland and Mountainous 0.134*** 0.132*** 0.139*** 0.127*** 0.055*
(0.027) (0.043) (0.041) (0.038) (0.031)
North Central & South Central Coast 0.006 0.064* 0.147*** 0.132*** 0.051*
(0.020) (0.034) (0.036) (0.028) (0.027)
Central Highland 0.054 0.009 0.126** 0.084 0.074
(0.048) (0.062) (0.059) (0.056) (0.054)
South East 0.461*** 0.222*** 0.222*** 0.123*** 0.118***
(0.023) (0.033) (0.035) (0.029) (0.026)
Mekong River Delta 0.129*** 0.044 0.065* 0.095*** 0.106***
(0.024) (0.035) (0.037) (0.029) (0.027)
Selection Term:
Lambda 1 0.036 0.064 0.070 0.091 0.050
(0.048) (0.089) (0.089) (0.100) (0.074)

14
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table A1 (continued)

(1) (2) (3) (4) (5)


Men 2002 Men 2008 Men 2010 Men 2012 Men 2016
Lambda 2 0.426*** 0.451* 0.327 0.567*** 0.343*
(0.068) (0.244) (0.205) (0.164) (0.205)
Constant 1.497*** 1.622*** 1.889*** 2.362*** 2.316***
(0.074) (0.164) (0.117) (0.141) (0.123)
Observations 6,651 2,349 2,736 2,838 3,035
R-squared 0.369 0.379 0.395 0.370 0.338

Notes: Robust standard errors are reported in parentheses; As this is a log-level regression, the precise estimates are calculated as exp(coefficient)-1; *Significant at 10%;
**Significant at 5%; ***Significant at 1%. Sample of non-farm and full-time employees.

Table A2
Returns to Education – Mincer Equations with additional controls for Women: 2002, 2008, 2010, 2012 and 2016.

(1) (2) (3) (4) (5)


Female 2002 Female 2008 Female 2010 Female 2012 Female 2016
Education and experience controls:
Years worked 0.058*** 0.039*** 0.099*** 0.069*** 0.085***
(0.009) (0.014) (0.014) (0.011) (0.009)
Years worked squared 0.001*** 0.001* 0.002*** 0.002*** 0.002***
(0.000) (0.000) (0.000) (0.000) (0.000)
Education attainment (Ref: Secondary level or below):
High school 0.072** 0.057 0.151** 0.095* 0.107***
(0.032) (0.073) (0.066) (0.051) (0.039)
Vocational training 0.295*** 0.120 0.512*** 0.425*** 0.362***
(0.053) (0.103) (0.084) (0.102) (0.066)
College 0.771*** 0.452*** 0.849*** 0.718*** 0.635***
(0.049) (0.118) (0.073) (0.055) (0.050)
University 0.697*** 0.567*** 0.959*** 0.863*** 0.759***
(0.061) (0.125) (0.096) (0.119) (0.082)
Graduate 1.164*** 1.099*** 1.320*** 1.100*** 1.064***
(0.146) (0.286) (0.160) (0.142) (0.112)
Marital status (Ref: Single):
Married 0.038 0.066** 0.120*** 0.073** 0.122***
(0.028) (0.033) (0.041) (0.031) (0.029)
Widowed 0.018 0.030 0.036 0.009 0.010
(0.054) (0.115) (0.100) (0.090) (0.068)
Divorced 0.091 0.003 0.037 0.029 0.086
(0.072) (0.088) (0.088) (0.064) (0.054)
Separated 0.097 0.145 0.011 0.003 0.134
(0.100) (0.240) (0.122) (0.113) (0.207)
Economic industries (Ref: Manufacturing):
Mining and quarrying 0.197*** 0.287*** 0.209 0.188* 0.047
(0.062) (0.090) (0.136) (0.112) (0.111)
Electricity 0.133* 0.153 0.017 0.047 0.163**
(0.073) (0.113) (0.180) (0.099) (0.078)
Construction 0.098** 0.235** 0.148 0.130** 0.173***
(0.039) (0.100) (0.093) (0.058) (0.056)
Commerce 0.051 0.101** 0.091* 0.032 0.005
(0.035) (0.044) (0.052) (0.038) (0.032)
Transport 0.206*** 0.293*** 0.266** 0.052 0.139**
(0.064) (0.100) (0.115) (0.071) (0.071)
Finance 0.070 0.424*** 0.314*** 0.377*** 0.371***
(0.071) (0.094) (0.113) (0.064) (0.065)
Other Services 0.053** 0.006 0.038 0.000 0.086**
(0.025) (0.042) (0.050) (0.039) (0.036)
Location (Ref: Rural):
Urban 0.142*** 0.026 0.117*** 0.062** 0.063***
(0.021) (0.028) (0.031) (0.024) (0.020)
Sector types (Ref: Private):
State 0.204*** 0.254*** 0.140*** 0.101*** 0.001
(0.026) (0.042) (0.050) (0.036) (0.034)
Collective 0.247*** 0.045 0.187 0.011 0.168*
(0.080) (0.082) (0.144) (0.210) (0.093)
Foreign 0.296*** 0.217*** 0.360*** 0.232*** 0.289***
(0.040) (0.035) (0.047) (0.030) (0.026)
Location (Ref: Red river Delta):
Northern Midland and Mountainous 0.013 0.076* 0.124** 0.082** 0.011
(0.032) (0.043) (0.054) (0.037) (0.033)
North Central & South Central Coast 0.002 0.032 0.146*** 0.130*** 0.094***
(0.027) (0.043) (0.044) (0.034) (0.029)

(continued on next page)

15
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table A2 (continued)

(1) (2) (3) (4) (5)


Female 2002 Female 2008 Female 2010 Female 2012 Female 2016
Central Highland 0.028 0.037 0.053 0.154*** 0.010
(0.058) (0.075) (0.074) (0.055) (0.045)
South East 0.515*** 0.224*** 0.323*** 0.180*** 0.122***
(0.030) (0.038) (0.044) (0.034) (0.029)
Mekong River Delta 0.174*** 0.006 0.064 0.092** 0.100***
(0.031) (0.039) (0.048) (0.038) (0.032)
Selection Term:
Lambda 1 0.200*** 0.130 0.326** 0.372*** 0.520***
(0.072) (0.101) (0.138) (0.128) (0.097)
Lambda 2 0.657*** 1.343*** 0.757*** 0.605** 0.644***
(0.083) (0.321) (0.210) (0.262) (0.213)
Constant 1.124*** 1.655*** 0.919*** 1.426*** 1.496***
(0.115) (0.205) (0.192) (0.204) (0.148)
Observations 4,405 1,738 2,013 2,176 2,403
R-squared 0.366 0.433 0.402 0.396 0.389

Notes: Robust standard errors are reported in parentheses; As this is a log-level regression, the precise estimates are calculated as exp(coefficient)-1; *Significant at 10%;
**Significant at 5%; ***Significant at 1%. Sample of non-farm and full-time employees.

Table A3
Returns to Education in the Red River Delta Region – Mincer Equations for Men: 2002, 2008, 2010, 2012 and 2016.

(1) (2) (3) (4) (5)


Men 2002 Men 2008 Men 2010 Men 2012 Men 2016
Education and experience controls:
Years worked 0.050*** 0.059*** 0.062*** 0.019 0.043***
(0.011) (0.020) (0.014) (0.017) (0.011)
Years worked squared 0.001*** 0.001** 0.001*** 0.000 0.001***
(0.000) (0.001) (0.000) (0.000) (0.000)
Education attainment (Ref: Secondary level or below):
High school 0.233*** 0.149** 0.216** 0.303*** 0.124**
(0.049) (0.075) (0.097) (0.071) (0.059)
Vocational training 0.393*** 0.305*** 0.380*** 0.289*** 0.277***
(0.065) (0.100) (0.066) (0.083) (0.060)
College 0.550*** 0.675*** 0.710*** 0.456*** 0.293***
(0.086) (0.122) (0.159) (0.116) (0.073)
University 0.728*** 0.823*** 0.805*** 0.696*** 0.694***
(0.069) (0.115) (0.103) (0.139) (0.076)
Graduate 1.116*** 1.265*** 1.217*** 1.257*** 1.173***
(0.130) (0.214) (0.187) (0.174) (0.155)
Selection Term:
Lambda 1 0.095 0.013 0.224* 0.331* 0.064
(0.083) (0.176) (0.125) (0.175) (0.101)
Lambda 2 0.809 0.065 2.552* 0.287 0.732
(0.553) (0.975) (1.520) (0.911) (1.072)
Constant 1.639*** 1.709*** 2.098*** 2.452*** 2.358***
(0.150) (0.284) (0.181) (0.255) (0.134)
Observations 1,700 603 705 670 749
R-squared 0.328 0.329 0.393 0.376 0.318

Notes: Robust standard errors are reported in parentheses; As this is a log-level regression, the precise estimates are calculated as exp(coefficient)-1; *Significant at 10%;
**Significant at 5%; ***Significant at 1%. Sample of non-farm and full-time employees.

16
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table A4
Returns to Education in the Red River Delta Region – Mincer Equations for Women: 2002, 2008, 2010, 2012 and 2016.

(1) (2) (3) (4) (5)


Women 2002 Women 2008 Women 2010 Women 2012 Women 2016
Education and experience controls:
Years worked 0.029 0.019 0.042 0.044*** 0.059***
(0.020) (0.022) (0.026) (0.016) (0.016)
Years worked squared 0.000 0.000 0.001 0.001** 0.002***
(0.000) (0.001) (0.001) (0.000) (0.000)
Education attainment (Ref: Secondary level or below):
High school 0.163*** 0.298*** 0.304*** 0.231*** 0.117
(0.055) (0.096) (0.098) (0.078) (0.086)
Vocational training 0.276*** 0.110 0.335*** 0.362*** 0.051
(0.072) (0.104) (0.125) (0.093) (0.078)
College 0.510*** 0.553*** 0.425*** 0.474*** 0.163
(0.084) (0.154) (0.147) (0.095) (0.099)
University 0.723*** 0.860*** 0.918*** 0.769*** 0.432***
(0.088) (0.136) (0.141) (0.100) (0.074)
Graduate 1.180*** 1.403*** 1.080*** 1.066*** 0.842***
(0.183) (0.116) (0.198) (0.128) (0.141)
Selection Term:
Lambda 1 0.307** 0.210 0.303 0.040 0.260*
(0.121) (0.160) (0.215) (0.125) (0.135)
Lambda 2 0.623 0.534 0.471 0.417 2.657***
(0.998) (0.750) (0.572) (0.475) (0.743)
Constant 1.740*** 1.938*** 1.884*** 1.971*** 2.392***
(0.285) (0.227) (0.317) (0.174) (0.171)
Observations 1,004 409 458 490 565
R-squared 0.353 0.388 0.353 0.309 0.217

Notes: Robust standard errors are reported in parentheses; As this is a log-level regression, the precise estimates are calculated as exp(coefficient)-1; *Significant at 10%;
**Significant at 5%; ***Significant at 1%. Sample of non-farm and full-time employees.

Table A5
Returns to Education for the Mekong River Delta Region – Mincer Equations for Men: 2002, 2008, 2010, 2012 and 2016.

(1) (2) (3) (4) (5)


Men 2002 Men 2008 Men 2010 Men 2012 Men 2016
Education and experience controls:
Years worked 0.038*** 0.020 0.033 0.050*** 0.037***
(0.012) (0.015) (0.023) (0.016) (0.013)
Years worked squared 0.001*** 0.000 0.001 0.001*** 0.001**
(0.000) (0.000) (0.001) (0.000) (0.000)
Education attainment (Ref: Secondary level or below):
High school 0.140* 0.091 0.499*** 0.331** 0.112
(0.073) (0.108) (0.187) (0.134) (0.125)
Vocational training 0.370*** 0.315*** 0.425*** 0.409*** 0.029
(0.078) (0.095) (0.133) (0.098) (0.131)
College 0.429*** 0.393** 0.842*** 0.687*** 0.350***
(0.104) (0.193) (0.181) (0.087) (0.133)
University 0.555*** 0.628*** 0.918*** 0.817*** 0.327**
(0.078) (0.132) (0.141) (0.100) (0.166)
Graduate 0.761*** 0.809*** 1.210*** – 0.544*
(0.183) (0.102) (0.245) (0.304)
Selection Term:
Lambda 1 0.023 0.311** 0.333 0.057 0.046
(0.092) (0.130) (0.273) (0.162) (0.125)
Lambda 2 0.386*** 0.441* 0.104 0.106 0.682**
(0.109) (0.227) (0.309) (0.187) (0.346)
Constant 1.860*** 2.279*** 2.008*** 1.942*** 2.517***
(0.158) (0.217) (0.290) (0.207) (0.237)
Observations 1,294 373 531 573 571
R-squared 0.181 0.343 0.346 0.373 0.272

Notes: Robust standard errors are reported in parentheses; As this is a log-level regression, the precise estimates are calculated as exp(coefficient)-1; *Significant at 10%;
**Significant at 5%; ***Significant at 1%. Sample of non-farm and full-time employees.Graduate variable omitted because of collinearity in Model (4).

17
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

Table A6
Returns to Education for the Mekong River Delta Region – Mincer Equations for Women: 2002, 2008, 2010, 2012 and 2016.

(1) (2) (3) (4) (5)


Women 2002 Women 2008 Women 2010 Women 2012 Women 2016
Education and experience controls:
Years worked 0.037*** 0.059*** 0.104*** 0.065*** 0.062**
(0.009) (0.013) (0.028) (0.017) (0.024)
Years worked squared 0.001*** 0.001*** 0.002*** 0.002*** 0.001*
(0.000) (0.000) (0.001) (0.000) (0.001)
Education attainment (Ref: Secondary level or below):
High school 0.076 0.036 0.375** 0.140 0.137
(0.093) (0.142) (0.158) (0.146) (0.110)
Vocational training 0.425*** 0.712*** 0.786*** 0.610*** 0.127
(0.102) (0.173) (0.193) (0.123) (0.287)
College 0.919*** 1.151*** 1.304*** 0.729*** 0.654***
(0.077) (0.132) (0.129) (0.123) (0.113)
University 0.897*** 1.380*** 1.196*** 1.094*** 0.332
(0.121) (0.147) (0.194) (0.120) (0.337)
Graduate – 2.474*** 1.794*** 1.422*** 0.453
(0.120) (0.290) (0.214) (0.410)
Selection Term:
Lambda 1 0.187* 0.570*** 0.469 0.484** 0.214
(0.098) (0.131) (0.307) (0.207) (0.327)
Lambda 2 0.545*** 0.512* 0.798** 0.480* 1.884***
(0.096) (0.299) (0.310) (0.269) (0.585)
Constant 1.586*** 1.024*** 0.819* 1.353*** 2.255***
(0.142) (0.205) (0.419) (0.231) (0.524)
Observations 923 295 387 436 447
R-squared 0.229 0.464 0.376 0.280 0.334

Notes: Robust standard errors are reported in parentheses; As this is a log-level regression, the precise estimates are calculated as exp(coefficient)-1; *Significant at 10%;
**Significant at 5%; ***Significant at 1%. Sample of non-farm and full-time employees.
Graduate variable omitted because of collinearity in Model (1).

Table A7
Descriptive Statistics for 2002, 2008, 2010, 2012, 2016.

2002 2008 2010 2012 2016


Variable Obs. Mean Obs. Mean Obs. Mean Obs. Mean Obs. Mean
Wages: Wage 11,085 11.34 4,408 13.72 4,822 16.29 5,077 17.53 5,514 21.50
Ln (Wage) 11,081 2.16 4,408 2.36 4,822 2.50 5,077 2.65 5,514 2.88
Gender: Men 25,672 0.52 9,847 0.50 9,662 0.50 9,715 0.50 9,669 0.50
Women 25,672 0.48 9,847 0.50 9,662 0.50 9,715 0.50 9,669 0.50
Age (in years) 25,672 29.47 9,847 30.34 9,662 31.23 9,715 31.75 9,669 33.39
Experience: Years worked 25,672 13.06 9,847 12.27 9,662 13.81 9,715 14.25 9,669 15.77
Years worked squared 25,672 367.76 9,847 313.18 9,662 375.45 9,715 394.48 9,669 454.39
Education: No qualifications 25,672 0.14 8,938 0.00 9,379 0.07 9,488 0.07 9,425 0.06
Primary 25,672 0.21 8,938 0.17 9,379 0.14 9,488 0.15 9,425 0.15
Secondary 25,672 0.31 8,938 0.31 9,379 0.27 9,488 0.25 9,425 0.27
High school 25,672 0.19 8,938 0.26 9,379 0.22 9,488 0.22 9,425 0.20
Vocational training 25,672 0.08 8,938 0.14 9,379 0.15 9,488 0.14 9,425 0.11
College 25,672 0.02 8,938 0.03 9,379 0.04 9,488 0.04 9,425 0.05
University 25,672 0.06 8,938 0.09 9,379 0.10 9,488 0.11 9,425 0.14
Graduate 25,672 0.00 8,938 0.00 9,379 0.01 9,488 0.01 9,425 0.01
Economic Industries: Manufacturing 11,098 0.31 4,520 0.30 4,956 0.30 5,174 0.31 5,613 0.33
Mining and quarrying 11,098 0.03 4,520 0.02 4,956 0.01 5,174 0.01 5,613 0.01
Electricity 11,098 0.01 4,520 0.02 4,956 0.01 5,174 0.02 5,613 0.02
Construction 11,098 0.15 4,520 0.13 4,956 0.14 5,174 0.13 5,613 0.12
Commerce 11,098 0.11 4,520 0.13 4,956 0.15 5,174 0.15 5,613 0.17
Transport 11,098 0.08 4,520 0.07 4,956 0.07 5,174 0.06 5,613 0.06
Finance 11,098 0.01 4,520 0.02 4,956 0.02 5,174 0.02 5,613 0.02
Other Services 11,098 0.30 4,520 0.31 4,956 0.29 5,174 0.30 5,613 0.27
Location: Urban 25,672 0.45 9,847 0.45 9,662 0.46 9,715 0.47 9,669 0.48
Sector Types: Private 11,084 0.50 4,520 0.53 4,956 0.57 5,174 0.59 5,613 0.60
State 11,084 0.45 4,520 0.36 4,956 0.34 5,174 0.31 5,613 0.27
Collective 11,084 0.01 4,520 0.03 4,956 0.01 5,174 0.01 5,613 0.00
Foreign 11,084 0.04 4,520 0.08 4,956 0.08 5,174 0.09 5,613 0.12
Region: Red river Delta 25,672 0.23 9,847 0.22 9,662 0.23 9,715 0.23 9,669 0.22
Northern Midland and Moutainous 25,672 0.11 9,847 0.13 9,662 0.11 9,715 0.10 9,669 0.11
North Central and South Central Coast 25,672 0.23 9,847 0.20 9,662 0.22 9,715 0.22 9,669 0.21
Central Highland 25,672 0.04 9,847 0.05 9,662 0.05 9,715 0.05 9,669 0.05
South East 25,672 0.16 9,847 0.20 9,662 0.18 9,715 0.19 9,669 0.20
Mekong River Delta 25,672 0.22 9,847 0.20 9,662 0.20 9,715 0.21 9,669 0.21
Children: Number less than age 6 25,672 0.28 9,847 0.29 9,662 0.35 9,715 0.36 9,669 0.39
Number of children aged 6–17 25,672 1.15 9,847 0.03 9,662 0.61 9,715 0.01 9,669 0.01

18
S. McGuinness, E. Kelly, Thi Thu Phuong Pham et al. World Development 138 (2021) 105205

References Nguyen D. T., Pham Thi Tuyet Trinh, Nguyen Thanh Tung. (2017). Minimum Wages
in Vietnam: Preliminary Observations. VEPR Policy Discussion PD-06.
Nguyen, N.N. (2004). ‘‘Trends in the Education Sector.” Economic Growth, Poverty,
Barro, R. J., & Lee, J. W. (2013). A new data set of educational attainment in the
and Household Welfare in Vietnam. P. Glewwe, N. Agrawal and D. Dollar (eds.).
world, 1950–2010. Journal of Development Economics, 104, 184–198.
Economic Growth, Poverty, and Household Welfare in Vietnam (Washington
Barro, R. J., & Lee, J. W. (2001). International data on educational attainment:
DC: World Bank), pp.425-66.
Updates and implications. Oxford Economic Papers, 53(3), 541–563.
Nguyen Viet Cuong. (2010). The impact of a minimum wage increase on
Becker, G. (1964). Human capital: A theoretical and empirical analysis, with special
employment, wages, and expenditures of low-wage workers in Vietnam.
reference to education. University of Chicago Press.
MPRA Paper no. 36751.
Brainerd, E. (1998). Winners and losers in Russia’s economic transition. American
Oostendrop, R. H., & Doan, Q. H. (2013). Have the returns to education really
Economic Review, 88(5), 1094–1116.
increased in Vietnam? Wage versus employment effect. Journal of Comparative
Doan, T., Le, Q., & Tran, T. Q. (2018). Lost in transition? Declining returns to
Economics, 41(2013), 923–938.
education in Vietnam. The European Journal of Development Research, 30(2),
Orazem, P. F., & Vodopivec, M. (1995). Winners and losers in transition: Returns to
195–216.
education, experience, and gender in Slovenia. The World Bank Economic Review,
Doan, T., & Stevens, P. (2011). Labour market returns to higher education in
9(2), 201–230.
Vietnam. Economics The Open-Access, Open-Assessment E-Journal, 5 (2011-
Orazem, P. F., & Vodopivec, M. (1997). Unemployment in Eastern Europe, Value of
2012).
Human Capital in Transition to Market: Evidence from Slovenia. European
Fang, H., Eggleston, K., Rizzo, J., Rozelle, S. and Zeckhauser, R.J. (2012). The returns to
Economic Review, 41(3–5), 893–903.
schooling: Evidence from the 1986 compulsory education law. NBER Working
Patrinos, H. A., Thang, P. V., & Thanh, N. D. (2018). The economic case for education in
Paper 18189. National Bureau of Economic Research.
Vietnam. World Bank Group: Education Global Practice.
Gallup, J. L. (2002). The wage labor market and inequality in Viet Nam in the 1990s.
Pham, H. T., & Reilly, B. (2007). The gender pay gap in Vietnam, 1993–2002: A
World Bank Policy Research, Working Paper No., 2896.
quantile regression approach. Journal of Asian Economics, 18(5), 775–808.
Gao, W., & Smyth, R. (2015). Education expansion and returns to schooling in urban
Phan, D., & Coxhead, I. A. (2013). Long-run costs of piecemeal reform: Wage
China, 2001–2010: Evidence from three waves of the China Urban Labor Survey.
inequality and returns to education in Vietnam. Journal of Comparative
Journal of the Asia Pacific Economy, 20(2), 178–201.
Economics, 41(2013), 1106–1122.
Glewwe, P., & Jacoby, H. G. (2004). Economic growth and the demand for education:
Psacharopoulos, G., & Patrinos, H. A. (2004). Returns to Investment in Education: A
Is there a wealth effect? Journal of Development Economics, 74(1), 33–51.
Further Update. Education Economics, 12(2), 111–134.
Grubb, W. N. (1993). The varied economic returns to postsecondary education: new
Psacharopoulos, G., & Patrinos, H. A. (2018). Returns to investment in education: A
evidence from the class of 1972. Journal of Human Resources, 28(2), 365–382.
decennial review of the global literature. Education Economics, 26(5), 445–458.
Harkness, S., & Machin, S. (1999). Graduate earnings in Britain, 1974–1995 (p. 95).
Quoc, C. H., & Thi, C. D. (2018). Analysis of foreign direct investment and economic
Research Report No: Department for Education and Employment.
growth in Vietnam. International Journal of Business, Economics and Law, 15(5),
Heckman, J., Lochner, J., Lance, J., & Todd, P. E. (2008). Earnings functions and rates
19–27.
of return. Journal of Human Capital, 2(1), 1–31.
Rumberger, R. W. (1980). The economic decline of college graduates: Fact or
Hung, N.M. (2004). Sector development strategy - Planning and priority program in
Fallacy? Journal of Human Resources, 15(1), 99–112.
Vietnam’s socioeconomic development strategy to the year 2010. Orientation
Rutkowski, J. (1996). High skills pay off: The changing wage structure during
Toward the Year 2020 and the System of Guiding Document. Hanoi: Statistical
economic transition in Poland. Economics of Transitions, 491, 89–111.
Publishing House.
Večerník, J. (1995). Changing Earnings Distribution in the Czech Republic: Survey
ILO & ILSSA. (2018). Labour and Social Trends in Vietnam 2012-2017. Hanoi:
Evidence from 1988–1994. Economics of Transition, 3, 355–371.
Institute of Labour Science and Social Affair & International Labour
Saner, R. and Yiu, L. (2014) ‘‘Learning to Grow: A Human Capital-focused
Organization.
Development Strategy, with Lessons from Singapore, in ‘‘Policy Debate |
Jensen, R. (2010). The (perceived) returns to education and the demand for
Learning to Grow Beyond the Middle-Income Trap - Singapore as an Export
schooling. Quarterly Journal of Economics, 125(2), 515–548.
Model?” by Raymond Saner, Lichia Yiu and S. Gopinathan, International
Katz, L. F., & Autor, D. H. (1999). Changes in the wage structure and earnings
Development Policy, Graduate Institute, Nr. 1803, University of Geneva.
inequality. In O. Ashenfelter & D. Card (Eds.). Handbook of Labor Economics (Vol.
Schmillen, A. D., & Packard, T. G. (2016). Vietnam’s Labor Market Institutions,
3A, pp. 1463–1555). Amsterdam: North Holland.
Regulations, and Interventions - Helping People Grasp Work Opportunities in a
Krueger, A., & Pischke, J. (1995). A comparison of east and west German labor
Risky World Policy Research Working Paper 7587. Washington DC: World Bank.
markets before and after unification. In R. B. Freeman & L. Katz (Eds.), Differences
Strauss, H., & De la Maisonneuve, C. (2009). The wage premium on tertiary
and Changes in Wage Structures (pp. 405–445). Chicago, USA: University of
education: New estimates for 21 OECD countries. OECD Journal: Economic
Chicago Press.
Studies, 2009, 183–210.
Liu, A. Y. C. (2006). Changing wage structure and education in Vietnam 1993–1998:
Stroup, R. H., & Hargrove, M. B. (1969). Earnings and education in rural South
The roles of demand. Economics of Transition, 14(4), 681–706.
Vietnam. Journal of Human Resources, 4(2), 215–225.
London, J. D. (2011). Education in Vietnam. Singapore: Institute of Southeast Asian
World Bank. (2019). ‘‘GDP Growth (annual %)” World Bank National Accounts Data.
Studies.
The World Bank Group. Accessed 09-04-2019. https://data.worldbank.org/
McGuinness, S., McGinnity, F., & O’Connell, P. J. (2009). Changing returns to
indicator/NY.GDP.MKTP.KD.ZG?locations=VN.
education during a boom? The case of Ireland. LABOUR, 23(1), 197–221.
World Bank. (2012). Gender and economic opportunity. Toward Gender Equality in
McGuinness, S., Kelly, E., O’Connell, P., & Callan, T. (2011). The impact of wage
East Asia and the Pacific – A companion to the World Development Report.
bargaining and worker preferences on the gender pay gap. European Journal of
Washington DC: World Bank.
Industrial Relations, 17(3), 277–293.
Yang, C. (2016). Relocating labour-intensive manufacturing firms from China to
Moock, P. R., Patrinos, H. A., & Venkataraman, M. (2003). Education and earnings in a
Southeast Asia: A preliminary investigation. Bandung: Journal of the Global
transition economy: The case of Vietnam. Economics of Education Review, 22,
South, 3(1), 3.
503–510.
Zhuang, H. (2017). The effect of foreign direct investment on human capital
Münich, D., Svejnar, J., & Terrell, K. (2005). Returns to human capital under the
development in East Asia. Journal of the Asia Pacific Economy, 22(2), 195–211.
communist wage grid and during the transition to a market economy. Review of
Economics and Statistics, 87(1), 100–123.

19

You might also like