Zhang - Wang and Wang China 2012

Financial Development and Economic Growth:
Evidence from China*
Jin Zhang,a Lanfang Wangb and Susheng Wangc
January, 2012
Abstract: Using data from 286 Chinese cities over the period 20012006, this paper investi-
gates the relationship between financial development and economic growth at the city level in
China. Our results from both traditional cross-sectional regressions and first-differenced and
system GMM estimators for dynamic panel data suggest that most traditional indicators of
financial development are positively associated with economic growth. This result runs con-
trary to the existing conclusion that a state-ruled banking sector, such as that in China, hin-
ders economic growth because of the distorting nature of the government. Since we focus on
the years after Chinas accession to the World Trade Organization (WTO) in 2001 while the
existing studies mainly covered the years before 2001, our finding suggests that the financial
reforms that have taken place after Chinas accession to the WTO are in the right direction.
To examine the sensitivity of our results, different conditioning information sets are experi-
mented with. Our results are shown to be robust.
Keywords: Financial development, Economic growth, Emerging market
Classifications: N2, O1, O43
* We gratefully acknowledge helpful comments and suggestions from anonymous referees and financial sup-
port from the Program of Humanities and Social Science of Chinese MOE (No. 11YJC790271), the Fundamental
Research Funds for the Central Universities (No. 11YJC790271) and the National Natural Science Foundation of
China (No. 71102134).
a, c Department of Economics, Hong Kong University of Science and Technology, Clear Water Bay, Hong Kong.
Email addresses: zhangjin@ust.hk and s.wang@ust.hk.
b Institute of Accounting and Finance, Shanghai University of Finance and Economics. Email address:
wang.lanfang@mail.shufe.edu.cn.
1. Introduction
This paper investigates the relationship between financial intermediation and economic
growth in China. China has been experiencing fast economic growth and rapid expansion of
financial intermediation in the last thirty years. Since the start of its reforms in 1978, the Chi-
nese economy has maintained an annual growth rate of 9.8% in real terms (China Statistical
Yearbook 2007), while the total loans outstanding in its financial institutions relative to GDP
has increased from 51% to 107% (China Compendium of Statistics, 19492004; China Statis-
tical Yearbook 2007). As the largest emerging market and with many years of uninterrupted
fast growth, China presents us with an interesting case for study. One fundamental question is:
what is the relationship between financial development and economic growth in China? A
unique feature of this paper is that our empirical investigation is based on a rich set of city-
level data, in contrast to existing studies that are based on national or provincial datasets.
In the finance-growth literature, China attracts great interest as a unique case. According
to Allen et al. (2005), China is an important counterexample to the common finding in the
finance-growth literature, since China has enjoyed fast economic growth for more than 30
years while its financial sector is very much under state control and is still quite under-
developed today. The literature on the relationship between finance and growth in China gen-
erally finds a negative relationship. Using provincial data over the period 19901999,
Boyreau-Debray (2003) found that financial inter-mediation has a negative impact on local
economic growth. She attributed the negative influence to the banking sectors support of
loss-making state-owned enterprises. Hasan, Wachtel and Zhou (2009) also found that the
financial sector has a negative influence on economic growth using provincial data over the
period 19862002. In contrast, using provincial data for the period 19851999, Chen (2006)
showed that China financial development contributes to economic growth. He further identi-
fied two channels for the financial sector to contribute to the economy: mobilization of savings
and the substitution of loans for budget appropriation. In addition, using provincial data for
the period 19952003, Cheng and Degryse (2007), who studied the impact of the develop-
ment of banks and non-bank financial institutions on local economic growth, found that bank-
ing development has a significant positive effect on economic growth. Guariglia and Poncet
(2008) used data from 1989 to 2003 and two different sets of indicators of financial develop-
ment to examine the relationship between finance and growth in China. They found that their
China-specific indicators measuring state intervention in finance are negatively associated
with economic growth, while the indicators measuring market-driven financing are positively
associated with economic growth. Finally, Ayyagari, Demirg-Kunt and Maksimovic (2008)
used firm-level data to examine the relationship between firm growth and firm financing pat-
terns, i.e., formal versus informal finance. They concluded that it is the formal financial sys-
tem that spurs firm growth, while fundings from informal channels do not. In addition, Park
and Sehrt (2001) found that policy lending by state banks did not fall during the period 1991
Page 2 of 36
1997, and consequently the financial reforms in the mid-1990s were not able to turn the trend
of worsening bank performance around.
We have a unique dataset from 286 Chinese cities over the period 20012006. Our data-
set has two features: (1) Unlike the existing empirical studies on Chinas finance and growth
that employ provincial data, we choose to use city-level data which have more local observa-
tions and information; (2) We focus on the period after Chinas accession to the World Trade
Organization (WTO) in 2001 so as to investigate the effect of recent financial reforms. Results
from both traditional cross-sectional regressions and first-differenced generalized method of
moments (GMM) and system GMM estimators for dynamic panel data suggest that most tra-
ditional indicators of financial development are generally positively associated with economic
growth. This result runs contrary to the existing conclusion that a state-ruled banking sector,
such as that in China, hinders economic growth because of the distorting nature of the gov-
ernment. Since we focus on the years after Chinas accession to the World Trade Organization
(WTO) in 2001 while the existing studies mainly covered the years before 2001, our finding
suggests that the financial reforms that have taken place after Chinas accession to the WTO
are in the right direction. To examine the sensitivity of our results, we experiment with differ-
ent conditioning information sets. In addition, we conduct a sensitivity analysis by introduc-
ing a dummy variable to indicate coastal cities, capital cities and the cities that have hosted
foreign bank entries. Our results are shown to be robust.
The rest of this paper is organized as follows. Section 2 presents a literature review. Sec-
tion 3 briefly describes the development of the Chinese financial sector and provides some
background information about financial intermediation in China. Section 4 describes the da-
taset, defines the variables, and presents summary statistics. Section 5 presents the results
using purely cross-sectional data, while Section 6 discusses and presents the first-differenced
and system dynamic panel results. Section 7 concludes the paper with a summary.
2. Literature Review
Financial intermediaries serve as the medium of the savings-investment process. One
fundamental question is: will development of financial intermediaries exert a positive effect
on economic growth? For a long period of time, economists have had very different views on
this. For example, Robert Lucas (1988) believed that the importance of financial matters is
very badly over-stressed in popular and even much professional discussion, while Merton
Miller (1998) countered with that financial markets contribute to economic growth is a prop-
osition almost too obvious for serious discussion. Amid such disagreements, the literature on
finance and growth continues to expand with new theoretical models and advanced empirical
methods. Recently, a large body of research, especially empirical work, suggests that devel-
Page 3 of 36
opment of financial intermediaries exerts a positive effect on economic growth, rather than
following economic growth passively.
A variety of theoretical models have been proposed to analyze the connection between fi-
nancial development and economic growth. Levine (2005) presented a survey of theories on
the issue and listed five possible channels through which finance may influence growth. These
channels include: (i) providing information about possible investments so as to allocate capi-
tal efficiently; (ii) monitoring firms and exerting corporate governance; (iii) ameliorating risk;
(iv) mobilizing and pooling savings; and (v) easing the exchange of goods and services.
There is also a vast empirical literature on the issue. Early cross-country studies based on
cross-sectional regressions documented a positive correlation between financial development
and economic activity (Goldsmith, 1969; King and Levine, 1993; Levine and Zervos, 1998; La
Porta et al., 2002). Goldsmith (1969) did a ground-breaking empirical study using data from
35 countries. Although a positive link between finance and economic growth was found, the
question on whether there is a causal relationship between financial development and growth
was not addressed. Besides, his work did not systematically control for other relevant explana-
tory variables. King and Levine (1993) added more control variables to their regression model
and employed a dataset containing more countries. They ran regressions on a cross-country
sample of 77 countries over the period 19601989, after controlling for other factors affecting
economic growth, such as trade, education and political stability. However, the causality issue
was again not formally dealt with. Levine and Zervos (1998) further added measures of stock
markets to their regression model and systematically controlled for other factors affecting
long-run growth including banking development. They showed that stock market liquidity and
banking development can predict economic growth. However, none of these cross-country
studies gave a satisfactory answer to the causality question.
To answer the question of whether financial development is a leading indicator or a fun-

damental factor of economic growth, instrumental variables were employed in several cross-
country studies. Levine (1998, 1999) and Levine, Loayza and Beck (2000) identified a coun-
trys legal origin as a valid instrumental variable and found that financial development has a
significant positive impact on economic growth. Levine, Loayza and Beck (2000) further ap-
plied a more advanced econometric technique, the generalized moments method (GMM) for
dynamic panel data, on a panel of 71 countries over the period 19601995. This advanced
technique yielded the same result as the traditional cross-sectional instrumental variable re-
gressions. That is, the exogenous component of financial development is positively associated
with economic growth. Beck, Levine and Loayza (2000) also used GMM estimators for dy-
namic panel data and found that financial development has a large and positive effect on total
factor productivity growth. Benhabib and Spiegel (2000) found that the indicators of financial
development that are correlated with total factor productivity growth are different from those
that stimulate investment using GMM. Dynamic panel models permit the use of instrumental
Page 4 of 36
variables for all the explanatory variables so that more precise estimates can be obtained.
Thus, quite a few studies have examined the relationship between finance and growth using
dynamic panel models in recent years. For example, Rousseau and Wachtel (2002) examined
whether the relationship between finance and growth varies with inflation. Rioja and Valev
(2004a) examined the effects of financial development on the sources of growth in different
groups of countries with panel data of 74 countries. Rioja and Valev (2004b) further found
that the impact of financial development on growth may be nonlinear. Rousseau and Wachtel
(2000) and Beck and Levine (2004) applied dynamic panel techniques to their regression
analyses after adding measures of stock markets to their models. Their results suggested that
some exogenous components of bank and stock market development can have a large impact
on economic growth. Finally, many time-series studies on the relationship between finance
and growth have also documented financial developments positive impact on economic
growth (Jung, 1986; Demetriades and Hussein, 1996; Neusser and Kugler, 1998; Arestis et al.,
2001; Xu, 2000; Christopoulos and Tsionas, 2004; Bekaert et al., 2005).
A few recent papers studied the relationship between finance and growth in individual
countries. Compared with cross-country studies, in studies of individual countries, research-
ers can design specific measures of financial development according to the particular charac-
teristics of the country. These studies can also avoid dealing with country-specific factors in
regression analysis. In a study of the United States, Jayaratne and Strahan (1996) found that
the branch deregulation boosted bank-lending quality and accelerated economic growth. They
also found evidence that financial development stimulated economic growth. By examining
individual states of the United States from 1900 to 1940, Dehejia and Lleras-Muney (2003)
also confirmed that a well-functioning banking system boosts economic growth through im-
proving capital allocation. Beck, Levine and Levkov (2010) assessed the impact of bank de-
regulation on the distribution of income in the United States from 1970s through 1990s. They
found that deregulation reduces income inequality in the U.S. Rousseau and Sylla (2005) set
up a set of multivariate time series models that relate banking and equity market activity to
investment, imports and business incorporations of the United States from 1790 to 1850. They
found strong support for the hypothesis of finance-led growth in the U.S. Rousseau (1999)
studied Japan over the period 18801913. Through a set of vector autoregressive models,
their results offered evidence that financial factors played a leading role in promoting Japans
rise to world power during the Meiji period. Guiso et al. (2004) studied the effect of local fi-
nancial development in Italy. Their results indicated that financial development promotes
firm growth and enhances the probability that an individual starts her own business.
To improve the understanding of the relationship between financial development and

economic growth, more detailed micro-level data have been employed in recent years, both at
the industry and firm levels. Rajan and Zingales (1998) influential paper tested the hypothe-
sis that, at the industry level, the sectors that are more dependent on external financing will
Page 5 of 36
have higher growth in the countries with more-developed financial markets. Their results
supported the hypothesis and confirmed that financial development has a positive influence
on industrial growth. Wurgler (2000) employed industry-level data to examine the impact of
financial development on economic growth through the channel of capital allocation. Cetorelli
and Gambera (2001) investigated the impact of bank concentration on industrial growth using
a sample of 36 industries in 41 countries. They found that bank concentration promotes the
growth of industries that are more dependent on external finance. Kumar et al. (1999) showed
that the average firm size in industries that are heavy users of external finance is larger in
countries with better financial markets. Therefore, they concluded that financial development
has a positive effect on external finance and hence on firm size. Claessens and Laeven (2003)
further examined the joint impact of financial development and property rights on growth in
different industries. They provided evidence that financial development improves financial
access and better property rights foster growth through better asset allocation. Beck at al.
(2005) used firm-level survey data covering 54 countries to evaluate the impact of financing
obstacles on firm growth and found that the negative impact of financial obstacles on growth
is more substantial for small firms. Beck et al. (2008) showed that industries with a larger
share of small firms grow faster in economies with well-developed financial systems.
3. Chinas Financial System: A Historical Review

China has been experiencing very rapid and stable economic growth since the start of its
reforms in 1978. Its economy is now the second largest just behind the U.S. and it may be-
come the largest economy in the world in 10 years based on the Purchasing Power Parity (PPP)
(Allen, Qian and Qian, 2005). Although China is playing an increasingly significant role in the
world economy and its financial system has been subject to substantial structural reforms, its
financial system is still quite underdeveloped, lagging far behind other parts of the economy
in terms of transformation from central planning to market-based operations. To understand
this situation, we need to go back to the history of Chinas financial system.
In this section, we review the institutional history, regulation evolution, and economic
environment of the Chinese financial system. We divide the history into three periods: before
1994, from 1994 to Chinas WTO entry in 2001, and after the WTO entry.
3.1. Before 1994
Before the start of its transition in 1978, there was no market-based financial system in
China. One single bank, the Peoples Bank of China (PBOC), functioned as both the central
bank and the only commercial bank for all banking transactions in China. This highly central-
ized financial system was transformed into a two-tier system when the four state-owned
Page 6 of 36
commercial banks (the Big Four) were formally established. Since 1984, the PBOC began to
function as the central bank and the Big Four state-owned commercial banks (SOCBs) took
over commercial banking business from the PBOC. During this period, these four banks were
known as specialized banks as shown in Table 1. They generally extended loans to state-owned
enterprises regardless of profitability. Because of this policy lending, there was very limited
competition among the Big Four until the mid-1990s.
Table 1. The Big Four and Their Designated Sectors Before 1994
Name of SOCB Designated sectors to serve
Bank of China (BOC) Foreign exchange, foreign trade and the national economy
China Construction Bank (CCB) Construction sector
Agricultural Bank of China (ABC) Rural banking business
Industrial and Commercial Bank of China (ICBC) Commercial and industrial activities in urban areas
Source: PBOC.
However, following economic reforms in the 1980s, the establishments of new banks and
other financial institutions became a source of competition in the financial sector. Bank of
Communications (BOCOM) was the first state-owned joint-equity bank, which was estab-
lished in 1986. Further, foreign banks were gradually allowed to become an integral part of
Chinas banking sector. In the meanwhile, some non-banking financial institutions started to
enter Chinas financial system in the mid-1980s, including trust and investment companies,
financial companies, financial leasing companies, and urban and rural credit cooperatives.
Also, starting from the mid-1980s, the old way of centrally planning the allocation of financial
resources was gradually phased out. To some degree, local governments began to have the
rights to decide on their own resource allocation via loans or self-raised funds. Cull and Xu
(2000 and 2003) showed that banks were more efficient in allocating resources than state
budgetary appropriation over the period 19801994 in China. However, during this period,
Chinas banking sector was overwhelmingly dominated by the Big Four. Their total assets ac-
counted for 64% of the total assets of Chinas entire financial system (Almanac of Chinas Fi-
nance and Banking, 1995). But, the Big Four were renowned for their low efficiency and had
been burdened with a large amount of non-performing loans (NPLs).
3.2. From 1994 to Chinas WTO Entry in 2001
Sweeping reforms on Chinas financial system were initiated in 1994. A series of financial
reforms from 1994 to 2000 entailed a progressive move toward less administrative and more
independent banking operations. First, in order to relieve the Big Four from policy lending,
three policy banks were established in 1994. Second, Chinas monetary policy was shifted to-
wards indirect monetary control. Starting in 1998, credit planning for SOCBs was abandoned.
Accompanied with the enhanced independence of the PBOC, asset-liability management and
indirect monetary instruments began to take over the role of credit planning. Third, in 1995,
Page 7 of 36
the Commercial Bank Law of China was passed and enacted, which provides details of the
requirements for operations of commercial banks. Lastly, the restructuring of the SOCBs was
initiated in late 1990s when four Asset Management Companies (AMCs) were established to-
gether with an injection of 270 billion yuan by the government into the Big Four. In 1999, the
state-owned AMCs bought 1.4 trillion yuan of NPLs from the Big Four, which amounted to
roughly 20% of their total loans (Almanac of Chinas Finance and Banking, 2000).
Meanwhile, more and more new banks appeared in the mid-1990s. China Minsheng Bank
Corporation (CMBC)s establishment in 1996 made it the first privately-owned national bank
in China. By the end of 1999, there were 11 national joint-equity commercial banks, with total
assets of 1,447.7 billion yuan (PBOC 2000). Together with two recently established banks,
there are 13 national joint-equity commercial banks in China as of the end of 2005. Their
names, headquarter locations and year of establishment are listed in Table 2. Established
mostly in the late 1980s and early 1990s, these joint-equity commercial banks have gradually
increased their collective market share while the Big Fours market share has gradually de-
creased.
Table 2. Joint-Equity Commercial Banks

Name of national joint-equity commercial banks Headquarter location Year of establishment
Bank of Communications Shanghai 1986
China Merchants Bank Shenzhen 1987
Hengfeng Bank Yantai 1987
Shenzhen Development bank Shenzhen 1987
CITIC Industrial Bank Beijing 1987
Guangdong Development Bank Guangzhou 1988
Industrial Bank (Formerly Fujian Industrial Bank) Fuzhou 1988
Huaxia Bank Beijing 1992
China Everbright Bank Beijing 1992
Shanghai Pudong Development Bank Shanghai 1993
China Minsheng Banking Corporation Beijing 1996
Huishang Bank Hefei 2005
Bohai Bank Tianjin 2005
Source: PBOCs and the respective banks websites.
In the mid-1990s, a few city commercial banks were established by consolidating local ru-
ral and urban cooperatives. They take the form of joint-equity banks with their business re-
stricted to their location cities. By the end of 1999, 90 such banks were operating in China,
with total assets of 554.7 billion yuan (PBOC 2000). Chinas city commercial banks are re-
ferred to as the third tier of Chinas banking industry. In a field survey of Chinese banks,
Ferri (2008) found that the New Tigers (including state-owned joint-equity banks and city
commercial banks) are better performing than the SOCBs. The unhealthy link between state-
owned entities (SOEs) and SOCBs still negatively affects the performance of SOCBs today.
Furthermore, the restriction on foreign bank entries was relaxed in 1994. During this pe-
riod, the PBOC began to make recommendations to improve bank risk controls and to follow
the Basel requirements. Lastly, two stock exchanges in Chinathe Shanghai Stock Exchange
Page 8 of 36
and the Shenzhen Stock Exchange were established in 1990 and 1991, respectively. Chinas
stock market began to take shape although it was renowned for its lack of transparency and
fairness throughout the 1990s.
3.3. After the WTO Entry
China formally entered the WTO on December 11, 2001. The years thereafter are charac-
terized by an impressive financial liberalization process, including more interest rate liberali-
zation, less restrictions on ownership takeovers, and greater freedom to foreign banks, etc.
Interest rate liberalization is an important element of Chinas effort in financial liberaliza-

tion and marketization of resource allocation. Before 1999, interest rates in money markets
and bond markets were first liberalized. After the WTO entry in 2001, China began to take
quick steps to relax restrictions on interest rates of loans and deposits.
Another supervisory institution, the China Banking Regulatory Commission (CBRC), was
established in 2003. With its establishment, there have been several improvements in asset
quality, capital adequacy, risk control and general supervision. Banks in difficulties or in cre-
dit crisis can be taken over or restructured by the CBRC. The CBRC issued the document
Chinese Banking Sectors Reform, Opening, and New Progress of Regulations on December
5, 2005, which pointed out that one of the major disadvantages or problems associated with
state ownership in the banking sector is that the lack of incentives in monitoring state-owned
banks creates black holes in corporate governance. To alleviate this problem, the CBRC dis-
closes individual bank data regularly, including their NPLs, and makes peer comparisons.
Foreign investment in domestic banks first appeared in 1996, when Asian Development
Bank (ADB) bought a 1.9% share in China Everbright Bank. This practice has been intensified
since 2003, when the CBRC announced guidelines to encourage and facilitate foreign share
holdings. This reflects the general perception that Chinese banks need improvements in cor-
porate governance, operation technologies and risk management through foreign strategic
investors. While Chinese banks have started to allow foreign ownership with minority owner-
ship, they are also taking initiatives to offer their shares to both domestic and foreign market
participants. This can be seen in the initial public offerings (IPOs) made by the Big Four. In
addition, according to the CBRCs regulation issued in February 2006, all newly-established
shareholding commercial banks are required to have at least one foreign strategic investor,
while any form of ownership by local governments is explicitly forbidden. Berger, Hasan and
Zhou (2009) found that minority foreign ownership strengthens the efficiency of Chinese
banks.
Since the accession to the WTO in 2001, foreign banks presence in China has increased
dramatically. At the end of 2006, 223 foreign banks from 42 countries and regions established
242 representative offices and 312 operational institutions, including branches, sub-branches
Page 9 of 36
and wholly-foreign-owned banks (Wang and Zhang, 2009). As part of the entry agreement,
China pledged a five-year time table to fully open up its domestic banking sector for foreign
competition, as shown in Table 3.
Table 3. The Opening-Up Schedule for the Banking Sector after the WTO Accession
2001 2002 2003 2004 2005 2006
Foreign Geography All of China
currency
business Customers All individuals and
enterprises
Local Geography Dalian, Shanghai, Guangzhou, Chengdu, Beijing, Ningbo, All of China
currency Shenzhen and Tianjin Nanjing, Chongqing, Kunming and Shantou,
business Qingdao, Fuzhou and Xiamen Shenyang
Wuhan and Jinan and Xi'an
Zhuhai
Customers All foreign individu- All Chinese All Chinese clients
als and enterprises enterprises
Source: World Trade Organization.
In sum, reforms during 19782006 are largely based on three main pillars: bank restruc-
turing, financial liberalization, and regulation and supervision strengthening.
4. Variables and Data

This section defines variables and describes data. We are going to use city-level data to
examine the relationship between financial development and economic growth in China.1 To
measure financial development, we construct a number of financial indicators. In the mean-
while, we control for different conditioning information sets in our growth regression model.
4.1. Variables
To investigate whether the exogenous component of financial development positively in-

fluences economic growth, a growth regression model is set up with the annual growth rate of
real per capita GDP as the dependent variable. The independent variables include a variable
representing financial development and a conditioning information set controlling for other
factors.
1 We carry out the Feldstein-Horioka (1980) test to test capital mobility among Chinese cities as a justification
for the use of city-level data. The results from both cross-sectional and panel regressions support the view of a low
degree of capital mobility which makes the analysis meaningful. As of the panel regression, we use a fixed-effects
estimator where city dummies and year dummies are included. Boyreau-Debray (2003) and Boyreau-Debray and
Wei (2004) found evidence that the degree of Chinas inter-provincial capital mobility is low.
Page 10 of 36
Indicators for development of financial intermediation
Traditional indicators of financial development have been employed by existing cross-

country studies, such as the value of credits provided by financial intermediaries to the private
sector divided by GDP (Levine, Loayza and Beck, 2000). However, Chinas statistical data do
not provide exact information for us to calculate such indicators at the city level. Hence, re-
searchers who study the finance-growth relationship of China have developed a set of indica-
tors to represent Chinas financial development, taking the data constraint into consideration.
This paper makes use of these indicators to examine the relationship between financial devel-
opment and economic growth in China. Here are the five indicators measured at the city level:
(1) Credit is the ratio of total loans in the financial system (banking institutions and non-
banking financial institutions) to GDP, which measures the overall depth of financial in-
termediation.
(2) Deposit is the ratio of total deposits in the financial system to GDP, which measures the
overall size of financial intermediaries.
(3) Savings is the ratio of total household savings deposited in the financial system to GDP,
which serves as a proxy of Chinas financial development in terms of mobilizing house-
hold savings.
(4) Loans Over Appro is the share of fixed asset investment financed by domestic loans
relative to that financed by state budgetary appropriation. Fixed asset investment comes
from different sources including domestic loans, state budgetary appropriation, foreign
investment and self-raised funds. Among these sources, loans are considered more effi-
cient than state budget appropriation in terms of capital allocation. Thus we follow the li-
terature (Liu and Li ,2001; Guariglia and Poncet, 2008; Chen, 2006) and make use of this
ratio to measure the substitution of more market and profit-oriented financial transac-
tions for state budget appropriation in order to allocate capital more efficiently.
(5) Corporate is the ratio of corporate deposits to total deposits in the financial system.
This measures Chinas financial development in providing corporate banking services.
Conditioning information sets
To use conditioning information sets to capture the influence of factors other than the fi-
nancial indicators on economic growth, we collect data for those control variables that are
traditionally used in the finance-growth literature. To examine the sensitivity of our empirical
results further, we divide these variables into four different conditioning information sets. The
sets are defined as follows:
Page 11 of 36
(1) Simple Set: the constant, the logarithm of the initial per capita GDP2 (Initial PCGDP)
to capture the convergence effect, and the logarithm of the initial level of education 3
(Education) to control for human capital accumulation.
(2) Medium Set: the simple set plus the share of state-owned entities in total fixed asset
investments (SOE) as an inverse proxy for the progress of economic reforms, and the
Consumer Price Index (CPI) to control for inflation.
(3) Policy Set: the medium set plus the ratio of foreign direct investment to GDP (FDI) to
measure the degree of openness of the local economy, and government expenditure over
GDP (Government) to control for city government size.
(4) Full Set: the policy set plus business volume of postal and telecommunication services
(Postal&Telecom) to indicate the status of information transmission, and the density of
roads4 (Infrastructure) as a proxy for local infrastructure.
Note that, due to the possibility of a nonlinear relationship between economic growth and
the explanatory variables, we use natural logarithms of these variables in regressions.
4.2. Data
By the Chinese government administrative classification, there are three levels of cities in
China: municipalities, prefecture-level cities,5 and county-level cities. There are 4 municipali-
ties: Beijing, Shanghai, Tianjin and Chongqing. They are governed directly by the central gov-
ernment and they are not subject to the administration of any provincial government. Each
province in China has about 10 prefecture-level cities,6 which are governed directly by the
provincial government. And, a county-level city is usually governed by a prefecture-level city.
In this paper, a city refers to either a prefecture-level city or a municipality.
We collect a set of panel data from 286 Chinese prefecture-level cities and municipalities
over the period 20012006. The data is from the China City Statistical Yearbook for various
years. The yearbook provides two kinds of statistical data for each variable. One set of data is
from municipalities and urban regions of prefecture-level cities only, and the other set of data
2 The initial per capita GDP is per capita GDP in yuan of the previous year.
3 The initial level of education is the percentage of students in the total population enrolled in secondary
schools in the previous year.
4 The density of roads is defined as the total length of roads in kilometers per one million square kilometers.
5 Some prefecture-level cities (15 in total by the end of 2006) are further defined as sub-province cities. How-
ever, in terms of administrative statistics, they are still considered as prefecture-level cities.
6 There are three special cases: by the end of 2006, there were 21 prefecture-level cities in Guangdong prov-
ince, while Qinghai province and Tibet each had only 1 prefecture-level city.
Page 12 of 36
is from municipalities, urban and rural regions of prefecture-level and county-level cities. We
use the latter so as to take into account any economic activity that might have occurred in any
region of China.
Due to data limitation, we use the provincial data for the variables Loans Over Appro
CPI and Government. This means that all the cities in the same province have the same value
for these variables. The provincial data is from the China Statistical Yearbook for various
years.
Table 4 presents the descriptive statistics and their correlations for the dependent vari-
able and financial indicators. We can see that there is considerable variation across cities.
Table 4. Descriptive Statistics and Correlations

Growth Credit Deposit Savings Loans over Appro Corporate
Descriptive Statistics
Mean 0.127 0.809 1.131 0.713 5.032 0.337
Maximum 0.377 3.288 5.590 1.797 23.127 0.965
Minimum -0.078 0.059 0.083 0.067 0.660 0.012
Stand.Dev. 0.127 0.426 0.510 0.224 4.110 0.129
Observations 1680 1644 1644 1648 1607 1644
Correlations
Growth 1.000
Credit 0.017* 1.000
Deposit 0.030 0.846*** 1.000
Savings -0.089*** 0.548*** 0.730*** 1.000
Loans over Appro 0.160*** -0.06** 0.048* -0.063** 1.000
Corporate 0.228*** 0.549*** 0.552*** -0.082*** 0.078*** 1.000
Notes: The significance levels at the 1%, 5% and 10% are identified by ***, ** and *, respectively.
5. Finance and Growth: Cross-Sectional Analyses

A purely cross-sectional analysis focuses on the initial value regressions, where we use
the values of the financial indicators and control variables contained in conditioning informa-
tion sets in 2001 and the average value of the dependent variable over 2001-2006. Conse-
quently, there is one observation per city. The basic cross-sectional regression model is:
Growthi = a + b Financei + g [Conditioning Information Set ]i + ei , (1)
where the dependent variable Growth is the real per capita GDP growth rate, Finance takes
each of the five financial indicators described in Subsection 4.1, and Conditioning Informa-
tion Set takes each of the four control sets defined in Subsection 4.1.
While this analysis does not resolve the issue of causality, the initial value regressions al-
leviate two critical weaknesses of the contemporaneous regressions where dependent and
explanatory variables are all averaged over the same period. First, it is possible that a common
shock to the dependent and explanatory variables during the same period drives the empirical
findings of the contemporaneous regressions. Second, the contemporaneous regressions
Page 13 of 36
overlook the potential endogenous determination of the dependent and the explanatory vari-
ables (Levine and Zervos, 1998).
Table 5 summarizes the least squares results with different conditioning information sets.
The results indicate a very strong positive association between the development of financial
intermediation and economic growth. For brevity, we report only the coefficients on the fi-
nancial development indicators. Each of the five financial indicators is significantly correlated
with economic growth at the 5% significance level in the simple, medium, policy and full con-
ditioning information set regressions, except that Corporate is significant at the 10% signifi-
cance level in the simple set regression. In addition, the results show that the strong relation-
ship between financial development and economic growth does not merely reflect contempo-
raneous shocks. Furthermore, financial development does not simply follow economic growth.
Table 5. Finance and Growth: OLS Estimators (Initial Value Regressions)

Conditioning Information Set Credit Deposit Savings Loans over Appro Corporate
Simple Coefficient 0.010** 0.008** 0.011** 0.010*** 0.007*
Standard error (0.004) (0.004) (0.005) (0.003) (0.004)
R2 [0.060] [0.049] [0.046] [0.081] [0.039]
Observation 236 236 239 234 235
Medium Coefficient 0.013*** 0.011*** 0.012** 0.010*** 0.010**
Standard error (0.004) (0.004) (0.005) (0.003) (0.004)
R2 [0.103] [0.089] [0.081] [0.100] [0.074]
Observation 236 236 239 234 235
Policy Coefficient 0.011*** 0.010*** 0.012** 0.008** 0.009**
Standard error (0.004) (0.003) (0.005) (0.003) (0.004)
R2 [0.138] [0.135] [0.131] [0.149] [0.121]
Observation 231 231 233 228 230
Full Coefficient 0.012*** 0.012*** 0.011** 0.008** 0.009**
Standard error (0.004) (0.004) (0.005) (0.003) (0.004)
R2 [0.141] [0.141] [0.129] [0.158] [0.124]
Observation 231 231 232 227 230
Notes: Cities in Tibet are excluded from the sample due to missing data. The significance levels at the 1%, 5% and
10% are identified by ***, ** and *, respectively.
6. Finance and Growth: Dynamic Panel Analyses
6.1. Model Specification
Our panel data analyses use data over the period 20012006 from 286 cities. To investi-
gate the relationship between financial development and GDP growth,7 our basic regression
model is:
Growthi ,t = a + b Financei ,t + g [Conditioning Information Set ]i ,t + hi + lt + ei ,t , (2)
7 We will explain how to derive this dependent variable using GMM estimation in the next subsection.
Page 14 of 36
where the subscript i is a city index and t is a time index. {hi } are unobserved city-specific
effects, {lt } are time fixed effects, and ei ,t is an idiosyncratic error term.
6.2. Methodology
The generalized-method-of-moments (GMM) estimators for dynamic panel data (Arella-

no and Bond, 1991; Arellano and Bover, 1995; Blundell and Bond, 1998) have been applied
widely in recent years, especially in deriving the impact of financial development on economic
growth. There are several advantages of using GMM panel estimators over purely cross-
sectional estimators. First, we are able to control for time fixed effects and city-specific effects.
Second, we can use appropriate lags of the independent variables as instrumental variables to
deal with possible endogeneity in the regressors. In our case of growth regressions, a simulta-
neity bias caused by the joint determination of financial development and economic growth
may produce inconsistent estimators. Also, the variables in conditioning information sets may
suffer from an endogeneity problem. The GMM panel estimators allow us to address these
econometric problems using lagged observations of the explanatory variables as instruments
(internal instruments). As such, we can reliably examine the impact of the exogenous compo-
nent of financial development on economic growth in China.
Specifically, let y be the logarithm of real per capita GDP and X be a set of explanatory va-
riables including one of the financial indicators and control variables contained in one of the
conditioning information sets but excluding the lagged dependent variable. Consider the fol-
lowing regression equation:
yi ,t - yi ,t = (a - 1) yi ,t-1 + b X i ,t + hi + lt + ei ,t . (3)
We can rewrite Equation (3) as follows:
yi ,t = a yi ,t-1 + b X i ,t + hi + lt + ei ,t . (4)
The existence of city-specific effects hi makes the within-group estimators inconsistent even if
ei ,t is not serially correlated, because hi is correlated with the lagged dependent variable yi ,t-1 .
Thus, to eliminate city-specific effects, we take the first difference of Equation (4) to obtain:
yi ,t - yi ,t = a( yi ,t-1 - yi ,t-2 ) + b ( X i ,t - X i ,t-1 ) + (lt - lt-1 ) + (ei ,t - ei ,t-1 ). (5)
Now, instrumental variables are needed to deal with two issues: (a) endogeneity of the regres-
sors; (b) correlation between the new error term ei ,t - ei ,t-1 and the lagged dependent variable
yi ,t-1 - yi ,t-2 of Equation (5).
The first-differenced GMM estimators use lagged explanatory variables as the instrumen-
tal variables under two assumptions: (a) the idiosyncratic error term ei ,t is not serially corre-
lated; (b) the variables contained in X i ,t are weakly exogenous. The following moment condi-
tions are used by the first-differenced GMM estimators:
Page 15 of 36
E yi ,t-s (ei ,t - ei ,t-1 ) = 0, for s 2; t = 3,..., T , (6)
E X i ,t-s (ei ,t - ei ,t-1 ) = 0, for s 2; t = 3,..., T . (7)
In our case, these moment conditions imply that the twice and further lagged values of the
real per capita GDP, the financial indicators and the variables contained in a conditioning
information set can be used as instrumental variables to obtain the first-differenced GMM
estimators.
However, as Alonso-Borrego and Arellano (1996) and Blundell and Bond (1998) pointed
out, the instruments available for the first-difference equation are weak instruments when the
explanatory variables are persistent over time. Weak instruments can result in serious finite
sample biases. The variance of the coefficients gets larger asymptotically. To deal with the
potential bias and imprecision of the first-differenced GMM estimators, additional moment
conditions are proposed for an equation expressed in levels (Arellano and Bover, 1995; Blun-
dell and Bond, 1998). When an equation in differences and an equation in levels are combined
as a system, the estimators based on the moment conditions associated with this system are
called system GMM estimators. The instruments for the equation in levels are the lagged dif-
ferences of the explanatory variables. One additional assumption needs to be made to ensure
the validity of the additional instrumental variables: the first differences of the independent
variables in Equation (4) are uncorrelated with city-specific effects hi . In this case, we have
the following moment conditions for the equation in levels:8
E ( yi ,t-s - yi ,t-s-1 ) ( hi + ei ,t ) = 0, for s = 1; t = 3,..., T , (8)

E ( X i ,t-s - X i ,t-s-1 ) ( hi + ei ,t ) = 0, for s = 1; t = 3,..., T . (9)

In our case, these moment conditions imply that the first lagged differences of the real per
capita GDP, the financial indicators and the variables contained in the conditioning informa-
tion sets can be used as additional instruments. Indeed, Bond et al. (2001) and Hauk and
Wacziarg (2009) pointed out that the system GMM estimators should be employed for growth
regressions to generate consistent and efficient parameter estimates.
Two specification tests suggested by Arellano and Bond (1991), Arellano and Bover (1995)
and Blundell and Bond (1998) are to be carried out: (1) The Sargan test 9 for the over-
8 As pointed out by Arellano and Bover (1995), only the most recent differences can be used as instrumental
variables for an equation in levels. Otherwise, we will have redundant moment conditions, since lagged variables
are already used as instruments for the equation in differences.
9 We report the Sargan test statistics rather than Hansen J tests, because Sargan and difference-in-Sargan
tests are not so vulnerable to instrument proliferation as they do not depend on an estimate of the optimal weight-
ing matrix (Roodmand, 2009). We acknowledge the drawback of the Sargan test that it assumes homoskedasticity.
But because we consistently find the Sargan test to be more conservative than the Hansen test which easily pro-
duces J statistics with implausibly perfect p-values of 1.000, we choose to report the Sargan test statistics.
Page 16 of 36
identification restrictions, which is to test the overall validity of the instruments. Under the
null hypothesis that the instruments are valid, the test statistic is asymptotically distributed as
chi-square with the degree of freedom being equal to the number of instruments minus the
number of parameters estimated; (2) a second-order serial correlation test, which is to exam-
ine the hypothesis that the error term ei ,t is not serially correlated. Under the null of no sec-
ond-order serial correlation, the test statistic is asymptotically distributed as standard normal.
6.3. Techniques for Reducing the Instrument Count
The instrument count easily grows large relative to the sample size as the time period
T or the number of explanatory variables rises when differenced GMM or system GMM esti-
mators are performed. Roodman (2009), Windmeijer (2005) and Arellano (2002) pointed out
that numerous instruments can overfit endogenous variables leading to biased estimators.
Roodman (2009) further discussed another problem of instrument proliferation. Specifically,
a high instrument count can make the Sargan test for instrument validity weak and mislead-
ing. He also pointed out weak Sargan tests are particularly misleading for system GMM esti-
mators.
To address the issue of too many instruments, we follow the techniques suggested by
Roodman (2009). The first is to use only one or two lags instead of all available lags for in-
struments. This strategy to limit the number of instruments generated from first-differenced
and system GMM estimators has been adopted by several researchers (Levine, Loayza and
Beck, 2000; Giedeman and Compton, 2009; Demir and Dahi, 2009). The second is to collapse
the instruments through additions into smaller sets. To effectively mitigate the problems
caused by instrument proliferation, Roodman (2009) suggested combining the two strategies.
In this paper, we follow Roodmans (2009) approach to both collapse instruments and limit
the lag depth in first-differenced and system GMM estimators.
In the meanwhile, we carry out difference-in-Sargan tests to examine the validity of the
system GMM instruments, i.e., the validity of the first lagged differences of the explanatory
variables. In addition, we also perform difference-in-Sargan tests for the validity of instru-
ments based on lagged growth, which are viewed as problematic by Roodman (2009).
6.4. Regression Results
Results Using First-differenced GMM Estimators
Although system-GMM estimators have advantages over first-differenced GMM

estimators as mentioned in Subsection 6.2, weak Sargan tests are less problematic in first-
Page 17 of 36
differenced GMM estimations. The results from first-differenced GMM estimations are
summarized in Table 6,10 which suggest that financial development exerts a large positive
impact on economic growth. Four financial development indicators (Credit, Deposit, Loans
Over Appro, and Corporate) are significantly positive at the usually acceptable levels of
significance with various conditioning information sets. There is only one exception: the
coefficient of Corporate is insignificant with the full set. The coefficient of Savings is positive
with each information set, while it is significant only at the 5% significance level with the
simple set.
Table 6. Finance and Growth: First-differenced GMM Estimators

Conditioning Information Set Credit Deposit Savings Loans over Appro Corporate
Simple Coefficient 0.082** 0.066** 0.095** 0.030* 0.094*
Standard error (0.035) (0.031) (0.048) (0.017) (0.053)
Sargan test (p-value) [0.474] [0.033] [0.846] [0.241] [0.288]
Instruments 10 10 10 10 10
Observation 1092 1092 1092 1050 1090
Medium Coefficient 0.079** 0.070*** 0.100 0.038* 0.094*
Standard error (0.037) (0.024) (0.064) (0.023) (0.056)
Observation 1092 1092 1092 1050 1090
Policy Coefficient 0.099** 0.071* 0.021 0.042** 0.125*
Standard error (0.033) (0.060) (0.063) (0.020) (0.060)
Observation 1092 1092 1092 1050 1090
Full Coefficient 0.098* 0.079* 0.020 0.046** 0.115
Standard error (0.077) (0.050) (0.052) (0.018) (0.106)
Observation 1063 1063 1063 1021 1061
Notes: The test statistics and standard errors are asymptotically robust to heteroskedasticity. Cities in Tibet are
excluded from the sample due to missing data. The significance levels at the 1%, 5% and 10% are identified by ***,
** and *, respectively.
The instrument count is quite low, ranging between 10 and 16. Most regressions pass the
Sargan test. There are only two exceptions: the coefficients of Deposit with the simple and
medium sets do not pass the test. For the models that pass the Sargan test, the p-values for
the validity of the full instrument set comfortably satisfy the conventional significance levels,
ranging between 0.241 and 0.846.11
10 We collapse instruments for each explanatory variable, including both the financial indicators and the vari-
ables contained in the conditioning information sets. We use all available lags to instrument the financial indica-
tors and use only one lag for each variable contained in the control sets. The purpose is to avoid exact identification
which makes the Sargan test unavailable.
11 All the first-differenced GMM regressions pass the difference-in-Sargan test for the validity of instruments
based on lagged growth. The p-values exceed the conventional significance levels substantially. These results are
available upon request.
Page 18 of 36
The regression estimates are also economically large. The signs and significance levels are
in line with the results from purely cross-sectional estimators.
Results Using System GMM Estimators
Tables 7812 report system GMM estimates of Equation (2) with the policy set and the full
set respectively. To further ensure the credibility of system GMM estimators, we carry out two
different regressions for each financial development indicator by collapsing and un-collapsing
the instruments.13 Tables 7-8 present the results from the collapsed regressions. Tables A1-
A4 in the appendix present the results from both collapsing and un-collapsing instruments
with four different conditioning information sets (Simple, Medium, Policy and Full).
The regression results reported in Tables 7-8 show a significant positive relationship be-
tween the four financial development indicators (Credit, Deposit, Loan Over Appro, and Cor-
porate) and economic growth. These positive relationships are consistent with the ones from
OLS estimators and first-differenced GMM estimators. All of the regressions pass the second-
order serial correlation test. The null hypothesis that the error term is not serially correlated
cannot be rejected.
The regressions for the three financial indicators (Credit, Deposit and Corporate) pass
the Sargan test, except the Deposit regression with the policy set. Compared with the un-
collapsed variants, the instrument count of the collapsed regressions is reduced to a large
extent, ranging from 19 to 23. Most p-values for the Sargan test comfortably satisfy the con-
ventional significance levels with an average value of 0.354. The p-values for the difference-in-
Sargan test for the validity of the instruments based on lagged growth, which is suspected by
Roodman (2009),14 have an average value of 0.425. The validity of the subsets of instruments
is established for these regressions. We hence confirm that the overall size and depth of the
financial sector spur economic growth and the development of financial intermediation in
China positively influences economic growth through the mobilization of corporate deposits.
12 We limit the lag depth to one for every explanatory variable, including both the financial indicators and the
variables in the control sets.
13 The difference-in-Sargan tests for system GMM instruments are not available when we collapse the instru-
ments and use only one lag per instrumental variable. The model is under-indentified without system GMM in-
struments.
14 Roodman (2009) claims that It seems likely that lagged growth is an invalid instrument in the LLB re-
gressions, that system GMM is invalid(page 155). Our results confirm his findings. When the p-values of the dif-
ference-in Sargan test for the validity of system GMM instruments and instruments based on lagged growth are
lower than the conventional levels simultaneously or only the latter is lower, the regression fails to pass the Sargan
test for joint validity of the instruments.
Page 19 of 36
Page 20 of 36
Table 7. Finance and Growth: System GMM Estimators (Policy Set)
(1) (2) (3) (4) (5)
Regressors Collapsed Collapsed Collapsed Collapsed Collapsed
Credit 0.037*
(0.021)
Deposit 0.047**
(0.019)
Savings -0.092
(0.085)
Loans Over
0.019**
Appro
(0.007)
Corporate 0.204*
(0.119)
Initial PCGDP 0.053*** 0.046*** 0.041 0.005 -0.023
(0.020) (0.014) (0.055) (0.015) (0.044)
Education 0.024 0.114** 0.044 0.015 0.134
(0.056) (0.054) (0.057) (0.017) (0.099)
SOE -0.037 -0.060*** -0.006 -0.012 -0.066
(0.043) (0.017) (0.049) (0.023) (0.044)
CPI 0.191 0.189 -0.143 -0.785 -5.003**
(0.833) (0.406) (2.655) (0.722) (2.147)
FDI -0.017* 0.002 -0.001 -0.007 0.007
(0.009) (0.005) (0.015) (0.005) (0.012)
Government -0.028 0.004 -0.075 -0.040* -0.139***
(0.039) (0.030) (0.060) (0.022) (0.051)
Dummy2002 -0.017 0.007 0.000 -0.042** -0.094**
(0.021) (0.011) (0.071) (0.017) (0.042)
Dummy2003 -0.008 0.003 0.010 -0.017*** 0.004
(0.007) (0.007) (0.023) (0.006) (0.012)
Dummy2004 0.008 0.013 0.024 0.024 0.142***
(0.022) (0.012) (0.058) (0.018) (0.053)
Dummy2005 0.002 0.003 0.006 -0.001 0.021**
(0.005) (0.003) (0.007) (0.003) (0.009)
Constant -0.610 -1.244*** 0.058 -0.014 -0.762
(0.516) (0.437) (0.815) (0.226) (0.742)
Observations 1381 1381 1381 1339 1379
Sargan test
(p-value) 0.220 0.047 0.689 0.005 0.802
Difference-in-
Sargan test for
instruments
based on
lagged growth
(p-value) 0.853 0.045 0.780 0.011 0.893
m2 test
(p-value) 0.624 0.813 0.624 0.522 0.466
Notes: The standard errors are in parentheses. The test statistics and standard errors are asymptotically robust to
heteroskedasticity. m2 is a test for a second-order serial correlation, which is asymptotically N(0,1) under the null
of no second-order serial correlation. Cities in Tibet are excluded from the sample due to missing data. The signifi-
cance levels at the 1%, 5% and 10% are identified by ***, ** and *, respectively.
Page 21 of 36
Table 8. Finance and Growth: System GMM Estimators (Full Set)
(1) (2) (3) (4) (5)

Credit 0.035*
(0.021)
Deposit 0.037*
(0.020)
Savings -0.162
(0.138)
Loans Over
Appro 0.014**
(0.007)
Corporate 0.102*
(0.062)
Initial PCGDP 0.054** 0.060*** 0.024 0.007 -0.001
(0.026) (0.021) (0.045) (0.013) (0.026)
Education 0.029 0.142** 0.045 0.022 -0.010
(0.056) (0.062) (0.059) (0.019) (0.090)
SOE -0.029 -0.053*** -0.012 -0.044
(0.041) (0.018) (0.057) (0.045)
CPI 0.186 -0.294 -1.732 -2.638
(0.866) (0.478) (2.762) (2.018)
FDI -0.019* -0.002 0.003 -0.001 -0.006
(0.010) (0.006) (0.021) (0.006) (0.011)
Government -0.008 -0.028 -0.013 -0.031 -0.056
(0.033) (0.032) (0.077) (0.024) (0.044)
Postal&Telecom 0.016 0.018 0.047 -0.013 0.009
(0.014) (0.013) (0.039) (0.013) (0.017)
Infrastructure 0.001 -0.006 -0.003 -0.007 -0.002
(0.011) (0.010) (0.026) (0.008) (0.015)
Dummy2002 -0.023 -0.001 -0.048 -0.021** -0.075*
(0.019) (0.014) (0.064) (0.008) (0.043)
Dummy2003 -0.011 0.006 0.001 -0.015** -0.007
(0.008) (0.009) (0.017) (0.006) (0.009)
Dummy2004 0.004 0.023 0.050 0.007 0.080*
(0.024) (0.014) (0.063) (0.005) (0.048)
Dummy2005 0.000 0.003 0.011 -0.003 0.013*
(0.006) (0.003) (0.010) (0.003) (0.007)
Constant -0.726 -1.420*** 0.331 -0.036 0.040
(0.539) (0.492) (0.838) (0.204) (0.656)
Observations 1377 1377 1377 1377 1375
Sargan test
(p-value) 0.217 0.300 0.883 0.011 0.369
Difference-in-
Sargan test for
instruments based
on lagged growth
(p-value) 0.367 0.247 0.485 0.091 0.478
m2 test
(p-value) 0.718 0.673 0.576 0.811 0.553
of no second-order serial correlation. Cities of Tibet are excluded from the sample due to missing data. The signifi-
Page 22 of 36
The Loan Over Appro regressions fail to pass the tests in any specification. Although we
are unable to infer a causality relationship between this financial indicator and economic
growth, their positive correlation is real. This implies that more active use of market-oriented
and profit-driven financial transactions, such as loans relative to state budget appropriation,
is positively associated with economic growth. In addition, this conclusion is in line with those
of Liu and Li (2001), Guariglia and Poncet (2008) and Chen (2006) who employ the same
measure with provincial data.15
However, the coefficient of Savings turns to be negative from system GMM estimations,
but loses its significance for economic growth. The difference-in-Sargan tests of the un-
collapsed variants indicate that the system GMM instruments are invalid. Considering the
positive relationship from the OLS and differenced GMM estimators, we consider the negative
coefficients from system GMM estimators as a fluke of problematic system GMM
instruments.16 However, it is difficult to conclude that Savings plays a significantly positive
role to promote growth from the above evidence. China has a very high savings rate of about
40%, which may cause low household consumption as a percentage of GDP and hurt eco-
nomic growth. This possibly explains the reported unclear relationship between Savings and
growth in China.
In summary, we find that the development of financial intermediation in China after the
WTO entry positively influences economic growth. 17 This finding is consistent with most
cross-country studies on the relationship between financial intermediation and economic
growth. However, this finding is contrary to several existing studies on the finance-growth
relationship in China (Boyreau-Debray, 2003; Hasan, Wachtel and Zhou, 2007; Guariglia and
Poncet, 2008). These studies argued that financial deepening in China did not contribute to
economic growth, since banks, especially the SOCBs, continued to support loss-making SOEs
and, as a government policy to reduce poverty, capital was channeled to slow-growing regions.
This inefficient allocation of capital was blamed as the main cause of the negative effect of
15 As a robustness check, we replace Loan Over Appro with the share of fixed asset investment that is fi-
nanced by domestic loans. The OLS, differenced GMM and system GMM regressions produce significantly positive
coefficients.
16 The difference-in-Sargan tests of the un-collapsed variants, which are reported in the appendix, indicate
that the system GMM instruments are invalid. If the system GMM instruments are dropped, the regressions are
brought back to differenced GMM where Savings positively influences growth.
17 In order to investigate the issue of this causality further, we apply a panel Granger causality test (Granger,
1969) to our data. Freeman (1983) developed a direct Granger method to assess the Granger causality and de-
termine the causality direction of a relationship between two variables in a direct way: regress one variable on the
lagged values of the other variable and itself, and then use an F-test to examine the null hypothesis that all the
coefficients on the lagged values of the other variable are zero. We find no evidence that economic growth Granger-
causes financial development. These results are available upon request.
Page 23 of 36
financial development on economic growth in China. However, these studies mainly covered
the years before Chinas accession to the WTO in 2001. Since our study focuses on the years
after 2001, the opposite results are not surprising. This suggests that the financial reforms in
China after 2001 are in the right direction, especially in accelerating financial deepening, bank
restructuring and financial liberalization.
Furthermore, the positive estimated coefficients are also economically significant (large
in value). For example, based on the coefficient reported in column (1) of Table 7 from the
collapsed system GMM estimation with the policy set, a city exogenously moving from the
25th percentile in the distribution of the ratio of total loans to GDP (53.96%) to the 75th per-
centile (92.99%) will have a 2.01% larger GDP growth rate.
Most control variables show expected signs although not always statistically significant.
We find evidence of convergence because the coefficient of the lagged per capita GDP, Initial
PCGDP, is significantly smaller than unity in most regressions. We find that human capital
has a positive impact on economic growth. The variable, SOE, which is the share of state-
owned entities in total investment, always shows a negative effect on economic growth. This
indicates that the relatively poor performance of state-owned entities hurts economic growth.
Also, inflation, CPI, negatively affects growth. Government size, Government, has a negative
impact on growth. The stock of a citys communication facilities, Postal&Telecom, influences
growth positively. The infrastructure variable, Infrastructure, is always insignificant. The in-
dicator of openness, FDI, is also insignificant in almost all the regressions. The last result is in
line with Carkovi and Levine (2005) and Boyreau-Debray (2003), who found that foreign in-
vestment has no significant impact on economic growth after controlling for endogeneity.
6.5. Sensitivity Analysis: a City Dummy Variable
To ensure robustness of our findings further, we conduct a sensitivity analysis by intro-

ducing a dummy variable to indicate coastal cities, capital cities and the cities that have hosted
foreign bank entries. The reasons for introducing this city dummy are as follows: First, in
China, there is a huge economic development imbalance between coastal and hinterland re-
gions. Soon after the Chinese economic reforms started in 1978, an open door policy for for-
eign investment and trade led to the establishment of five special economic zones in coastal
areas (Shenzhen, Zhuhai, Shantou, Xiamen, and Hainan province) in 1980. Later, 14 coastal
cities (Dalian, Qinghuangdao, Tianjin, Yantai, Qingdao, Lianyungang, Nantong, Shanghai,
Ningbo, Wenzhou, Fuzhou, Guangzhou, Zhanjiang, and Beihai) were opened up for foreign
investment and designated as Economic and Technical Development Zones. The central gov-
ernment invested directly in many projects in these cities and offered many favorable policies,
including more autonomy and lower taxes. Song, Chu and Cao (2000) found that these poli-
cies exacerbated inter-city disparities in China in terms of per capita GDP and per capita in-
come. Boyreau-Debray (2003) introduced a coastal dummy to her growth-finance analysis to
Page 24 of 36
take into account the fact that the banking sector in coastal provinces is characterized by a
lower share of SOCB credit and less credit from the central bank. Second, each province in
China has a provincial capital city. Usually, the capital city enjoys the dominant status in
economic development and is the most populous city in the province. Lastly, by the end of
2006, a total of 31 cities in China have hosted foreign bank entries. A citys banking sector
becomes more diversified after foreign bank entries. Thus, introducing a dummy variable to
indicate coastal cities, capital cities and cities having foreign banks18 allows us to check for any
omitted variable biases created by the special characteristics of these cities.
We conducted the sensitivity analysis using system GMM estimators with the policy set.
The results are presented in Table 11. We can see that adding this city dummy does not change
our conclusions. That is, the strong connection between financial development and economic
growth is not associated with whether a city is a coastal city, a capital city or a city with foreign
banks, a characteristic that was not taken into account in our baseline regressions. Moreover,
the collapsed Loans Over Appro regression passes the Sargan test and this result strength-
ens the conclusion in Subsection 6.4. This indicates that the development of financial inter-
mediation has a causal and positive impact on economic growth through the substitution of
loans for state budget appropriation.
Table 9. Finance and Growth with a City Dummy: System GMM Estimators (Policy Set)
(1) (2) (3) (4) (5)
Credit 0.060**
(0.028)
Deposit 0.038
(0.028)
Savings -0.033
(0.085)
Loans Over
Appro 0.030***
(0.010)
Corporate 0.171*
(0.103)
Initial PCGDP 0.050** 0.045*** 0.049 -0.034 -0.018
(0.025) (0.016) (0.059) (0.027) (0.052)
Education 0.019 0.124* 0.042 -0.007 0.090
(0.063) (0.070) (0.044) (0.020) (0.114)
SOE 0.001 -0.066** 0.023 0.091*** -0.026
(0.035) (0.032) (0.046) (0.025) (0.041)
CPI 0.694 0.205 0.200 -1.210 -4.443*
(0.903) (0.448) (2.733) (0.769) (2.471)
FDI -0.011 0.000 0.012 0.004 0.007
(0.010) (0.007) (0.017) (0.010) (0.016)
Government -0.031 0.000 -0.069 -0.061* -0.130**
(0.039) (0.036) (0.052) (0.033) (0.053)
City Dummy
(costal, capital,
with foreign
banks) -0.007 0.006 0.018 -0.062*** -0.087*
(0.024) (0.012) (0.074) (0.020) (0.050)
Dummy2002 -0.011 0.003 0.015 -0.030*** 0.002
(0.009) (0.009) (0.023) (0.011) (0.014)
Dummy2003 -0.007 0.011 0.019 0.025 0.128**
(0.024) (0.014) (0.060) (0.020) (0.060)
Dummy2004 -0.000 0.002 0.005 -0.002 0.020**
18 We can alternatively introduce three dummy variables for the three types of cities. However, applying three
separate dummies does not change our main conclusions. In fact, the dummies are insignificant in most cases.
Page 25 of 36
(0.005) (0.003) (0.008) (0.004) (0.010)
Dummy2005 -0.101 0.046 -0.266* -0.167* -0.075
(0.078) (0.059) (0.159) (0.094) (0.089)
Constant -0.632 -1.248** -0.229 0.500 -0.445
(0.607) (0.542) (0.822) (0.321) (0.766)
Observations 1381 1381 1381 1339 1354
Sargan test
(p-value) 0.232 0.153 0.600 0.526 0.835
Difference-in-
Sargan test for
instruments
based on
lagged growth
(p-value) 0.904 0.076 0.221 0.355 0.821
m2 test
(p-value) 0.524 0.885 0.737 0.368 0.480
heteroskedasticity. The city dummy takes the value of 1 on 44 cities (coastal cities, capital cities and cities with
foreign banks). m2 is a test for a second-order serial correlation, which is asymptotically N(0,1) under the null of
no second-order serial correlation. Cities of Tibet are excluded from the sample due to missing data. The signifi-
7. Conclusions
This paper examines the relationship between financial intermediation and economic
growth in China, using data from 286 Chinese cities over the period 20012006. We investi-
gate the exogenous component of financial development on economic growth using system
GMM estimators for dynamic panel data. This technique yields the same results as the tradi-
tional cross-sectional estimators and the simpler first-differenced GMM estimators. We study
an empirical relationship between various measures of financial development and economic
growth with a unique city-level dataset. Our results suggest that traditionally used indicators
of financial development are generally positively associated with economic growth after con-
trolling for many factors associated with growth. The size and depth of the financial sector
spur economic growth. With more use of markets and profit-oriented financial transactions
and mobilization of corporate deposits, the development of financial intermediation in China
after the WTO entry positively influences economic growth in China. These results are consis-
tent with most cross-country studies on the relationship between financial intermediation and
economic growth, but run contrary to existing studies on China that suggests that financial
development hinders economic growth due to the distorting nature of the state-ruled banking
sector. Thus, our findings suggest that the banking reforms after Chinas accession to the
WTO are in the right direction.
However, household savings are found to have an unclear effect on economic growth. The
results from OLS and differenced GMM estimators suggest a positive relationship between
household savings and economic growth, while results from system GMM estimators suggest
a negative but insignificant effect, which we consider may be a by-product of problematic in-
struments.
Page 26 of 36
We pay particular attention to the issue of too many instruments when first-differenced
GMM and system GMM estimators are employed. Techniques for reducing the instrument
count are applied, including limiting the lag depth for instruments and collapsing instruments.
Furthermore, we test for the validity of the full instrument set and subsets of instruments rig-
orously. The conclusions are drawn carefully based on both complicated estimators and sim-
ple estimators. To examine the sensitivity of our results, different conditioning information
sets are experimented with. We have also carried out a sensitivity analysis by adding a city
dummy to differentiate between coastal cities, capital cities and cities with foreign bank en-
tries. These analyses show that our results are robust.
More insights into the linkage between financial development and economic growth in
China can be obtained if more detailed city-level data are available, especially data on shares
of SOCBs and joint-equity commercial banks in total credit. Also, a firm-level study of the
impact of external finance on firm growth may produce more insightful results. These are on
our agenda for future research.
Page 27 of 36
Appendix
Table A1. System GMM Estimators: Two Variants (Simple Set)
(1a) (1b) (2a) (2b) (3a) (3b) (4a) (4b) (5a) (5b)
Un- Un- Un- Un- Un-
Regressors Collapsed Collapsed Collapsed Collapsed Collapsed Collapsed Collapsed Collapsed Collapsed Collapsed
Credit 0.026* 0.040*
(0.016) (0.019)
Deposit 0.050* 0.075*
(0.027) (0.045)
Savings -0.114 -0.079**
(0.114) (0.033)
Loans Over
Appro 0.012* 0.013*
(0.007) (0.004)
Corporate 0.120** 0.072*
(0.055) (0.039)
Initial PCGDP 0.021** 0.018 0.062*** 0.059*** 0.056** 0.026 0.015 0.010* 0.041* 0.035**
(0.009) (0.014) (0.022) (0.021) (0.028) (0.016) (0.013) (0.006) (0.021) (0.015)
Education 0.037 0.088* 0.093 0.108 0.188 0.195** 0.019 0.018 0.091 0.089*
(0.023) (0.052) (0.108) (0.085) (0.142) (0.082) (0.030) (0.017) (0.071) (0.051)
Dummy2002 -0.026*** -0.029*** 0.005 0.007 0.008 -0.008 -0.022*** -0.024*** 0.008 -0.001
(0.007) (0.012) (0.013) (0.014) (0.015) (0.010) (0.006) (0.003) (0.011) (0.009)
Dummy2003 -0.018*** -0.023*** 0.006 0.005 0.010 -0.004 -0.013** -0.014*** 0.010 0.003
(0.007) (0.011) (0.011) (0.01) (0.011) (0.007) (0.006) (0.003) (0.009) (0.008)
Dummy2004 0.005 0.000 0.021** 0.020*** 0.017 0.008 0.007 0.005* 0.023*** 0.018***
(0.005) (0.006) (0.009) (0.007) (0.010) (0.007) (0.005) (0.003) (0.007) (0.005)
Dummy2005 0.005 -0.003 0.004 0.004 0.002 0.000 -0.001 -0.000 0.008* 0.005
(0.003) (0.003) (0.005) (0.004) (0.006) (0.004) (0.003) (0.002) (0.005) (0.003)
Constant -0.410** -0.762* -1.277* -1.459** -1.129 -1.037** -0.205 -0.150 -1.261** -1.019**
(0.161) (0.394) (0.755) (0.725) (1.146) (0.495) (0.206) (0.116) (0.516) (0.430)
Observations 1381 1381 1381 1381 1381 1381 1339 1339 1379 1379
Instruments 11 30 11 30 11 30 11 30 11 30
Sargan test
(p-value) 0.013 0.117 0.748 0.689 0.269 0.239 0.001 0.000 0.981 0.244
Difference-in-
Sargan test for
system GMM
instruments
(p-value) NA 0.050 NA 0.680 NA 0.094 NA 0.000 NA 0.928
Difference-in-
Sargan test for
instruments
based on
lagged growth
(p-value) 0.010 0.034 0.519 0.427 0.384 0.542 0.012 0.000 0.641 0.080
m2 test
(p-value) 0.817 0.923 0.877 0.737 0.613 0.607 0.707 0.674 0.992 0.949
heteroskedasticity. m2 is a test for second-order serial correlation, which is asymptotically N(0,1) under the null of
no second-order serial correlation. Cities in Tibet are excluded from the sample due to missing data. The signifi-
Page 28 of 36
Table A2. System GMM Estimators: Two Variants (Medium Set)
(1a) (1b) (2a) (2b) (3a) (3b) (4a) (4b) (5a) (5b)
Un- Un- Un- Un- Un-
Credit 0.033** 0.037***
(0.015) (0.012)
Deposit 0.039* 0.045*
(0.023) (0.026)
Savings -0.127 -0.066*
(0.080) (0.083)
Loans Over
0.021* 0.014*
Appro
(0.011) (0.007)
Corporate 0.195** 0.115*
(0.094) (0.062)
Initial PCGDP 0.022** 0.020* 0.064*** 0.045*** 0.032 0.029* 0.045*** 0.004 0.015 0.036
(0.011) (0.011) (0.019) (0.017) (0.033) (0.017) (0.017) (0.010) (0.030) (0.027)
Education 0.049 0.071* 0.062 0.135** 0.033 0.142** 0.041 0.027 0.176 0.092
(0.042) (0.038) (0.074) (0.058) (0.036) (0.069) (0.047) (0.028) (0.143) (0.107)
SOE -0.030 -0.029* -0.042* -0.052*** 0.010 -0.021* -0.041 -0.005 -0.060 -0.081***
(0.027) (0.015) (0.021) (0.019) (0.057) (0.012) (0.035) (0.041) (0.047) (0.029)
CPI -0.527 -0.381 0.626 -0.146 0.072 0.824 -0.833 -0.918 -1.663 -0.669
(0.857) (0.724) (0.765) (0.693) (2.081) (0.817) (0.791) (0.897) (2.150) (2.303)
Dummy2002 -0.035 0.034* 0.019 -0.003 -0.008 0.011 -0.020 -0.048** -0.025 -0.007
(0.022) (0.018) (0.022) (0.021) (0.058) (0.022) (0.020) (0.019) (0.049) (0.055)
Dummy2003 -0.018** -0.020*** 0.009 0.000 0.001 0.000 0.001 -0.017*** 0.006 0.009
(0.007) (0.006) (0.010) (0.008) (0.016) (0.009) (0.008) (0.006) (0.012) (0.011)
Dummy2004 0.020 0.012 0.008 0.018 0.013 -0.009 0.038** 0.026 0.061 0.039
(0.021) (0.019) (0.016) (0.017) (0.045) (0.020) (0.018) (0.023) (0.048) (0.052)
Dummy2005 0.003 0.000 0.004 0.002 0.007 -0.001 0.009** 0.001 0.014* 0.011
(0.004) (0.004) (0.004) (0.004) (0.008) (0.004) (0.004) (0.005) (0.007) (0.007)
Constant -0.510 -0.652** -1.036* -1.347*** 0.154 -0.777 -0.649* -0.143 -1.788* -1.170*
(0.325) (0.301) (0.544) (0.459) (0.576) (0.473) (0.330) (0.207) (0.985) (0.700)
Observations 1381 1381 1381 1381 1381 1381 1339 1339 1379 1379
Instruments 15 46 15 46 15 46 15 46 15 46
Sargan test
(p-value) 0.210 0.198 0.408 0.113 0.330 0.111 0.015 0.000 0.599 0.408
Difference-in-
Sargan test for
system GMM
instruments
Difference-in-
Sargan test for
instruments
based on
lagged growth
(p-value) 0.451 0.478 0.537 0.148 0.829 0.877 0.590 0.000 0.641 0.977
m2 test 0.618 0.484

(p-value) 0.942 0.958 0.761 0.815 0.662 0.474 0.592 0.918
Notes: The standard errors are in parentheses. Test statistics and standard errors are asymptotically robust to
Page 29 of 36
Table A3. System GMM Estimators: Two Variants (Policy Set)
(1a) (1b) (2a) (2b) (3a) (3b) (4a) (4b) (5a) (5b)
Un- Un- Un- Un- Un-
Credit 0.037* 0.032*
(0.021) (0.018)
Deposit 0.047** 0.049**
(0.019) (0.024)
Savings -0.092 -0.099**
(0.085) (0.039)
Loans Over
0.019** 0.018**
Appro
(0.007) (0.009)
Corporate 0.204* 0.084*
(0.119) (0.046)
Initial PCGDP 0.053*** 0.025 0.046*** 0.064** 0.041 0.026* 0.005 -0.006 -0.023 0.039
(0.020) (0.020) (0.014) (0.030) (0.055) (0.025) (0.015) (0.020) (0.044) (0.021)
Education 0.024 0.139* 0.114** 0.033 0.044 0.158 0.015 0.016 0.134 -0.027
(0.056) (0.076) (0.054) (0.063) (0.057) (0.123) (0.017) (0.020) (0.099) (0.074)
SOE -0.037 -0.046 -0.060*** -0.040 -0.006 -0.009 -0.012 0.007 -0.066 -0.049*
(0.043) (0.035) (0.017) (0.033) (0.049) (0.038) (0.023) (0.018) (0.044) (0.026)
CPI 0.191 -0.675 0.189 -0.223 -0.143 -0.176 -0.785 -1.207 -5.003** 0.174
(0.833) (0.839) (0.406) (0.770) (2.655) (1.401) (0.722) (0.911) (2.147) (0.784)
FDI -0.017* 0.000 0.002 -0.006 -0.001 -0.003 -0.007 -0.004 0.007 -0.007
(0.009) (0.006) (0.005) (0.007) (0.015) (0.008) (0.005) (0.007) (0.012) (0.006)
Government -0.028 -0.021 0.004 -0.055* -0.075 -0.043 -0.040* -0.047** -0.139*** -0.078*
(0.039) (0.037) (0.030) (0.033) (0.060) (0.055) (0.022) (0.022) (0.051) (0.042)
Dummy2002 -0.017 -0.028 0.007 -0.011 0.000 -0.007 -0.042** -0.055** -0.094** 0.008
(0.021) (0.025) (0.011) (0.029) (0.071) (0.035) (0.017) (0.022) (0.042) (0.020)
Dummy2003 -0.008 -0.012 0.003 0.011 0.010 0.002 -0.017*** -0.019*** 0.004 0.014*
(0.007) (0.037) (0.007) (0.014) (0.023) (0.011) (0.006) (0.007) (0.012) (0.008)
Dummy2004 0.008 0.023 0.013 0.020 0.024 0.016 0.024 0.032 0.142*** 0.024
(0.022) (0.022) (0.012) (0.017) (0.058) (0.038) (0.018) (0.023) (0.053) (0.021)
Dummy2005 0.002 0.002 0.003 0.003 0.006 -0.002 -0.001 0.000 0.021** 0.007
(0.005) (0.004) (0.003) (0.004) (0.007) (0.008) (0.003) (0.004) (0.009) (0.005)
Constant -0.610 -1.057* -1.244*** -0.737 0.058 -0.587 -0.014 0.117 -0.762 -0.117*
(0.516) (0.602) (0.437) (0.562) (0.815) (1.039) (0.226) (0.300) (0.742) (0.631)
Observations 1381 1381 1381 1381 1381 1381 1339 1339 1379 1379
Instruments 19 62 19 62 19 62 19 62 19 62
Sargan test
(p-value) 0.220 0.179 0.047 0.758 0.689 0.116 0.005 0.000 0.802 0.158
Difference-in-
Sargan test for
system GMM
instruments
Difference-in-
Sargan test for
instruments
based on
lagged growth
(p-value) 0.853 0.317 0.045 0.665 0.780 0.953 0.011 0.134 0.893 0.127
m2 test
(p-value) 0.624 0.965 0.813 0.849 0.624 0.574 0.522 0.392 0.466 0.490
Page 30 of 36
Table A4. System GMM Estimators: Two Variants (Full Set)
(1a) (1b) (2a) (2b) (3a) (3b) (4a) (4b) (5a) (5b)
Un Un- Un- Un- Un-
Credit 0.035* 0.032*
(0.021) (0.016)
Deposit 0.037* 0.008
(0.020) (0.036)
Savings -0.162 -0.085*
(0.138) (0.048)
Loans Over
Appro 0.014** 0.019**
(0.007) (0.009)
Corporate 0.102* 0.086**
(0.062) (0.040)
Initial PCGDP 0.054** 0.005 0.060*** 0.043** 0.024 0.046* 0.007 -0.005 -0.001 0.022
(0.026) (0.012) (0.021) (0.021) (0.045) (0.027) (0.013) (0.017) (0.026) (0.014)
Education 0.029 0.031 0.142** 0.037 0.045 0.060 0.022 0.020 -0.010 -0.052
(0.056) (0.050) (0.062) (0.059) (0.059) (0.065) (0.019) (0.020) (0.090) (0.069)
SOE -0.029 -0.014 -0.053*** -0.019 -0.012 -0.043 0.001 -0.044 -0.066***
(0.041) (0.020) (0.018) (0.043) (0.057) (0.039) (0.019) (0.045) (0.024)
CPI 0.186 -0.920 -0.294 0.736 -1.732 0.468 -1.015 -2.638 -1.582
(0.866) (0.839) (0.478) (0.905) (2.762) (0.932) (0.791) (2.018) (0.968)
FDI -0.019* 0.000 -0.002 -0.008 0.003 -0.005 -0.001 0.002 -0.006 -0.003***
(0.010) (0.006) (0.006) (0.009) (0.021) (0.011) (0.006) (0.006) (0.011) (0.008)
Government -0.008 -0.020 -0.028 -0.018 -0.013 -0.010 -0.031 -0.045* -0.056 -0.071*
(0.033) (0.026) (0.032) (0.035) (0.077) (0.045) (0.024) (0.026) (0.044) (0.037)
Postal&Telecom 0.016 0.009 0.018 0.028* 0.047 0.042* -0.013 -0.013 0.009 0.024
(0.014) (0.012) (0.013) (0.016) (0.039) (0.023) (0.013) (0.014) (0.017) (0.017)
Infrastructure 0.001 0.005 -0.006 0.010 -0.003 -0.003 -0.007 -0.006 -0.002 -0.006
(0.011) (0.007) (0.010) (0.012) (0.026) (0.017) (0.008) (0.008) (0.015) (0.009)
Dummy2002 -0.023 -0.040 -0.001 -0.004 -0.048 -0.007 -0.021** -0.046** -0.075* -0.046
(0.019) (0.014) (0.014) (0.013) (0.064) (0.019) (0.008) (0.020) (0.043) (0.029)
Dummy2003 -0.011 -0.014* 0.006 -0.001 0.001 0.000 -0.015** -0.016** -0.007 0.003
(0.008) (0.007) (0.009) (0.005) (0.017) (0.007) (0.006) (0.007) (0.009) (0.009)
Dummy2004 0.004 0.022 0.023 0.005 0.050 0.004 0.007 0.030 0.080* 0.064**
(0.024) (0.014) (0.014) (0.014) (0.063) (0.019) (0.005) (0.020) (0.048) (0.031)
Dummy2005 0.000 0.002 0.003 0.002 0.011 0.001 -0.003 0.002 0.013* 0.012**
(0.006) (0.003) (0.003) (0.004) (0.010) (0.004) (0.003) (0.004) (0.007) (0.006)
Constant -0.726 -0.203* -1.420*** -0.551 0.331 -0.323 -0.036 0.122 0.040 0.205
(0.539) (0.378) (0.492) (0.516) (0.838) (0.573) (0.204) (0.263) (0.656) (0.475)
Observations 1377 1377 1377 1377 1377 1377 1377 1377 1375 1375
Instruments 23 78 23 78 23 78 23 78 23 78
Sargan test
(p-value) 0.217 0.260 0.300 0.310 0.883 0.215 0.011 0.001 0.369 0.272
Difference-in-
Sargan test for
system GMM
instruments
Difference-in-
Sargan test for
instruments based
on lagged growth
(p-value) 0.367 0.870 0.247 0.122 0.485 0.798 0.091 0.098 0.478 0.572
m2 test
(p-value) 0.718 0.744 0.673 0.692 0.576 0.906 0.811 0.495 0.553 0.787
of no second-order serial correlation. Cities of Tibet are excluded from the sample due to missing data. The signifi-
Page 31 of 36
References
Allen, F., Qian, J., Qian, M. (2005). Law, Finance and Economic Growth in China.
Journal of Financial Economics, 77(1), 57116.
Alonso-Borrego, C., Arellano, M. (1996). Symmetrically Normalized Instrumental-

variable Estimation Using Panel Data. Journal of Business and Economic Statistics, 17: 36-
49.
Arellano, M. (2002). Modelling Optimal Instrumental Variables for Dynamic Panel Data
Models. Econometrics Invited Lecture, European Meeting of the Econometric Society,
CEMFI Working Paper no. 0310.
Arellano, M., Bond, S. (1991). Some Tests of Specification for Panel Data: Monte Carlo
Evidence and an Application to Employment Equations. Review of Economic Studies, 58(2),
277297.
Arellano, M., Bover, O. (1995). Another Look at the Instrumental Variable Estimation of
Error-components Models. Journal of Econometrics, 68(1), 2951.
Arestis, P., Demetriades, P.O., Luintel, K.B. (2001). Financial Development and Eco-
nomic Growth: the Role of Stock Markets. Journal of Money, Credit and Banking, 33(1), 16
41.
Ayyagari, M., Demirg-Kunt, A., Maksimovic, V. (2008). Formal versus Informal Fi-
nance : Evidence from China. The World Bank, Policy Research Working Paper Series 4465.
Beck, T., Levine, R., Loayza, N. (2000). Finance and the Source of Growth. Journal of
Financial Economics, 58(1-2), 261300.
Beck, T., Levine, R., Stock Markets, Banks, and Growth: Panel Evidence, Journal of
Banking & Finance, vol. 28(3), 423-442, 2004.
Beck, T., Demirg-Kunt, A., Maksimovic, V. (2005). Financial and Legal Constraints to
Firm Growth: Does Size Matter? Journal of Finance, 60(1), 137177.
Beck, T., Demirg-Kunt, A., Laeven, L., Levine, R. (2008). Finance, Firm Size, and
Growth. Journal of Money, Credit and Banking, 40(7), 13791405.
Beck, T., Levine, R., Levkov, A. (2010). Big Bad Banks: The Winners and Losers From
Bank Deregulation in the United States. Journal of Finance, forthcoming.
Bekaert, G., Harvey, C.R., Lundblad, C. (2005). Does Financial Liberalization Spur
Growth? Journal of Financial Economics, 77(1), 355.
Benhabib, J., Spiegel, M.M. (2000). The Role of Financial Development in Growth and
Investment. Journal of Economic Growth, 5(4), 341360.
Page 32 of 36
Berger, A.N., Hasan, I., Zhou, M. (2009). Bank Ownership and Efficiency in China: What
will Happen in the Worlds Largest Nation? Journal of Banking and Finance, 33(1), 113130.
Blundell, R., Bond, S. (1998). Initial Conditions and Moment Restrictions in Dynamic
Panel Data Models. Journal of Econometrics, 87(1), 115143.
Bond, S., Bowsher, C., Windmeijer, F. (2001). Criterion-based Inference for GMM in Au-
toregressive Panel Data Models. Economic Letters, 73(3), 379388.
Boyreau-Debray, G. (2003). Financial Intermediation and Growth: Chinese Style, The

World Bank, Policy Research Working Paper Series 3027.
Boyreau-Debray, G., Wei, S. (2004). Can China Grow Faster? A Diagnosis on the Frag-
mentation of the Domestic Capital Market. International Monetary Fund, IMF Working
Papers 04/76.
Carkovic, M., Levine, R. (2005). Does Foreign Direct Investment Accelerate Economic
Growth? Does Foreign Direct Investment Promote Development, Washington DC, U.S., In-
stitute for International Economics, 195220.
Cetorelli, N., Gambera, M. (2001). Banking Market Structure, Financial Dependence and
Growth: International Evidence from Industry Data, Journal of Finance, 56(2), 617648.
Chen, H. (2006). Development of Financial Intermediation and Economic Growth: the

Chinese Experience. China Economic Review, 17(4), 347362.
Cheng, X., Degryse, H. (2007). The Impact of Banks and Non-Bank Financial Institu-
tions on Local Economic Growth in China. Bank of Finland Institute for Economies in Tran-
sition, BOFIT Discussion Papers 22/2007.
Christopoulos, D.K., Tsionas, E.G. (2004). Financial Development and Economic

Growth: Evidence from Panel Unit Root and Cointegration Tests. Journal of Development
Economics, 73(1), 5574.
Claessens, S., Laeven, L. (2003). Financial Development, Property Rights, and Growth.
Journal of Finance, 58(6), 24012436.
Cull, R., Xu, L.C. (2000). Bureaucrats, State Banks, and the Efficiency of Credit Alloca-
tion: The Experience of Chinese State-Owned Enterprises. Journal of Comparative Econom-
ics, 28(1), 131.
Cull, R., Xu, L.C. (2003). Who Gets Credit? The Behavior of Bureaucrats and State Banks
in Allocating Credit to Chinese State-owned Enterprises. Journal of Development Economics,
71(2), 533559.
Dayal-Gularti, A., Huissain, A. (2002). Centripetal Forces in Chinas Economic Takeoff.

International Monetary Fund, IMF Staff Papers, 42, 364-394.
Page 33 of 36
Dehejia, R., Lleras-Muney, A. (2003). Why Does Financial Development Matter? The
United States from 1900 to 1940. National Bureau of Economic Research, Working Paper
No.9551.
Demetriades, P.O., Hussein, K.A. (1996). Does Financial Development Cause Economic
Growth? Time-series Evidence from 16 Countries. Journal of Development Economics, 51(2),
387411.
Demir, F., Dahi, O. (2009), Asymmetric Effects of Financial Development on South-

South and South-North Trade: Panel Data Evidence from Emerging Markets. Munich Per-
sonal RePEc Archive (MPRA), Working Paper No. 19177.
Giedeman, D., Compton, R. (2009) ''A note on finance, inflation, and economic growth'',
Economics Bulletin, 29(2), 749-759.
Feldstein, M., Horioka, C. (1980). Domestic Savings and International Capital Flows.
The Economic Journal, 90 (358), 314329.
Ferri, G. (2009). Are New Tigers Supplanting Old Mammoths in Chinas Banking System?
Evidence from a Sample of City Commercial Banks. Journal of Banking & Finance, 33(1),
131140.
Freeman, J.R. (1983). Granger Causality and the Times Series Analysis of Political Rela-
tionships. American Journal of Political Science, 27(2), 327358.
Garca-Herrero, A., Gavil, S., Santabrbara, D. (2006). China's Banking Reform: An As-
sessment of its Evolution and Possible Impact, CESifo Economic Studies, 52(2), 304363.
Goldsmith, R.W. (1969). Financial Structure and Development, New haven, U.S., Yale
University Press.
Granger, C.W.J. (1969). Investigating Causal Relations by Econometric Models and

Cross-spectral Methods. Econometrica, 37(3), 424438.
Guariglia, A., Liu, X., Song, L. (2008). Internal Finance and Growth: Microeconometric
Evidence on Chinese Firms. Institute for the Study of Labor, Discussion Papers 3808.
Guariglia, A., Poncet, S. (2008). Could Financial Distortions be No Impediment to Eco-

nomic Growth After All? Evidence from China. Journal of Comparative Economics, 36(4),
633657.
Guiso, L., Sapienza, P., Zingales, L. (2004). Does Local Financial Development Matter?
Quarterly Journal of Economics, 119(3), 929969.
Hasan, I., Wachtel, P., Zhou, M. (2009). Institutional development, financial deepening
and economic growth: Evidence from China, Journal of Banking & Finance, 33(1), 157170.
Page 34 of 36
Hauk, W., Wacziarg, R. (2009). A Monte Carlo study of growth regressions, Journal of
Economic Growth, 14(2), 103147.
Holtz-Eakin, D., Newey, W., Rosen, H.S. (1989). The Revenues-expenditure Nexus: Evi-
dence from Local Government Data. International Economic Review, 30(2), 415429.
Jayaratne, J., Strahan, P.E. (1996). The Finance-Growth Nexus: Evidence from Bank
Branch Deregulation. Quarterly Journal of Economics, 111(3), 639670.
Jung, W.S. (1986). Financial Development and Economic Growth: International Evi-
dence, Economic Development and Cultural Change, 34(2), 333346.
King, R.G., Levine, R. (1993). Finance and Growth: Schumpeter Might Be Right, Quar-
terly Journal of Economics, 108(3), 717738.
Kumar, K. B., Rajan, R.G., Zingales, L. (1999). What Determines Firm Size? National
Bureau of Economic Research, Working Paper No.7208.
La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2002). Government Ownership of Com-

mercial Banks. Journal of Finance, 57(1), 265301.
Lardy, N. (1998). Chinas Unfinished Economic Revolution. Brookings Institution Press.
Levine, R. (1998). The Legal Environment, Banks, and Long-Run Economic Growth.
Journal of Money, Credit and Banking, 30 (3), 596613.
Levine, R., Zervos, S. (1998). Stock Markets, Banks, and Economic Growth. American
Economic Review, 88(3), 537558.
Levine, R. (1999). Law, Finance, and Economic Growth. Journal of Financial Interme-
diation, 8(1-2), 835.
Levine, R. (2005). Finance and Growth: Theory and Evidence. Handbook of Economic
Growth, Amsterdam, Netherlands, Elsevier, Chapter 12, 1(1), 865934.
Levine, R., Loayza, N., Beck, T. (2000). Financial Intermediation and Growth: Causality
and Causes. Journal of Monetary Economics, 46(1), 3177.
Liu, L. (2002). Sequencing China's Banking Sector Reform after the WTO: Options and
Strategies. Asian Development Bank Institute, Working Paper.
Liu, T., Li, K.-W. (2001). Impact of Financial Resources Liberalization in Chinas Eco-
nomic Growth: Provincial Evidence. Journal of Asian Economics, 12, 245-262.
Lucas Jr., R.E. (1988). On the Mechanics of Economic Development. Journal of Mone-
tary Economics, 22, 342.
Miller, M.H. (1998). Financial Markets and Economic Growth. Journal of Applied Cor-
porate Finance, 11(3), 815.
Page 35 of 36
Neusser, K., Kugler, M. (1998). Manufacturing Growth and Financial Development: Evi-
dence from OECD Countries. Review of Economics and Statistics, 80(4), 638646.
Park, A., Sehrt, K. (2001). Tests of Financial Intermediation and Banking Reform in
China. Journal of Comparative Economics, 29(4), 608644.
Peng, A. (2006). China: Export Opportunities for Foreign Commercial Banks. The U.S.
Commercial Service.
Rajan, R. G., Zingales, L. (1998). Financial Dependence and Growth. American Eco-
nomic Review, 88(3), 559586.
Rioja, F., Valev, N. (2004a). Finance and the Sources of Growth at Various Stages of
Economic Development. Economic Inquiry, 42(1), 127140.
Rioja, F., Valev, N. (2004b). Does One Size Fit All?: A Reexamination of the Finance and
Growth Relationship. Journal of Development Economics, 74(2), 429447.
Roodman, D. (2009). A note on the Theme of Too Many Instruments. Oxford Bulletin
of Economics and Statistics, 71(1), 135-158.
Rousseau, P.L. (1999). Finance, investment, and growth in Meiji-era Japan. Japan and
the World Economy, 11(2), 185198.
Rousseau, P.L., Wachtel, P. (2000). Equity Markets and Growth: Cross-country Evi-
dence on Timing and Outcomes, 1980-1995. Journal of Banking & Finance, 24(12), 1933
1957.
Rousseau, P.L., Wachtel, P. (2002). Inflation Thresholds and Finance-Growth Nexus.

Journal of International Money and Finance, 21(6), 777793.
Rousseau, P.L., Sylla, R. (2005). Emerging Financial Markets and Early US Growth.
Explorations in Economic History, 42(1), 126.
Song, S., Chu, G.S.F., Cao, R. (2000). Intercity Regional Disparity in China. China Eco-
nomic Review, 11(3), 246261.
Wang, S., Zhang, J. (2009). Location Strategies under Risk and Asymmetric Information,
with an Application to Multinational Banking. Working Paper.
Windmeijer, F. (2005). A Finite Sample Correction for the Variance of Linear Efficient
Two-step GMM Estimators. Journal of Econometrics, 126, 25-51.
Wurgler, J. (2000). Financial Markets and the Allocation of Capital. Journal of Finan-
cial Economics, 58(1-2), 187214.
Xu, Z. (2000). Financial Development, Investment, and Growth. Economic Inquiry,

38(2), 331344.
Page 36 of 36

Zhang - Wang and Wang China 2012

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Zhang - Wang and Wang China 2012

Uploaded by

Copyright:

Available Formats

Financial Development and Economic Growth:

Evidence from China*

Jin Zhang,a Lanfang Wangb and Susheng Wangc

Keywords: Financial development, Economic growth, Emerging market

Classifications: N2, O1, O43

To answer the question of whether financial development is a leading indicator or a fun-

To improve the understanding of the relationship between financial development and

3. Chinas Financial System: A Historical Review

3.1. Before 1994

3.2. From 1994 to Chinas WTO Entry in 2001

Table 2. Joint-Equity Commercial Banks

Source: PBOCs and the respective banks websites.

3.3. After the WTO Entry

Interest rate liberalization is an important element of Chinas effort in financial liberaliza-

Source: World Trade Organization.

4. Variables and Data

To investigate whether the exogenous component of financial development positively in-

Traditional indicators of financial development have been employed by existing cross-

Conditioning information sets

Table 4. Descriptive Statistics and Correlations

5. Finance and Growth: Cross-Sectional Analyses

Growthi = a + b Financei + g [Conditioning Information Set ]i + ei , (1)

Table 5. Finance and Growth: OLS Estimators (Initial Value Regressions)

6. Finance and Growth: Dynamic Panel Analyses

6.1. Model Specification

Growthi ,t = a + b Financei ,t + g [Conditioning Information Set ]i ,t + hi + lt + ei ,t , (2)

The generalized-method-of-moments (GMM) estimators for dynamic panel data (Arella-

We can rewrite Equation (3) as follows:

yi ,t - yi ,t = a( yi ,t-1 - yi ,t-2 ) + b ( X i ,t - X i ,t-1 ) + (lt - lt-1 ) + (ei ,t - ei ,t-1 ). (5)

tions are used by the first-differenced GMM estimators:

E X i ,t-s (ei ,t - ei ,t-1 ) = 0, for s 2; t = 3,..., T . (7)

E ( yi ,t-s - yi ,t-s-1 ) ( hi + ei ,t ) = 0, for s = 1; t = 3,..., T , (8)

6.3. Techniques for Reducing the Instrument Count

6.4. Regression Results

Results Using First-differenced GMM Estimators

Although system-GMM estimators have advantages over first-differenced GMM

Table 6. Finance and Growth: First-differenced GMM Estimators

Results Using System GMM Estimators

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

6.5. Sensitivity Analysis: a City Dummy Variable

To ensure robustness of our findings further, we conduct a sensitivity analysis by intro-

m2 test 0.618 0.484

Alonso-Borrego, C., Arellano, M. (1996). Symmetrically Normalized Instrumental-

Boyreau-Debray, G. (2003). Financial Intermediation and Growth: Chinese Style, The

Chen, H. (2006). Development of Financial Intermediation and Economic Growth: the

Christopoulos, D.K., Tsionas, E.G. (2004). Financial Development and Economic

Dayal-Gularti, A., Huissain, A. (2002). Centripetal Forces in Chinas Economic Takeoff.

Demir, F., Dahi, O. (2009), Asymmetric Effects of Financial Development on South-

Granger, C.W.J. (1969). Investigating Causal Relations by Econometric Models and

Guariglia, A., Poncet, S. (2008). Could Financial Distortions be No Impediment to Eco-

La Porta, R., Lopez-de-Silanes, F., Shleifer, A. (2002). Government Ownership of Com-

Lardy, N. (1998). Chinas Unfinished Economic Revolution. Brookings Institution Press.

Rousseau, P.L., Wachtel, P. (2002). Inflation Thresholds and Finance-Growth Nexus.

Xu, Z. (2000). Financial Development, Investment, and Growth. Economic Inquiry,

You might also like