WtPSo 3o
POLICY RESEARCH WORKING PAPER             3027
Financial Intermediation and Growth
Chinese Style
Genevieve Boyreau-Debray
The World Bank
Development Research Group
Investment Climate
April 2003



POLICY RESEARCH WORKING PAPER 3027
Abstract
Boyreau-Debray analyzes the relationship between            indicators of local banking development into the
growth and financial intermediation at the subnational       traditional growth regression framework using the
level within China. Does the quality of the banking          GMM-system estimator. The results suggest that credit
sector in a province affect its rate of growth? Do state     extended by the banking sector at the state level has a
and nonstate banking sectors perform differently? Does       negative impact on provincial economic growth. This
the structure of the local banking sector affect the rate of  negative effect appears to be attributable to the
provincial economic growth? To answer these questions,       burden of supporting the state-owned corporate sector
the author first uses evidence on the fragmentation of       rather than to the poor performance of state-owned
regional capital markets to justify the existence of local   banks. Moreover, provinces with more diversified
credit channels. Second, using a dataset of 26 provinces     banking sectors appear to grow faster.
between 1990 and 1999, she defines and introduces
This paper-a product of Investment Climate, Development Research Group-is part of a larger effort in the group to
understand China's economic and financial development. Copies of the paper are available free from the World Bank, 1818
H Street NW, Washington, DC 20433. Please contact Paulina Sintim-Aboagye, room MC3-300, telephone 202-473-8526,
fax 202-522-1155, email address psintimabaogye@worldbank.org. Policy Research Working Papers are also posted on the
Web at http://econ.worldbank.org. The author may be contacted at gboyreaudebray@worldbank.org. April 2003. (56
pages)
The Policy Research Worknzg Paper Series dssePuc ates the fbtdgs of ivork in progress to encoarage the exchange of ideas about
developnmeitt isstues. An objective of the serres is to get the findings ouit quickly, even if the presentations are less than fully polished The
papers carry the iiames of the aiitbors ayid 5bolwd be cited accordnzigly. The findings, interpretations, aiid conclzfsions expressed in this
paper are entirely those of the authors They do not necessarily represent the viewv of the World Batik, Its Executive Directors, or the
countries they represent
ProdLuced by the Research Advisory Staff



Financial Intermediation and Growth: Chinese Style
Genevieve Boyreau-Debray
World Bank, Development Research Group
1818 H Street, N.W. Washington, D.C.
Washington, DC 20433
gboyreaudebravg worldbank.org.
The author thanks Robert Cull, David Dollar, Patrick Honohan and Shang-Jin Wei for
useful comments and suggestions as well as the participants of China Center of Economic
Research seminar series (Beijing University).






Table of contents
1.  Introduction ..............................................1
2.   Financial Intermediation in China: International Comparison ..................................2
3.  Capital Market Fragmentation in China .   ..............................................5
4.  Financial Intermediation and Growth: Subnational Evidence ....................................7
4.1    Empirical Framework ..............................................7
4.1.1    Econometric Method: GMM System Estimator .........................................8
4.1.2    The Data ...............................................9
Control Variables (Conditioning Information Set)             ..9
Local Indicators of Financial Intermediation              ..9
4.2    Descriptive Statistics and Correlations .11
4.3 Econometric Estimation .14
5. Conclusion .20
References .21
Appendix: definition of the variables and statistical sources .25



List of Figu,res
Figure 1a: Liquidity Rate - International Sample (1999) ................................................. 26
Figure Ib: Liquidity Rate - International Sample (1979) ................................................. 27
Figure 2a: Commercial-Central Bank - International Sample (1999) .............................. 28
Figure 2b: Commercial-Central Bank - International Sample (1985) .............................. 29
Figure 3a: Private Credit - International Sample (1999) .................................................. 30
Figure 3b: Private Credit - International Sample (1985) .................................................. 31
Figure 4: Provincial Investment and Saving Rates ........................................................... 32
Figure 5.la: Size and Real Growth of per Capita GDP (1990-1994, average values) ..... 33
Figure 5.2a: Size and Real Growth of per Capita GDP (1995-1999, average values) ..... 34
Figure 5. lb: Size and Initial Level of Capita GDP -(1995-1999, average values) ........... 35
Figure 5.2b: Size and Initial Level of Capita GDP (1990-1994, average values) ............ 36
Figure 6a: Sob and Real Growth of per Capita GDP ........................................................ 37
Figure 6b: Sob and Initial Level of Capita GDP ............................................................  38
Figure 7a: Central and Real Growth of per Capita GDP .................................................. 39
Figure 7b: Central and Initial Level of Capita GDP ......................................................... 40
Figure 8a: Concentration and Real Growth of per Capita GDP ....................................... 41
Figure 8b: Concentration and Initial Level of Capita GDP .............................................. 42
Figure 9.1: Size and Non-State Production .............................................................   43
Figure 9.2: SIZE and Non-State Production ............................................................   43
Figure 10: Sob and Non-State Production (1990-1999, average values) .......................... 44
Figure 11: CENTRAL and Non-State Production (1990-1999, average values) ................. 44
Figure 12: Concentration and Non-State Production (1990-1999, average values) ......... 45
List of Tlables
Table I a: Feldstein-Horioka test for Capital Mobility ...................................................... 46
Table lb : Feldstein-Horioka test for Capital Mobility .................................................... 46
Table 2a: Descriptive Statistics - Average Values (1990-99) ........................................... 47
Table 2b: Descriptive Statistics: Correlations .............................................................   48
Table 3a: Financial Intermediation and Economic Growth: Augmented Solow Model .. 49
Table 3b: Financial Intermediation and Economic Growth: Controlling for the State
Sector .............................................................                 50
Table 3c: Financial Intermediation and Economic Growth: Controlling for Foreign Direct
Investment .........................................................                 51
Table 4a: Financial Intermediation and Economic Growth: Alternative Estimators (Diff-
GMM    and Within) .........................................................         52
Table 4b: Financial Intermediation and Economic Growth: Alternative Estimators (OLS-
Level) .............................................................                 53
Table 5a: Financial Internediation and Economic Growth: Without Investment ............ 54
Table 5b: Financial Intermediation and Economic Growth: With Coastal Dummy ........ 55
Table 6: Financial Intermediation and Economic Growth: Two 5-years Periods ............ 56



1.    Introduction
China has maintained a high rate of economic growth since the early 1980s-
averaging 9.5 percent per year-while also implementing reforms aimed at transitioning
from a planned economy to a market economy. During this time, the non-state sector has
steadily expanded. In 2000, non-state enterprises accounted for more than 80 percent of
production. The financial deepening of the Chinese economy has also been impressive.
The total liquidity of the banking sector has increased from 30 percent of GDP in 1979 to
148 percent of GDP in 1999, one of the highest ratios in the world. After two decades of
reform, however, the transition to a modem and profit-oriented banking sector is far from
being achieved. Four big state banks dominate the market and allocate most of their
financial resources to the inefficient and loss-making state-owned enterprise sector. In
2000, three-quarters of all bank lending was channeled toward state-owned enterprises,
despite their plummeting contribution to national output.
The coexistence of fast economic development and financial deepening with a
massive misallocation of financial resources in China is puzzling, as it does not follow
the expected growth and finance relationship. According to this literature, both
theoretical and empirical evidence suggests a positive relationship between financial and
economic development, and that the development of financial markets and institutions is
a critical and inextricable part of the growth process (Levine 1997). A crucial question is,
however, what is the direction of causality between the two processes, e.g., whether
financial development passively follows economic development or whether it is an
important determinant of economic growth. Given the endogenous nature of the
relationship, cross-countries studies have usually relied on instrumentation techniques to
extract the exogenous component of financial development. The availability of new
econometric techniques for analyzing panel data that control for endogeneity, however, is
providing impetus to the empirical literature on finance and growth.'
This paper analyzes the relationship between growth and financial intermediation
in China by applying the traditional cross-country empirical framework to a panel of
Chinese provinces. A number of studies have analyzed the growth patterns of the Chinese
provinces, with a focus on the role of openness, foreign direct investment, or
infrastructure2. The impact of financial intermediation on growth has received little
attention, which is surprising given the importance of the issue in terms of policy
implications and the challenges posed by the increase of competition in the financial
' See Levine, Loayza, and Beck. (1999).
2 See among others, Chen and Fleisher (1996), Jian, Sachs, and Warner (1996), Raiser (1998),
Demurger and others (2002).
1



sector that will follow China's accession to the World Trade Organization (WTO).3 One
could argue that studying the relationship between finance and growth at the intranational
level is meaningless because intranational capital markets are assumed to be perfectly
integrated. Available evidence, however, suggests the existence of a credit channel at the
local level.4 For instance, Hansen, McPherson, and Waller (2000) show that even in a
financially developed economy such as the United States local banks affect the
performance of the local economy. In the case of China, the assumption of perfect capital
mobility is even more questionable, given the strong presumption of local market
fragmentation regarding goods, labor, and capital.
This paper proceeds as follows: Section 1 characterizes China's banking
development by comparing the value of financial intermediation indicators in this country
with an international sample of countries. Section 2 documents the evidence of capital
market fragmentation in China, relying on previous evidence and on a test for capital
mobility between the Chinese provinces. Section 4 presents the methodology used to
analyze the impact of the local banking sector development on local economic
performance and estimates the impact of the local banking sector development on local
economic performance in China. Section 5 concludes.
2.     FinanciaR lEnternmiediation in China: Mltnernnanfional Comparison
According to Levine (1997), the financial sector contributes to economic growth
through five main channels: (1) facilitating the trading, hedging, diversifying, and
pooling of risk; (2) allocating resources; (3) monitoring managers and exerting corporate
control; (4) mobilizing savings; and (5) facilitating the exchange of goods and services.
Ideally, we want to select indicators that reflect the quality of financial services when
analyzing the impact of financial intermediation on economic growth. However, only
quantitative indicators are available in a wide enough range to make cross-country
comparisons. The three indicators of banking sector development are traditionally
defined as follows.5 The first indicator, usually used as a measure of financial depth, is
the ratio of liquid liabilities to GDP, including currency and demand and interest-bearing
deposits of banks and non-bank financial intermediaries. The underlying assumption is
that the quality of financial services is positively correlated with the size of the financial
intermediary sector. The second indicator of financial intermediation is the ratio of
commercial bank assets to commercial bank and central banks assets. It measures the
degree to which commercial banks versus the central bank allocate society's savings, and
3 See Li and Liu (2001), who study the impact of investment financing sources on industrial growth;
Dayal-Gulati and Husain (2000), who introduce in their growth regression financial variables among other
variables of interest; and Park and Sehrt (2001), who develop empirical tests of financial intermediation.
4 See Hansen McPherson and Waller (2000), Samolky (1994) and Neely and Wheelock (1997).
5 See Beck, Demirgil9-Kunt, and Levine (2000).
2



assumes implicitly that the banks perform better than the central bank in allocating
financial resources. The third- and last indicator of financial intermediation is the ratio of
credit extended by financial intermediaries to the private sector to GDP, as opposed to
governments, government agencies, and public enterprises. The assumption is that
financial systems that allocate more credit to private firms are more engaged in
researching firms, exerting control, providing risk management services, mobilizing
savings, and facilitating transactions than financial systems that simply funnel credit to
the government or state-owned enterprises.
We calculate these three indicators for China and the 78 countries included in the
financial dataset of the World Bank (Beck, Demirguic-Kunt, and Levine 2000) for the
years 1979 (the beginning of the reform period in China), 1985 (the year following the
central bank creation People's Bank of China), and 1999 (the most recent year of data
availability6). The 79 observations are subsequently sorted by ascending order for each
year in order to locate China within the sample.
China's financial deepening is impressive. Financial depth measured by the ratio
of total bank liquidity to GDP has increased from 33 percent of GDP in 1979 to 148
percent in 1999. When compared to an international set of countries, China's liquidity
rate ranks among the highest in the world in the late 1990s-only Switzerland and Malta
have higher liquidity ratios (Figure la). By contrast, China ranked 37th among the 79
countries in 1979 (Figure 1 b). What are the factors accounting for such a dramatic
increase in real monetary 6alances? First, an increase in monetary transactions has been
generated by the monetization of the economy, defined as the increase of the amount of
transactions going through the market pushed by a combination of factors: the de-
collectivization of agriculture, the development of rural industry, and the development of
free markets together with the gradual withdrawal of state monopolies. Hence, the ratio
of money (MI) to GDP has increased from 19 percent in 1978 to 53 percent in 1999. Real
balances for savings have grown even faster. The ratio of quasi-money has increased
from 10 percent in 1978 to 81 percent in 1999. This financial deepening can be explained
by the change in income distribution from the government to the households over the
transition period because of dismantling state monopolies and the rapid development of
the non-state economy. The increase in the household saving propensity can be explained
by several factors. First, economic growth increased the saving rate along with higher
expectations of future income. Another growth-related factor is the diversification of
needs stemming from improved living standards and the availability of new consumption
6 In the World Bank database, end-of-year financial balance sheet items are deflated by indices and
the GDP series is deflated by average consumer prices. For comparing China's financial indicators with
international values, we simply use nominal figures, thereby implicitly assuming a common deflator for
numerators and denominators.
7 Friedman (1957) and Kraay (2000) find that expectations of future income growth are an important
determinant of household saving in China.
3



goods. Credit constraint plays an important role also. Consumer credit is just starting to
develop in China and households face a cash-in-advance constraint for consuming or
investing. Demographic factors such as an increase in life expectancy and an aging
population also matter. In addition, growing employment opportunities in the non-state
sector with limited social welfare by contrast with the state sector have increased future
income uncertainties and precautionary savings. Finally, the guarantee of deposits in the
state banks has kept household savings in the form of monetary assets despite the
emergence of alternative ways of saving (stock market, government bonds, and more
recently private housing). The state banks have been asked to channel savings to the loss-
making state-owned enterprises (SOEs). However, their ability to do so depends critically
on the savers' behavior and their willingness to place funds in the state-banking sector.
As a result, the government commitment to bailout the state-owned enterprises has
resulted in an implicit guarantee of the deposits in the state-owned banks (SOB).
Turning to the second indicator, China's share of commercial bank credit in total
credit to the economy equals 94 percent of GDP in 1985 and 98 percent of GDP in 1999,
situating the country in the upper quartile of the sample, at a level similar to
industrialized countries such as Canada or United States (Figures 2b and 2a,
respectively). In other words, the role of the People's Bank of China in credit allocation
appears relatively modest. Interpreting this result as a signal of efficiency can be however
misleading given the two weaknesses associated with this indicator (see Levine 1997).
First, banks are not the only financial intermediaries providing financial functions and,
second, banks may lend to the government or public enterprises. The first pitfall is not
likely to create serious biases in the case of China, as the financial system remains mostly
bank-based. In the late 1 990s the non-banking financial institutions market share of loans
was barely 16 percent. The second weakness is of bigger concern for China, as the four
state-owned banks dominate the banking sector and have a preferential policy toward
lending to state-owned enterprises. The third and last indicator of financial
intermediation-the ratio of credit going to the private sector-might thus be more
relevant for characterizing China's banking sector.
Chinese SOEs are famous for inefficiency. Despite two decades of incremental
reform, the role of state enterprises in China's economy has shrunk in most ways except
the proportion of bank lending they consume. In 2000, SOEs accounted for less than one-
quarter of industrial output and just over one-third of urban formal employment. Yet
conservative estimates suggest they absorbed three-quarters of all bank lending in the late
1990s.8 This high figure reflects the use of the state-owned bank credit by the
8 Data is from the Economist Intelligence Unit, "China: Grossly Distorted Product," February 18,
2002. Chinese statistics do not provide direct figures for loans to SOEs. Estimates are calculated by
assuming that all lending aside that recorded for foreign invested enterprises, township and village
enterprises, and "other" goes to the state sector.
4



government as a policy instrument to support the state sector.9 In other words, the state-
owned banks have been obliged to support the SOEs, despite and because of the low
profitability of the latter and their inability to repay their debts.' 0 Using an estimate of the
share of total credit directed to state enterprises of 90 percent in 1985 and 77.5 percent in
1999, the ratio of private credit to GDP would be 7 percent in 1985 and 14 percent in
1999, which ranks China as 4th and 15th among 79 countries respectively (Figure 3b and
3a). Hence, the allocation of credit shows little improvements during the reform period.
At the end of the 1 990s, the bulk of credit is still directed to the public sector in China,
despite the growing evidence of its structural inefficiency.
To summarize, at the national level China's financial intermediation can be
characterized on the one hand by a dramatic financial deepening and on the other hand by
a massive misallocation of financial resources. In other words, while the financial sector
has managed to facilitate the exchange of goods and services and mobilize savings, it has
not yet succeeded in allocating resources efficiently. In the following section, we justify
the existence of a local channel between financial intermediation and growth using
evidence from capital market fragmentation within China. In section 4, we analyze the
impact of banking development on local performance.
3.     Capital Market Fragmentation in China
In a fully integrated financial system, households should be able to deposit or
invest their savings and firms should be able to borrow anywhere in the economy either
through the banking system or through the financial markets. As a result, and after
controlling for aggregate shocks, lending by a region's banks and local economic
performance should be uncorrelated. Therefore, as a prerequisite for analyzing the impact
of local banking development on the local economy in China, we wish to evaluate the
degree of capital market fragmentation in China.
China's spatial economic organization is described as a de facto federalism,
involving a decentralized economic system in which each region can be considered as an
autonomous economic entity in terms of both goods and production factors.' Several
studies have pointed out that the level of inter-provincial trade in China resembles a loose
federation of sovereign states rather than a unified country.'2 In contrast to the well-
documented patterns of intra-national trade, no formal study is available on the degree of
9 With the deterioration of its fiscal position in the 1980s, the government increasingly shifted the
fiscal burden onto banks, thereby converting direct subsidies to firms into bank loans.
10 It is now commonly estimated that non-performing loans account now for 30 to 40 percent of the
state banks' total lending.
'" See Qian and Xu (1993).
12 See Poncet (2001).
5



intra-national capital mobility in China, although the available information supports the
view of a high degree of capital market fragmentation.'3
We estimate the degree of capital mobility within the Chinese provinces by
applying the test proposed by Feldstein and Horioka (1980) for international capital
mobility'4. The Feldstein-Horioka (F-H) test for capital mobility relies on the idea that,
under the null hypothesis of a perfectly integrated capital market, investment in one
region should not be constrained by the available savings in that region and the
correlation between local savings and local investment should be low. Conversely, if
regional capital markets are fragmented, domestic investment may be closely related to
domestic saving as a source of finance. The F-H test may not be ideal in a cross-country
context, as it does not give conclusive evidence on the degree of international capital
market integration. Furthermore, a low saving-investment correlation can be consistent
with various alternatives other than a low degree of capital market integration across
countries, such as the presence of currency devaluation risk premiums, and government's
efforts to target the level of current account balance by manipulating the exchange rate.
However, within a country, the F-H test turns out to be a reasonable indicator of the
degree of capital market integration across different regions, as the alternative
interpretations mentioned above are not operative within a country. For example, out of
six papers that have looked at countries that are known to have an integrated capital
market internally (Canada, Germany, Japan, the United Kingdom, and the United States),
all have found a very low savings-investment correlation across the regions.'5 This
provides the justification for using the F-H test to examine capital market integration
within China.
For our sample we use 26 Chinese provinces between 1985 and 2000. The
investment rate is the share of gross capital formation in GDP. The saving rate is defined
as the ratio of the difference between GDP and total consumption over GDP. 16 Figure 4
shows the relation between saving and investment rates averaged over the period. Two
features deserve attention: first, no clear relationship emerges from the cross-sectional
dimension. Second, Beijing, Shanghai, and Tianjin appear as outliers.'7 We therefore
" The World Bank (1994) finds no evidence of price or returns to capital convergence across
provinces as would be the case if provinces were financially integrated. Also Park and Sehrt (2001) find
that deposits are a key determinant of lending given the poor intermediation across institutions and between
and within provinces. The importance of deposits in determining the volume of local lending has also
increased over time between 1991 and 1997.
14 See Boyreau-Debray and Wei (2002).
15 Bayoumi and Rose (1993), Dekle (1996), Iwamoto and van Wincoop (2000), Sinn (1992), Thomas
(1993), Yamori (1995).
16 We have tested for the order of integration of the series using panel unit-root tests (Im, Pesaran, and
Shin 1995). Both investment and saving rates are stationary.
7 Moreover, including municipalities for testing capital mobility might not be relevant as capital
mobility between the cities and the surrounding provinces is likely to be high.
6



estimate the correlation between investment and saving with and without the three
municipalities.
Turning to the panel regressions, we use a simple within estimator with individual
and time-fixed effects. In a subnational context, time dummies control for national shocks
or macro policies that may increase simultaneously saving and investment for a given
degree of capital mobility. Table 1 reports the results. Unlike the results of the previous
subnational studies, the investment rate is significantly positively related to savings (row
I a, Table 1 a). We check the robustness of the results by first using an alternative measure
of investment, by subtracting the share of investment financed by the central government
(row 2a, Table I a). Although lower, the investment-saving correlation remains
significant. Second, potential biases arising from saving endogeneity with investment are
controlled by running 2SLS regressions, using the share of food expenditure in total
consumption expenditure as an instrument (Kraay 2000). The positive correlation
between investment and saving remains significant (Table Ib) . Overall, the results
contrast with the usual finding of the literature, which finds no investment-saving
correlation at the subnational level. Instead, the correlations appear closer to cross-
country investment-saving correlations, supporting the view of a low degree of capital
mobility within Chinese provinces. The results show that China still has a long way to go
to reach financial integration. But more importantly for our purpose, they provide a
justification for analyzing the impact of local financial intermediation on local growth
performance.
4.     Financial Intermediation and Growth: Subnational Evidence
This section presents an empirical analysis of the impact of local banking sector
on provincial growth performance. First, we present the growth equation to estimate and
discuss the econometric method as well as the economic and banking indicators used on
the right-hand side. Second, we present and discuss the results.
4.1    Empirical Framework
The data set consists of economic and financial statistics for 26 Chinese provinces
and 3 municipalities directly under central government control between 1990 and 1999.
All variables are averaged over two years, providing five observations per province. 18 We
estimate the following growth equation:
y,t =ao Yt-T + a,X,1 + a2F,1 + u, + t+ e,                                 (1)
18 Averaging all the variables over two years results from a compromise between controlling for short-
term shocks on the one hand and on the other hand keeping enough observations given the length of the
time series (10 years). For variable definitions and statistical sources see the appendix. Due to missing
values, Tibet is excluded from the sample.
7



where y is the real GDP per capita (in logarithm), T is the period length, X is a vector of
control variables (or the conditioning information set), F is a vector of financial
intermediation indicators, m is a province fixed effect, t is a time fixed effect, and e is the
error term, and i and t are, respectively, the provincial and time subscripts.
4.1.1 Econometric Method: GMM System Estimator
Equation (1) confronts us with two econometric issues. First, introducing the lagged
dependent variable among the regressors together with fixed individual effects renders
the OLS estimator biased and inconsistent even if the e,t are not serially correlated, as the
lagged dependent variable is correlated with the error term. Second, most of the
explanatory variables can be expected to be endogenous with economic growth. We thus
need to control for endogeneity arising either from the dynamic specification of the
equation or from reverse causation. The Generalized Method of Moments (GMM) is
usually used to control for endogeneity arising in panel data models. The first difference
GMM estimator, as proposed by Arellano and Bond (1991), involves as a first step taking
the first difference of the proposed equation in order to remove the fixed individual
effects from the equation. However, in the differenced equation, the error term is still
correlated with the lagged dependent variable. The second step consists of instrumenting
the explanatory variables. Under the assumption that there is no serial correlation in the
error term, the lagged levels of the explanatory variables can be used as instruments of
the first differenced variables. In the context of economic growth models, this method has
the advantage of avoiding biases related to omitted specific individual effects and to
control for endogeneity arising from bi-directional causality. One critical assumption is,
however, that lagged levels of variables are good instruments for explaining subsequent
first differences. Hence, when the time-series are persistent and the number of time-series
is small, which is typically the case in the empirical growth models, the first-differenced
GMM estimator is shown to behave poorly (Bond, Hoeffler, and Temple, 2001). In
particular, these authors show that the coefficient of the lagged dependent variable tends
to be below the corresponding coefficient estimate in the within groups, suggesting
GMM first-differences are biased. In the case of growth models, Bond and others (2001)
recommend the use of the so-called System-GMM estimator (Blundell and Bond 1998),
which uses the information contained in the initial conditions to generate efficient
estimators when T is small and variables are highly persistent. The basic idea of this
estimator is to use lagged first differences of the variables as instruments for the equation
in levels in combination with the usual approach. In the next section, the System-GMM
estimator is used for estimating growth equations with financial intermediation
indicators. Next, we use the System-GMM method to estimate equation (1).
8



4.1.2 The Data
Control Variables (Conditioning Information Set)
The vector of control variables Xis defined according to the augmented Solow
model as proposed by Mankiw, Romer, and Weil (1992). We introduce the investment
rate as a proxy for physical capital and the share of population with more than secondary
schooling as a proxy for human capital (schooling). '9 Here we depart from the traditional
specification used in the empirics on finance and growth, which usually do not introduce
the investment rate, and use the initial rather than the contemporary value of human
capital in order to avoid biases arising from potential endogeneity of factor accumulation
with economic growth.20 This method, however, gives rise to another bias due to omitted
variables as long as physical and human capital accumulation is considered a relevant
determinant of economic growth. As emphasized by Bond, Hoeffler, and Temple (2001),
the System-GMM methodology allows parameters to be estimated consistently in models
with endogenous right-hand side variables by using instrumental variables. We therefore
include the current values of physical and human capital accumulation while controlling
for endogeneity. The share of non-state production (non-state production) and the ratio of
foreign direct investment to GDP (Fdi) are introduced as control variables. The former is
an indicator of the macro environment of the local economy (or of the progress of the
transition process at the local level) whereas the latter captures the provincial degree of
integration to the world economy.
Local Indicators of Financial Intermediation
At the subnational level, none of the three banking development indicators
traditionally used in cross-country studies as described above is available. We therefore
build four indicators of financial intermediation corresponding as closely as possible to
the cross-country ones. Given that information on cash distribution among regions is not
available, we simply use the ratio of total deposits of the banking system to GDP as an
indicator of the size of the local banking sector (SIZE). Similarly, central bank credit is
not available at the province level. Following Lardy (1998) and Dayal-Gulati and
Hussain (2002), we use the ratio of loans to deposits of the state-owned banks as a proxy
for central bank lending to the provinces (CENTRAL). In China, while the volume of
deposits is determined by economic activity, the volume of lending was largely
determined by policy objectives, through the credit plan and independently of the ability
of branch banks in each region to finance the lending target from local deposits (Lardy
1998). Hence, some rapidly growing provinces could have a low credit quota and be
19 We thank Colin Xu for kindly providing us with the schooling series.
20 This specification is proposed by Barro and Sala-I-Martin (1992) and justified by the need to
control for differences in steady states of the economies depending on their structural features and initial
conditions.
9



constrained in their lending relative to the rapid growth of their deposits. Alternatively,
branch banks in slower growing regions could be assigned high quotas with insufficient
local deposits to finance their lending; and these provinces depended on the central banks
to lend them additional funds. Hence, the ratio of SOB credit to SOB deposits provides a
measure of the central bank credit to the local branch banks to meet their lending quotas.
In the recent years, the administrative targets have been phased out and replaced by a
maximum ratio between loans and deposits. The ratios apply to total national lending by
individual banks but allow the headquarters to alter credit allocation for specific
provinces.21 Therefore, the ratio of loans to deposits can also be interpreted as a measure
of interregional fund allocation, as the state banks are provided with greater flexibility to
use within bank transfers to adjust regional needs.
The third indicator of financial intermediation is the ratio of state-owned banks
credit to GDP (SOB). As already noted, Chinese statistics do not provide any information
on credit allocation between state and non-state enterprises. However, given that the state
banks' primary function was to channel savings to state-owned enterprises, the ratio of
the state-owned banks credit to GDP can be interpreted as a proxy for the credit
channeled to the state-owned sector. For instance, 80 percent of the total amount of credit
by the state-owned bank was extended to the state-owned enterprises in the late 1 990s.
Even with the recent emphasis on profit maximization and management responsibility in
the state banking sector, the state banks may still favor the SOEs with which they have a
long customer history and which are more likely to be bailed out by the government than
non-state enterprises in the case of financial troubles. By contrast, projects in the non-
state sector are perceived as more risky because of higher information costs and moral
hazard.
Finally, we are interested in assessing whether for a given size of the banking
sector its structure matters for local economic growth in China. Since 1984 the initially
specialized state banks have been allowed to compete for deposits and loans in each
other's previously monopolized markets, and enterprises have been allowed to open
accounts with more than one bank.22 Existing evidence suggests that all the state banks
have remained largely involved in the same specialized business areas. However, the
development of new financial institutions including national and regional non-state
banks, urban and rural credit cooperatives, and non-banking financial institutions has
increased the competition for deposits. More importantly, following China's recent WTO
accession, foreign banks will be entitled to national treatment without geographic or
21 It is questionable whether state banks are actually conforming to these ratios, as the ratios of
outstanding loans to total deposits remain well above the authorized ceiling.
22 When the People's Bank of China was granted the authority of a central bank in 1984, its
commercial operations were transferred to four specialized banks: the Agricultural Bank of China for the
rural sector, the Industrial and Commercial Bank of China for the industrial sector, the People's Bank
Construction Bank of China for long-tcrm investment; and the Bank of China for foreign exchange.
10



customer restrictions within five years. Hence, the banking sector is about to face strong
and intensive competitive pressures from large foreign financial institutions. However, as
highlighted by Huang and Qian (2001), the banking and sector is different from the real
sector where more competition is considered better, as an increase in competition on the
financial side creates a tradeoff between banking stability and banking efficiency. Indeed
two main outcomes can result from the weakening of the monopolistic position of the
state-banking sector. First, increased competition may result in a drop of saving deposits
in the state banks, in turn threatening their ability to finance the state-owned sector and
overall financial stability, as the system rests on the continuing willingness of savers to
deposit much of their income with state banks. Second, increased bank competition may
result in improved efficiency of financial resources allocation and improved access to
credit of the non-state sector, thereby fostering economic growth.23 We use a fourth
indicator to account for the banking market structure in the provinces, by calculating an
Herfindahl index of bank concentration. If n is the number of banks, i the province and t
the period, the index is computed as follows:
H,, = i      D1,                                                          *(2)
J=     / J=1
Where Dj,,,t, is the deposits for ba kj, the index equals one in case of monopoly and 1/n
in case of equal shares among the n banks. The available data allows us to distinguish
between n=7 financial institutions, e.g., the four state-owned banks (Agriculture Bank of
China, Industrial and Commercial Bank of China, Bank of China, Construction Bank of
China), the Bank of Communication, the Rural Credit Cooperatives and the "other
financial institutions." Using the deposits of each financial institution, we compute the
bank concentration index for each of the 29 regions (26 provinces and 3 municipalities)
and the 5 periods (CONCENTRATION).
4.2    Descriptive Statistics and Correlations
Table 2a reports the average value of the control variables and the four banking
indicators. The latter exhibit considerable variation across provinces. For instance, SIZE
ranges from 60 percent of GDP in Hunan or Anhui to 274 percent of GDP in Beijing
municipality or 115 percent of GDP in Shanxi province. SOB is as high as 107 percent in
Qinghai but only of 41 percent of GDP in Zhejiang. The pattern of central bank credit to
the provinces is also uneven. In Jilin state bank, credit is 1.7 time higher than deposits,
whereas in Beij in, Guangdong, or Zhejiang credit outstanding does not even match
23 Theoretical arguments can be found to support either a positive or a negative effect of bank
concentration on economic growth. On the one hand, a lower bank concentration results in a higher amount
of credit available for the economy as a whole. Banks with monopoly power would determine an
equilibrium with higher loan rates and a smaller quantity of loanable funds. On the other hand, the positive
effect derives from the greater incentive for monopolistic banks to establish lending relationships, which in
turn promotes firms' access to investment funds. Using a cross-country data set over the period 1989-1996
and several indicators of bank concentration, Cetorelli and Gambera (2001) find, however, that bank
concentration has a negative effect on industrial growth.



deposits (0.86). Finally, the market structure of the banking sector in the provinces varies.
The provinces associated with the lowest degree of concentration are Shanxi and Jiangsu
(0.18), with 0.14 as a benchmark of equal shares among the seven banks. At the opposite
extreme come Beijing (0.29) municipality, and Nei Mongolia, Qinghai, Jilin, and
Liaoning provinces (0.25).
Table 2b reports the correlations between the set of control variables and the
banking indicators. All control variables are positively and highly correlated with each
other. The investment rate in GDP is likely to be associated with a high level of human
capital (correlation of 0.94), a large share of non-state sector production (correlation of
0.85), as well as a relatively higher ratio of foreign investment in GDP (correlation of
0.96). Similarly, a relatively high share of non-state production is accompanied by a high
level of foreign investment in GDP. Similarly, two of the four financial indicators are
highly positively correlated: a high rate of SOB loans in GDP is associated with a high
level of central bank credit to the province, which is not surprising given the importance
of this financing for SOB. SOB is also highly correlated with CONCENTRATION. A
relatively high rate of SOB loans also indicates a high degree of concentration of the
banking sector. Finally, Table 2b shows the correlation between the four financial
indicators and the set of control variables: Size does seem to be related to any of the
growth determinants while SOB, Central and Concentration are all highly negatively
correlated with the share of non-state production, Fdi, and, to a lower extent, investment
rate.
Figures 5 to 8 show the relations between the four averaged banking indicators
and the growth rate of GDP per capita and the initial level of GDP per capita,
respectively. Regarding the overall size of the local banking sector, the municipalities of
Beijing, Tianjin and Shanghai have the highest levels of deposit to GDP ratio, of initial
GDP per capita and of per capita GDP growth rate. In contrast to cross-country studies
that highlight a positive correlation between the size of the banking sector and growth of
GDP per capita, a slightly negative relationship emerges between financial depth and the
growth of per capita GDP (Figures 5. la and 5.2a). Fast growing provinces such as Fujian,
Zhejiang, Shandong Jiangsu Hebei, Hubei, and Anhui are associated with a low ratio of
total deposits to GDP, whereas provinces with the highest values of deposits to GDP have
been growing at a somewhat lower rate (Guangdong, Shaanxi, Shanxi, Ningxia, Gansu
and Xinjiang). No clear relation can be seen from Figures 5.1b. and 5.2b between the size
of the banking sector and the initial level of GDP per capita. One has to keep in mind that
a key assumption underlying the use of the ratio of deposits to GDP as an indicator of
financial development is that the size of financial intermediary sector should be positively
correlated to the provision and quality of financial services. This assumption is
questionable for China, given the quasi-monopoly of the state-owned banks and their
12



poor performance in allocating resources efficiently. The other indicators that consider
the state sector might prove more relevant for China.
Central bank lending seems to be preferably allocated in provinces with lower
growth of per capita GDP (Figure 6.a). Similarly, there is a negative correlation between
the ratio of SOB lending to GDP and the real growth rate of GDP per capita (Figure 7.a).
No clear relation can be detected between central bank or state bank lending and the
initial GDP per capita (Figures 6.b and 7.b, respectively). Provinces such as Jilin, Hubei,
Qinghai, and Nei Mongolia benefit the most from central bank lending but are neither the
poorest nor the richest provinces. By contrast, some of poorest provinces such as
Guangxi or Yunnan receive little credit from the central bank. Beijing's deposits exceed
by far its corresponding value of credit, suggesting that the capital city is a net contributor
to the central bank resources.
Depending on the causality between those two negative relationships, two
different interpretations are proposed. In the first case-GDP growth to central bank SOB
lending-provinces where the state sector dominates the local economy such as Qinghai,
Jilin or Nei Mongolia may suffer from higher rigidities in their labor and product markets
which could be in turn be reflected in lower economic performance. The government may
compensate those provinces for this structural handicap by softening their budget
constraint and providing them with easier access to credit. The other interpretation
involves a reverse causality-from SOB lending to economic growth-which would
suggest that SOBs are less efficient at allocating resources than other financial
institutions, leading to lower growth performance in the provinces where SOBs dominate
the local financial sector.
A negative relationship between CONCENTRATION and the growth of per
capita GDP emerges from Figure 8.a, suggesting that a monopolistic banking sector is
less efficient than a banking sector where banks and enterprises have to compete for
scarce financial resources. Bank concentration does not appear to be related with the level
of economic development (Figure 8.b). Provinces with similar levels of initial GDP per
capita can have very different banking structures with, for instance, middle-income
provinces of Qinghai, Nei Mongolia, Jilin Gansu having a relatively more concentrated
market than the richer provinces of Shandong, Shanxi, Fujian, Hainan, Hubei, or Hebei.
This statistical description suggests several preliminary results. First, the size of
the intermediation sector does not appear to be related to the level of development or to
local growth performance. Second, provinces with a lower rate of economic growth
receive larger credit flows from the central government. Third, higher market
concentration is associated with lower growth performance at the local level. Fourth, the
three municipalities exhibit outlying features. Finally, according to the correlations
between all the explanatory variables, it seems that a higher share of the non-state sector
13



is associated with mnore foreign investment, a lower ratio of state bank loans in GDP, less
credit from the central bank, and less concentration in the banking sector.
These bivariate correlations do not, however, allow us to discriminate between the
different interrelated hypotheses such as the causality between financial development and
growth, the state banking overall performance, the impact of the central bank lending and
the effect of bank concentration on local real growth. In the next section, we estimate the
significance of financial intermediation on growth by estimating growth equations and
controlling for the potential endogeneity of the regressors.
4.3   Econometric Estimation
Table 3a reports the estimates of equation (1), using only the investment rate and
the schooling variables as control variables. As suggested by the statistical evidence
provided in the previous section, we treat Beijing, Tianjin, and Shanghai as outliers by
excluding them from the sample. Moreover, including municipalities in the context of the
growth and finance relationship would be questionable as capital mobility is likely to be
high between these cities and their surrounding provinces.24
Surprisingly, the liquidity indicator has a significant negative impact on economic
growth, which contrasts sharply with the usual cross-country result of a positive
relationship between the size of the financial sector and economic growth (Table 3a,
column la). As mentioned before in section 2, financial deepening during the reform
period has been impressive at the national level, although not accompanied by an
improvement in savings allocation. At the regional level, interior provinces with a
relatively lower growth performance are associated with a high ratio of deposits to GDP.
Therefore, the implicit assumption that the size of the financial sector is positively
correlated to the quality of financial services is questionable in the case of China. A high
level of this indicator may not prefigure a high level of financial intermediation
development. Park and Sehrt (2001) reach a similar conclusion: they find an inverse
relationship between the rate of financial intermediation and the level of economic
development among Chinese provinces, suggesting factors other than economic
fundamentals play an important role in lending decisions.
State-owned bank credit has a negative and significant impact on local economic
growth, supporting the hypothesis that the state-banking sector does not allocate savings
efficiently and that provinces with a more developed non-state financial sector may
benefit from more efficient resource allocation. Similarly, the more credit a province
receives from the central bank, the lower its growth performance (columns 2a and 3a
24 Indeed, when the three municipalities are included, none of the financial indicators are significant
(results available from author).
14



respectively).25 Keeping in mind that we control for the endogeneity of the regressors, the
direction of causality can be interpreted as running from the central bank lending to
economic growth. At first glance, this result seems puzzling: increased availability of
financial resources can be thought of as having a positive or insignificant impact on
economic growth, but can hardly be thought of as worsening local economic
performance. Central bank lending can, however, affect local growth performance in a
negative way by softening the local budget constraint. Following Komai (1980), an
organization is said to have a soft budget constraint when it expects to be bailed out in
case of financial trouble.26 For instance, some unprofitable enterprises will become
profitable only if they undergo restructuring. But since it is costly for an enterprise to
restructure, it will only do so if the enterprise would otherwise go bankrupt. Therefore, if
the enterprise anticipates that it will be bailed out by the government, it will not
restructure. In the case of China, the state-banking sector has been used by the
government as a quasi-fiscal instrument to bailout loss-making state enterprises or to
deliver specific policy loans without consideration for efficiency. The result has been a
growing accumulation of non-performing loans. In compensation for their lack of
autonomy, the SOEs were implicitly guaranteed bailouts by the central bank. Given the
easy access to central-bank refinancing and the government's commitment to support the
state sector, the budget constraint of the state banks is likely to be soft. Thus, the finding
of a negative impact of central bank lending on local economic growth can be interpreted
as a lack of incentives for the state banks to improve their management or base their
lending decisions on efficiency criteria, as they expect the central bank to fulfill their
losses. Furthermore, this result supports the idea that the state-owned bank management
has not improved in the recent years despite reforms aimed at transforming them into
commercial banks responsible for their losses.27 Greater bank concentration is
significantly associated with lower economic growth and remains robust when the size of.
the intermediary sector is also introduced (columns 4a and 5a, respectively). This result
supports the idea that bank competition is likely to: improve economic performance as
found by Cetorelli and Gamberra (2001) using a cross-country sample. It is also
consistent with the evidence shown at the micro level by Cull, Shen, and Xu (2002) that
greater entry into the banking sector improves bank performance, which in turn improves
resource allocation in the province and leads to higher economic growth. Table 3a also
presents the Sargan test for instrument validity, where the null hypothesis states that the
instrumefital variables are uncorrelated with the residuals, and the serial correlation test,
23 When introducing the ratio of loans to deposits, Dayal-Gulati and Hussain (2002) find the same
result during the 1988-97 period.
26 See Maskin and Xu (2001) for a review of soft budget constraints.
27 Park and Sehrt (2001) analyze financial intermediation efficiency in China between 1991 and 1997
and draw a similar conclusion: "financial intermediation in China is far from efficient and that financial
reforms in'the midl1990s have not reversed a worsening trend."
15



where the null is that the errors exhibit no second order serial correlation. There is no
evidence of second order serial correlation and the validity of the instruments cannot be
rejected, as shown by the insignificance of test statistics.
The difficulty here is to interpret the three indicators of financial intermediation
separately, as they are all closely interrelated. For instance, one can argue that a high
level of state-owned bank credit is likely to be explained by the predominance of the
state-owned enterprises on the real side of the local economy. As the state-owned banks
lend mainly to state-owned enterprises, the negative growth impact of the credit they
extend can simply be explained by the poor performance of their clients. A high level of
state-bank credit is also likely to be translated into a higher dependence on central-bank
lending used to finance the shortfalls between loans and deposits in the SOBs. Similarly,
a high ratio of SOB loans to GDP in the local economy is likely to indicate a high degree
of banking sector concentration, as the state banks dominate the banking sector. Ideally,
we want to distinguish the state banks' negative impact on economic growth explained by
their intrinsic performance from the negative impact arising from the importance of the
state-owned corporate sector and more generally from a poor economic climate. To do
this we run the same regressions while controlling for the significance of the state sector
in the local economy.
Figures 9 to 12 show the average levels of SIZE, SOB, CENTRAL and
CONCENTRATION related to the share of non-state enterprises in total gross industrial
output value (sgovns). With the exception of SIZE, each of these indicators is clearly
negatively correlated to the size of the non-state sector. For instance, provinces such as
Qinghai, Jilin, Nei Mongolia, Ningxia, Jilin, Gansu, or Guizhou are characterized by a
large share of state-industry and a high level of SOB credit, a high level of central-bank
lending, and a high bank concentration index. By contrast, the provinces of Zhejiang,
Jiangsu, Shandong, Fujian, Henan have a higher share of non-state production, rely less
on SOB credit, have a more diversified banking sector, and receive less central-bank
financing. These provinces are thus less likely to receive a large amount of policy loans
to support state industry. In addition to such arguments, the share of non-state sector
production is often used in studies of regional economic growth in China as an indicator
for structural macroeconomic differences, such as a differences in the degree of goods
and labor market flexibility, differences in the progress of reforms, and more generally
for the extent to which a market climate prevails in the province.
Table 3b reports the results when the share of industrial production by non-state-
enterprises in total industrial production is added among the regressors. As found in
previous studies, provinces more advanced in the transition process benefit from a higher
rate of real growth, as shown by the positive and significant coefficient of the share of the
non-state production. Interestingly, the financial deepening variable is now insignificant
(column 2b). As a result its negative impact in the base model comes from the fact that
16



the provinces with the highest liquidity rates are also the ones associated with the highest
share of state production, meaning that a high ratio of deposits to GDP is more the result
of a preferential credit policy than the result of financial deepening in the local economy.
Another interesting result is that the ratio of the share of SOB credit ceases to be
significant when we control for the importance of state-owned industry at the local level.
Its negative impact in previous regressions was more related to the bad performance of
the state enterprises than the misallocation of credit by the SOB per se (column 2b). The
negative effect of central-bank lending to the provinces and the bank concentration index
both remain significant when we control for the share of state-owned industry (columns
3b and 4b).
We are also interested in checking whether the good growth performance in some
Chinese provinces comes from easy access to foreign savings, e.g., mostly in the form of
foreign direct investment. Non-state enterprises face different financing constraints
depending on their access to foreign capital. Some provinces such as Guangdong, Fujian,
or Hainan have received foreign direct investment ranging from 1 1 to 14 percent of GDP
during the 1 990s, while more remote areas such as Ningxia, Nei Mongolia, Guizhou or
Yunnan provinces received almost nothing.28 Foreign investment is usually found to be a
general indicator of openness and a robust determinant of provincial growth.29 However,
when we introduce the ratio of foreign investment into our growth regressions (Table 3c),
foreign investment enters none of the regressions significantly while the other coefficient
estimates remain unchanged.30 This latter result is in line with Carkovi and Levine (2002)
who find no evidence of a causal impact of foreign investment on economic growth once
the endogeneity of Fdi with economic growth is taken into account.
Now turning to the significance of the control variables, the results are somewhat
disappointing. Physical capital accumulation enters only three regressions significantly
(column 4b in Table 3b and columns lc and 2c in Table 3c) whereas human capital
accumulation is only significant in one of the regressions (column 3b in Table 3b). The
same result can be found in other empirical studies of provincial growth in China. For
instance Li, Liu, and Rebelo (1998) and Aziz and Duenwald (2001) find no evidence of a
significant contribution of physical capital accumulation to provincial growth in China.
Most of the previous studies do not introduce a proxy for human capital accumulation or
use the secondary school enrolment, which is only a rough proxy for the stock of human
capital in the provinces.3' Although the share of population in the province with more
28 See Table 2.
29 For more on the impact of China's open door policy see Lee (1994), Mody and Wang (1997), Chen
and Feng (2000), and Demurger (2000).
30 The ratio of exports and imports to GDP was also introduced but never appeared to be significant.
3' Demurger (2001) uses the share of population with more than secondary schooling, which is
calculated from the permanent inventory method.
17



than secondary schooling is a more accurate indicator, it may still suffer from
measurement error. More generally, such indicators only capture the level of capital
accumulation and do not provide information about the ability of the economy to allocate
factors efficiently. In their analysis of the sources of economic growth in China, Wang
and Yao (2001) find rapid human capital accumulation since 1952, and that both physical
and human capital contributed to China's growth performance during the reform period.
These authors also point out that given the rapid expansion of capital base, the relative
importance of factor accumulation may be declining in favor of total factor productivity
as the driving force for growth. Increasing total factor productivity depends on improving
the allocation of factors by reforming the state and financial sector and allowing both
capital and labor to move freely.32
Little evidence of conditional (or beta) convergence is found. As reported at the
bottom of Tables 3a and 3b, the coefficient estimates of the lagged dependent variable are
significantly smaller than unity only in four regressions. How can this weak result of
conditional convergence be explained? First, because variables are averaged over two
years, our estimates might capture both short-term (business cycles) and long-term
(structural) variations. Second, the general consensus among studies of income per capita
convergence among Chinese provinces is that the relative dispersion of income per capita
has decreased over the 1 980s and increased over the 1 990s, while beta-convergence is
significant over both periods.33 However, forces of beta-convergence were stronger in the
pre- 1990 period and weaker in the 1990s. For instance, estimates over the whole reform
period range from 4.3 to 7.5 percent per year. When the period starts in the late 1980s or
early 1990s, the convergence coefficients range from 1.6 to 2.5 percent per year (see for
instance Aziz and Duenwald 2001 and Dayal-Gulati and Hussain 2002). When
significant, our estimates of GDP per capita convergence range at comparable levels,
from 2.6 to 4.6 percent per year (bottom of Table 3a and 3b).
Finally, another reason for an absence of convergence may lie in the econometric
technique. Quite surprisingly and despite the well-known inefficiency of OLS or within
estimators for panel data when the lagged dependent variable is introduced among the
regressors, few previous studies on growth empirics in China have relied on GMM
estimators.34 As Bond, Hoeffler, and Temple (2001) point out, using a OLS level
estimator for AR(1) models gives an estimate of the lagged dependent variable biased
upward in the presence of individual fixed effects, whereas within estimators give an
32 This conclusion is similar to the one of Easterly and Levine (2001) who find on a cross-section of
countries that the level of investment is not as important as its quality, and that TFP accounts for most of
the variation in output.
33 See Aziz and Duenwald (2001) for a review of results on the convergence of provincial income per
capita in China.
34 One exception is the study of Aziz and Duenwald (2001), which implements the GMM-System
methodology.
18



estimate biased downward in short panels. Therefore, finding out that the GMM estimate
of the coefficient on the lagged dependent variable lies close to the corresponding within
parameter estimate can be considered as a signal that biases due to weak instruments may
be important. Tables 4a and 4b report the estimates using first-differenced GMM, Within
Groups (Table 4a) and OLS (Table 4b) estimators, respectively. The estimates of the
lagged dependent variable using a first-differenced GMM estimator are close to the
corresponding within ones, which are likely to be biased downward in a short panel. In
contrast, the estimate of the coefficient of the initial income using the GMM system lies
well above the corresponding within and below the OLS estimates. Overall, the results
suggest a bias problem caused by weak instruments in the first-differenced GMM
estimates and by the correlation of the lagged dependent variable with the error term in
the within estimates. Hence, the previous results on provincial growth convergence using
within estimators most likely overstate the beta-convergence process between Chinese
provinces.
We also try out alternative specifications in order to check the robustness of the
results. First, we drop the investment rate from the set of control variables and estimate a
reduced form of equation (1) where investment rate is implicitly supposed to be a
function of the GDP growth rate. As mentioned earlier, the rate of investment is highly
correlated with the schooling variable, so estimating the reduced form of the growth
equation can help eliminate colinearities. The results in Table 5a are very close to results
when the investment rate is part of the conditioning set of variables. Hence estimating the
equation with or without investment does not seem to make a difference and colinearity is
unlikely to be important, as the schooling variable remains insignificant even when the
investment rate is left out of the regression.
Second, we introduce a coastal dummy among the regressors, the idea here being
coastal provinces have been growing more rapidly and their banking sector is
characterized by a lower share of SOB credit, less credit received by the central bank, and
a more diversified banking sector. Hence introducing a coastal dummy checks for
omitted variables that can create a spurious relation between the growth rate of per capita
GDP and the indicators of financial intermediation. The results in Table 5b with a coastal
dummy introduced remain similar to that without any dummy, supporting the view that
the significance of the relations between the financial intermediation indicators and local
growth performance does not come from any common feature of coastal areas not taken
into account in the regression.
Finally, we estimate equation (1) over two five-years periods rather then five two-
years periods in order to evaluate sensitivity to cyclical issues. Table 6 reports the
estimate using successively within (fixed individual effects) and Generalized Least
Squares (random individual effects) estimators. Both estimators lead to qualitatively
19



similar results for the financial indicators35: Size has either the "wrong" sign or is
insignificant while SOB has a negative sign and is significant when fixed effects are
used. CENTRAL is significantly negative while CONCENTRATION has a negative
sign, although significant only in the GLS specification. Overall, using a longer period to
avoid cyclical trends leads to similar results.
5.     Conclusion
This paper examines the relationship between financial intermediation and local
economic growth in China. It appears that because of capital market fragmentation, the
unevenness in banking development is an important factor in local economic
performance. We find that China's financial deepening, as exceptional as it is, does not
contribute to local economic performance, as shown by the insignificance of the ratio of
deposits to GDP. Furthermore, the banking sector's continued support of loss-making
state-sector enterprises over non-state enterprises is reflected in the negative impact of
state and central-bank lending on economic growth. More precisely, the negative impact
of state-bank credit is not a matter of financial performance per se, but more the result of
the burden of supporting the state-owned sector. Indeed, when we control for the state
corporate size in the local economy, the state banks' negative impact on growth ceases to
be significant. This finding suggests that improving state bank performance and more
generally financial resource allocation would necessitate first reforming the state
corporate sector and improving the economic climate at the local level. Another
important result with policy implications is that provinces with a more diversified
banking sector have performed better in terms of economic growth. In the short term, it
would be advisable for China to relax restrictions on entry into the banking sector in
order to prepare the economy for the strong competitive pressures likely to come from
foreign banks with China's accession to the WTO.
Finally, traditional determinants of economic growth do not appear to explain
local economic growth in China over the 1990s. For instance, foreign direct and domestic
investments are insignificant most of the time. Another example is the somewhat
worrisome no-convergence result of per capita income over the period. Poorer provinces
do not catch up to richer ones, even after controlling for a set of conditioning factors.
While part of the reason for these unconventional results may be the period under study
or the short panel dimension, a more important explanation lies in the econometric
technique. The GMM-System estimator that we use provides efficient estimates for
empirical growth models, while standard estimators used in previous studies are
35 The two estimates lead, however, to very different results regarding the set control variables: the
initial income per capita variable becomes insignificant when the within estimator is uses, suggesting that
the fixed individual effects capture the convergence effect. Similarly, the share of non-state production is
insignificant. Overall, this suggests that these variables may not vary enough to remain significant in the
presence of fixed provincial effects.
20



inefficient for panel autoregressive models. Moreover, GMM-system estimators also
outperform the first generation GMM-estimators when the time-series is persistent as is
typically the case in economic growth models.
References
The word "processed" describes informally reproduced works that may not be commonly
available through library systems.
Arellano, M., and S. Bond. 1991. "Some Tests of Specification for Panel Data: Monte
Carlo Evidence and an application to employment equations." Review of
Economic Studies 58: 277-297,
Aziz, J., and C. Duenwald. 2001. "China Provincial Growth Dynamics." WUSTL
Economics Working Paper-Development and Comp Systems, 12004,
Barro, R. J., and X. Sala-i-Martin. 1992. "Convergence." Journal of Political Economy
100(2): 223-51.
Bayoumi T. A. and A. K. Rose. 1993. "Domestic Savings and Intra-National Capital
Flows." European Economic Review Review 37: 1197-1202.
Beck, T., A. Demirguic-Kunt, and R. Levine. 2000. "A New Database on the Structure
and Development of the Financial Sector." World Bank Economic Review 14(3):
597-605.
Blundell, R., and S. Bond. 1998. "Initial Conditions and Moment Restrictions in
Dynamic Panel Data Models." Journal of Econometrics 87(1): 115-43.
Bond, S., A. Hoeffer, and J. Temple. 2001. "GMM estimation of Empirical Growth
Models." Economics Group Papers W21, Nuffield College, University of Oxford.
Boyreau-Debray, G., and S.J. Wei, 2002, "How Fragmented is the Capital Market in
China?", presented at the Chinese Economic Society Conference, "The
Integration of the Greater Chinese Economies: Causes, Consequences, and
Implications", Hong Kong, June26th-28th.
Carkovi M. and R. Levine, 2002, "Does Financial Investment Accelerate Economic
Growth?", Working Paper presented at the Center for Financial Studies and
World Bank Joint Conference, May 30-31, Washington DC.
Cetorelli, N., and M. Gambera. 2001. "Banking Market Structure, Financial Dependence
and Growth: International Evidence from Industry Data." Journal of Finance
56(2, April): 617-48.
Chen, B., and Y. Feng. 2000. "Determinants of Economic Growth in China: Private
Enterprise, Education and Openness." China Economic Review 11(1): 1-15.
Chen, J., and B. M. Fleisher. 1996. "Regional Income Inequality and Economic Growth
in China." Journal of Comparative Economics 22: 141-64.
Cull, R., M. Shen, and L. C. Xu. 2002. "How much is Lack of Finance Holding Back
China's Interior Provinces." World Bank, Washington, D.C. Processed.
21



Dayal-Gulati, A. and A. M. Husain. 2002. "Centripetal Forces in China's Economic
Takeoff." IMFStaff Papers 42(3): 364-94.
Dekle, R. 1996. "Saving-Investment Associations and Capital Mobility: On the Evidence
from Japanese Data." Journal of International Economics 41: 53-72.
Demurger S., J. D. Sachs, W. T .Woo, S. Bao, G. Chang, and A. Mellinger. (2002).
"Geography, Economic Policy, and Regional Development in China." NBER
Working Paper 8897, April.
Demurger, S. 2000. Economic Opening and Growth in China. Paris: OECD Deveiopment
Center Studies.
Demurger, S. 2001, "Infrastructure Development and Economic Growth: An Explanation
of Regional Disparities in China." Journal of Comparative Economics 29: 95-117.
Easterly W. and R. Levine; 2001, "It's not Factor Accumulation: Stylized Facts and
Growth Models", World Bank Economic Review, 15(2):177-219.
Feldstein, M., and C. Horioka. 1980. "Domestic Savings and Intemational Capital
Flows." Economic Journal 90: 314-29.
Friedman, M. 1957. "Discussion of the Inflationary Gap." In Essays in Positive
Economics. Chicago: Univ. Chicago Press.
Hanson, McPherson S. and C. Waller. 2000. "Do Local Banks Matter for the Local
Economy ? In Search of a Regional Credit Channel." In G. D. Hess and E. van
Wincoop, eds., Intranational Macroeconomics. Cambridge University Press.
Huang, H., and Y. Qian. 2001. "Financial Stability and Development in China after the
WTO accession." University of California, Berkeley. Processed.
Im, K., M.. Pesaran H, and Y. Shin. 1995. "Testing for Unit Roots in Heterogenous
Panels." Working Paper 9526 (June). Department of Applied Economics,
University of Cambridge.
Iwamoto, Y., and E. van Wincoop. 2000. "Do Borders Matter? Evidence from Japanese
Regional Net Capital Inflows." International Economic 41 (1, February) : 241-69.
Jian, T., J. D. Sachs, and A. M. Warner. 1996. "Trends in Regional Inequality in China."
NBER Working Paper 5412. National Bureau of Economic Research, Cambridge,
Mass.
Kornai, J. 1980. Economics of Shortage. Amsterdam: North-Holland.
Kraay, A. 2000. "Household Saving in China." World Bank Economic Review 14(3,
September): 545-70.
Lardy, N. 1998. China's Unfinished Economic Revolution. Brookings Institution Press.
Lee, J. 1994. "Regional Differences in the Impact of Open Door Policy on Income
Growth in China." Journal of Economic Development 19(1): 215-34.
Levine, R. 1997. "Financial Development and Economic Growth: Views and Agenda."
Journal of Economic Literature 35(2): 688-726.
22



Levine, R., N. Loayza, and T. Beck. 1999. "Financial Intermediation and Growth.
Causality and Causes." World Bank Country Economics Department Papers
2059. Washington, D.C.
Li, H., Z. Liu, and I. Rebelo. 1998. "Testing the Neoclassical Theory of Economic
Growth: Evidence form the Chinese Provinces." Economics of Planning 31: 117-
32.
Li, K., and T. Liu. 2001. "Impact of Financial Resources Liberalization in China's
Economic Growth: Provincial Evidence." Journal ofAsian Economics 12(2): 245-
62.
Mankiw, N. G., D. Romer, and D. Weil. 1992. "A Contribution to the Empirics of
Economic Growth." Quarterly Journal of Economics 107(2): 407-37.
Maskin, E., and C. Xu. 2001. "Soft Budget Constraint Theories." Economics of
Transition 9(1) :1-27.
Mody, A., and F. Wang. 1997. "Explaining Industrial Growth in Coastal China:
Economic Reforms... and What Else?." World Bank Economic Review 11 (2):
293-325.
Neely, M., and D. Wheelock. 1997. "Why does bank performance vary across states?"
Review/Federal Reserve Bank of St. Louis 79(2):27-41.
Park, A., and K. Sehrt. 2001. "Tests of Financial Intermediation and Banking Reform in
China." Journal of Comparative Economics 29(4, December):608-44.
Poncet, S. 2002. "Is China Disintegrating? The Magnitude of Chinese Provinces'
Domestic and International Integration." CERDI Etudes et Documents 5.
Qian, Y., and C. Xu. 1993. "Why China's Economic Reform Differ: the M-form
Hierarchy and Entry/Expansion of the Non-State Sector." The Economics of
Transition 1(2): 135-70.
Raiser, M. 1998. "Subsidising Inequality: Economic Reforms, Fiscal Transfers and
Convergence Across Chinese Provinces." Journal of Development Studies 34(3):
1-26.
Samolyk, K. A. 1994. "Banking Conditions and Regional Economic Performance."
Journal of Monetary Economics 34: 259-78.
Sinn, S. 1992. "Saving-Investment Correlations and Capital Mobility: On the Evidence
from Annual Data." The Economic Journal 102: 1062-1170.
Thomas, A. H. 1993. "Saving, Investment and the Regional Current-Account: An
Analysis of Canadian, British, and German Regions." IMF Working Paper 93/62.
International Monetary Fund, Washington, D.C.
Wang, Y., and Y. Yao. 2001. "Sources of China's Economic Growth. 1952-99:
Incorporating Human Capital Accumulation." Policy Research Working Paper
2650. World Bank, Development Research Group, Washington, D.C.
World Bank. 1994. China: Internal Market Development and Regulation. Washington
D.C.: World Bank.
23



Yamori, N. 1995. "The Relationship Between Domestic Savings and Investment: The
Feldstein-Horioka Test Using Japanese Regional Data." Economic Letters 48:61-
66.
24



Appendix: definition of the variables and statistical sources
Variable definition
*  GDP per capita: logarithm of real GDP per capita.
*  Investment rate: ratio of fixed investment to GDP.
*  Schooling: share of population with more than secondary schooling.
*  Non-state production: Share of non-state gross industrial output value in total gross industrial output
value.
*  Fdi: ratio of foreign direct investment to GDP.
*  SIZE: ratio of total deposits of the banking system to GDP.
*  SOB: ratio of total state-owned bank credit to GDP.
*  CENTRAL: ratio of loans to deposits of the state-owned banks.
* CONCENTRATION: Herfindqahl index of banking deposit concentration (H,,t).
H,.,t=    E       iSL  Si,,,,,
Where Sj,,,t isthe sum of deposits for bankj, i the province, t the period, and n the number of banks.
List of provinces and municipalities
Beijing, Tianjin, Hebei, Shanxi, Nei Mongolia, Liaoning, Jilin, Heilongjiang, Shanghai, Jiangsu,
Zhejiang, Anhui, Fujian, Jiangxi, Shandong, Henan, Hubei, Hunan, Guangdong, Guangxi, Hainan,
Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang, Sichuan+Chongqing.
Statistical sources
*  State Statistical Bureau, various years, Almanac of China Foreign Relations and Trade, China
Economics Publishing House.
*  Research and Statistics Department of the People's Bank of China, various years, Almanac of China
Finance and Banking, Chinese version, People's China Publishing House, Beijing.
*  State Statistical Bureau, various years, China Statistical Yearbook, China Statistical Bureau, Beijing.
*  State Statistical Bureau, 1996, China Regional Economy, a profile of 17 years of reform and opening-
up, China Statistical Bureau, Beijing,
*  Al l China Marketing Research, 2001, 1949-1999 China Statistical Data Compilation, China
Statistical Bureau, Beijing.



Malta
Switzerland
E! 2-1=ti1Ei^;X_*z$*Q*$ -                    China
Japan o
Cyprus
Thailand                    4)
Malaysia
United Kingdom
Israel
New Zealand
Euro Area
Panama
Mauritius
Egypt, Arab Rep.
Dominica
Australia
Nicaragua
Korea, Rep.                a
Philippines
Syrian Arab Republic
Canada
United States
South Africa                0
Indonesia
Denmark
Norway
India
Chile                       E
Bolivia                     0
Trinidad and Tobago         4
Jamaica                    4)
Uruguay
Nepal
El Salvador                 ,
Algeria                        a
Sweden                         EL)
Honduras
Pakistan  Gn                0
Arentia 
Iran, Islairic Rep.            a
@                                        l - |       | ~~~~~~~~    ~      ~     ~~~~~Sri Lanka 
Paraguay                    m0'
V=4~~~~~~~~~~ Peru, 
Mexa ic                    0
~~~,                  .                   , ~~~~~~~~~~:   | ~~~~~Haiti            Q  v
C   mGarbia,The
v3                                        r   ,   - W ~~~~~~~~~~~~~~~   ~    ~~Papua New Guinea  0o 
Argentina
T_'                                         I l'    ' ~~~~~~      ~       ~      ~~~~~~Lesotho  |'
a                                           & ;     l ~~~~~~       ~      ~      ~~~~~~Ecuador  8 U;
T=-,                                        I,      ' ~~~~~       ~      ~      ~~~~~~~~~~~Mexico  00t_c 
~~~>                                        l ,,    l ~~~~~~~~~~~Colombia         o  o~
im,                                         ',      ' ~~~       ~        ~      ~~~~~~~Togo  i _ 2
.                                           ' l    ' ~~~~~~~~    ~       ~      ~~~~~~Senega,'l  cB 
Guatemala                  co
Zimbabwe
b                                              t    ' ~~~~       ~       ~      ~~~~~~~~Ghana  c 
^  r  t Venezuela                     .   c
.,                                                                X   ' Central African Republic           .
LZZ) Sierra Leone
:j                                                                      Cameroon                         c
Congo, Rep.                                                                                        0  c   .2
Rwanda
.0  W  ]Malawi
- Sudan
-      -           -        C-         |       Niger
cr i.o .e         c1   c  o  cc    s     .!  s                                           lu
;_       -    -     -     -     -     0     0     o     0     o0c 



-             -          -          ,        X Malta
-              __ Switzerland
- =   =       i    Japan
*  -  -  -! ____ ~~~Italy
*  -  -             ~~~~~~Portugalo
-      -         _        t_s Spain
-      -           -= . France r
-     z_     _     Austria                  >
-       =N= Cyetherlands
*  -           ~~~~Dominica
* -      -     =    Germanv
Al&eria'
-_ ___             United States
South Africa
-  -         ~~~~Ireland
- - = ~~Sweden
-            ~~~~~Guyana               EZ
-                 ' Egypt, Arab Rep.
-     -   iii Norwa        R
-- _ i.EA Iran, Is amic Rep.
Greece
Canada
Belgium                  ei
Ma[aysia
Syrian Arab Republic      y
Pakistan                 S
Costa Rica                "
_..Denmark 
z = ~~Finlandc
LesothoE
Uruguay
Australia                 °
Panama
Thailand
India
o _ Togo                  0
Papua New Guinea
Trinidad and Tobago       E
.-                                                      Kenya                   ,.-
.2                                       ~~~~~~~~~~~~~~~Jamaica 
China
United Kingdom
Venezuela
Korea, Rep.
Sri Lanka
cis                                   =s-      ~~~~~      ~       ~     ~~~~~~Mexico  ,
ts esD ~~~~~~~~~~~~~~~~~Nicaragua                                           C a
oo ~~~~~~~~~~~~~ ~~~New Zealand 
El Salvador                  o
Haiti 
Senegal
| - Sudan w-
Zimbabwe                  o .2
C2                                                    G   _    Argentina
Phl ippines              .   ,
Ecuaaor
o                                                     z  _    Honduras
Paraguay                  8 -   ,
Israei l
Chile
Guatemala                .r ^
t                                      l =    ~~~~~~~~~~~~~~~Nenal            -
Ghiana
Gambnia, The
Cameroon
Bolivia
Sierra Leone       c
Peru
Malawi           r              .2
Central African Republic        a
:                                       ra=   ~~~~~~~~~~~Congo,Rep.            c
Colombia                  0     . 
Indonesia                 8 
..                                        r   ~~~~~~~~~~Brazil          .-  8  Ei
.0                                                       ii    Rwanda                      .
|   i Niger
-      -                   c   Congo, Dem. Rep.          c $ -
;~ ~        -   -     -    -      o     o     6     0     0                          c. 



Figure 2a: Commercial-Central Bank - International Sample (1999)
Banks (1999)
1.2
1.0
,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~                 EL
0.8
0.4       -   -   -   -   -   -   -   -   -               -   -   -   -   -   -   -   -   -   -
0.2    -   -   -   -   -                                                          3
0.0                                                                                    l i l l
-.   -. .              _             2         = = =        g 
_. ~ ~ ~ ~ ~~C  r                                           oQ
~~~ _  2  ~~~~~~."                                      2 2s   C
Commrcia-Cetralisclcua.e as th Mai of asset of deoi mone bank (line 22-d toGP(in9)
ID  " _.~     ~   ~   ~  ~   ~~~~~                       ~N.
CD
~~~~~~0                                           0~~~~~~~~~~~~~
The sample is composed of China and the 78 countries of the World Bank dataset of financial development indicators (see Beck, Demirgiig-Kunt, and Levine 2000).
Commercial-Central is calculated as the ratio of assets of deposit money banks (lines 22a-d) to GDP (line 99b).
Source: International Financial Statistics, IMF.
28



Malta
Austria
Netherlands
Germany                      g
Belgium 
Switzerland
France
South Africa>
Venezuela
Trinidad and Tobago
Finland
Japan
_ ~~~~~~~~~~~~Ireland3
-  United Kingdom
Malaysia                       .
Korea, Rep.
Denmark                      2
Norway
China
New Zealand
Canada
Cameroon
Australia
Papua New Guinea
United States
Indonesia                    o   o
Peru
__                5  Spain ._ <,,~§~pa
. __ __ C~~~~~yprus ].
Sweden
Congo, Rep.                  2 
Greece
Thailand                       0
Israel                       6   °
Italy                         >  'a
Algeria                      'o  <
_  E                   _       _   _    Lesotho                      i 
00                                s                                 _ _=Rwanda
Zimbabwe                         A                O
_n                                    .                                     . .,,^. _                 _ _Philippines  a      4
x0                                    z                                     _ _"Paraguay
cri       I                            _        _        _         _      Costa Rica                  001
U _ ~~~~~~~~~~~~~~Niger ZS
U                                      _                      _          Honduras                      f   ),
Portugal                     .    c
Senegal                          0
Panama
Kenya
_                                        E       _             _          Pakistan
Brazil
Dominica
.                                    _   _    _    _   ~~~~~~~~~~~~~Colombia             o
es                                      E    _lD         _India
El Salvador
Gambia, The                   0  I
X ES . -                              ................ Togo
_  s          _          Mauritius ................ Muitu
Argentina           ..
Uruguay                       o  L3
Central African Republic     0x
Ecuador                      r
_ _ --                                .- Sri Lanka
Mexico                               U
Guatemala
eU                                    ==-_ _; Chile                                                     Ce.
Q _ _ __ ~~~~~~~~~~~~Malawi ........ 
-                                    _                                    __ Jamaica                         ce
Nepal Arab Rep.                      U
Egypt, Arab Rep.              o  . 
X _ _ ~~~~~~~~~~~~~~Sudan ........ 
a _ ~~~~~~~~~~~~~~~~Guyana                                       ......... n 
Haiti
Ghana
Nicaragua                    0 ,
Sierra Leone .O
Syrian Arab Republic                 E
_____ _Bolivia                                                    c
00                                    0 oE
_        _         0        o        6        6         o6)



Figunre 3z: Private Caredit - lunterrnational Saimple (1999)
Private (1999)
2.0 -
1.8
1.6
1.4
1.2
1.0-                                                                                                                               - - - - - -  -
0.8
0.6                                                                                                           -    
0.4                                                                                     -    
0 .2-                                                                              - - - -  - - - - - -  - - - - -  - - - -
0.0                                                                                   ---------- ~  ~
0  -~ ~~q               .0    .                 L-
_ A
B ~ ~ ~ ~ ~   ~~~~~i                               r -o.3BEo                                                   >    z  wD -  ZCC 
The samnple is composed of China and the 78 countries of the World Bank dataset of financial development indicators (see Beck, DemirgtlS-Kunt, and Levine 2000).
Private Credit is calculated as the ratio of credit by deposit money banks and other financial institutions to the private sector (lines 22d + 42d) to GDP (line 99b).
For China, Private credit is calculated using an estimate of the share of total credit directed to state enterprises of 90% in 1985 and 77.5% in 1999 (from Economic
Intelligence Unit, February 18, 2002, "China: Grossly Distorted Product.").
Source: International Financial Statistics, IMF.
30



Figure 3b: Private Credit - International Sample (1985)
Private (1985)
0.7 -
0.6                                                                                                                                          S
0.5 
0.4-
0. -  n¢     -tnrz      -3                                        - -.r2  -no < z  - - - -  - -  ----  -     - - ----c  - - - -  ---  -tz
0.1 ~ ~      ~     ~      N   t                     iZi-'-           -uD1
=r  E CD  =,.0 CDm  C yd~- CD                               D            -  --. D . 0 0
~~  -~~~~   CD ~~~~~       n                                o                                                   CD    CD -
ForC China, Pia  c  i car                                                     CD t  t  e  o.                      i 1999.
PI  r-                   9   , ~~~~~~~~~~~~~~~~~~~DCD                                    >o
*~~~~~~~~~~~~~~~~~~ 0
0~~~~~~~~~~~~~~~~~)0
T'he sample is composed of China and the 78 countries of the World Bank dataset of financial development indicators (see Beck, Demirgii~-Kunt, and Levine 2000).
Private Credit is calculated as the ratio of credit by deposit money banks and other financial institutions to the private sector (lines 22d + 42d) to GDP (line 99b)
For China, Private credit is calculated using an estimate of the share of total credit directed to state enterprises of 90% in 1985 and 77.5% in 1999.
Source: International Financial Statistics, IMF.
31



IFigure 4: Provinciae Investment and Saving Rates
(Chinese provinces, 1985-2000, average values)
0.63                                                                      Bejing
MngKua Xinjiang
Qinghai                        T~~~~~~~~~~~~~~lanjin
Investment                                                               Shanghai
rate
Shaan,da
Shaan~ ~ ~ ~   ~~~had          Jiangsu
Shanxd
Gansu NM                        iang
I-eaHam* i
_~~~~~~~~ii Fujian Guargldo
Yunnan     Hbei
G kzhotu         iAmp)a u    fu] Hviiaw   Ua"J
Sachugj          liaonfing
Guangn
0.31    -              Hunan
0.23                                                               0.59
Saving rate
Variable definition:
Investment rate: share of gross capital formation in GDP
Saving rate: ratio of GDP minus total consumption to GDP.
Source: Al I China Marketing Research, 2001, 1949-1999 China Statistical Data Compilation, China Statistical
Bureau, Beijing. China Statistical Yearbook, 2001, China Statistical Bureau, Beijing.



Figure 5.1a: Size and Real Growth of per Capita GDP (1990-1994, average values)
Full Sample
.17979 -                                      Hainan
Zhejiang
Shandong
QL                            Hebei
w
g
Q   knhui              Shanghai
Henan
CD  ~ ~   ~   icun     Xinjiang
iunan S      infng                                                         Beijing
Tianjin
Nei Mong
G&anxi
Guizhou
Heilongj
.056381                     Ningxia
.555932                                                                  2.42567
dgdp
Restricted Sample (without the three municipalities of Beijing, Tianjin and Shanghai)
.108796 -          Fujian
Jiangsu   Hebei
AhBndong
nudel        Zhejiang
Anhui
tlunan      Henan
Nel M8^ggXI          JDm         Ningxia
Hgliongj                      S haanx i
Q           _              Sichuan                   Gansu                  Guangdon
Liaoning
0)                                                         Shanxl
Guangxi           Yunnan
ulZhOU0inghai
Xinjiang
.053677                                                                               Hainan
.636271                                                                  1.71433
dgdp
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
33



Figure 5.2a: Size aImdl Real Growth of per Capita GD1P (l995-1999, average valRes)
Full Sample
.108796 -       Fujian
Tianjin
JiangaMebel
Sa don
S     lua  ePang              Shanghai
Anhui
tunan Henan
tU                Ne    8ii       Ningxia
w                            Jilin
O                                    Shaanxi
Q.                     Heilongj
'O                   Sichuan      Gansu     Guangdon
0)                             Liaoning
0)                                  Shanxi                                                      Beijing
Guangxi Yunnan
Gull*fthai
Xinjiang
.053677 -                                       Hainan
I 
.636271                                                                        3.04873
dgdp
Restricted Sample (without the municipality of Beijing)
.108796             Fujian
Tianjin
Jiangsu    Hebei
AFlU el      Zhejiang                                                  Shanghai
Anhul
unan       Henan
Nei M8J§gxi                         Ningxia
a                                            Jilin
O                                                                Shaanxi
Heilongj
V                            Sichuan                     Gansu                    Guangdon
i)                                                 Liaoning
0)                                                             Shanxi
Guan&nxi            Yunnan
uizhouainghai
Xinjiang
.053677                                                                                      Hainan
.636271                                                                        1.71433
dgdp
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
34



Figure 5.1b: Size and Initial Level of Capita GDP -(1995-1999, average values)
Full Sample
3.04873                                                                Beijing
0.
Q
X                                            1lHainan                                      Shangha
Guangdon
Shaanf,anxi                           Tianjin
Yunnan        Xinjiang       Liaoning
Jilin
Qinghai      Hebei         Zhejiang
G uizhou                    Heilong          Jangsu
Guan4ai91Rqqn 
e' Mi°beiSha"R         ins
.636271                  Hun6phul
6.36362                                                                    9.03762
gdpcapO
Restricted Sample (without the three municipalities of Beijing, Tianjin and Shanghai)
1.71433                                                Hainan
Guangdon
Shaanxi Shanxi                                         Tianjin
Ningxia
CL                                     Gansu
Xinjiang              Liaoning
XJ                            Yunnan
Jilin
Qinghai           Hebei
Zhejiang
6 u:zhou                                Heilongi
Guangxi Sichuan                                   Jiangsu
Henal4angxJ
Nei Mon0f  b  Shandopn
.636271                        Hunanhui
6.36362                                                                    8.22575
gdpcapO
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
35



IFigure 5.2b: Size aindl Ilnitial Level of¢ Capits GDP (1L990-1994, average values)
Full Sample
2.42567                                                                  Beijing
Hainan
Shangha
Tianjin
_             ~~~~~~~~~G a n Ms ngJ I a n g
Jlin         LLiaoning
G uizhou     83n fin
.555932
6.01976                                                                     8.48298
gdpcapO
Restricted Sample (without the three municipalities of Beijing, Tianjin and Shanghai)
1.71433                                                          Hainan
Guangdon
Shaanxi
Shanxl
Ningxia
QL                                           Gansu
Xinjiang                   Liaoning
Yunnan
Jilin
Qinghai               Hebel
ZheJiang
f uizhou                                         HellongJ
Guangxi Sicihuan                                             Jiangsu
H,n,rflJiangxi
Nei Mong          Shandong
Hubeil         n? 
.636271 -                           Hunan Anhui
6.36362                                                                     7.90601
gdpcapO
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
36



Figure 6a: Sob and Real Growth of per Capita GDP
(full sample, 1990-1999, average values)
Full Sample
.140419 - Fujian
Z3ejiang
Guangdon
Hebel
Hubel                    ~~~~~~~Shanghai  Hainan
O                                  Hubigiagx
Henan           Jiangxi                                    Tianjin
0)                        Hunan
Sichuan                                          Jilin   Beijing
Liaoning      Nei Mong
_                            ~~~~~~~~Xin!!A l
Yunnan                       Gansu
HeilongJ
Ningxia
Guizhou
.063143                                                                                Qinghai
.411136                                                                   1.06752
Isobgdp
Restricted Sample (without the three municipalities of Beijing, Tianjin and Shanghai)
.140419 - Fujian
Z ejiang
Iitgrimg
Guangdon
Hebei
CL                         AiUgxi
Hubei                            ~~~~~~~~~Hainan
0.                          Henan           Jlangxi
0)                        Hunan
Jilin
Sichuan
Llaoning      Nei Mong
Yunnan                       Gansu
Heilongj
Nlngxia
Gulzhou
.063143                                                                                Qlnghai
.411136                                                                   1.06752
Isobgdp
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
37



lFigure 6b: Sob and Ilnitial Level of Capita GDD1P
(restricted sample, 1990-1999, average values)
Full Sample
1.06752 -                    Qinghal
Ningxia                          Tianjin
Hainan                                 Beijing
Jilin
Shangha
Gansu
Sbanxi
0Q                           Shaanxi
_Xnjiang
Ne M Ong
0
_            G uizhou      Jiangxi
Heilongj
Uaeoning
_Vunnan     Hubei     Guangdon
Hebei
G uangxi    ONEn
|Ic uan
.411136                        Str'gRng ZhejiarBangsu
6.01878                                                                   8.48298
gdpcapO
Restricted Sample (without the three municipalities of Beijing, Tianjin and Shanghai)
1.06752                                      Qlnghal
Ningxia
Hainan
Jilin
Gansu
0.                                         Shaanxi Shanxi
X1                                                       xinjiang
Nei Mong
.0
_              JG uizhou                Jiangxi
Hellongj
Liaonlng
Yunnan                 Hubel               Guangdon
G uangxi              plHn               e
.411136 -1                                         S   &ng           Zhejiang  Jiangsu
6.01878                                                                      7.311
gdpcapO
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
38



Figure 7a: Central and Real Growth of per Capita GDP
(full sample, 1990-1999, average values)
Full Sample
.140419-                              Fujian
Zhejiang
JiangsuShandong
Guangdon
Hebei
CL                                          Guangxi               Anhui
H aifTnghai                     Hubei
0.                                                   Henan         Jianjin
0)                                                             Hunan
E eijing                                          Sichuan
LiaoningNei Mong
Xlnjlang Shanxi   Shaanxi
Yunnan     Gansu
Hellongj
Ningxia
Guizhou
.063143                                                                  Qinghai
.531215                                                                         1.6879
Idsob
Restricted Sample (without the three municipalities of Beijing, Tianjin and Shanghai)
.140419 -       Fujian
Zhejiang
Jiangsu   Shandong
G angdon
Hebei
CL                      Guangxl                        Anhul
Hainan                                      Hubei
0.                     Henan             ~~~~~~~~~~~~Jiangxi
0)
0)                                                Hunan
Sichuan                                      Jilin
Liaoning  Nei Mong
_      Xlnj,ang  Shanx.          Shaanxi
Yunnan         Gansu
Heilongj
Ningxia
Guizhou
.063143 -                                                        Qinghai
I                   I                                        II
.859324                                                                         1.6879
Idsob
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
39



Figure 7b: Central and linitial Level of Capits G1DP
(restrieted sample, 1990-1999, average values)
Full Sample
1.6879                              Jilin
Qing&Hbel
Nel Mong
ulzh   ;|an      Heilongi  Linoning     Tianjin
Sheanaringxia
Co                ~~~~~Henan 0~dn
Shanxi                                                 Shangha
G angxi   Yn          Vang         Jiangsu
Yunnan      Fujian  ZhejIang
Guangdon
Beijing
.531215 6                                                                         8.48298
6.01878                                                                  8.48298
gdpeapO
Restricted Sample (without the three municipalities of Beijing, Tianjin and Shanghai)
1.6879                                                    Jilin
Qinghai Hubei
Nei Mong
o                                 Anhui Jiangxi
U)          Guizhou                Skhuan                            Heilongj           Liaoning
__                       ~~~~~~~Hunan
Shaanxl   Ningxia
Henan
_GansuShandong
Shanxi
G uangxi                               Haina                       Jiangsu
HeeXinjlang
Yunnan                Fujian           Zhejiang
.8593241                                                               Guangdon
6.01878                                                                     7.311
gdpcapO
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
40



Figure 8a: Concentration and Real Growth of per Capita GDP
(full sample, 1990-1999, average values)
Full Sample
.140419 -      Fujian
Zhejiang
J anghandong
Guangdon
Hebei
QL              Guang)Anhui
0                  Haininb                     Shanghai
Q           _    JLxanu              Tianjin
-o
0U)
Hunan
Sichuan                                                   Beijing
Nei Mong Liaoning
Shanxl                  Shang
Yunnan              Gansu
Hellongj
Ningxia
Guizhou
.063143                                          Qlnghal
.18045                                                                 .294679
cond
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
41



lFigure $b: Concentration and lRnitial Level of Capita GDP
(restricted sample, 1990-1999, average values)
Full Sample
.294679                                                                  Beijing
Liaoning
'O                               Qinghai  Jilin
O             -                    Nei Mong
0                                Gansu                                                       Shangha
Xinjiang   qdon
Shaanxi
G uizhou              Ningxia                           Tianjin
Hunan
Sichuan
Yunnan
Hebei
Anhui      HubeO    Zhejiang
Guangxi      Heangxl       Fing
.18045    -                                   Jiangsu
6.01 878                                                                     8.48298
gdpcapO
Restricted Sample (without the three municipalities of Beijing, Tianjin and Shanghai)
.246836                                                                                    Liaoning
Qinghai          Jilin
Nei Mong
Gansu
HeOtiOua ngdo n
Xlnjiang
Shaanxi
Ningxia
G uizhou
Hunan
Sichuan
Yunnan
Hebei
Anhui                Hubei            Zhejiang
Henan             Hainan
G uangxi                    Jiangxi        "Vng
Shanxi
.18045 -                                                                        Jiangsu
6.01878                                                                         7.311
gdpcapO
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
42



Figure 9.1: Size and Non-State Production
(1990-1994, average values)
1.45918                        Hainan
0)              Ningxia           Shaanxi                          Hebe
X nJiarQansu
Liaoning
Jilin
eiilongj
Henan                           Zhejiang
Sichuan                 Shandong
.555932                                      Hunan      Anhui
.217838                                                                      .750653
sgovns
Figure 9.2: SIZE and Non-State Production
(1995-1999, average values)
1.71433                                          Hainan
Guangdon
Shaanxi
Shanxi
Ningxia
Q.                                 Gansu
)            _ Xinjiang                                     Liaoning
Yunnan
Jilin
C inghai                                                  Hebei
Guizhou Heilongj
Sicm3A,gxi                  Jiangsu
Jiangxi          Henan
Nei Mong                 Hubei Shandong          Fujian
.636271 -                                                      Hunafnhui
.280456                                                                      .869722
sgovns
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
43



Figure 10: Sob and Non-State Production (1990-1999, average values)
1.06752 j inghai
Ningxia
Hainan
Jilin
Gansu
Shanxi
0.                                 Shaanxi
Xinjiang
Nei Mong
0
'A                        Guizhou                 Jiangxl
HeilongJ
Liaoning
Yunnan                      Hubei       Hebei        Guangdon
Guangxi Hunan .ne gy
Sichuan
.411 136                                                            Shandong    JIanWa,epang
.22114                                                                   .810187
sgovns
Figure 11: CENTRAL and Non-State Production (1990-1999, average values)
1.6879                     Jilin
nghai                                     Hubei
Nei Mong
O                                                 Jiangxi      Anhui
0   -   Gtjr8z9I          $u9nan
Ningxia   Shaanxi
Henan
Gansu                                    Shandong
Shanxi
XiHainauangxi           Hebel           Jiangsu
Xmbjaann  Yunnan
Fuiliahejiang
.859324                                                                       Guangdon
.22114                                                                   .810187
sgovns
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
44



Figure 12: Concentration and Non-State Production (1990-1999, average values)
1 .6879 -                Jlin
C inghai                               Hubel
Nei Mong
Q                                             Jiangxi     Anhui
__                                              &MA"MuRnan
Ningxia  Shaanxi
Henan
Gansu                                 Shandong
Shanxi
Hainarbuangxi         Hebei         Jiangsu
Xlnjleng  Yu~nanFuJ'a&ejiang
.859324 -                                                               Guangdon
.22114                                                              .810187
sgovns
Source: China Statistical Yearbook, Almanac of China Finance and Banking (various issues).
45



T'able la: Feldstein-Horioka Test for Capital Mobility
(Within)
Explanatory variables:
Saving rate               r2   #provinces  # obs
Dependent variables:
(l.a) Investment rate                            0.532
(0.059)***             0.79     25       400
(2.a) Investment rate - central budget financing  0.518
(0.065)F**             0.78     25       362
TablIe lb : Feldsteiin-Hlorioka Test for Capital Mobility
(2SLS)
Explanatory variables:
Share of
food in
Saving rate  consumption   r2  #p rovinces  # obs
Dependent variables:
(l.b) investment rate                            1.371                  0.03     25       400
(0.565)**
(2.b) Investment rate - central budget financing  1.340                 0.12     25       362
(0.601)**
(3.b) Saving rate *                                           -0.218    0.84     25       400
(0.094)**
All the regressions include individual and time fixed effects.
Robust standard errors are in parentheses, ¢ (t) (t¢) indicate significance at the 10 (5) (1) percent levels
* Instrumental regression.
* The r2 levels are lower in the 2SLS regressions because they account only for the explanatory variables.
In the OLS regressions, the r2 account for both the explanatory variables and the time fixed effects.
Period: 1985-2000, 26 provinces. The three municipalities of Beijing, Tianjin and Shanghai are excluded.
Due to data non-availability, Tibet is excluded from the sample.
46



Table 2a: Descriptive Statistics - Average Values (1990-99)
GDP per capita                                                                            Non-state
Growth rate  Initial (log)  SIZE  SOB CENTRAL CONCENTRATION Investmentrate Schooling Fdi       production
Beijing                 9%          7.93       274%  98%     0.53        0.29           50%         69%     6%        43%
Tianjin                 10%         7.73       117%  104%    1.29        0.22           40%         55%     9%        62%
Hebei                   12%         6.67       96%   58%     0.95        0.20           31%         43%     2%        64%
Shanxi                  8%          6.66      129%* 79%      1.03        0.18           29%         47%     1%        53%
Nei Mongolia            8%          6.71       71%   74%     1.35        0.24           29%         43%     0%        38%
Liaoning                9%          7.31       99%   61%     1.25        0.25           29%         53%     4%        53%
Jilin                   9%          6.84       92%   91%     1.69        0.24           28%         51%     2%        35%
Heilongjiang            7%          6.99       81%   65%     1.26        0.23           25%         50%     1%        33%
Shanghai                11%         8.48       145%  89%     1.03        0.23           48%         66%     7%        52%
Jiangsu                 13%         7.16       83%*  42%     0.96        0.18           32%         44%     5%        76%
Zhejiang                14%         7.01       83%   41%     0.89        0.20           36%         40%     2%        81%
Anhui                   11%         6.37       60%   52%     1.27        0.19           24%         34%     1%        60%
Fujian                  14%         6.71       74%*  42%     0.90        0.19           28%         31%     11%       78%
Jiangxi                 10%         6.46       81%* 68%      1.28        0.19           22%         35%     1%        50%
Shandong                13%         6.72       71%   43%     1.07        0.19           27%         39%     3%        67%
Henan                   10%         6.38       75%   54%     1.09        0.19           27%         44%     1%        61%
Hubei                   10%         6.71      74%*   59%     1.42        0.19           27%         41%     2%        54%
Hunan                   9%          6.38       60%   52%     1.23        0.21           23%         40%     2%        55%
Guangdong               12%         7.08      155%* 59%      0.86        0.23           37%         39%    12%        75%
Guangxi                 11%         6.02       83%*  53%     0.97        0.19           25%         34%     3%        50%
Hainan                  11%         6.68      171%* 98%      0.98        0.19           47%         41%    14%        45%
Guizhou                 7%          6.02       78%   68%     1.25        0.22           25%         23%     0%        32%
Yunnan                  8%          6.34      1t0%* 59%      0.91        0.20           30%         23%     0%        33%
Shaanxi                 8%          6.53       115%  78%     1.20        0.22           31%         42%     2%        39%
Gansu                   8%          6.61       106%  83%     1.06        0.23           29%         32%     1%        34%
Qinghai                 6%          6.60      98%* 107%      1.41        0.24           39%         27%    0%*        22%
Ningxia                 7%          6.69       111%  103%    1.21        0.22           40%         38%     0%        31%
Xinjiang                8%          6.76       103%  75%     0.94        0.23           42%         37%    0%*        26%
Sichuan+Chongging**     9%          6.41       76%   51%     1.25        0.21           27%         34%     1%        52%
* averaged over 1995-1999
**Chongqing city was given a municipality status in 1997, and was before this date part of Sichuan province. The statistics of Sichuan province and
Chongqing city were therefore aggregated from 1997 onwards.
Due to data unavailability, Tibet is excluded from the sample.
Source: see data description in appendix.



Table 21b: ]Descriptive Statistics: Correlations
Banking Indicators        Control Variables
z
0
H 
co  o  _      S    _     DS -  o
Investment rate                                     0.94  0.85 0.96
Control Variables  Schooling                                                0.64 0.81
Non-state production                                          0.93
SOB                       0.33  0.96 0.95     -0.76 -0.50 -0.92 -0.91
Banking            SIZE                           0.04 0.03      0.32  0.59  -0.16 0.04
Indicators         Central                             0.99      -0.91 -0.72 -0.93 -0.98
CONCENTRATION                                 -0.92 -0.74 -0.96 -0.99



Table 3a: Financial Intermediation and Economic Growth: Augmented Solow Model
(GMM-System)
la              2a              3a              4a              5a
Initial GDP per capita       1.059   ***     0.950  ***      0.979   ***     1.008   **      1.015   ***
0.038           0.029           0.043           0.078           0.053
Investment rate              0.208           0.317           0.132           0.373           0.393
0.164           0.170           0.206           0.242           0.335
Schooling                    0.343          0.235            0.344           0.030           0.130
0.457           0.186           0.220           0.522           0.427
SIZE                        -0.148  **-0.044
0.033                                                           0.053
SOB                                         -0.207  ***
0.045
CENTRAL                                                     -0.243  ***
0.075
CONCENTRATION                                                               -2.151   ***    -1.625   ***
0.653           0.529
Sargan Test                21.580          22.030          17.740          19.490           19.260
AR(2) Test                 -1.046          -1.248          -0.966           -1.401          -1.199
Observations               98             104             104             101               98
Provinces                  26              26              26              26               26
Beta-convergence rate*:    ns               2.6%      *    ns              ns               ns
ns: non significant.
The regressions are panel regressions, which include time, fixed effects, with data averaged over 2-year period from 1990-
1999, and using lagged differences and level values as instruments, as described in the text.
The null hypothesis of the Sargan test is that the instruments are not correlated with the residuals. The null hypothesis of the
serial correlation test is that the errors in the first-difference regression exhibit no second-order serial correlation.
Robust standard errors are in parentheses, * (**) (***) indicate significance at the 10 (5) (1) percent levels.
*Beta convergence significance is assessed by testing whether the coefficient of the initial GDP per capita is significantly
smaller than unity. The convergence rate is calculated by applying the following formula: -ln(oco)/T, where T=2 is the time
spell and ao is the coefficient of the initial GDP per capita.
49



Table 3b: Financial lIntermediation and Economic Growth: Controlling for the State Sector
(GMM-System)
lb             2b              3b             4b              5b
Initial GDP per capita         0.976 *        0.907 ***       0.908   *       0.930 *c*      0.935    *
0.039           0.035          0.029           0.046          0.045
Investmentrate                 0.172          0.208           0.148           0.354 *'*      0.231  *'*
0.180           0.158          0.216           0.136          0.199
Schooling                     -0.025          0.312           0.552 t         0.191          0.269
0.323          0.206           0.230           0.395           0.445
Non-state production           0.237 *        0.226 *         0.208           0.198 ***      0.209 *
0.078          0.086           0.077           0.096           0.069
SIZE                          -0.010                                                         0.021
0.051                                                          0.053
SOB                                          -0.084
0.083
CENTRAL                                                      -0.150  ' 
0.046
CONCENTRATION                                                                -1.479 **       -1.213  *
0.702           0.462
Sargan Test                  20.130          20.360          17.830          17.120         17.400
AR(2) Test                    -1.020         -1.245          -1.218          -1.262          -1.180
Observations                  98             104             104             101            98
Provinces                     26             26              26             26              26
Beta-convergence rate:        ns              4.9%   '*       4.8%           ns              ns
ns: non significant.
The regressions are panel regressions, which include time, fixed effects, with data averaged over 2-year period from 1990-
1999, and using lagged differences and level values as instruments, as described in the text.
The null hypothesis of the Sargan test is that the instruments are not correlated with the residuals. The null hypothesis of the
serial correlation test is that the errors in the first-difference regression exhibit no second-order serial correlation.
Robust standard errors are in parentheses, * (**) (e**) indicate significance at the 10 (5) (1) percent levels.
*Beta convergence significance is assessed by testing whether the coefficient of the initial GDP per capita is significantly
smaller than unity. The convergence rate is calculated by applying the following formula: -ln(ao)/T, where T=2 is the time
spell and ao is the coefficient of the initial GDP per capita.
50



Table 3c: Financial Intermediation and Economic Growth: Controlling for Foreign Direct
Investment
(GMM-System)
Ic            2c             3c             4c            5c
Initial GDP per capita         0.984  ***    0.903   ***    0.929  ***    0.906   ***    0.970   ***
0.036          0.039         0.052          0.058         0.050
Investment rate                0.431    *    0.377     *    0.269          0.242         0.320
0.243          0.176         0.210          0.163          0.233
Schooling                     -0.013         0.279          0.575    *    0.126          0.057
0.202          0.253         0.307          0.385          0.329
Non-state production           0.368  ***    0.154   ***    0.219  ***     0.227  ***    0.280   ***
0.067          0.073         0.076          0.071          0.076
Foreign direct investment     -0.833        -0.069         -0.314          0.193         -0.628
0.426          0.378         0.437          0.456          0.640
SIZE                          0.046                                                      0.093
0.057                                                      0.064
SOB                                         -0.197    **
0.092
CENTRAL                                                    -0.182    **
0.071
CONCENTRATION                                                             -0.926         -1.356   **
0.619          0.666
Sargan Test                  19.780         19.490         17.020         16.480        15.960
AR(2) Test                   -1.304         -1.265         -1.207         -1.198         -1.289
Observations                 96            102            102            99             96
Provinces                    26             26             26            26             26
Beta-convergence rate:       ns              5.1%          ns             4.9%          ns
ns: non significant.
The regressions are panel regressions, which include time, fixed effects, with data averaged over 2-year period from 1990-
1999, and using lagged differences and level values as instruments, as described in the text.
The null hypothesis of the Sargan test is that the instruments are not correlated with the residuals. The null hypothesis of the
serial correlation test is that the errors in the first-difference regression exhibit no second-order serial correlation.
Robust standard errors are in parentheses, * (**) (***) indicate significance at the 10 (5) (1) percent levels.
*Beta convergence significance is assessed by testing whether the coefficient of the initial GDP per capita is significantly
smaller than unity. The convergence rate is calculated by applying the following formula: -ln(a0)/T, where T=2 is the time
spell and cN is the coefficient of the initial GDP per capita.
51



Tsable 4a: IFksmcial Thmntermnedisadoin and EconomAe GgrowtDn: ARtenmadve Estnmatows (DiDf-GMM  nmd Watlhi)
Diff     Diff      Diff      Diff      Diff      within    within    within    within    within
1         2         3         4         5         1'        2'        3'        4'        5'
InitialGDPpercapita          0.521 *** 0.538*** 0.597*o*    0.563*** 0.527*** 0.536o**   0.523t**   0.582*** 0.5564**  0.570***
0.092      0.071     0.067     0.086     0.076     0.074     0.081     0.064     0.082     0.064
Investnent rate              0.251     0.243     0.204     0.260     0.193     0.310**   0.361t*   0.279*t   0.283"t   0.295""'
0.226      0.174     0.099     0.190     0.153     0.132     0.095     0.054     0.091     0.093
Schooling                    0.288     -0.039    0.077     0.096     0.225     0.383     0.183     0.283     0.222     0.211
0.467      0.409     0.335     0.441     0.344     0.443     0.303     0.249     0.350     0.349
Non-state production        -0.018     -0.043    0.016    -0.005      0.013    -0.040   -0.050     -0.024    -0.029   -0.032
0.095      0.098     0.088     0.120     0.086     0.105     0.098     0.087     0.109     0.097
SIZE                        -0.056                                   -0.060    0.043                                   0.039
0.072                                    0.057     0.083                                   0.071
SOB                                    -0.152**                                          -0.198 t
0.062                                             0.065
CENTRAL                                          -0.133 t                                          -0.165 "t4
0.004                                             0.047
CONCENTRATION                                              -0.565    -0.480                                  -0.738 **  -0.823 *
0.349     0.309                                   0.319     0.379
Sargan Test                 21.000    19.000    20.300    20.540     19.630
AR(2) Test                  -1.494     -1.684    -1.516   -1.528     -1.448
Observations                  98         104       104       101        98        98       104       104       101        98
Provinces                     26          26        26        26        26        26        26        26        26        26
R2                                                                              0.98       0.98      0.99      0.99      0.98
Beta-convergence rate:       33%  t     31%**     26%***    29%***    32%***    31%***    32%t**    27%***    29%***    28%***
The regressions are panel regressions, which include time fixed effects., with data averaged over 2-year period from 1990-1999
Robust standard errors are in parentheses, * (**) (***) indicate significance at the 10 (5) (1) percent levels
*Beta convergence is assessed by testing whether the coefficient of the initial GDP per capita is significantly smaller than unity. The convergence rate is calculated by
applying the following formula: -ln(cO)/T, where T=2 is the time spell and aO is the coefficient of the initial GDP per capita.



Table 4b: Financial Intermediation and Economic Growth: Alternative Estimators
(OLS-Level)
I            2"          31"          4"            5"t
Initial GDP per capita         0.963 *      0.952 ***    0.959 ***   0.988 ***     0.989
0.019        0.019        0.019       0.014         0.015
Investment rate                0.224***     0.294***     0.134***    0.169***      0.190***
0063         0.100        0.047        0.048        0.059
Schooling                      0.099        0.102       0.105        0.062         0.083
0.081        0.092        0.105        0.075        0.072
Non-state production           0.260 ***    0.204 ***    0.236 ***   0.196 ***     0.186 ***
0.027        0.043        0.036        0.032        0.034
SIZE                          -0.022                                             -0.014
0.016                                               0.011
SOB                                        -0.094 ***
0.047
CENTRAL                                                -0.056 ***
0.045
CONCENTRATION                                                        -0.822 ***   -0.830 ***
0.223        0.242
Observations                     98          104          104          101           98
Provinces                        26           26           26           26           26
R2                             0.99         0.99         0.99         0.99         0.99
Beta-convergence rate:         1.9%        2.5%          2.1%           ns           ns
The regressions are panel regressions, which include time fixed effects., with data averaged over 2-year period from 1990-
1999
Robust standard errors are in parentheses, * (**) (***) indicate significance at the 10 (5) (1) percent levels.
ns: non significant
*Beta convergence is assessed by testing whether the coefficient of the initial GDP per capita is significantly smaller than
unity. The convergence rate is calculated by applying the following formula: -ln(aO)/T, where T=2 is the time spell and aO is
the coefficient of the initial GDP per capita.



TablRe 5a: IFinancizl llnteirmediation and Economic Growth: Without ]Investment
1             2             3            4              5
Initial GDP per capita               0.985 **     0.967*        0.937         0.977 **      0.921
0.0411         0.033         0.033         0026          0.052
Schooling                           -0.045        0.154         0.446 *      -0.098         -0.132
0243          0.127         0.210         0.185         0.476
Non-state production                 0.244**      0.244***      0. 1 80***    0.175         0.215
0.095         0.093         0.046         0.079         0.104
SIZE                                0.015                                                   0.045
0.069                                                   0.066
SOB                                               0.033
0.077
CENTRAL                                                        -0.146 4*
0.051
CONCENTRATION                                                                 -1.045        -1.529
0.382         0.579
Sargan Test                        22.450        21.720        19.060        21.300        21.450
AR(2) Test                         -1.168        -1.320        -1.146         -1.131        -1.140
Observations                           98           104           104           101            98
Provinces                              26            26            26            26            26
R2                                   0.99          0.99          0.99          0.99          0.99
beta-convergence rate:                           -1.7%
t(10%,5%, 1%)= 1.66 (2.275, 2.625)  -0.365       -1.000     *  -1.909         -0.885        -1.519
Initial GDP per capita < 1         -0.015        -0.033        -0.063         -0.023        -0.079
ns: non significant.
The regressions are panel regressions, which include time fixed effects., with data averaged over 2-year period from 1990-
1999
Robust standard errors are in parentheses, ° (¢) (8"t) indicate significance at the 10 (5) (1) percent levels.
54



Table 5b: Financial Intermediation and Economic Growth: With Coastal Dummy
1           2           3           4          5
Initial GDP per capita             0.887***    0.800***    0.876***    0.896      0.914
0.079       0.077       0 086       0.077      0.070
Investment rate                    0.237       0.261 **    0.130       0.290      0.251
0.207       0.115       0.057       0 229      0234
Schooling                          0.354       0.434       0.483       0.091      0.265
0.081       0.298       0.353       0.291      0.399
Non-state production               0.217***    0.176       0.226***    0.199**    0.197
0.081       0.118       0.067       0.078       0.081
SIZE                              -0.046                                          0.016
0.074                                          00160
SOB                                           -0.101
0.075
CENTRAL                                                   -0.140 **
0.057
CONCENTRATION                                                         -0.858*    -1.074
0.506      0538
Coastal dummy                     0.030       0.078        0.009       0.042      0.013
0.053       0057        0 057       0.063       0.061
SarganTest                       20.320       16.540      21.350      19.240     18.760
AR(2) Test                       -1.222       -1.181      -1.197      -1.097     -1.185
Observations                         98         104         104         101          98
Provinces                            26          26          26          26         26
R2                                 0.99        0.99         0.99        0.99       0.99
beta-convergence rate:                       -11.2% *
t (10%,5%, 1%) = 1.66 (2.275, 2.625)  -1.430  -2.597      -1.442      -1.351     -1.229
Initial GDP per capita < 1       -0.113       -0.200      -0.124      -0.104     -0.086
ns: non significant.
The regressions are panel regressions, which include time fixed effects., with data averaged over 2-year period from 1990-
1999
Robust standard errors are in parentheses, * (**) (***) indicate significance at the 10 (5) (1) percent levels.
55



Table 6: IFinancial Iitermedi aion and Ecoiomnic Growth: Two 5-years Periods
within   within   within  within   within    gis      gis      gis      gis     gis
la       2a       3a      4a       5a       lb       2b      3b       5b       6b
Initial GDP per capita          -0.034    0.147    0.245 **  0.228  0.001    0.922 **  0.922 *** 0.894 ** 1.056 *** 1.070
0.114    0.094    0.098   0.140    0.154    0.073    0.064    0.059   0.054    0.062
-r.7109   .       t.241    t.14%   -.M      t.A%     Z.M9     %.3S    - O\%     A.M
1.030    0.948    0.811   1.045    0.220    0.302    0.270   0.263    0.199    0.224
Non-state production            -0.543   -0.224   -0.227  -0.206   -0.528    0.484*** 0.515*** 0 444**  0.299*  0 191
0.329    0.273    0.260   0.325    0.344    0.128    0.151   0.107    0.105    0.122
Time fixed effect                0.670** 0.337     0.253   0.324**  0.633   -0.075   -0.091 *** -0.015*** -0.170*** 4.4173
0.116    0.071    0.080   0.095    0.158    0.039    0.034    0.037   0 034    0.041
SIZE                            40.634**                            40.587** 0.018                             -0.006
0.203                              0.249    0.078                              0.057
SOB                                      -0.325                                      -0.061
0.142                                       0.119
CENTRAL                                           -0.254                                      -0.226
0.090                                       0.075
CONCENTRATION                                              -1.036  -0.414                              -2.305  -2.660
0.941    1.173                              0.706    0.767
Observations                       43       52       52       50       43      43       52       52       50       43
Provinces                          26       26       26       26       26      26       26       26       26       26
R2                                0.00    0.46     0.55     0.46    0.00     0.96     0.96     0.96     0.97    0.97
ns: non significant.
The regressions are panel regressions, which include time fixed effects., with data averaged over 5-year period between 1990 and 1999
Robust standard errors are in parentheses, * (**) (***) indicate significance at the 10 (5) (1) percent levels.



Policy Research Working Paper Series
Contact
Title                           Author                  Date             for paper
WPS2996 Bank Competition, Financing      Thorsten Beck           March 2003        K. Labrie
Obstacles, and Access to Credit  Asl1 Demirgu9-Kunt                       31001
Vojislav Maksimovic
WPS2997 Financial and Legal Institutions  Thorsten Beck          March 2003        K. Labrie
and Firm Size                   Asl1 Demirgui-Kunt                        31001
Vojislav Maksimovic
WPS2998 Does Micro-Credit Empower Women" Mark M. Pitt            March 2003        P. Kokila
Evidence from Bangladesh        Shahidur R. Khandker                      31001
Jennifer Cartwright
WPS2999 Explaining Liberalization         Philipp Harms          March 2003        P Flewitt
Commitments in Financial        Aaditya Mattoo                            32724
Services Trade                  Ludger Schuknecht
WPS3000 International Climate Regime beyond  Franck Lecocq       March 2003        V Soukhanov
2012 Are Quota Allocations Rules  Renaud Crassous                         35721
Robust to Uncertainty?
WPS3001 An Introduction to Financial and  Antonio Estache        March 2003        G Chenet-Smith
Economic Modeling for Utility   Martin Rodriguez Pardina                  36370
Regulators                      Jose Maria Rodriguez
German Sember
WPS3002 Information-Based Instruments for  Uwe Deichmann         March 2003        Y D'Souza
Improved Urban Management       Somik V. Lall                             31449
Ajay Sun
Pragya Rajoria
WPS3003 The Investment Climate and the    Mary Hallward-Driemeier  March 2003      P. Sintim-Aboagye
Firm Firm-Level Evidence from China Scott Wallsten                        37644
Lixin Colin Xu
WPS3004 Institutions, Trade, and Growth   David Dollar           March 2003        A. Bonfield
Revisiting the Evidence         Aart Kraay                                31248
WPS3005 Vouchers for Basic Education in  Varun Gauri             March 2003        H. Sladovich
Developing Countries: A Principal-  Ayesha Vawda                          37698
Agent Perspective
WPS3006 Social Rights and Economics: Claims Varun Gauri          March 2003        H. Sladovich
to Health Care and Education in                                           37698
Developing Countries
WPS3007 The Impact of Urban Spatial Structure Antonio M. Bento   March 2003        V. Soukhanov
on Travel Demand in the United  Maureen L Cropper                         35721
States                          Ahmed Mushfiq Mobarak
Katja Vinha
WPS3008 Technology, Computers, and Wages  Chris N Sakellariou    March 2003        N. Vergara
Evidence from a Developing Economy Harry A Patrinos                       30432
WPS3009 Agriculture in the Doha Agenda    Patrick Messerlin      April 2003        P Flewitt
32724
WPS3010 Vulnerability in Consumption,     Edmundo Murrugarra     Apnl2003          E Murrugarra
Education, and health Evidence from  Jose Signoret                        34452
Moldova during the Russian Crisis



Policy Research Working Paper Series
Contact
Title                           Author                  Date             for paper
WPS3011 Renegotiation of Concession      J. Luis Guasch          April 2003        J. Troncoso
Contracts in Latin America      Jean-Jacques Laffont                      37826
Stephane Straub
WPS3012 Just-in-Case Inventories. A Cross-  J. Luis Guasch       April 2003        J Troncoso
Country Analysis                Joseph Kogan                              37826
WPS3013 Land Sales and Rental Markets in  Klaus Deininger        April 2003        M. Fernandez
Transition: Evidence from Rural  Songqing Jin                             33766
Vietnam
WPS3014 Evaluation of Financial Liberalization: Xavier Gine      April2003         K. Labrie
A General Equilibrium Model with  Robert M. Townsend                      31001
Constrained Occupation Choice
WPS3015 Off and Running? Technology, Trade, Carolina Sanchez-Paramo April 2003     H. Sladovich
and the Rising Demand for Skilled  Norbert Schady                         37698
Workers in Latin America
WPS3016 Partisan Politics and            Stuti Khemani           April 2003        H. Sladovich
Intergovernmental Transfers in India                                      37698
WPS3017 Why is Unemployment so High      Jan Rutkowski           April 2003        J. Rutkowski
in Bulgaria?                                                              84569
WPS3018 Integrating the Least Developed   Paul Brenton           Apnl 2003         L. Tabada
Countries into the World Trading                                          36896
System: The Current Impact of EU
Preferences under Everything but Arms
WPS3019 Foreign Aid, Conditionality, and Ghost Thilak Ranaweera  April 2003        P. Rathan Raj
of the Financing Gap A Forgotten                                          33705
Aspect of the Aid Debate
WPS3020 Consumption, Health, Gender,     Anne Case               April 2003        D. Ballantyne
and Poverty                     Angus Deaton                              87198
WPS3021 Institutional Arrangements for Public  Elizabeth Currie  April 2003        M Rosenquist
Debt Management                 Jean-Jacques Dethier                      82602
Eriko Togo
WPS3022 Sending Farmers Back to School   Gershon Feder           April 2003        P. Kokila
The Impact of Farmer Field Schools  Rinku Murgai                          33716
in Indonesia                    Jaime B. Quizon
WPS3023 Productivity Growth and Product  Douglas M Addison       April 2003        D. Addison
Variety: Gains from Imitation and                                         31188
Education
WPS3024 R&D and Development               Daniel Lederman        April 2003        P. Soto
William F. Maloney                        37892
WPS3025 Trade Structure and Growth        Daniel Lederman        April 2003        P. Soto
William F. Maloney                        37892
WPS3026 Strategic Approaches to Science   Robert Watson          April 2003        R Watson
and Technology in Development   Michael Crawford                          36965
Sara Farley