the effect of ownership on the prudential behavior of...
TRANSCRIPT
1
The effect of ownership on the prudential behavior of banks—the case of China
Chunxin Jia*
（October 30, 2006）
Abstract
Although the relationship between bank ownership and performance is the current
focus of much research, this paper investigates the relationship between ownership
and the prudential behavior of banks. Using Chinese data, I show that lending by
state-owned banks has been less prudent than lending by joint-equity banks, but has
improved over time. This is consistent with the hypothesis that accountability to
shareholders and depositors gives joint-equity banks a better incentive than
state-owned banks to engage in prudent lending, and with the hypothesis that the
reform of the banking system has improved the incentive for state-owned banks to
behave more prudently in their lending.
JEL classification: G21; G28, G34;O53
Keywords: Bank; Governance; Bank portfolio allocation; Bank prudence.
Address for correspondence: Guanghua School of Management, Peking University, 5 Yiheyuan Road,
Beijing, 100871, China. Phone: 86-10-62757795; Fax: 86-10-62751463; E-mail address:
[email protected]. This work was supported by Grant 70373012 from the National Natural Science
Foundation of China. I thank Hongbin Cai, Fengqi Cao, Philip Dybvig, for their helpful comments.
2
1. Introduction
Many countries have experienced the state ownership of banks at some point in
their history, and some have experimented with the privatization of state-owned banks
in recent decades. The relationship between bank ownership and performance is well
studied in the literature, but the findings are mixed. Clarke et al. (2005), for example,
find that although some measures of bank performance have improved as a result of
privatization, others have registered little change or may even have deteriorated. This
may be explained by the fact that privatization usually takes a long time to yield gains,
because more time may be required by management to overcome the organizational
inertia and resistance to change that are common characteristics of newly privatized
firms (Otchere, 2005). Another possible reason is that bank performance measures are
notoriously noisy, leading to less than robust results.
Using Chinese data, I attempt to investigate the effects of ownership on bank
lending and prudential behavior. I compare state-owned banks and joint-equity banks
in China, which together make up the main part of the Chinese banking sector.
Commercial banks are supposed to operate prudently, and failure to obey this basic
rule may result in long-term disastrous outcomes, such as mounting non-performing
loans and even closure. Hence, risk-taking behavior measures maybe better indicators
of bank long-term improvement than short-term performance measures, such as
annual returns. Moreover, risk-taking behavior can be much easier measurement than
performance measures, especially in developing countries, like China. The
3
identification of a difference in the level of risk-taking behavior between
government-owned and privatized banks would provide strong evidence of the
benefits of bank privatization. Nevertheless, despite the potential informativeness of
this relationship, previous research seems to focus only on performance.
China’s state-owned banks did not begin to be privatized until 2005, but
joint-equity banks were in existence as early as 1986. As stock-holding companies,
joint-equity banks have a different ownership structure than state-owned banks, but
are quite similar to privatized state-owned banks on several counts. First, bank
privatization in transition economies is usually organized by the government, and
joint-equity banks in China were set up by the government. Second, governments
often retain control over or a minority stake in privatized banks (Beck, Cull, and
Jerome, 2005; Clarke et al., 2005), and similarly in China most of the controlling
shareholders of the joint-equity banks are state-owned enterprises (SOEs). Thus, a
study of the difference in the level of prudence between state-owned banks and
joint-equity banks will contribute to the literature on privatization.
To identify whether joint-equity banks are likely to be more prudent than
state-owned banks, I use portfolio allocation data, such as the bank excess reserves
ratio, loan/asset ratio, and deposit/loan ratio as measures of bank prudence. The
empirical results show that joint-equity banks tend to be significantly more prudent
than state-owned banks, and regardless of the control variables that are used, the
difference is significant at the one-percent level. Annual firm-level panel data and
quarterly macro data show exactly the same pattern. In addition, the four state-owned
4
banks in China have been carrying out reforms in line with the economic reform
policy, and I find that they have become more prudent over the years as a result1.
The relationship between the constraints on corporate decision-makers and
behavior that maximizes firm value is a popular topic in corporate governance
research, but few studies focus on the way in which these constraints affect
managerial risk taking. Yeung, Litov and John (2006) examine the relationship
between investor protection and the incentive of corporate insiders to take
value-enhancing risks. Studying the banking sector, Saunders, Strock, and Travlos
(1990) find stockholder-controlled banks to have an incentive to take greater risks
than managerially controlled banks. Gorton and Rosen (1995) discuss the relationship
between insider stock ownership and bank risk taking, and Berger and Udell (1994),
Hellmann, Murdock, and Stiglitz (2000), Milne (2002), and Shrieves and Dahl (2003)
focus on the relationship between risk-based capital regulation and bank portfolio
choice (prudential behavior). However, all of these studies focus on private banks, and
it was not until recently that scholars began to investigate prudential behavior and
bank government ownership. Berger et al. (2005) test bank portfolio difference in
Argentina in the 1990s, and although they find that banks did lend more prudently
after privatization, this is not the main thrust of their research. Haber (2005) finds that
banking sector reform in Mexico during the period 1997-2003 resulted in less bank
lending, but uses only total loan numbers as the proxy for prudential behavior and
1 Many of the studies on bank ownership and performance investigate the selection effect to identify the kinds of state-owned banks that tend to be privatized. However, in China state-owned banks did not begin to be privatized until 2005, and thus insufficient data are available to test the selection effect.
5
regards bank prudence as a negative, rather than a positive, sign of bank reform.
Why are joint-equity banks likely to be more prudent than state-owned banks?
One possible reason is that joint-equity banks have better corporate governance.2
Berger et al. (2005) and others explicitly refer to bank ownership and performance as
a corporate governance topic. State-owned banks, in contrast, are less monitored by
their owners, as is the case with most SOEs. Alchian (1965) argues that all citizens
can be considered SOE owners, but that as they have no way of selling their “share,”
the level of monitoring in the public sector is sub-optimal. Vining and Boardman
(1992) further argue that the monitoring of SOEs is no more effective than in the
private sector, and can in fact be worse. Furthermore, state-owned banks may be less
subject to monitoring by depositors, because in a financial crisis they are more likely
to be bailed out by the government, which gives depositors less inventive to pay
attention to what is going on. Joint-equity banks, in contrast, are faced with a higher
probability of bank runs, and thus maintain more reserves and lend less, which is a
moral hazard issue.3 The possibility of a takeover, such as the takeover in 2004 of the
Shenzhen Development Bank by a foreign investor, also promotes efficient and
prudent management behavior in joint-equity banks, a threat that does not exist for
state-owned banks.
China has a very large banking sector, but has been largely ignored as a context
2 Unlike studies of US corporations, studies of governance in developing nations often focus on the role of ownership in reducing agency problems, because weak legal infrastructures often do not adequately protect investors (Berger et al., 2005) 3 As most controlling holders of joint-equity banks are SOEs, the difference in bailing out may not be serious. For example, when the Hainan Development Bank was closed in 1998 – the only joint-equity bank to do so – all of the deposit was guaranteed by the government.
6
for studies, such as on bank ownership and performance,4 even though it is the largest
of the transition economies. Chen, Li, and Moshirian (2005) studied ownership and
performance in Chinese banking firms, but limited their investigation to the Bank of
China in Hong Kong. The possible reason for the lack of research on the Chinese
banking sector may be the scarcity and poor quality of the data, which are limited in
availability and widely doubted in the areas of nonperforming loans, return on assets,
and return on equity.5 I have tried my best in this paper to minimize the effect of poor
data on the quality of the research by hand-collecting most of the data and using bank
assets and liability data to probe the differences in risk-taking behavior. Data that
might be doubtful, such as non-performing loans, returns on assets, and returns on
equity, are omitted from the empirical survey.
This paper makes three main contributions. First, it adds to the literature on the
effect of ownership on the prudential behavior of banks and highlights that differences
in the prudential behavior of banks provide an alternative to performance to test the
effects of a change in bank ownership. Second, it is the first paper to study ownership
4 Journal of Banking and Finance 29, 2005, is a special issue on this topic that covers many countries. 5 For example, according to the Web site of the China Banking Regulatory Commission, the proportion of non-performing loans in the domestic banking sector at the end of 2003 was 17.8%, but in its Asia Pacific Banking Outlook 2004, S&P announced its estimate of this proportion to be 45% (People’s Daily, September 9, 2003). Furthermore, the writing-off of non-performing loans seems quite arbitrary. Taking the Bank of China as an example, in its annual reports the adjusted operating profit before provisions was stated to have risen from 53,043 million RMB in 2002 to 58,505 million RMB in 2003, yet at the same time its returns on assets dropped from 0.28% to 0.12% and its return on equity dropped from 4.34% to 2.26%. The main reason for this discrepancy is that for future listing purposes, the bank wrote off many non-performing loans that had been accumulating for some time. In 2002, only 30,100 million RMB was written off, but in 2003 the figure was 85,045 million RMB.
7
and bank portfolio allocation using China as the background, and to report a
significant difference in the level of prudence between state-owned banks and
joint-equity banks. Third, differences in ownership structure notwithstanding, the
difference in bank portfolio allocation between banks is an interesting topic, and this
paper gives insight into the main reasons for this difference.
The remainder of the paper is organized as follows. Section 2 presents a brief
introduction to the Chinese banking sector. Section 3 describes the data and the
methodology. Section 4 presents the annual firm-level panel data regression results.
Section 5 provides the quarterly macro data regression results. Section 6 concludes the
paper.
2.China’s Banking Sector
Until the initiation of economic reform in 1978, there was only one bank in China –
the People’s Bank of China (PBOC) – which combined the roles of central and
commercial banking. Between 1979 and 1984, the Agricultural Bank of China (ABC),
the People’s Construction Bank of China (PCBC), the Bank of China (BOC), and the
Industrial and Commercial Bank of China (ICBC) split off from the PBOC, which
retained its central bank role. The four banks were known as specialized banks
because they had their own designated business, the ICBC focusing on financing in
urban areas, for example, and the ABC concentrating on rural areas. Since 1984, the
banking system has been undergoing reform, with the four banks aiming to become
8
real enterprises, and since 1985 they have been permitted to engage in business
outside of their designated economic sector. In 1994, three policy banks were
established to take over the policy loans of the four state-owned banks, as they were
now known as commercial banks and were supposed to be operating according to
market principles. The state-owned banks have also been subject to managerial and
mechanistic reform. For example, by increasing the managerial responsibility for risk,
the importance of risk management has been reinforced, and the lending behavior of
the banks has become more prudent. Until 2004, the four banks were still SOEs with
one owner – the Chinese government – but in 2005 they began to be privatized
through the recruitment of strategic investors and listing on the stock exchange.
To increase competition in the banking sector, the Chinese government began to
establish new banks, known as joint-equity banks, in 1986, the first of which was the
Bank of Communication. By the end of 2004, five of the eleven joint-equity banks
were listed on China’s stock exchanges. However, despite their joint-equity status,
most of these banks are still indirectly controlled by the government, because the
largest shareholders are usually SOEs. For example, before 2003, the local
government of Shenzhen controlled the Shenzhen Development Bank, which was
founded in 1987, through the Shenzhen City Bureau of National Assets. The only
genuinely private bank that is wholly owned by private shareholders in China is the
China Minsheng Bank.
Although most of the large shareholders of joint-equity banks are still SOEs, the
difference between the two types of banks is obvious. First, whereas the government
9
controls 100% of the state-owned banks, the stock control of the joint-equity banks is
fairly diversified. At the end of 2004, the controlling ratio of the five largest
shareholders of the five listed banks, namely, the Shenzhen Development Bank (listed
in 1991), the Pudong Development Bank (listed in 1999), the China Minsheng Bank
(listed in 2000), the China Merchant Bank (listed in 2002), and Huaxia Bank (listed in
2003), were 25%, 26%, 29%, 38%, and 49%, respectively. Second, the procedure for
nominating the governor is quite different. The Central Committee of the Communist
Party of China nominates the governors of the state-owned banks, whereas the
governors of the joint-equity banks are nominated by the board of directors. Third,
jobs in the state-owned banks are fairly stable, whereas in the joint-equity banks they
are highly unstable.
At the end of 2004, the assets of depositary institutions in China totaled
31,598.98 billion RMB, of which 53.6% was held by the four state-owned banks and
only 14.9% was held by the eleven joint-equity banks. The remainder was held by city
commercial banks, policy banks, city credit unions, and rural credit unions.
Commercial banks in China are regulated by the Commercial Banking Law (1995),
which is enforced by the People’s Bank of China and the China Banking Regulatory
Commission (founded in 2003). There is not much difference between the
state-owned banks and the joint-equity banks as regards regulation and supervision.
All bank interest rates in China must be within a range that is designated by the PBOC,
although this range has been substantially broadened of late. Furthermore, although
founded in accordance with commercial banking law, banks in China must achieve a
10
minimum capital adequacy ratio of 8%, currently the ratios of both the state-owned
and the joint-equity banks may be less than 8%. The only regulatory difference
between the state-owned and joint-equity banks relates to credit control. Before 1998,
the central banks controlled the credit of the state-owned banks by setting mandatory
credit quotas, which did not apply to joint-equity banks. This might have led the
state-owned banks to have higher excess reserves ratios, deposit/loan ratios, and lower
loan/asset ratios, but in this paper I find lower excess reserves ratios and deposit/loan
ratios and higher loan/asset ratios.
3.Data and methodology
This paper reports an annual firm-level panel data analysis and a quarterly macro
data analysis. Firm-level data is from 1985 to 2004 for the 4 state-owned banks and
10 joint-equity banks6, which gives a total of 14 banks and 20 years, 214 observations.
Most of the banking and interest rate data were hand collected from China’s Banking
and Finance Almanacs of 1986 to 2004, which are published by China’s Finance
Publishing House, including hard copies and reprinted text documents from the China
Statistics Database of the China Info Bank Web site. The first sample year is 1985, as
the first Almanac was published in 1986 and contains banking data for 1985, and the
last sample year is 2004, as the state-owned banks began to be privatized in 2005. I
looked at related bank firm Web sites to reduce the amount of missing data. Quarterly
6 As the eleventh joint-equity bank – Hengfeng Bank – is very small and was only founded in 2003, this paper focus on the other ten banks.
11
macro data is for the period from quarter 1, 1993 to quarter 4, 2004, a total of 96
observations. Each observation is either for the aggregate of four state-owned banks
or the aggregate of the joint-equity banks. The number of joint-equity banks changes
due to the founding of new banks and the closure of the Hainan Development Bank,
but as I use only ratio variables, this may not be a big problem. The quarterly banking
and interest rate data were collected from text documents from the China Statistics
Database of the China Info Bank Web site. I took great care to make the observations
accurate and comparable. This sample begins in 1993, as no quarterly data before
1992 could be found. I was unable to carry out tests at the quarterly firm level, as I
could identify very little quarterly firm-level banking data.
I calculate three ratios of bank portfolio allocation as proxies for bank prudence:
bank excess reserves ratio, loan/asset ratio, and deposit/loan ratio. If a bank operates
more prudently, then it will have larger excess reserves to safeguard against deposit
runs, and will have a smaller liquidity risk. I use bank total deposit times required
reserves ratio to calculate the required reserves, and then calculate excess reserves by
subtracting the required reserves from the bank reserves. I then divide bank excess
reserves by total deposits to calculate the bank excess reserves ratio. In calculating the
quarterly macro data, I have made sure that claims on the Central Bank before 1997
are counted as bank reserve, but after that are excluded, which is important because
claims on the Central Bank are growing rapidly. Total deposits include demand
deposits plus savings deposits plus time deposits plus other deposits. I exclude foreign
currency deposits from liabilities to non-financial institutions for comparability, as
12
foreign currency deposits were not reported before 2002. Commercial bank loans are
usually more risky than other assets, such as government debt and corporate bonds,
and thus a higher loan/asset ratio indicates that the bank is less risk averse. Loan and
assets data for the annual firm-level analysis were retrieved from bank balance sheets,
and some calculation was needed. For the macro quarterly data, I use claims on
non-financial institutions as a proxy for total loans, not including loans to financial
institutions. China’s macro banking statistics data did not include total assets until
2002, and thus I calculate total assets before 2002 by adding overseas assets, reserve
assets, claims on the Central Bank (after 1997), claims on the government, claims on
non-financial institutions, and claims on other financial institutions together. Total
assets data are available for the period after 2002, but I do not use them, as they are
not completely comparable with my calculations for the period before 2002, when
other assets were not reported in the balance sheet. Thus I delete other assets from
total assets to make the data more comparable. The deposit/loan ratio is quite close to
the loan/asset ratio as a proxy of prudence, and prudential banks are likely to disburse
fewer loans based on the same deposits, where deposit and loan are defined as before.
There are other proxies that are commonly used in studying the behavior of
commercial banks. Shrieves and Dahl (2003), Berger and Udell (1994) use loan and
asset growth and Gorton and Rosen (1995) use the ratio of non-performing loans to
estimate the risk of a bank portfolio. Other possible proxies of bank prudence are loan
structure and the standard deviation of the bank firm stock returns. China’s
joint-equity banks first emerged in 1986 and the last of the 10 that are included in this
13
sample was founded in 1996, and thus the loans of the joint-equity banks grow from a
starting point of 0. Although prudent banks tend to have lower loan growth rate, a
higher loan and asset growth rate cannot be regarded as a proxy for imprudence. The
ratio of non-performing loans, as mentioned, suffers from serious data quality
problems. Chinese banking statistics fail to provide enough loan structure data. As
only five banks went public before 2004, the standard deviation of stock returns
cannot be used.
Based on the macro quarterly data, Figure 1 shows the difference in bank excess
reserves ratio, loan/asset ratio, and deposit/loan ratio between the state-owned banks
and joint-equity banks. Although all three variables changed over the sample period, a
significant difference between the state-owned and joint-equity banks can be
discerned. The bank excess reserves ratios and deposit/loan ratios of the state-owned
banks are lower than those of the joint-equity banks, whereas the loan/asset ratios of
the state-owned banks are higher than those of the joint-equity banks, which coincides
with my hypotheses. The state-owned banks improved in all three aspects over the
sample period, which demonstrates the positive effect of bank reform. However, I am
cautious of coming to the conclusion that there is a difference in the level of prudence,
as this may be caused by other variables, such as different sources of funds.
To test whether there is difference in the level of prudence between the
state-owned and joint-equity banks, I use the following basic regression model:
Bank prudence measures = α+β1 * Bank dummy+β2 * List dummy,
+β3 * GDP growth+β4 * Interest rate measures,
14
+β5 *Bank fund source measures+β6* Bank assets +β7*Regulatory measures,
+β8 * State-owned bank* Time + Error term. (1)
The dependent variables are the bank prudence measures, that is, bank excess
reserves ratio, loan/asset ratio, and deposit/loan ratio.
I use a bank dummy to capture the difference in prudence between the state-owned
and joint-equity banks, in which the state-owned banks are assigned the value 0 and
the joint-equity banks are assigned the value 1. I hypothesize that the joint-equity
banks will be more prudential, and will thus have a higher bank reserves ratio, a lower
loan/asset ratio, and a higher deposit/loan ratio. Managers and directors of Chinese
joint-equity banks rarely hold stocks from their own bank, and insiders hold no stocks
in four of the five listed banks. Non-listed joint-equity banks do not usually produce
annual reports, and thus insufficient data on large shareholders could be collected.
Thus, rather than inside and outside equity, I use a list dummy to determine whether
listing on the stock exchange makes a bank behave more prudently. A listed bank is
assigned the value 1, and non-listed banks are assigned the value 0.
I control other variables to better understand the effect of the bank dummy. As
bank prudential behavior is rarely studied, these variables are interesting in
themselves. The first control variable is GDP growth. When the economy is growing
faster, banks tend to hold smaller reserves and disburse more loans, which should
entail a smaller bank excess reserves ratio, a higher loan/asset ratio, and a lower
deposit/loan ratio.7 The annual real GDP growth data for this variable are taken from
7 In terms of the loan/asset ratio, faster economic growth may lead to growth in the denominator beyond loans, such as government debt and corporate bonds. However, as loans are more risky,
15
the Almanac of Chinese Statistics 2005, which is published by the China Statistics
Publishing House in Beijing. The quarterly GDP data are taken from the CCER of
Peking University.
I use two interest rate variables: interest rate spread and real excess reserves
interest rate. When the interest rate spread is greater, banks can make more money
through lending, and will hence have a lower bank reserves ratio, a higher loan/asset
ratio, and a lower deposit/loan ratio. I use the one-year bank loan rate and deposit rate
to calculate the spread. In the annual firm-level test, there is difference between the
one-year working capital loan and the one-year fixed asset loan. I calculate the
average of these two interest rates as the loan rate. If there was a change in interest
rates in a given year, then I calculate the average before and after the change, and use
the same method for the quarterly macro data test. China’s bank deposits and loan
interest rates do not reflect the market rate most of the time, and therefore I question
the explanatory power of this variable. A higher excess reserves interest rate will
induce banks to deposit more in the central bank and thus lend less, which should
produce higher bank reserves ratios, lower loan/asset ratios, and higher deposit/loan
ratios. For the annual firm-level test, I subtract the annual CPI from the excess
reserves interest rate to obtain real data, and when an adjustment is made to the excess
reserves interest rate I average them. For the quarterly data, I use the end of quarter
excess reserves interest rate minus the quarterly CPI. The annual CPI data are taken
from the Almanac of China’s Statistics, 2005, and the quarterly data come from
economic growth may lead to riskier behavior, and thus loans should grow faster than other assets.
16
Shanghai Gildata Service Co., Ltd.
Bank portfolio ratios may be affected by fund source characteristics. For the
annual firm-level data, I use data on short- and long-term deposits from the relevant
bank sheets (short deposit/long deposit) to capture the effect of this difference. For the
quarterly macro data, I use demand deposits to depict short-term deposits, and savings
deposits plus time deposits to illustrate long-term deposits. Until 2003 Chinese
financial data did not distinguish individual demand deposit from savings deposit. It
may therefore be incorrect to include savings deposit in the proxy term for long-term
deposits. To solve this problem, I again calculate demand deposit to time deposit. A
higher short- to long-term and demand deposit to time deposit ratio means that the
bank faces a greater liquidity risk, which should lead to a higher bank reserves ratio, a
lower loan/asset ratio, and a higher deposit/loan ratio.
I control bank assets in the firm-level regression to make sure that the results are
not driven by differences in bank size using a natural logarithm for normality. There is
no theory to predict whether bigger banks will be more or less prudent, but given that
banks differ markedly in size the controlling of this variable should determine
whether an effect could be identified.
I control two regulatory dummy variables – credit quota and WTO – to capture
the regulatory impacts on the portfolio decisions of banks8. The credit quota variable
takes the value 1 for state-owned banks for the period after 1998, and 0 for the
state-owned banks for the period before 1997 and the joint-equity banks. In December
8Since banks first reported standardized capital adequacy ratio in 2004, I don’t use this measure.
17
2001, China acceded to the WTO, which meant the opening up of the banking market .
Thus, for the WTO dummy the banks are assigned the value 1 after 2002, and 0 before
2001. In 1994, three policy banks were set up to take the policy loans away from the
state-owned banks, and thus I use a 1994-2004 firm-level sub-sample to ascertain
whether has been any difference in their level of prudence subsequent to this event.
I use state-owned bank*time to capture the change in the level of prudence that is
displayed by the state-owned banks. For the annual firm-level data, I use only
state-owned bank data from 1985-2004 as a sub-sample, and calculate state-owned
bank*time as the number of years since 1985. For the quarterly macro data,
state-owned bank*time is the number of quarters since the beginning of the sample
period (March 1993), and joint-equity banks are assigned the value 0. I hypothesize
that the state-owned banks will become increasingly prudent as a result of the banking
reforms. Table 1 shows the annual firm-level data and macro data regression variables
that are specified in (1) and their descriptive statistics.
4. Firm-level regression results
I use firm-level panel data from 1985-2004 with three dependent variables to
explain the portfolio characteristics of the state-owned and joint-equity banks. The
results in Table 2 show that the bank dummy is the most robust variable in explaining
bank prudence measures. In most of the regressions with the three dependent variables
and a different complement of independent variables, the bank dummy is significant
18
at the one-percent level and has the correct sign. This means that joint-equity banks
tend to hold more reserves than state-owned banks，and to have higher deposit/loan
ratios and lower loan/asset ratios, which are an obvious sign of greater prudence.
State-owned banks have a less efficient system of corporate governance, and tend to
be less concerned with liquidity risk and default risk and more eager to disburse loans.
The only two regressions in which the bank dummy is not significant are the bank
excess reserves regressions 3 and 4. When bank assets are added into the regressions,
the bank dummy becomes insignificant, whereas the new control variable – bank
assets – is negative and significant at the one-percent level in both regressions. This
can be explained by the correlation between the bank dummy and bank assets, which
is -0.718 and significant at the one-percent level. All four state-owned banks are big
banks, whereas all of the joint-equity banks are medium-sized or small banks. The
negative coefficient of bank assets suggests that smaller banks tend to hold greater
excess reserves, which indicates that joint-equity banks tend to be more prudent, a
result that is coincident with my hypothesis. In the loan/asset ratio and deposit/loan
ratio regressions, the inclusion of bank assets does not make the bank dummy
insignificant, and the high correlation suggests this result to be robust. The list dummy
seems to have little explanatory power, and is not significant in the loan/asset ratio
and deposit/loan ratio regressions. It is significant in the excess reserves regression,
but has the incorrect sign. It can thus be concluded that listed banks are not
necessarily more risk averse than the other joint-equity banks. So I follow Berger et al
(2005) and others focusing on ownership.
19
GDP growth also seems to be a good explanatory variable for bank portfolio
allocation, especially when I control for the ratio of short to long deposits. This
indicates that during an economic boom, banks tend to hold less reserves and lend
more. The performance of the two interest variables is not satisfactory, and they seem
only to explain bank excess reserves well. I discuss this further in section 5.
The ratio of short to long deposits appears to explain the change in the loan/asset
ratio and the deposit/loan ratio very well, which indicates that when the liquidity risk
is greater, banks with more short-term deposits and less long-term deposits tend to
disburse fewer loans. However, the short/long deposit ratio fails to explain bank
excess reserves, giving a significant result but with the opposite sign. This may be a
consequence of the absence of 99 values from the short/long deposit data, or of the
definition of the variable, as I discuss in section 2.
To identify the change in the prudential behavior of the state-owned banks, I run a
regression on a sub-sample that includes only the state-owned banks. The dynamic
change measure of state-owned bank* time has the correct sign as hypothesized, and
is significant in five of the six regressions for all three dependent variables, and in
four at the one-percent level. These fairly robust results support the theory that the
reform of the banking system has improved the incentive for state-owned banks to
behave more prudently, and paints a positive picture of the last 20 years of banking
reform in China, which, due to the accumulating non-performing loan problem, is
often viewed negatively. However, the result does not mean that the state-owned
banks have solved their corporate governance problem and now display rational
20
behavior; indeed, the significant bank dummy coefficients demonstrate that they are
still significantly less prudent than the joint-equity banks. Nevertheless, viewed
dynamically, the state-owned banks are showing a sustained improvement.
The credit quota dummy is not significant in any of the three regressions. This
may be because since the credit quota of the state-owned banks was cancelled in 1998,
they have been better able to constrain their loan-extending attitude. Entry to the
WTO seems to have induced higher loan/asset ratios and lower deposit/loan ratios, but
has had no significant effect on excess reserves. As China’s banking market would not
be fully open for five years, I am cautious about these results.
One may turn to a social view of state ownership to explain the difference in bank
prudence, in that state-owned banks may be trying to maximize broader social
objectives. In addition, it must be remembered that the joint-equity banks were just
come into being early in the sample period, and thus the dominance of the
state-owned banks, which might be insensitive to interest rates, may cause the interest
variables to be insignificant. To test these ideas, I shrink my sample to the period
1994-2004. The results (shown in Table 2) are quite close to those for the sample
period 1984-2004, and the bank dummy is again significant at the one-percent level in
most of the regressions. As policy loans were split off from the state-owned banks
during the 1994-2004 period and the state-owned banks were supposed to start
operating like “real banks,” the results seem not to support the social view.9 The real
9 I use bank excess reserves, the loan/asset ratio, and the deposit/loan ratio as measures of bank prudence. Social objectives may lead state-owned banks to lend to SOEs, but there is no theory as to why such ratios would be affected. Furthermore, as the largest shareholders of joint-equity banks are usually SOEs, they also suffer from political interference. These factors lead me to support the corporate governance view.
21
excess reserves interest rate and interest rate spread are still not good explanatory
variables.
5. Macro data regression results
In this section, I use the quarterly data from the Chinese banking sector to explain
the difference in the portfolio characteristics of the state-owned and joint-equity banks.
The data are for all four state-owned banks and all joint-equity banks, respectively,
and the empirical results are shown in Table 3. Coincident with the annual firm-level
data, the bank dummy is significant in all cases at the one-percent level, even after I
control for bank fund source characteristics, macro variables, and regulatory variables,
and has the correct sign. It is again the most powerful explanatory variable for the
bank allocation and prudence measures of bank excess reserves, loan/asset ratio, and
deposit/loan ratio. In the regressions of state-owned bank*time, it is significant at the
one-percent level in the loan/asset ratio and deposit/loan ratio regressions, and has the
correct sign. However, it is insignificant in the bank excess reserves regression. I
believe that this is driven by the change in China’s money market. China began to
develop an organized money market in 1996 through the establishment of a national
inter-bank lending market, and, in 1997, an inter-bank bond market. Both markets
have flourished, and have given China’s commercial banks increasing numbers of
money market instruments from which to choose as a secondary reserve. This
explains why I find no significant response from reserves to the better management of
22
state-owned banks over time.
When I put more independent variables in the regressions of all three dependent
variables, there is a considerable rise in the coefficients of the bank dummy variable.
For the bank excess reserves regression, the coefficient increases from 0.654 in the
first regression to 1.538 in the last regression; for the loan/asset ratio regression it
increases from -0.743 to -2.763; and for the deposit/loan ratio regression it increases
from 0.358 to 2.816. The coefficient of state-owned bank*time follows the same
pattern. This means that the other independent variables to some extent mask the
effect of bank ownership on bank prudence and the change in prudence of the
state-owned banks. Figure 2 clearly shows that the state-owned banks tend to have
much larger demand deposit to time deposits, which means that the joint-equity banks
should have much lower bank excess reserves and deposit/loan ratios and much
higher loan/asset ratios. This is directly contrary to that which is depicted in Figure 1,
which indicates that the difference in prudence between the state-owned banks and the
joint-equity banks is much more serious than Figure 1 shows.
Excess reserves interest rate also seems to be a very good explanatory variable,
being significant in most of the regressions at the one-percent level with the correct
sign. This shows that a higher excess reserves interest rate can successfully induce
banks to hold more reserves, and when the excess reserves interest rate is lower,
banks tend to lend more based on their total assets or deposits. In China, this is
understandable. In the United States and most other countries, the central bank does
not pay interest on reserves, whereas China has traditionally paid a fairly high interest
23
rate on bank reserve. The excess reserves interest rate is not significant in the
firm-level regressions, but this may be due to the variance in the behavior of the banks
in the firm-level panel data, such as different banks having different attitudes toward
changes in the excess reserves interest rate. However, on the whole state-owned banks
and joint-equity banks do respond to changes in the excess reserves interest rate.
Interest rate spread performs well in the bank excess reserves regressions, but is not
as robust as excess reserves interest rate. As is discussed in section 3, China has been
loosening its control over bank interest rates, and as the restrictions for commercial
banks are diminishing, the official interest rates that are set by the PBOC are no
longer a good way of calculating the interest rate spread. The two deposit component
proxies are both very good exogenous variables, as is also the case with the firm-level
regression results, and are significant in most of the regressions and have the correct
sign. If the total deposits of a bank consist of a greater proportion of demand deposits,
then the bank will hold more reserves to mitigate deposit runs, and will also disburse
fewer loans based on its assets and deposits. Consistent with the discussion in section
2, demand deposits as a proportion of time deposits seems to work better as a measure
than short deposits as a proportion of long deposits. The two regulatory variables
show the same results in the macro data regressions as in the firm-level regressions.
The results for GDP growth seem to be less robust. This may be because I use real
GDP data in the firm-level regressions, but no quarterly real GDP data are available.
5.Conclusion
24
This paper focuses on the relationship between bank ownership and bank prudence
in China, which is seldom discussed even as a general topic, and uses bank excess
reserve, loan/asset ratio, and deposit/loan ratio as proxies of bank prudence. Through
the carefully selection of dependent and independent variables, I have tried my best to
reduce the effects of poor data quality, with the result that the empirical annual
firm-level panel data and quarterly macro data results are all fairly robust. Joint-equity
banks tend to have higher excess reserves, higher deposit/loan ratios, and lower
loan/asset ratios, which shows that they are significantly more prudent than the
state-owned banks. However, although their level of prudence is incomparable with
that of the joint-equity banks, the state-owned banks have been carrying out fairly
efficient reforms and are becoming more prudent as a result. Many people have
criticized the state-owned banks for simply being producers of non-performing loans,
but the empirical results that are reported in this paper offer a more positive view.
GDP growth, excess reserves interest rate, and bank fund source characteristics are
also found to affect bank portfolio allocation. However, whether the reform of the
state banking system without privatization will eliminate the identified difference in
prudence with the joint-equity banks remains an open question, and as China has only
four state-owned banks and eleven main joint-equity banks, the relatively small
sample that this affords may affect the results.
Reference
Alchian, A., 1965. Some economics of property rights. Politico 30, 816–829.
25
Beck, T., Cull, R., Jerome, A., 2005. Bank privatization and performance: Empirical
evidence from Nigeria. Journal of Banking and Finance 29, 2355-2379.
Berger, A.N., Clarke, G.R.G., Cull, R., Klapper, L., Udell, G.F., 2005. Corporate
governance and bank performance: A joint analysis of the static, selection, and
dynamic effects of domestic, foreign, and state ownership. Journal of Banking
and Finance 29, 2179-2221
Berger, A.N., Udell, G. F., 1994. Did risk-based capital allocate bank credit and cause
a “credit crunch” in the United States? Journal of Money, Credit, and Banking 26,
585-628.
Chen, Z., Li, D., Moshirian, F., 2005. China’s financial services industry: The
intra-industry effects of privatization of the Bank of China Hong Kong. Journal
of Banking and Finance 29, 2291-2324.
Clarke, G.R.G., Cull, R., Shirley, M.M., 2005. Bank privatization in developing
countries: A summary of lessons and findings. Journal of Banking & Finance 29,
1905–1930.
Gorton, G., Rosen R., 1995. Corporate control, portfolio choice, and the decline of
banking. Journal of Finance 50, 1377-1420.
Haber, S., 2005. Mexico’s experiments with bank privatization and liberalization,
1991–2003. Journal of Banking and Finance 29, 2325-2353.
Hellmann, T. F., Murdock, K. C., Stiglitz, J. E., 2000. Liberalization, moral hazard in
banking, and prudential regulation: Are capital requirement enough? The
American Economic Review, Volume 90, No.1,147-165.
26
Milne, A., 2002. Bank capital regulation as an incentive mechanism: Implications for
portfolio choice. Journal of Banking and Finance 26, 1-23.
Otchere, I., 2005. Do privatized banks in middle-and low-income countries perform
better than rival banks? An intra-industry analysis of bank privatization. Journal
of Banking and Finance 29, 2067-2093.
Saunders, A., Strock, E., Travlos, N. G., 1990. Ownership structure, deregulation, and
bank risk taking. Journal of Finance 45, 643 -654.
Shrieves, R. E., Dahl, D., 2003. Discretionary accounting and the behavior of
Japanese banks under financial duress. Journal of Banking and Finance 27,
1219-1243.
Vining, A.R., Boardman, A.E., 1992. Ownership versus competition: Efficiency in
public enterprise. Public Choice 73, 205-239.
Yeung, B. Y., Litov, L. P., John, K., 2006. Corporate governance and corporate risk
taking: Theory and evidence. AFA 2007 Chicago Meetings Paper. Available at
SSRN: http://ssrn.com/abstract=890915.
Figure 1. Macro quarterly data on the difference in bank prudence (a) Bank excess reserves ratio
27
bank excess r eser ves r at i o
00. 050. 1
0. 150. 2
0. 25
Mar
-93
Mar
-94
Mar
-95
Mar
-96
Mar
-97
Mar
-98
Mar
-99
Mar
-00
Mar
-01
Mar
-02
Mar
-03
Mar
-04
t i me
bank
exc
ess
rese
rves
rat
io
st at e-ownedj oi nt - equi t y
(b) Loan/asset ratio
l oan/asset
00. 20. 40. 60. 8
1
Mar
-93
Mar
-94
Mar
-95
Mar
-96
Mar
-97
Mar
-98
Mar
-99
Mar
-00
Mar
-01
Mar
-02
Mar
-03
Mar
-04
t i me
loan
/ass
et
st at e-ownedj oi nt - equi t y
(c) Deposit/loan ratio
deposi t / l oan
0
0. 5
1
1. 5
2
Mar
-93
Mar
-94
Mar
-95
Mar
-96
Mar
-97
Mar
-98
Mar
-99
Mar
-00
Mar
-01
Mar
-02
Mar
-03
Mar
-04
t i me
depo
sit/
loan
st at e-ownedj oi nt -equi t y
This figure shows the difference in prudence between China’s state-owned banks and joint-equity banks. The data are for the aggregate of all state-owned banks or all joint-equity banks, respectively; (a) shows that the state-owned banks have lower bank excess reserves ratios than the joint-equity banks; (b) shows that the state-owned banks have higher loan/asset ratios; (c) shows that the state-owned banks have lower deposit/loan ratios.
28
Figure 2. Ownership and bank fund source difference
demand deposi t / t i me deposi t
02468
10
Mar
-93
Mar
-94
Mar
-95
Mar
-96
Mar
-97
Mar
-98
Mar
-99
Mar
-00
Mar
-01
Mar
-02
Mar
-03
Mar
-04
t i me
dem
and
depo
sit/
time
depo
sit
st at e-ownedj oi nt - equi t y
This figure shows that the state-owned banks generally have higher demand deposit/time deposit ratios than the joint-equity banks, and therefore should have higher bank excess reserves ratios and deposit/loan ratios and lower loan/asset ratios. This figure differs markedly from Figure 1, and indicates that the difference in the level of prudence between the state-owned banks and the joint-equity banks is much more serious than Figure 1 shows.
29
N Mi ni mum Maxi mum Mean St d. Dev. N Mi ni mum Maxi mum Mean St d. Dev.Excess r eser ves 162 0. 02 0. 55 0. 14 0. 09 96 0. 04 0. 22 0. 10 0. 04
Loan/asset 210 0. 05 0. 90 0. 53 0. 14 96 0. 54 0. 80 0. 68 0. 07Deposi t / l oan 209 0. 37 2. 73 1. 31 0. 38 96 0. 73 1. 50 1. 17 0. 21Bank dummy 217 0. 00 1. 00 0. 63 0. 48 96 0. 00 1. 00 0. 50 0. 50Li st dummy 217 0. 00 1. 00 0. 14 0. 35
GDP growt h 217 103. 80 114. 20 109. 32 2. 53 80 1. 01 1. 64 1. 13 0. 12Excess reserve interest rate 217 -14. 92 4. 89 -1. 12 5. 70 96 -18. 12 6. 82 -0. 88 6. 10
Interest rate spread 217 -0. 12 3. 60 2. 15 1. 30 96 0. 00 3. 60 2. 42 1. 31Shor t deposi t / l ong depost 118 0. 56 72. 00 7. 07 10. 97 96 0. 40 2. 52 0. 93 0. 53
Demand deposi t / t i me deposi t 96 1. 83 8. 59 3. 62 1. 64Bank assets 214 1. 45 10. 93 7. 58 1. 97Credit quota 217 0. 00 1. 00 0. 13 0. 34 96 0. 00 1. 00 0. 58 0. 50
WTO 217 0. 00 1. 00 0. 18 0. 39 96 0. 00 1. 00 0. 25 0. 44St at e-owned bank*Ti me 80 0. 00 20. 00 10. 50 5. 80 96 0. 00 48. 00 12. 25 15. 77
Table 1.Summary of descr i pt i ve st at i st i cs Firm-level data Macro data
Firm-level data are the annual data for each state-owned and joint-equity bank between 1985 and 2004. The macro data are the quarterly data for all of the state-owned banks and joint-equity banks, respectively, between 1993 and 2004. Excess reserves is the bank excess reserves ratio. Bank dummy is a dummy variable that takes the value 0 for state-owned banks and 1 for joint-equity banks. List dummy is a dummy that takes the value 1 if a bank is listed and 0 otherwise. Bank assets are input in the form of a natural logarithm. Credit quota is a dummy variable that takes the value 1 for state-owned banks after 1998 (as China abolished the credit quota for state-owned banks on January 1, 1998) and 0 for state-owned banks before 1997 and joint-equity banks. WTO is a dummy that takes the value 0 for the years before 2001 (the year of China’s accession to the WTO), and 1 for the years after 2002. State-owned bank*time is the number of years since the beginning of the sample period (1985) for the firm-level data and the number of quarters since Quarter 1 1993 for the macro data.
30
0.825 3.935 0.824 0.656 4.720 4.697 3.930 0.317 0.713(2.14)** (3.64)*** (2.18)** (1.71)* (4.78)*** (4.74)*** (2.95)*** (1.04) (2.56)**
0.557 0.519 0.208 0.282 0.550 0.455 0.222(7.82)*** (6.07)*** (1.41) (1.54) (7.55)*** (2.87)*** (1.53)
-0.152 -0.221(-2.10)** (-2.69)***
-0.135 -0.605 -0.100 -0.073 -0.721 -0.712 -0.579 -0.057 -0.035(-1.68)* (-3.49)*** (-1.25) (-0.92) (-4.62)*** (-4.53)*** (-2.87)*** (-0.49) (-0.34)
0.182 -0.011 0.152 0.118 -0.198 -0.213 -0.231 0.166 0.212(2.04)** (-0.08)* (1.72)* (1.34) (-1.32) (-1.41) (-1.53) (1.29) (1.88)*-0.551 -0.584 -0.329 -0.243 -0.511 -0.464 -0.121 -0.621 -0.865
(-6.12)*** (-4.38)*** (-2.71)*** (-1.69)* (-4.45)*** (-3.43)*** (-0.70) (-2.38)** (-3.71)***-0.195 -0.167
(-2.37)** (-2.14)**-0.370 -0.308 -0.104 -0.997
(-2.67)*** (-2.01)** (-0.67) (-4.47)***-0.030 -0.377(-0.29) (-2.77)***-0.105 -0.120(-1.25) (-1.04)
State -0.114 0.998bank*Time (-0.49) (3.12)***
Observations 162 111 162 162 128 128 109 64 64Adjusted R2 0.303 0.297 0.329 0.349 0.365 0.362 0.403 0.363 0.518
1.699 -2.787 1.637 1.861 -2.297 -2.222 -1.963 0.619 1.229(4.15)*** (-2.67)*** (4.10)*** (4.75)*** (-1.96)* (-1.90)* (-1.22) -0.9 (1.74)*
All sample Sub sample (1994-2004) Sub sample (state banks)
(Constant)
Panel B. Loan/asste regressions
WTO
Credit quota
Bank assets
Excess reserveinterest rate
Interest rate spread
Short deposit/longdeposit
GDP growth
List dummy
(Constant)
Bank dummy
Table 2. State-owned banks are less prudent:Firm Level evidence Panel A. Bank excess reserves ratio regressions
All sample Sub sample (1994-2004) Sub sample (state banks)
31
-0.375 -0.621 -0.724 -0.743 -0.677 -0.491 -0.583(-6.24)*** (-8.91)*** (-6.28)*** (-5.57)*** (-10.29)*** (-3.65)*** (-4.56)***
-0.051 -0.002(-0.83) (-0.03)**
-0.167 0.440 -0.120 -0.145 0.371 0.348 0.280 0.018 0.036(-2.49)** (3.22)*** (-1.80)** (-2.23)** (2.41)** (2.26)** (1.58) (0.14) (0.30)
0.066 0.281 0.040 0.073 0.159 0.179 0.284 0.089 0.119(0.89) (2.57)** (0.55) (1. 03) (1.09) (1.23) (2.14)** (0.65) (0.90)-0.385 0.298 -0.144 -0.264 0.279 0.189 0.150 0.482 0.305
(-5.18)*** (2.64)*** (-1.44) (-2.36) (2.47)** (1.50) (0.97) (1.81)* (1.15)-0.316 -0.310
(-4.68)*** (-4.49)***-0.410 -0.519 0.205 -0.654
(-3.52)*** (-4.31)*** (1.57) (-2.54)**0.070 0.036(0.88) (0.30)0.257 0.110
(3.74)*** (1.07)-0.739 0.000
(-3.14)*** (-0.00)***Observations 210 114 210 210 146 146 111 78 78Adjusted R2 0.298 0.509 0.335 0.375 0.418 0.425 0.512 0.115 0.176
-1.822 7.654 -1.804 -2.127 6.449 6.125 7.131 0.472 0.139(-1.96)* (2.31)** (-1.93)* (-2.26)** (1.74)* (1.67)* (1.45) (0.56) (0.16)0.315 0.365 0.350 0. 437 0.461 0.193 0.579
(6.35)*** (5.07)*** (3.58)*** (3.76)*** (6.51)*** (1.34) (4.30)***-0.021 -0.027(-0.38) (0.35)
(Constant)
Bank dummy
List dummy
State bank*Time
Panel C. Deposit/loan regressionsAll sample Sub sample (1994-2004) Sub sample (state banks)
WTO
Credit quota
Bank assets
GDP growth
Excess reserve interestrate
Interest rate spread
Short deposit/longdeposit
Bank dummy
List dummy
32
0.154 -0.275 0.149 0.16 -0.234 0.348 -0.219 0.016 0.011(2.79)*** (-2.01)** (2.64)*** (2.83)*** (-1.45) (2.26)** (-1.25) (0.22) (0.15)
0.024 0.011 0.027 0.013 -0.016 0.179 0.026 -0.039 -0.047(0.39) (0.10)* (0.44) (0. 21) (-0.11)* (1.23) (0.20) (-0.47) (-0.57)0.616 0.070 0.592 0.607 0.119 0.189 -0.038 0.277 0.325
(10.06)*** (0.61) (7.00)*** (6.23)*** (1.00) (1.50) (-0.25) (1.73)* (1.96)*0.396 0.404
(5.70)*** (5.64)***0.041 0.111 0.205 0.176(0.42) (1.06) (1.57) (1.10)
0.047 0.201(0.68) (1.57)-0.121 -0.003
(-2.02)** (-0.03)**State 0.602 0.404
bank*Time (4.26)*** (1.76)*Observations 209 113 209 209 145 145 110 78 78Adjusted R2 0.524 0.483 0.522 0.527 0.331 0.348 0.479 0.681 0.682
Short deposit/longdeposit
GDP growthExcess reserve
interest rate
Interest rate spread
Credit quota
Bank assets
WTO
The data are annual firm-level data for each state-owned and joint-equity bank in the period 1985 to 2004. The dependent variables are bank excess reserves ratio (panel A), loan/asset ratio (panel B), and deposit/loan ratio (panel C). Regressions 1-4 are conducted on the whole sample; regressions 5-7 are conducted on the sub-sample of the period 1994-2004, which contains only data after the policy loans were split off from the state-owned banks in 1994, and regressions 8-9 are conducted on a sub-sample that contains all of the state-owned banks for the period 1985-2004 only. The coefficients are standardized. Bank dummy is a dummy variable that takes the value 0 for state-owned banks and 1 for joint-equity banks. List dummy is a dummy that takes the value 1 if a bank is listed and 0 otherwise. Bank assets are input in the form of a natural logarithm. Credit quota is a dummy variable that takes the value 1 for state-owned banks after 1998 (as China abolished the credit quota for state-owned banks on January 1, 1998) and 0 for state-owned banks before 1997 and joint-equity banks. WTO is a dummy that takes the value 0 for the years before 2001 (the year of China’s accession to the WTO), and 1 for the years after 2002. State bank*time is the number of years since 1985. Observations vary due to missing data. The functional form is OLS. The t statistics are in parentheses; * significant at the 10% level; ** significant at the 5% level; *** significant at the 1% level.
33
0.115 -0.068 -0.037 -0.101 0.762 0.886 1. 149 1.189 0.776 0.841 -0.077 -0.143(7.04)*** (-0.81) (-0.43) (-1.29) (38.52)*** (6.91)*** (13.25)*** (16.21)*** (12.93)*** (2.04)** (-0.32) (-0.780)
0.654 1.536 1.266 1. 538 -0.743 -1.194 -2.597 -2.763 0.358 0.713 2.636 2.816(4.09)*** (7.63)*** (4.17)*** (5.69)*** (-6.26)*** (-6.37)*** (-14.19)*** (-17.86)*** (2.84)*** (3.02)*** (13.32)** (18.54)***
-0.132 -0.137 -0.069 0.019 -0.008 -0.029 0.062 0.098 0.106(-1.21) (-1.26) (-0.72) (0.19) (-0.12)* (-0.52) (0.48) (1.38) (1.94)*
0.225 0.516 0.459 0.391 -0.025 -0.16 -0.458 -0.37 0.102 0.054 0.462 0.332(1.74)** (4.66)*** (3.81)*** (3.53)*** (-0.26) (-1.56) (-6.31)*** (-5.84)*** (0.99) (0.42) (5.89)*** (5.33)***-0.453 0.002 0.04 0.085 -0.149 -0.17 0.029 -0.073 0.435 0.199 -0.073 0.092
(-3.35)*** (0.01)*** -0.29 -0.66 (-1.48) (-1.37) (0.35) (-0.99) (4.07)*** (1.28) (-0.82) (1.27)1.097 0.996 1.055 -0.456 -0.979 -0.954 0.011 0.727 0.655
(4.71)*** (4.03)*** (4.79)*** (-2.10)** (-6.57)*** (-7.57)*** (0.04)** (4.51)*** (5.29)***-0.048 -0.107 0.454(-0.27) (-0.82) (3.27)***
0.166 0. 029 -0.129(1.72)* (0.53) (-2.37)**-0.376 0.256 -0.297
(-4.88)*** (5.81)*** (-6.87)***-0.217 0.042 -1.128 -1.343 1.545 1.82(-1.18) (0.24) (-10.22)*** (-13.63)*** (12.95)** (18.82)***
Obs. 96 80 80 80 96 80 80 80 96 80 80 80Adjusted R2 0.441 0.635 0.637 0.724 0.691 0.683 0.868 0.91 0.652 0.498 0.846 0.913
Statebank*Time
WTO
Credit quota
Interest ratespread
Demand/time
Short/long
(Constant)
Bankdummy
Reserveinterest rate
GDP growth
Table 3. State-owned banks are less prudent: Macro data regressionsBank excess reserves ratio regressions Loan/asset regressions Deposit/loan regressions
The data are macro quarterly data on state-owned banks and joint-equity banks between 1993 and 2004. The dependent variables are bank excess reserves ratio (Regressions 1-4), loan/asset
ratio (regressions 5-8), and deposit/loan ratio (regressions 9-12). The coefficients are standardized. Bank dummy takes the value 0 for state-owned banks and 1 for joint-equity banks. Reserves
interest rate is the excess reserves interest rate. Demand/time is the demand deposit/time deposit ratio. Short/long is the ratio of short-term to long-term deposits. Credit quota is a dummy that
takes the value 1 for state-owned banks after 1998 and 0 for state-owned banks before 1997 and joint-equity banks. WTO is a dummy that takes the value 0 for the years before 2001 and 1 for
the years after 2002. State bank*time is the number of quarters since March 1993. The functional form is OLS. The t statistics are in parentheses; * significant at the 10% level; ** significant at
the 5% level; *** significant at the 1% level.