Comparing the performance of Chinese banks: A principal component approach

This article was originally published in a journal published by
Elsevier, and the attached copy is provided by Elsevier for the
author’s benefit and for the benefit of the author’s institution, for
non-commercial research and educational use including without
limitation use in instruction at your institution, sending it to specific
colleagues that you know, and providing a copy to your institution’s
administrator.
All other uses, reproduction and distribution, including without
limitation commercial reprints, selling or licensing copies or access,
or posting on open internet sites, your personal or institution’s
website or repository, are prohibited. For exceptions, permission
may be sought for such use through Elsevier’s permissions site at:
http://www.elsevier.com/locate/permissionusematerial

---

China Economic Review 18 (2007) 15 – 34
Comparing the performance of Chinese banks:
A principal component approach ☆
Victor SHIH a,⁎, Qi ZHANG a, Mingxing LIU b
a Department of Political Science, Northwestern University, United States
b The School of Government, Beijing University 03/2006, China
Abstract
Using previously unavailable central bank data, this paper first uses principal component analysis to
derive four measures of a bank's ability to perform the core task of financial intermediation. This study then
compares the performance of China's state banks, joint-stock banks, and city commercial banks along these
measures. In terms of overall performance and in credit risk management, joint-stock banks perform
significantly better than both the state banks and the city commercial banks. In China, unlike in other
developing countries, the size of the bank is not correlated with their performance. Mid-size national joint-
stock banks perform considerably better than the Big Four banks and smaller city commercial banks
(CCBs). We further conduct regional and jurisdictional analysis of the CCBs, which indicates that a mix of
geographical and historical legacies drives the substantial variation in CCB performance.
© 2006 Elsevier Inc. All rights reserved.
JEL classification: G21; G28; P31; P34
Keywords: China; Banks; Finance; Transition economy; Principal component analysis
1. Introduction
Since China launched its sweeping economic reform, state intervention in many sectors of the
economy has been substantially reduced or eliminated entirely. Yet, a series of recent studies,
beginning with Lardy's seminal book, point out that the state continues to play a dominant role in
the financial Author's personal copy
sector (Huang & Bonin, 2001; Lardy, 1998; Park & Sehrt, 2001; Park & Shen, 2002).
The high degree of state intervention has resulted in, among other things, a high non-performing
☆ We would like to thank Tom Rawski for generously providing extensive comments to a preliminary draft of this paper.
We would additionally like to thank anonymous reviewers for providing helpful and constructive comments. Of course, all
errors are our own.
⁎ Corresponding author.
1043-951X/$ - see front matter © 2006 Elsevier Inc. All rights reserved.
doi:10.1016/j.chieco.2006.11.001

---

16
V. Shih et al. / China Economic Review 18 (2007) 15–34
loan ratio in the Chinese banking system. Estimates of China's non-performing loan ratio vary
between the official 20% to Western estimates of 40–50% (Berger, Nast, & Raubach, 2002; Chan,
Mukherji, Chew, & Napier, 2001; Roubini & Setser, 2005). Although these estimates provide a
general sense of the Chinese banking sector's performance, especially the performance of the Big
Four state banks, there have been few comprehensive studies of the other segments of the banking
sector. This study aims to provide evidence for the on-going discussion on the causes of China's
banking problem, as well as to fill in the empirical gaps in our understanding of the city
commercial banks.
Both the general literature on banking and the literature on China provide plausible theories
about the determinants of Chinese bank performance, but bank data have been scarce for most
banks, thereby preventing rigorous testing of these frameworks. Up to this point, analysts were
not even sure how Chinese banks are performing on a number of important dimensions. As a first
step toward parsing the determinants of bank performance, we use internal People's Bank of
China (PBOC) data to derive four core measures of bank performance: overall performance,
liquidity management, credit risk management, and capital profitability. Deriving these four
scores with principal component analysis (PCA), we compare the performance of city commercial
banks (CCBs) with that of the state banks and joint-stock banks. We further compare CCB
performance in different regions and administrative jurisdictions. These inter-segmental and
interregional comparisons constitute an important first step toward identifying the causal
mechanisms that drive Chinese bank performance.
2. Domestic determinants of bank performance
The main approaches to explain the problems in the Chinese banking sector include the
incomplete legal and regulatory environment, the government's concern for unemployment,
ownership of banks, and external political interference (Bowles & White, 1993; Huang & Bonin,
2001; Lardy, 1998; Park & Sehrt, 2001). Despite the rich array of explanations for the banking
sector's poor performance, it has been difficult to isolate the causes of banking problems,
primarily due to the incompleteness of banking data. Although we do not have all the necessary
data, our dataset enables us to eliminate some of the most common explanations, thus clarifying
our understanding of the problem.
In the literature on financial intermediation, a main issue is how the legal and regulatory
environment affects the relative performance of different kinds of banks. In general, intermediaries
(banks) enhance the efficiency of capital allocation by lowering the cost of monitoring
entrepreneurs (borrowers) for the investors (depositors) (Diamond, 1984; Yanelle, 1997). Banks
that can do this well would most likely yield a non-negative return. Given a relatively complete set
of legal and credit monitoring institutions, larger institutions with a more diversified holding of
independent borrowers are most able to reduce the average cost of monitoring each borrower
(e.g. Diamond,Author's personal copy
1984).
In developing countries with incomplete legal and monitoring institutions, however, there are
compelling reasons to think that smaller, local financial institutions might have an advantage.
Stiglitz (1990) argues that where legal and credit monitoring systems are incomplete, local
financial institutions gain an advantage by having borrowers monitor each other, a mechanism
which he calls “peer monitoring.” Empirical evidence from Korea reveals that this mechanism
might have saved the vast majority of small Korean credit association from financial woes after
the Asian Financial Crisis (Bongini, Ferri, & Kang, 2002). Moreover, because smaller institutions
often serve as the exclusive financial institutions for small and medium enterprises (SMEs), it

---

V. Shih et al. / China Economic Review 18 (2007) 15–34
17
makes defaulting on loans extremely costly for SMEs, thus fostering a long-term cooperative
relationship between local banks and SMEs (Berger & Udell, 1995; Keeton, 1995; Strahan &
Weston, 1998). The findings of this paper, however, support neither point of view. While small
city commercial banks (CCBs) have a highly varied performance record, the best performing
banks in China are mid-size joint-stock banks (JSBs) instead of the big state behemoths.
In cross-national research on bank performance, economists find that ownership structure has
a significant impact on bank performance (Grigorian & Manole, 2002). There is good reason to
think that the same logic applies to China. With private and foreign shareholders, national joint-
stock banks (JSBs) and city commercial banks (CCBs) may be less prone to make policy loans
than their state-owned counterparts (Han, 2000; Li, 2000). In fact, this logic has created
momentum within the government to privatize parts of the state banking system (Bloomberg,
2002). Based on this hypothesis, we would expect both CCBs and JSBs to outperform the Big
Four state banks in credit risk and profitability. Again, the findings of the paper cast serious
doubt on this claim. While joint-stock banks in China perform the best, CCBs, which are also
structured as share-holding companies, run the gamut between star performers to the worst banks
in China.
Finally, since socialist economies operate on the basis of hierarchical power rather than of the
market, economic outcomes invariable stem from political incentives (Kornai, 1992: 33). Within
this rubric, several political factors can play a vital role in determining banking performance.
First, poor banking performance might simply be the product of the government's obligation to
keep SOEs alive for social stability reason (Brandt & Zhu, 2000; Huang & Bonin, 2001; Lardy,
1998; Pei, 1998). If that were the case, we would expect banks located in China's northeastern
rustbelt and the west to perform worse than those on the prosperous coast. Also, we would expect
banks located in places with a high concentration of SOE workers to perform worse than banks in
areas dominated by the private sector. Finally, we would again expect the Big Four banks to lag
behind because they traditionally bore the greatest policy burden. Our findings provide some
evidence to support these hypotheses, as CCBs in the northeast rustbelt indeed perform worse
than their peers on the east coast. However, CCBs in western China, especially the northwest, are
among the best banks in China.
Similarly, according to Yang, inland provinces suffer from a vicious cycle of extraction by the
eastern provinces, reliance on central subsidies, and malnourished private sector (Yang, 1997). If
that were true, inland banks and bank branches would suffer from frequent local governmental
intervention, high credit risk, and low profitability because local bank illiquidity is one of the few
ways whereby the local government can blackmail the central government for more funding.
Likewise, prosperous coastal provinces with a vibrant private sector and a fiscal surplus should
have thriving banks with few problems. Our findings clearly suggest strong regional patterns in
bank performance, although the precise causal mechanism remains unclear.
Another approach within this rubric examines the relationship between different levels of
governments Author's personal copy
and the various kinds of banks. Here, however, dissimilar assumptions can yield
drastically different predictions. For example, if we assume that the central government sees the
Big Four state banks as vital policy tools, we would expect better short-term performance among
the Big Four state banks than among the small city commercial banks because Big Four banks are
“too big to fail” (Bongini et al., 2002). Moreover, as national-level bureaucratic entities, state
banks directly participate in the policy making process at the State Council level, which allows
them to lobby for preferential policies. Local commercial banks can only lobby indirectly through
the local governments. Thus, under these assumptions, we would expect the Big Four state banks
to perform much better than the CCBs, with the JSBs perhaps in the middle.

---

18
V. Shih et al. / China Economic Review 18 (2007) 15–34
The findings presented in this paper suggests an even more complicated mechanism than the
“too big to fail” story. To be sure, the Big Four state banks are too big to fail and enjoy numerous
forms of government subsidies. As a result, their overall performance and credit management are
within reasonable range. However, they do not perform nearly as well as the JSBs because of the
moral hazard created by their strategic importance in the economy. Our findings suggest —
although by no means conclusively — that public scrutiny from listing and the threat of closure
drive JSBs to excel.
Another political economy approach examines the impact of administrative location on banking
performance. Administratively, bank branches and banks are scattered in centrally administered
cities like Beijing, Tianjin, and Shanghai, in provincial cities, or in county and prefecture cities.
Presumably, banks in major metropolitan areas are under the most regulatory scrutiny because of
their strategic importance. Provincial branches of the Big Four or city commercial banks located in
provincial capitals would also be closely monitored by central regulators and would presumably
have a close working relationship with the provincial government.1 Meanwhile, banks at lower
levels of government are on average less regulated.
Given these conditions, a bank's administrative location is expected to have two effects on
bank performance. First, a bank's location in an administratively important city suggests tighter
regulation, which would result in less credit and capital risk. Second, depending on the mix of
policies, close ties with the provincial government can either increase the share of policy loans in
a bank's portfolio or artificially lower a bank's risk level and inflate its profitability due to
subsidies from the provincial government. Meanwhile, a prefecture level bank is further away
from the regulators' scrutiny and is subjected to more direct intervention by cash-strapped sub-
provincial governments. The conflicting logics at work here mirror our ambiguous finding in this
regard. While CCBs located in the three municipalities definitely perform better, their
counterparts in provincial capitals and in prefecture level cities perform at about the same level.
The limited data we have are far from sufficient to rigorously test all of the above hypotheses.
For example, both the social stability perspective and the fiscal blackmail perspective predict poor
bank performance for northeastern China. Without good longitudinal data on CCB performance,
SOE concentration, and fiscal balance, it would be impossible to parse out which explanation holds
more weight. Despite these drawbacks, there is sufficient data to raise serious doubts about some of
the conventional wisdom mentioned above. Moreover, many of the following comparisons further
add to our confidence about some commonly asserted notions about Chinese bank performance.
3. Data
Our data come from the 2002 People's Bank of China Banking Survey provided to us by
PBOC officials, which includes bank-level data on the Big Four state banks, 11 joint-stock banks
and 112 city commercial banks (see Appendix for the names of these banks). This data are
collected by
Author's personal copy
the PBOC through its regional offices and are aggregated at the Beijing headquarters.
Some indicators are collected quarterly, while others are collected annually. The dataset contains
numerous indicators unavailable in public sources. For example, since the PBOC has data on the
risk level of various loans, they are able to calculate risk adjusted capital adequacy ratio, whereas
an analyst using publicly available data is only able to calculate a simple capital adequacy ratio.
Because of PBOC restrictions, we are only allowed to conduct indirect analysis on the data.
1 This is especially the case after the formation of CBRC, which has provincial level jurisdiction rather than inter-
provincial jurisdiction that the PBOC has.

---

V. Shih et al. / China Economic Review 18 (2007) 15–34
19
By 2002, the major state banks and joint-stock banks have begun publishing data on credit risk
and asset quality. Thus, we are able to compare published data with our data for the major banks to
ascertain the discrepancies between the two data sources. According to PBOC officials,
discrepancies between the two data sources are expected since the PBOC collects many of its data
from banks during the first quarter of the year, while major banks publish their data later in the
year. In the interim, both the PBOC and the banks themselves have had an opportunity to verify
and revise some of the figures. Moreover, published NPL figures are calculated on the basis of
both the new five-category classification and the old “one overdue, two stagnant” (yiyu liangdai)
system, which classifies loans as normal (zhengchang), overdue (yuqi), doubtful (daizhi), and lost
(daizhang). In the mean time, NPL ratios reported in the PBOC dataset are solely calculated in the
four-category “one overdue, two stagnant” system.
Up to 2002, the four-category system was still the standard measure of NPLs in the PBOC
dataset because statistics departments in many banks still used the four-category system.2 We are
fully aware that the “one overdue, two stagnant” system reports a lower NPL ratio than the five-
category system, but we are mainly interested in the relative performance of banks, not the absolute
level of NPL ratio.3 As Table 1 reveals, banks that adopted the five-category system, including
Bank of China and the China Construction Bank, reported much higher NPL ratios than the NPL
ratios stated in the PBOC dataset. The discrepancies between published and PBOC NPL ratios are
much less for banks that published four-category NPL ratios.
For capital adequacy ratios (CAR), the discrepancies mainly stem from the difference between
the more accurate risk adjusted CARs reported in the PBOC data and the simple CARs calculated
by the authors on the basis published capital and asset figures (Table 1). If we take into account
the effects of the two different loan classification systems and the different ways of calculating
CARs, the PBOC data more or less conform to published data, at least in the ten banks for which
published data are available.
4. Methodology
In order to analyze the data without breaking our agreement with PBOC officials, we compare
bank performance along four dimensions using principal component analysis (PCA). Following
Canbas, Cabuk and Bilgin Kilic (2005), we use PCA to jointly take into account the information
provided by 10 financial ratios (Table 4) and generate four orthogonal indexes to measure banks'
performances. Factor scores were then calculated for each of the bank, and these scores were used
for comparing banking performance. Although our approach departs from the conventional
“frontier analysis” favored by the bank efficiency literature, we follow the prevailing wisdom in
the literature that bank performance is best measured by a group of indicators rather than by one
single indicator (Berger & Humphrey, 1997, Fraser, Phillips, & Rose, 1974). Our PCA approach
closely follows researchers who want to gauge the relative health of banks rather than the relative
efficiency of Author's personal copy
banks (Canbas et al., 2005).
As a first step, we determine how many factors we make use in our analysis. Table 2 reports the
estimated factors and their eigenvalues. Only those factors accounting for greater than 10% of the
2 Although we did not receive the latest PBOC data, we were told by PBOC officials that the four-category system still
applied to 2003 PBOC data.
3 Internal PBOC studies reveal that the five-category system increases NPL ratio by as much as 14%. See Geng & Qu,
2002. Xi guoyou shangye yinhang ‘yiyu liangdai' fahe ‘wuji fenlei' fa huafen buliangdaikuan chabie (Analyzing the
difference in categorizing non-performing loans between the ‘one overdue, two stagnant' and the ‘five-category' method
in state banks). Jinrong Cankao (Financial Reference) 2002.

---

20
V. Shih et al. / China Economic Review 18 (2007) 15–34
Table 1
A comparison of non-performing loan ratios and capital adequacy ratios between published data and PBOC data in 2002
for a selected group of banks
Published NPL
PBOC NPL
Published capital
PBOC capital
ratios
ratios
adequacy ratios
adequacy ratios
Industrial and Commercial Bank of China
22.21
22.47
5.54
5.8
Bank of China
22.49
18.79
8.15
8.75
China Construction Bank
15.17
11.89
6.91
7.03
China Merchant Bank
4.7
5.08
12.57
14.69
CITIC Industrial Bank
15.16
8.97
5.85
5.76
Fujian Industrial Bank
3.13
3.23
8.14
8.23
Bank of Communication
NA
14.22
8.83
5.95
Huaxia Bank
4.24
5.56
NA
8.37
Shenzhen Development Bank
10.29
10.29
9.49
9.78
Minsheng Bank
2.2
2.11
8.22
8.48
BOC and CCB only published five-category NPL ratios. CITIC Bank only published NPL amount and short-term and
long-term loan amount. Sources for the published figures include:Bank of China, 2003. 2002 Annual Report. Beijing:
BOC, Bank of Communication, 2003. Annual Report 2002. Shanghai: BOCO, China Construction Bank, 2003. Annual
Report 2002. Beijing: CCB, China Merchant Bank, 2003. Annual Report 2002. Shenzhen: CMB, CITIC Industrial Bank,
2003. Annual Report 2002. Beijing: CITICIB, Fujian Industrial Bank, 2003. Annual Report 2002. Fuzhou: FIB, Huaxia
Bank, 2003. Huaxia Gufen Youxian Gongsi 2003nian disan jidu baogao (Huaxia Share HoldingCompany's third quarter
report in 2003). Beijing: Huaxia Bank, Industrial and Commercial Bank of China, 2003. Annual Report 2002. Beijing:
ICBC, Minsheng Bank, 2003. Zhongguo Minsheng Yinhang gufen youxian gongsi 2002niandu nianbao (China Minsheng
Bank Share-holding Company's 2003 Annual report. Beijing: MSB, Shenzhen Development Bank, 2004. Shenzhen
Fazhan Yinhang gufen youxian gongsi 2003 nian niandubaogao (Shenzhen Development Bank Share-holding Company's
2003 Annual Report). Shenzhen: SDB.
variance (eigenvalues N1) are kept in the analysis. As a result, only the first four factors are finally
retained (Table 2). Among them, the first principal component factor (F1) accounts for 28% of the
variance of the 10 financial ratios. The other three component factors (F2, F3, F4) account for
14%, 11% and 10% of the total variance respectively. F1 through F4 altogether explain 64% of the
total variance of the financial ratios.
Table 3 presents the factor score coefficient matrix (wjk) estimated by PCA. To enhance these
factors' interpretability, we use the varimax factor rotation method to minimize the number of
variables that have high loadings on a factor. In other words, varimax rotation produces results which
make it the most likely to identify each variable with a single factor. This approach greatly enhances
Table 2
Eigenvalues of factors
Component
Eigenvalue
Proportion
Cumulative
F1
Author's personal copy
2.833
0.283
0.283
F2
1.418
0.142
0.425
F3
1.136
0.114
0.539
F4
1.001
0.100
0.639
F5
0.910
0.091
0.730
F6
0.803
0.080
0.810
F7
0.692
0.069
0.879
F8
0.565
0.057
0.936
F9
0.475
0.048
0.983
F10
0.168
0.017
1

---

V. Shih et al. / China Economic Review 18 (2007) 15–34
21
Table 3
Factor score coefficient matrix (wjk)
SLALR
SLDR
OLR
SLR
LLR
CCR
CAR
CRR
AP
CP
F1
0.136
0.167
−0.316
−0.359
−0.159
0.482
0.116
0.509
0.434
0.088
F2
0.423
−0.360
0.383
−0.293
0.246
0.044
0.573
−0.053
−0.057
0.250
F3
−0.466
0.143
0.333
−0.095
0.647
0.331
−0.016
0.211
−0.068
−0.255
F4
−0.216
0.440
−0.034
−0.176
0.174
−0.144
−0.064
−0.146
−0.024
0.807
SLALR: Asset turnover ratio; SLDR: Ratio of long-term debt to short-term debt; OLR: Overdue loan ratio;SLR: Stagnant
loan ratio; LLR: Lost loans ratio; CCR: Core capital ratio; CAR: Capital adequacy ratio; CRR: Capital risk ratio; AP: Asset
profitability ratio; CP: Capital profitability ratio.
our ability to make substantive interpretation of the main factors. Table 4 presents the factor loadings,
where variables with large loadings (N0.5) for a given factor are highlighted in bold.
As seen on Table 4, F1 reflects overall solvency of the banks and is highly correlated with core
capital ratio, capital risk ratio, asset profitability, and doubtful loan ratio (daizhi). As this factor
explains the most variance in the data, it constitutes the most informative indicator of a bank's
overall health. A high score in F1 suggests that a bank is doing well overall (low credit risk, high
capital adequacy, and high profitability). F2 likely reflects liquidity in a bank, as it is closely
correlated with the ratio of liquid asset to liquid debt, risk adjusted capital adequacy ratio, and
ratio of short- to long-term debt. However, high liquidity does not necessarily imply good overall
performance; it may also be a manifestation of a bank's low efficiency in asset utilization such
that it keeps a high ratio of liquid assets and maintains high risk-adjusted capital adequacy ratio by
acquiring treasury bonds instead of lending (Jiao, 2002: 85). Because of the ambiguity
surrounding this variable, it is valuable to consult other indicators in conjunction with this
liquidity indicator to evaluate the overall risk level of a bank.
F3 in all likelihood reflects credit risk because it is highly correlated with bad loan (daizhang)
and overdue loan ratios ( yuqi). Under the four-tier loan classification system, overdue loans are
loans that are overdue by less than 3 years, while bad loans are loans owed by bankrupt entities
(Fei, 1998). Here, it is puzzling that doubtful loan ratio, which is the proportion of loans overdue
by over 3 years, is so weakly correlated with bad and overdue loan ratios. This perhaps has
something to do with Ministry of Finance rules on bad debt provisioning which compel some
banks to shift all emerging bad loans into the doubtful loan category, leaving bad loans and
doubtful loans weakly correlated (Bureau of Economic Prediction of the State Information Center,
1999). Contrary to the first two factors, since this factor is positively correlated to credit risk, an
Table 4
Factor loadings of ten financial ratios after varimax rotation
F1
F2
F3
F4
Core capital
Author's personal copy
ratio
0.895
0.020
0.023
−0.045
Capital risk ratio
0.885
−0.018
−0.151
−0.042
Asset profitability
0.646
0.041
−0.334
0.115
Doubtful loan ratio
−0.590
−0.221
−0.076
−0.353
Short-term liquid asset-liability ratio
0.071
0.719
−0.250
0.121
Risk adjusted capital adequacy ratio
0.210
0.605
0.250
0.188
Ratio of short to long-term debt
0.249
−0.566
−0.128
0.285
Lost loan ratio
0.003
−0.134
0.803
0.025
Overdue loan ratio
−0.335
0.198
0.678
−0.082
Capital profitability
−0.039
0.044
−0.020
0.913

---

22
V. Shih et al. / China Economic Review 18 (2007) 15–34
Table 5
2002 Market share of State banks, Joint-stock banks, and CCBs in lending, deposits, and assets
Lending share
Deposit share
Asset share
Big Four
0.716
0.730
0.712
JSBs
0.209
0.200
0.207
CCBs
0.075
0.070
0.081
The remaining market share is divided by policy banks, rural credit cooperatives, and trust and investment companies.
Source: 2002 PBOC Banking Survey.
increase in this factor indicates an unhealthy bank. Finally, F4 consists of one profitability ratio,
capital profitability, which measures net profit as a share of bank capital base.
In order to facilitate analysis and interpretation, we further standardize the factor scores
assigned to each bank along a 1–10 scale. Suppose that Xi represents the F1 factor score for the ith
bank. We calculate the standardized F1 factor score for the ith bank using the following formula,
where Xmin is the lowest F1 score in the sample and Xmax is the highest F1 score in the sample:
Xi−Xmin  10
Xmax−Xmin
In the case of F3, because the factor score has an inverse relationship with bank health, we
employ the slightly different equation below to standardize it. Now, higher scores for the
standardized F3 represent better health in that category:
Xmax−Xi  10
Xmax−Xmin
The findings below are all reported in the standardized scores with higher values indicating
better performance which allow readers not only to gauge a bank's performance in a given
dimension relative to another bank, but also relative to the entire spectrum of banks.
5. Comparing different types of banks
The banking sector in China is essentially divided into four major categories. First and
foremost, the state banks, composed of both the policy banks and the Big Four state banks, make
up the bulk of the market (see Table 5).4 Because policy banks are policy agencies concerned with
objectives other than profitability, this study confines itself to the analysis of the Big Four state
banks. Second, national level joint-stock banks (JSBs) have mixed ownership and command an
increasing share of the market.5 Third, city commercial banks (CCBs) are joint-stock banks that
limit their operation to one city and surrounding area. Most of them were founded in the mid to
late 90s when Author's personal copy
city governments consolidated urban credit cooperatives (Han, 2000). In 2002,
there were a total of 112 CCBs scattered around China. Although CCBs command only a small
slice of the Chinese banking market (Table 5), they still control resources in the hundreds of
4 The Big Four State banks comprise of the Agricultural Bank of China (ABC), the Industrial and Commercial Bank of
China (ICBC), the Bank of China (BOC), and the Construction Bank of China (CCB).
5 According to official classification, national joint-stock banks include Bank of Communication (BOCO), CITIC
Industrial Bank (CIB), China Everbright Bank (CEB), Huaxia Bank (HXB), Minsheng Bank, Guangdong Development
Bank (GDB), Shenzhen Development Bank (SDB), China Merchant Bank(CMB), Shanghai Pudong Development Bank
(SPDB), Fujian Industrial Bank (FIB), and the Yantai Housing and Savings Bank (YHSB).

---

V. Shih et al. / China Economic Review 18 (2007) 15–34
23
billions of RMB and play an important role in local development (Han, 2000). The fourth type of
financial institutions is the rural credit cooperatives (RCCs), but due to the unavailability of data,
they remain beyond the scope of this study.
5.1. Risk and profit comparison between different segments
Table 6 compares the mean factor scores of the three segments of the banking sector along F1
to F4. The factor scores are continuous variables ranging from 0 to 10, with 10 equivalent to the
performance of the best bank in the group. First, although both joint-stock banks and city
commercial banks have share-holding structure, joint-stock banks perform much better on some
important dimensions. Second, joint-stock banks on average perform better than both state banks
and CCBs in almost every category. Third, the performance of CCBs is highly varied, a topic
explored in further detail later.
For F1, which measures overall health, the Big Four state banks and the CCBs both have an
average score around 7.4, while joint-stock banks do better at 8.14. This accords with anecdotal
evidence that national joint-stock banks perform better than both the Big Four banks and the
CCBs. It is worthwhile to note that the overall performance of CCBs varies greatly, producing a
standard deviation of 1.5 on a scale of 10.
For F2, which reflects liquidity of the banks, joint-stock banks (JSBs) once again out-perform
other types of banks, although not significantly so. There is, however, considerable variation in
JSB performance in this category, again calling into question whether low liquidity risk
necessarily reflects healthy performance. In contrast to F2, JSB's performance truly stands out in
F3, which is highly correlated with overdue and bad debt ratios. JSB performance in this category
is substantially higher than that of the Big Four state banks and of the CCBs. In combination with
its performance in F1, JSB performance in this category suggests much better credit risk
management than the other types of banks. JSB performance in this category is also uniformly
stellar, as the standard deviation for JSB is relatively small at 0.29. While on average CCBs have
higher credit management scores than the Big Four banks, they also have much greater variance
in performance. Finally, for F4, which is closely correlated with capital profitability ratio, all three
types of banks on average perform at the same level, although the variation of performance among
CCBs is again much higher.
The most encouraging finding in this comparison is the spectacular performance of the joint-
stock banks, which perform better than the other two types of banks over all and in credit risk
management. Several factors may well have contributed to their outstanding performance.
Although it is at first tempting to attribute the joint-stock banks' success to their ownership
structure, one has to remember that JSBs are nominally structured the same way as CCBs, which
on average do not perform much better than the Big Four banks. One reason for their stellar
performance is that many of them are now under public scrutiny. SPDB, Minsheng Bank, China
Table 6
Mean and
Author's personal copy
standard deviation of performance scores for State banks, Joint-stock banks, and City Commercial Banks
F1
F2
F3
F4
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Big Four
7.49
0.46
4.26
0.66
6.36
0.55
7.22
0.21
JSBs
8.14
0.62
5.29
1.35
7.10
0.29
7.17
0.46
CCBs
7.36
1.50
4.61
1.50
6.57
1.26
7.32
0.93

---