Journal of Business Research 57 (2004) 1341 – 1351

Determinants of capital structure of Chinese-listed companies

Jean J. Chen*
School of Management, University of Surrey, Guildford, Surrey GU2 7XH, UK
Accepted 21 March 2003

Abstract

This paper develops a preliminary study to explore the determinants of capital structure of Chinese-listed companies using firm-level panel
data. The findings reflect the transitional nature of the Chinese corporate environment. They suggest that some of the insights from modern
finance theory of capital structure are portable to China in that certain firm-specific factors that are relevant for explaining capital structure in
developed economies are also relevant in China. However, neither the trade-off model nor the Pecking order hypothesis derived from the
Western settings provides convincing explanations for the capital choices of the Chinese firms. The capital choice decision of Chinese firms
seems to follow a ‘‘new Pecking order’’—retained profit, equity, and long-term debt. This is because the fundamental institutional assumptions
underpinning the Western models are not valid in China. These significant institutional differences and financial constraints in the banking
sector in China are the factors influencing firms’ leverage decision and they are at least as important as the firm-specific factors. The study has
laid some groundwork upon which a more detailed evaluation of Chinese firms’ capital structure could be based.
D 2003 Elsevier Inc. All rights reserved.

1. Introduction examined characteristics of firms that were not similarly

Over the past 40 years, much of the capital structure
research has advanced theoretical models to explain the
capital structure pattern and also to provide empirical evid-
ence concerning whether the theoretical models have explan-
atory power when applied to the real business world. The
focus of both academic research and practical financial
analysis has been on those large corporations with publicly
traded debt and equity securities that dominate economic life
throughout the developed world.
Although the majority of the capital structure research has
focused on understanding the forces that influence corporate
financing behaviour of the U.S. firms, capital structure
research has become increasingly internationalised in recent
years, which provides researchers the opportunity to make
cross-sectional comparisons between countries and between
various industries around the world. In particular, Rajan and
Zingales (1995) applied the capital structure models derived
from a U.S. setting to firms in the G-7 countries and found
that the variables that were found to have correlation with
leverage in the United States were also correlated with
leverage of firms in other G-7 countries. Wald (1999)
* Tel.: +44-1483-689666; fax: +44-1483-686301.
E-mail address: j.j.chen@surrey.ac.uk (J.J. Chen).
correlated with leverage across countries. He demonstrated
that institutional differences could contribute to differences
in capital structure. His results indicate that institutions may
significantly influence firms’ capital structure decision and
that agency and monitoring problems, while existing in
every country, may create different outcomes.
While the majority of the research results has been
derived from the experience of developed economies that
have many institutional similarities (Hodder and Senbet,
1990; Rajan and Zingales, 1995; Wald, 1999; Ozkan,
2001; Chui et al., 2002; Bevan and Danbolt, 2002), little
work has been done to further our knowledge of capital
structure within developing countries that have different
institutional structures. Recently, Booth et al. (2001) pro-
vided the first empirical study to test the explanatory power
of capital structure models in developing countries. The
study used data from 10 developing countries to assess
whether capital structure theory was portable across coun-
tries with different institutional structures. It investigated
whether the stylised facts, which were observed from the
studies of developed countries, could apply only to these
markets or whether they had more general applicability. The
results were somewhat sceptical of this premise. They
provided evidence that firms’ capital choice decisions in
developing countries were affected by the same variables as
they were in developed countries. Nevertheless, there were

0148-2963/$ – see front matter D 2003 Elsevier Inc. All rights reserved.
doi:10.1016/S0148-2963(03)00070-5
1342 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351

persistent differences of institutional structure across coun- structure for individual firms that results from a trade-off
tries indicating that specific country factors were at work. between the tax benefits of increasing leverage and increas-
Their findings suggest that although some of the insights ing agency and bankruptcy costs that higher debt entails.
from modern finance theory are portable across countries, Although remaining as the mainstream theory of capital
much remains to be done to understand the impact of structure, the trade-off theory has failed to explain the
different institutional features on capital structure choices. observed corporate behaviour particularly witnessed with
Booth et al. (2001) selected countries operating a market- the stock market reaction to leverage-increasing and lever-
orientated economic system, which bore many similarities to age-decreasing transactions, which consistently yields stock
developed countries. It is interesting and important to know price increases and decreases, respectively. As an alternative
how capital structure theories work in a transitional economy to the trade-off model, the Pecking order hypothesis of
environment within which institutional structures differ not corporate leverage emerged based on asymmetric informa-
only from developed countries but also from developing tion problems (Myers and Majluf, 1984). It predicts that
economies. The People’s Republic of China is the largest firms will prefer internal financing to issuing security, and,
developing and transitional economy in the world, and if forced to resort to external financing, will use debt before
therefore is chosen as the focus of this study. Since this is equity. This model explains many observed patterns in
the first examination of its type, the aim of this study is to corporate finance including the tendency of firms not to
develop some preliminary groundwork that a more detailed issue stock and their choice to hold surprisingly large cash
evaluation could be based. It is hoped to answer the question reserves and other forms of ‘‘financial slack.’’
whether, and how closely, does the determinants of Chinese Also derived from asymmetric information problems,
capital structure support the Western finance theory? More various signalling models of capital structure have also been
specifically: proposed, which suggest that managers use leverage to
signal firm prospects to poorly informed outside investors
1. Are firm-specific factors correlated with leverage that who believe these signals because they are prohibitively
have been identified in the Western settings also similarly costly for weak firms to mimic (Ross, 1977).
correlated in China? There have been many empirical studies attempting to
2. Does the institutional structure in China affect Chinese test the explanatory power of capital structure models on
firms’ capital choice decision? corporate behaviour in developed countries, particular in a
3. Do the Western capital structure models have robust ex- U.S. setting. Most of the work has been to identify the
planatory power for Chinese companies in the Chinese determinants of capital structure. The main determinants of
economy? capital structure tested include profitability, size, growth
opportunity, asset structure, costs of financial distress, and
The remainder of this paper is organised into five sec-
tions. Section 2 covers a brief literature review of the capital Table 1
structure debate. Section 3 provides background to the Summary of the implications of capital structure theories and empirical
Chinese institutional environment. The data collection and evidences on the relationship of capital structure determinants with gearing
research method are presented in Section 4. Section 5 Determinants Predicted sign Sample empirical evidence
discusses the regression results, and finally Section 6 con- by the theories
cludes the study with implications of the findings and Profitability
(Pecking Kester (1986); Friend and Lang
suggestions for future research. order) (1988); Baskin (1989); Griner
and Gordon (1995); Shyam-
Sunder and Myers (1999)
+(trade-off, Bowen et al. (1982); Dammon
2. Review of capital structure debate signalling) and Senbet (1988); Givoly
et al. (1992)
In their landmark paper in 1958, Modigliani and Miller Size
(Pecking Kester (1986); Titman and
(MM) showed that if a company’s investment policy was order) Wessels (1988)
taken as given, then in a world of perfect markets—a world +(trade-off, Marsh (1982); Rajan and
signalling) Zingales (1995); Chittenden
without taxes, perfect and credible disclosure of all informa- et al. (1996)
tion, and no transaction costs associated with raising money Growth
(trade-off) Long and Malitz (1985)
or going bankruptcy—the extent of debt in a company’s opportunities ± (signalling, Titman and Wessels (1988);
capital structure would not affect the firm’s value. The Pecking order) Lang et al. (1996)
perfect capital markets they assumed have attracted a wide Asset structure +(trade-off) Long and Malitz (1985);
+(Pecking order) Chung (1993); Walsh and Ryan
variety of research of somewhat-less-than-perfect capital (1997)
markets. The development of agency theory in the 1980s, Cost of financial
(trade-off) Bradley et al. (1984); Friend
coupled with detailed research into the extent and effects of distress and Lang (1988); Walsh and
bankruptcy costs, has lead to the current mainstream view Ryan (1997)
that corporations act as if there is a unique, optimal capital Tax shields effects +(trade-off) Bradley et al. (1984)
 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351 1343

tax shields effects. Table 1 summarises the implications and Table 2

Summary for stock markets
empirical evidences of various capital structure theories on
the relationship between each of the above determinants and Item 1994 1995 1996 1997 1998 1999 2000
the level of gearing. No. of listed 291 323 530 745 851 949 1088
companies
No. of listed 345 381 599 821 931 1029 1174
stocks (100m)
3. China’s corporatisation and its institutional A shares 287 311 514 720 825 922 1060
environment B shares 58 70 85 101 106 108 114
Total market 3691 3474 9842 17,529 19,506 26,471 48,091
The most recent approach to reform large and medium- capitalisation
(100 million
sized state-owned enterprises (SOEs) in China is corporati-
yuan)
sation. At present, the government still holds the majority of A shares 3516 3311 9449 17,154 19,299 26,168 47,456
shares in the corporatised SOEs by direct shareholding and/ B shares 175 164 394 375 206 304 635
or indirect shareholding through state-owned institutions Negotiable market 969 938 2867 5204 5746 8214 16,088
such as state investment companies, state holding compan- capitalisation
(100 million
ies, and state asset management agencies. The major non-
yuan)
state ownership is individual shareholding; independent A shares 814 791 2514 4856 5550 7937 15,524
nonstate institutional investors are very rare. B shares 155 147 353 348 196 276 563
The corporatisation of SOEs has driven the change of Total turnover 8128 4036 21,332 30,722 23,544 31,320 60,827
the country’s financial system. As a result, two capital (100 million
yuan)
markets, the Shanghai Stock Exchange (SHSE) and the
A shares 8003 4319 21,052 30,295 23,418 31,050 60,279
Shenzhen Stock Exchange (SZSE), began to emerge in the B shares 125 78 280 427 127 270 548
early 1990s and have developed rapidly since the late Trading volume 101 705 2533 2560 2154.11 2932 4758
1990s. By the mid-2002, there were 1088 companies listed (100 million)
in the two stock exchanges with the majority being the A shares 99 681 2465 2471 2093 2810 4558
B shares 25 24 68 89 62 123 200
former SOEs (China Securities Regulatory Commission
web site). Source: China Statistical Bureau (2002).
There are some special features of the Chinese stock
markets. Shares are classified as A shares, which are des- assets over the past 3 years exceeds 30%, with an average
ignated for domestic investors, and B shares, which are annual return of net assets in each of these 3 years not less
designated for overseas investors, although the B share than 6%. The ceiling on the amount of new share issuing has
market opened to the local citizens in March 2001. The A also been cancelled.
shares typically consist of state shares owned by either the While stock markets are developing, China’s financial
central government or the local governments, legal-person sector has been under the strong grip of the state. The
shares owned by the state-owned institutions, or negotiable state monopoly of the financial sector has hindered the
shares owned by individual domestic investors. State shares development of China’s capital markets and the growth of
and legal-person shares account for almost two third of the nonstate financial institutions, in particular the bond mar-
total share issues, and they are not tradable in the stock ket. The access of Chinese firms to long-term debt
markets. Negotiable shares are the only class of A shares that provided by state banks has been strictly controlled by
can be publicly traded in stock exchanges. Table 2 summar- the state and the risk of default is high. In such a
ies some important features of the development of the situation, bankruptcy, even if enforced, may not be very
Chinese stock markets. Regulations governing the listed efficient.
companies of new equity issuance after the initial public China’s legal and institutional framework is still imma-
offering (IPO) are also striking. Before 1998, all new issues ture and incomplete. For example, the company law is
after IPO were implemented through ‘‘share allotment’’ ambiguous about the debtholders’ rights. It is seriously
(rights offer) to the existing shareholders proportionally at flawed in giving shareholders and government agencies
prices below market prices. The amount of share allotment too much power in bankruptcy procedures. Debtholders
was also limited to at most 30% of the existing equity capital are not given any control rights in liquidation. There is a
of the company once a year. Due to the high demand for lack of clearly defined private property rights and effective
shares, the existing shareholders usually accepted all offers property rights markets, and a lack of effective capital
available to them. To apply for new share issuance, a market for external corporate control. The above overview
company must ensure that its annual return on net assets in casts doubt on whether the Western models of capital
the past 3 years exceeded an average of 10%. Since 2001, the structure have explanatory power on the capital choices of
restrictions have been somewhat relaxed. A company can the Chinese corporations that are still largely owned or
issue new shares not only by share allotment but also by controlled by a state that is under economic transition. The
public offering in the stock markets if its total return on net determinants of Chinese firms’ capital structure may reflect
1344 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351

institutional structures and financial constraints rather than Table 4

Three different estimators of LEV equation
the result of economic rationale such as cost.
Dependent variable: LEV
Independent Fixed effects Random effects Pooled OLS
4. Research methodology variables
PROF
0.7954* * *
0.8255* * *
0.9768* * *
4.1. Sample set (0.0498) (0.0499) (0.0723)
SIZE 0.0249 0.0328 * * 0.0489* * *
(0.0188) (0.0157) (0.0125)
This study uses data from the annual report of 88 GROWTA 0.0772* * * 0.0721* * * 0.0551* *
Chinese public-listed companies for the period 1995 – (0.0168) (0.0169) (0.0251)
2000. The data set is called the Dow – China 88 Index, TANG 0.0991 * * 0.0713 * 0.0006
which is based on the entire Chinese stock market structure (0.0431) (0.0395) (0.0381)
EVOL
0.0024 *
0.0022 *
0.0011
created by the Dow – Jones in May 1996. The financial
(0.0013) (0.0013) (0.0021)
statements are prepared following the U.S. GAAP. Among NDTS 0.1062
0.0278
0.4276* * *
the 88 firms, 63 are from the SHSE and 25 are from the (0.0930) (0.0868) (0.0875)
SZSE. The listed companies represent the driving industrial DUM –
0.0232
0.0139
force in China so the sample may do well in capturing (0.0278) (0.0137)
_CONS 0.2730 * 0.2505 * 0.2010 *
aggregate leverage in the country.
(0.1626) (0.1383) (0.1096)
Since the balance sheets of the firms in financial sector No. of 462 462 462
(banks, insurance companies, and investments trusts) have a observations
strikingly different structure from those of nonfinancial R2 .5337 .5167 .5292
companies, financial firms are excluded from the sample. F statistic 61.8000 – 63.6600
Probability>F (.0000) – (.0000)
Those firms with any missing observations for any variable
Wald
2 124.7 484.3000 241.2
in the model during the period 1995– 2000 are also dropped. Probability>
2 (.0000) (.0000) (.0000)
As a result, the final sample set consists of a balanced panel AR(1) test: 4.078 10.9700 5.991
of 77 firms over a period of 6 years. (0.0000) (0.0000) (0.0000)
AR(2) test:
4.029 1.2910 5.343
(0.0000) (0.1970) (0.0000)
4.2. Variables
Root MSE 0.0783 0.0800 0.1333
Standard errors in parentheses are for coefficient and P values for
According to the research objectives and the research
diagnostic test.
questions this study has set, the variables used in this study * Significant at 10% level.
and their measurement are largely adopted from existing ** Significant at 5% level.
*** Significant at 1% level.
Table 3
Measurement of variables literature. This will allow me to highlight the similarities as
Variables Measurement well as the differences in the determinants of capital structure
Dependent variables
in other countries. The dependent variables are total leverage
Overall leverage (LEV) Ratio of book value of total debt to total and long-term leverage; the explanatory variables include
assets profitability, size, growth opportunities, tangibility, earnings
Long-term leverage (LLEV) Ratio of book value of long term debt to volatility, and nondebt tax shields. Their definitions are listed
total assets in Table 3. The book value is used for the calculation of
Independent variables
variables wherever applicable due to the fact that only about
Profitability (PROF) Ratio of earnings before interest, tax, 30% of the shares issued is tradable and there are extraord-
and depreciation to total assets inary capital gains resulting from secondary share trading.
Size (SIZE) Logarithm of total assets These two facts might introduce biases. In order to reflect the
Growth opportunities Sales growth/total asset growth (due to characteristics of firms listed on the SHSE and the SZSE, a
(GROWTA and GROWNO) the absence of R&D and advertising
expenditure data)
dummy group is introduced to control for any group-specific
Asset structure (TANG) Tangibility—ratio of tangible assets (the effect that may not be captured by the explanatory variables.
sum of fixed assets and inventories) to
total assets 4.3. Summary statistics and correlation matrix of variables
Cost of financial distress Earning volatility—absolute value of the
(EVOL) first difference of percentage change of
operating income
The summary statistics of dependent and independent
Tax shields effects (NDTS) Nondebt tax shields—ratio of variables, including mean, standard deviation, minimum and
depreciation to total assets (due to maximum, and a correlation matrix, are reported in Tables
depreciation is the most significant A3 and A4 in the Appendices. It can be seen that most cross-
element among nondebt tax shield) correlation terms for the independent variables are fairly
 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351 1345

small, thus, giving little cause for concern about the problem 4.5. Comparison of the models
of multicollinearity among the independent variables.
The estimation results are reported in Tables 4 and 5. The
4.4. Regression models fixed effects model has a slight statistical advantage over the
other two models. It has the lowest RMSE and the highest R2
Since the sample contains data across firms and over time, for both the total and long-term leverage equations. For the
the panel data method is employed. The basic regression joint test, all of the three models for LEV and LLEV are
model can be specified as follows: significant at a 5% critical level and they produce quite
similar results for profit, growth opportunities, and earning
yit ¼  þ X0it B þ uit i ¼ 1; . . . . . . ; 77; t ¼ 1; . . . . . . ; 6 volatility variables.
The Hausman specification test is employed to test the
where i denotes the cross-section dimension and t indicates fixed effects model versus the random effects model. The
the time dimension, Xit0 is a 1  k vector of observations on k test statistic for the total leverage equation is 15.77. The
explanatory variables for the ith firm in the tth period, B is a test is asymptotically
2 distributed with seven degrees of
k  1 vector of parameters, uit is a disturbance term and is freedom. The random effects model can be rejected in
defined as favour of the fixed effects model at a 5% critical level.
For the long-term leverage equation, the Hausman stat-
uit ¼ i þ vit istic is 9.78, which indicates that the null hypothesis
where i denotes the unobservable individual effect and vit cannot be rejected at any conventional level of signific-
ance. However, as it can be seem from Table 5, the R2
denotes the remainder disturbance. Three methods, pooled
among the three models is extremely low although the
OLS, fixed effects, and random effects, are used.
Wald joint tests for the models are significant at a 5%
critical level.
Table 5
Three different estimators of LLEV equation
Dependent variable: LLEV 5. Results and discussion
Independent Fixed effects Random effects Pooled OLS
variables 5.1. Regression results
PROF
0.0566 *
0.0624 * *
0.0878 *
(0.0306) (0.0303) (0.0501) The empirical evidence obtained has suggested that the
SIZE
0.0312* * *
0.0230 * *
0.0059 coefficients of profitability and growth opportunities are
(0.0115) (0.0101) (0.0087)
GROWTA 0.0278* * * 0.0249 * * 0.0019
significant for the total leverage regression (Table 4). The
(0.0103) (0.0103) (0.0174) coefficients of profitability, growth opportunities, tangib-
TANG 0.0837* * * 0.0862* * * 0.0755* * * ility, and size are significant for the long-term leverage
(0.0265) (0.0246) (0.0264) (Table 5). It shows that the three models offer quite similar
EVOL 0.0003 0.0002
0.0002 results but slightly different levels of significance in both the
(0.0008) (0.0008) (0.0014)
NDTS
0.0840
0.0604 0.0601
total and long-term leverage estimators. The significant
(0.0571) (0.0539) (0.0606) exception is that the coefficient of size is negative and
DUM / 0.0120 0.0146 highly significant in the long-tem leverage estimation but
(0.0209) (0.0095) positive in the total leverage estimation. However, it can be
_CONS 0.3014* * * 0.2204 * * 0.0819 argued that the positive coefficient of size for the total debt
(0.0999) (0.0891) (0.0760)
No. of 462 462 462
ratio is not significant in the fixed effects model. Therefore,
observations it could be concluded that large firms use more short-term
R2 .0698 .0550 .0562 finance and less long-term finance.
F statistic 4.0500 / 3.3700 Having further corroborated the relationships between
Probability>F (.0003) / (.0009) the significant explanatory variables and the dependent
Wald
2 15.5000 26.3400 16.5600
Probability>
2 (.0500) (.0010) (.0350)
variables, it is found that:
AR(1) test: 2.9560 10.0400 2.8610
(0.0030) (0.0000) (0.0040) 1. there is a negative relationship between profitability and
AR(2) test:
3.0310 0.4374 2.6440 debt;
(0.0020) (0.6620) (0.0080) 2. a positive relationship exists between growth opportunity
Root MSE 0.0481 0.0482 0.0924
and debt;
Standard errors in parentheses are for coefficient and P values for 3. there is a positive relationship between tangibility and
diagnostic test.
* Significant at 10% level.
debt;
* * Significant at 5% level. 4. a negative relation exists between a firm’s size and long-
* * * Significant at 1% level. term debt.
1346 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351
5.2. Discussion
5.2.1. Descriptive statistics: total leverage and long-term
leverage
Fig. 1 shows the striking difference between the ratio of
firms’ total leverage level and their long-term leverage
level in both stock exchanges, which implies that Chinese
firms usually prefer short-term loans rather than long-term
debt.
The 46% total book-debt level in China is below the
figure in most developed countries but the difference is
unremarkable. For example, in 1991, the mean of total
book-debt level in the G-7 countries was 66% (58% in
the United States) (Rajan and Zingales, 1995). It is also
close to the average total book-debt level of 51% in
developing countries (Booth et al., 2001). However, the
difference between total book-debt and long-term debt
ratio is much more pronounced in China. The long-term
book-debt level is only 7% in China compared with the
mean of 41% in the G-7 countries and 22% in devel-
oping countries. The substantially low amount of long-
term debt reflects the fact that the Chinese-listed com-
panies are mainly financed by share capital rather than
debt. Bank loans provide short-term financing for
working capital; equity is the main source of finance
for capital investment.
The main reasons may be because of the incomplete
institutional structure and legal system governing secondary
share trading and the defect of corporate governance struc-
ture in the corporatised SOEs inheriting from the public
ownership. The capital gains resulting from secondary share
trading are substantially high, usually about six to eight
times of the IPO prices (China Securities Regulatory Com-
mission). The inefficient corporate governance structure
provides the managers opportunities to personalise those
benefits or even take asset-stripping behaviour. Short-term
finance allows suppliers of capital to monitor and control
the borrowing of the firms more effectively due to the fact of
less default risks. This is particularly important considering
the prevailing agency problems in most corporatised SOEs.
To the extent that capital structure theories explain capital
Fig. 1. Average LEV and LLEV ratios (book value).
choices of firms in developed countries (e.g., assuming
well-developed legal systems and effective corporate gov-
ernance structures), this difference, in long-term versus
short-term debt, might limit their explanatory power in
China.
5.2.2. Profitability
Intuitively, the negative relationship between profitability
and debt in Chinese firms seems to support the Pecking
order model. However, upon taking another look, there may
be other reasons for this negative relationship rather than
those proposed by the Pecking order hypothesis such as to
avoid underinvestment problems and new projects being
mispriced.
As far as leverage is concerned, although banks are
willing to provide long-term bank loans to the listed firms
because of the influence of government-directed credit
policy, their capital resources are very much stretched.
The bond market is very much underdeveloped. Further,
listed firms are also attracted by equity finance due to the
substantial capital gains in the secondary markets. In addi-
tion, due to the corporate governance problems and the lack
of enforcement of company laws, individual shareholders do
not have adequate investment protection. Share capital has
become somewhat a ‘‘free’’ source of finance. The manage-
ment prefers equity financing rather than debt financing
because the former is not binding. Tax effects predicted by
the trade-off model are rather limited in China. This is
because the state is still the controlling stakeholder of firms
and the owner of banks as well as the beneficiary of tax,
which reflects China’s status as a centrally planned eco-
nomy. This induces firms to use equity finance as much as
possible. Therefore, equity finance is preferred to debt.
Retained profit is the quickest and easiest source of finance
for most companies compared with new equity issuance due
to the transaction costs associated with share issuance and
the restrictions on firms’ operating performance for apply-
ing for new equity issuance. Furthermore, since the majority
of new equity is issued through share allotments, new issues
usually lead to a decline in the firm’s stock price. Therefore,
retained profit is the preferred primary method of raising
 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351
additional capital. A new Pecking order of Chinese-listed
firms’ leverage thus appears—retained profit, then equity
finance, and lastly debt.
5.2.3. Growth opportunities
Another factor, which is supposed to affect capital struc-
ture, is growth potential. A positive relationship between
growth opportunities and debt in China is found, which
confirms the same relationship found in developed countries
except the United States (Wald, 1999).
According to the trade-off theory, firms holding future
growth opportunities, which are a form of intangible assets,
tend to borrow less than firms holding more tangible assets
because growth opportunities cannot be collateralised. Fur-
ther, agency theory argues that firms have a tendency to
expropriate wealth from debtholders (Myers, 1977; Jensen,
1986). Firms with greater growth opportunities have more
flexibility to invest suboptimally, and thus expropriate
wealth from debtholders to shareholders because of the asset
substitution effect. Ongoing growth opportunities imply a
conflict between debt and equity interests. Therefore, a
negative relationship is expected between debt and growth
opportunities.
However, the trade-off model does not apply to the
Chinese firms. One reason may be that most of the listed
firms are in the manufacturing and heavy industry sectors.
They possess more tangible assets and less intangible assets
such as good will, R&D, and advertising, and thus have
limited growth opportunities. This is a reflection of the
generally low technology level of Chinese firms.
Another reason may be that the equity market recog-
nises the value of growth opportunities, so do banks. This
recognition has been reflected in share prices. According
to the signalling model, high-value firms are able to use
more debt financing because debt has its dead weight
costs, which make less valuable companies more likely to
fall into bankruptcy (Ross, 1977). The signalling model
generally predicts that the firms with the best earnings and
growth prospects will employ the most leverage. Lang et
al. (1996) further argued that leverage was negatively
related to growth opportunities only for firms whose
growth opportunities were not recognised by the capital
market. The high market capitalisation in China may
indicate that the growth opportunities associated with
listed firms have been recognised by the capital market;
therefore, banks are willing to assign higher valuations to
highly levered firms and issue more long-term debt to
finance the firms’ growth opportunities.
5.2.4. Tangibility
Much research examining the correlation between lever-
age and tangibility in developed and developing countries has
proved that a positive relationship exists because tangible
assets are easy to collateralise for debt. From the viewpoint of
transaction cost economics, tangible assets usually have less
asset specificity, thus, increasing their use as collateralisation
1347
for debt to reduce lender’s risk (Williamson, 1988, 1975).
Agency theory also predicts the same relationship (Jensen
and Meckling, 1976). The agency costs of equity lead to
underinvestment problems. Further, the information asym-
metry results in new equity being underpriced. Issuing debt
secured by tangible assets reduces these agency costs. This
study confirms this positive relationship between a firm’s
leverage, particular long-term debt, and the tangibility of its
assets. It shows that asset tangibility is an important criterion
in banks’ credit policy, and this is particularly true for long-
term loans. This result is consistent with both the trade-off
model in terms of financial distress and bankruptcy costs and
the Pecking order hypothesis in terms of asset mispricing.
5.2.5. Size
Theoretically, the relationship between size and lever-
age is unclear. According to the trade-off model, large
firms are expected to have a higher debt capacity and are
able to be more highly geared. Large firms are more
diversified, thus, less exposed to the risk of bankruptcy.
They may also be able to reduce transaction costs asso-
ciated with long-term debt issuance. Another possibility is
that larger firms may have a more dilute ownership, and
thus have less control over individual managers. Managers
may then issue debt to reduce the risk of personal loss
resulting from bankruptcy (Friend and Lang, 1988).
Marsh’s (1982) survey of the literature concluded that
large firms more often chose long-term debt while small
firms chose short-term debt. Rajan and Zingales (1995)
and Wald (1999) suggested that size was positively
correlated with debt based on the data from developed
countries with Germany as an exception.
However, when size is used as a proxy for the (inverse)
probability of default, it should not be strongly positively
related with leverage in countries where costs of financial
distress are low. Furthermore, according to the Pecking
order hypothesis, informational asymmetries between
insiders within a firm and capital markets are expected to
be lower for large firms so large firms should be more
capable of issuing informationally sensitive securities like
equity (Kester, 1986). Titman and Wessels (1988) both
found evidence to support the negative hypothesis between
size and leverage.
The relationship of firm size to total debt in Chinese
firms produces a positive value, but this result is not
statistically significant in the fixed effects model. However,
the coefficient of size to long-term debt is negative and
highly significant, which suggests that firm size has a
negative relationship with long-term debt. This negative
relationship, however, may not be the result of informational
asymmetries suggested by Myers and Majluf (1984)
because the market capitalisation of equity in China is very
high; thus, undervalution of new equity may not be a
concern. The negative relationship between size and long-
term debt may be due to the fact that large firms have better
access to capital markets for equity finance because of their
1348 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351
reputation in the markets and the attraction of the capital
gains in the secondary markets. It may also be because the
bankruptcy costs are low in China since legal system is
incomplete and yet to be implemented. There is no protec-
tion of debtholders especially in the event of default. The
asset substitution problem prevails. Public bond market
virtually does not exist.
It can be further argued that agency theory (Jensen, 1986)
and transaction cost economics (Williamson, 1988) suggest
that large firms issue more long-term debt because they use
debt to better control management behaviours due to more
dilute ownership. However, smaller firms are more subject
to shareholder intervention in the case of mismanagement
because a reasonably small group of shareholders can gain a
controlling interest in the firm, as with small U.S. firms. The
corporate governance in the Chinese firms, in this respect, is
much like small U.S. companies in this way. The Chinese
government still owns a significant proportion of the com-
panies’ shares, and thus is the major stakeholder of the
firms. Therefore, in China, small numbers of professional
managers, usually selected by the state, control a sizeable
percentage of the listed firms’ stock and can force manage-
ment to act in the shareholders’ interests. If size proxies for
the relative dilution of control, the Chinese firms would
appear more like small U.S. companies. The result may
suggest that the centralised state control (state controlled
shares) is responsible for the negative relationship between
size and long-term debt.
6. Conclusion
A remarkable difference between the capital choices of
Chinese firms and firms in developed economies is that
Chinese firms prefer short-term finance and have substan-
tially lower amounts of long-term debt. To the extent that
theories of capital structure explain capital choices of firms
in developed countries, this difference, in long- versus short-
term debt, might limit their explanatory power in China. It
suggests that the theoretical underpinnings of the observed
correlations are still largely unresolved.
The results of this empirical study suggest that some of
the insights from modern finance theory are portable to
China in that certain firm-specific factors that are relevant
for explaining capital structure in the Western countries are
also relevant in China. This is true despite profound insti-
tutional differences that exist between China and the West-
ern countries. Knowing these factors could help predict the
financial structure of a firm. The Pecking order model along
with asymmetric information theory seems to provide partial
explanations.
However, a further investigation of firm-specific factors
correlated with leverage has shown that neither the trade-
off model nor the Pecking order hypothesis derived from
the Western settings has robust explanatory power in
explaining the capital choice preference of Chinese firms.
This is because the fundamental institutional assumptions
underpinning the Western models are not valid in China.
The capital choice decisions of Chinese firms seem to
follow a ‘‘new Pecking order’’—retained profit, then
equity, and lastly debt. The management of the firms
prefers equity financing rather than debt financing
because the former is not binding. The results imply that
significant institutional differences such as the legal
system governing companies’ operation and banking and
securities markets, ownership concentration and the cor-
porate governance structure of the listed firms, the agency
problems inheriting from public ownership, and the fin-
ancial constraints in the banking sector are all factors
influencing the roles of firm-specific factors on firms’
leverage decision. Knowing these differences are at least
as important as knowing the firm-specific factors they
measure.
The above findings have reflected the transitional nature
of the Chinese corporate environment. On the one hand, the
trade-off model has limited explanatory power in China in
that, for example, the effects costs of financial distress
(earning volatility, bankruptcy costs) are not significant.
This is because the Chinese environment still keeps some
features of a centrally planned economy. The state is still the
principal stakeholder of firms and the owner of banks as
well as the beneficiary of tax. If the state does not change its
controlling behaviour towards corporatised SOEs, those
firms are less likely to run into financial crisis compared
with their counterparts in private sectors, so the costs of
financial distress is likely to have much less effect on firms’
capital structure. On the other hand, certain firm-specific
factors that affect firms’ leverage in the Western countries
also affect Chinese companies’ leverage. This has shown
that Chinese-listed firms have followed the basic rules of a
market economy despite the state controlling ownership.
The business operation of these companies has shown a
profit-oriented nature.
This paper has laid some groundwork to explore the
determinants of capital structure of Chinese-listed compan-
ies upon which a more detailed evaluation could be based.
Further work is required to develop new hypotheses for the
capital choice decisions of Chinese firms and to design new
variables to reflect the institutional influence. A larger,
comprehensive, and detailed database is also required for
a further detailed capital structure study. In addition, this
study has also established a correspondence between cor-
porate governance of the firms and capital markets, though
preliminary, which should be further studied.
Acknowledgements
I would like to thank the two anonymous referees for
their detailed comments and suggestions, which improved
the quality of the paper. I am also grateful to Mr. Lichun
Zhang for his assistance on data collection.
 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351 1349

Appendix A

Table A1
LEV ratios
Ratios 1995 1996 1997 1998 1999 2000
LEV—Shenzhen
Mean 0.47131 0.46499 0.465352 0.454139 0.482438 0.55289
Median 0.533086 0.565761 0.498039 0.479479 0.522476 0.492388
S.D. 0.185249 0.187243 0.185411 0.240659 0.233648 0.369384
Minimum 0.113138 0.101311 0.141902 0.123601 0.117825 0.091919
Maximum 0.732885 0.699301 0.727202 1.041552 1.009489 1.800815
Count 23 23 23 23 23 23

LEV—Shanghai
Mean 0.447135 0.443107 0.417574 0.440641 0.452509 0.453046
Median 0.446125 0.435627 0.449763 0.413973 0.444713 0.449983
S.D. 0.177024 0.16367 0.17293 0.155476 0.176448 0.162636
Minimum 0.056372 0.083162 0.044308 0.085425 0.061965 0.030021
Maximum 0.81683 0.825065 0.790175 0.763757 0.920417 0.717421
Count 54 54 54 54 54 54

LEV—all sample
Mean 0.454356 0.449644 0.431845 0.444673 0.461449 0.482869
Median 0.46318 0.452391 0.460732 0.418361 0.460271 0.45828
S.D. 0.178639 0.170093 0.176891 0.18347 0.194177 0.245069
Minimum 0.056372 0.083162 0.044308 0.085425 0.061965 0.030021
Maximum 0.81683 0.825065 0.790175 1.041552 1.009489 1.800815
Count 77 77 77 77 77 77

Table A2
LLEV ratios
Ratios 1995 1996 1997 1998 1999 2000
LLEV—Shenzhen
Mean 0.06575 0.058462 0.05669 0.052137 0.055831 0.066501
Median 0.042836 0.026637 0.026523 0.0341 0.027317 0.052858
S.D. 0.110109 0.09114 0.084637 0.069144 0.085694 0.072182
Minimum 0 0
0.01801
0.01825
0.04251
0.04213
Maximum 0.530289 0.425444 0.3792 0.30965 0.348233 0.256729
Count 23 23 23 23 23 23

LLEV—Shanghai
Mean 0.086394 0.075027 0.074953 0.068855 0.064982 0.062883
Median 0.062835 0.037136 0.032973 0.017354 0.026104 0.029058
S.D. 0.101583 0.100604 0.10513 0.100353 0.097055 0.083504
Minimum
1.4e
05 0 0
0.00294
0.01706
0.00527
Maximum 0.493803 0.480632 0.423549 0.484013 0.458667 0.404255
Count 54 54 54 54 54 54

LLEV—all sample
Mean 0.080227 0.070079 0.069498 0.063862 0.062248 0.063964
Median 0.055644 0.035425 0.032828 0.020193 0.027077 0.035286
S.D. 0.103905 0.097575 0.099257 0.092012 0.09334 0.079835
Minimum
1.4e
05 0
0.01801
0.01825
0.04251
0.04213
Maximum 0.530289 0.480632 0.423549 0.484013 0.458667 0.404255
Count 77 77 77 77 77 77
1350 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351

Table A3
Summary statistics
Variable Observations Mean S.D. Minimum Maximum
LEV 462 0.4541394 0.1926236 0.0300211 1.800815
LLEV 462 0.0683129 0.0943057
0.042512 0.530289
PROF 462 0.0509647 0.1028456
1.466707 0.310276
SIZE 462 8.807845 0.535916 6.827678 10.31107
GROWTNO 462 0.2690713 1.427961
0.942656 22.13457
GROWTA 462 0.1701884 0.2688143
0.545685 1.838353
TANG 462 0.4946659 0.1767652 0.0155384 0.909136
EVOL 462 1.122254 3.313855 0.0006839 37.98547
NDTS 462 0.772868 0.0774104 0.0002099 0.742314
DUM 462 0.7012987 0.4581847 0 1

Table A4
Correlation matrix
LEV LLEV PROF SIZE GROWTA TANG EVOL NDTS
LEV 1.000
LLEV .3291 1.0000
(7.4722)* * *
PROF
.5213
.1181 1.0000
(
11.8329)* * * (
2.5486)* * *
SIZE .0843 .0032 .1711 1.0000
(1.8080) * * (0.0643) (3.7224)* * *
GROWTA
.1131
.0434 .3633 .0622 1.000
(
2.4392)* * * (
0.9231) (8.3554)* * * (1.3323) *
TANG .0854 .1751
.0644 .1953
.0812 .0233 1.0000
(1.8297) * * (3.8122)* * * (
1.3754 )* (4.2641)* * * (
1.7430) * * (0.4934)
EVOL .1801 .0372
.4013
.0082
.1881 1.0000
(3.9247)* * * (0.7941) (
9.3883)* * * (
0.1716) (
4.1053)* * *
NDTS
.0751 .1023
.1041 .2321
.1112 .0972 .3231 1.0000
(
1.6131) * (2.1991) * * (
2.2427) * * (5.1154)* * * (
2.3955)* * * (2.0903) * * (7.3199)* * *
* Significant at 10% level.
* * Significant at 5% level.
* * * Significant at 1% level.

References
Baskin J. An empirical investigation of the pecking order theory. Financ
Manage 1989;18:26 – 35.
Bevan AA, Danbolt J. Capital structure and its determinants in the UK—a
decompositional analysis. Appl Financ Econ 2002;12(3):159 – 70.
Booth L, Aivazian V, Demirguc-Kunt A, Maksimovic V. Capital structures
in developing countries. J Finance 2001;56(1):87 – 130.
Bowen RM, Daley LA, Huber CC. Evidence on the existence and determi-
nants of inter-industry differences in leverage. Financ Manage 1982;
11(4):10 – 20.
Bradley M, Jarrell GA, Kim EH. On the existence of an optimal capital
structure. J Finance 1984;39(3):857 – 78.
Chittenden F, Hall G, Hutchinson P. Small firm growth, access to capital
markets and financial structure: review of issues and an empirical in-
vestigation. Small Bus Econ 1996;8:59 – 67.
Chui ACW, Lloyd AE, Kwok CCY. The determinants of capital structure: is
national culture a missing piece to the puzzle? J Int Bus Stud 2002;
33(1):99 – 128.
Chung KH. Asset characteristics and corporate debt policy: an empirical
test. J Bus Finance Account 1993;20(1):83 – 98.
Dammon R, Senbet L. The effect of taxes and depreciation on corporate
investment and financial leverage. J Finance 1988;43:3 – 29.
Friend I, Lang L. An empirical test of the impact of managerial self-interest
on corporate capital structure. J Finance 1988;43:271 – 81.
Givoly D, Hayn C, Ofer R, Sarig O. Taxes and capital structure: evidence
from firm’s response to the tax reform act of 1986. Rev Financ Stud
1992;5:331 – 55.
Griner EH, Gordon LA. Internal cash flow, insider ownership and capital
expenditures: a test of the pecking order and managerial hypothesis.
J Bus Finance Account 1995;22:179 – 97.
Hodder JE, Senbet LW. International capital structure equilibrium. J. Fi-
nance 1990;45:1495 – 516.
Jensen MC. Agency costs of free cash flow, corporate finance and take-
overs. Am Econ Rev 1986;76:323 – 9.
Jensen MC, Meckling WH. Theory of the firm: managerial behaviour,
agency costs and the ownership structure. J Financ Econ 1976;3:305 – 60.
Kester CW. Capital and ownership structure: a comparison of United
States and Japanese manufacturing corporations. Financ Manage 1986;
15:5 – 16.
Lang L, Ofek E, Stulz RM. Leverage, investment and firm growth. J Financ
Econ 1996;40:3 – 29.
Long MS, Malitz J. The investment financing nexus: some empirical evi-
dence. Midl Corp Financ J 1985;3:53 – 9.
Marsh P. The choice between equity and debt: an empirical study. J Finance
1982;37(1):121 – 44.
Modigliani F, Miller MH. The cost of capital, corporation finance and the
theory of investment. Am Econ Rev 1958;48(3):261 – 97.
Myers SC. Determinants of corporate borrowing. J Financ Econ 1977;5:
147 – 76.
 J.J. Chen / Journal of Business Research 57 (2004) 1341–1351
Myers SC, Majluf N. Corporate financing and investment decisions when
firms have information that investors do not have. J Financ Econ 1984;
13:187 – 221.
Ozkan A. Determinants of capital structure and adjustment to long run
target. J Bus Finance Account 2001;28:175 – 98.
Rajan R, Zingales L. What do we know about capital structure?—Some
evidence from international data. J Finance 1995;50:1421 – 60.
Ross SA. The determination of financial structure: the incentive signalling
approach. Bell J Econ 1977;8:23 – 40.
Shyam-Sunder L, Myers SC. Testing static trade-off against pecking order
models of capital structure. J Financ Econ 1999;51:219 – 44.
1351
Titman S, Wessels R. The determinants of capital structure choice. J Finance
1988;43(1):1 – 19.
Wald JK. How firm characteristic affect capital structure: an international
comparison. J Financ Res 1999;22(2):161 – 88.
Walsh EJ, Ryan J. Agency and tax explanations of security issuance deci-
sions. J Bus Finance Account 1997;24(7):941 – 59.
Williamson OE. Market and Hierarchies: Analysis Antitrust Implications.
New York: The Free Press; 1975.
Williamson OE. Corporate finance and corporate governance. J Finance
1988;43(3):567 – 91.