Honor Thy Creditors Beforan Thy Shareholders

Honor Thy Creditors Beforan Thy Shareholders
Honor Thy Creditors Beforan Thy Shareholders

Giovanni Ferri
University of Bari
Via C Rosalba 53
70124 Bari, Italy
g.ferri@dse.uniba.it

Li-Gang Liu
ANZ Banking Group Limited
31/F, Exchange Square 1
8 Connaught St.
Central, Hong Kong
ligang.liu@anz.com

Honor Thy Creditors Beforan Thy
Shareholders: Are the Proªts of
Chinese State-Owned Enterprises
Real?*

Abstract
Chinese state-owned enterprises (SOEs) have become quite
profitable recently. As the largest shareholder, the state has not
asked SOEs to pay dividends in the past. Therefore, some have
suggested that the state should ask SOEs to pay dividends. Indeed,
the Chinese government has adopted this policy advice and
started to demand back dividend payments starting from 2008.
Although we do not question the soundness of the dividend poli-
cy, the point we raise is whether those profits are real if all costs
owed by SOEs are properly accounted for. Among others, we are
interested in investigating whether the profits of SOEs are still as
large as they claim if they were to pay a market interest rate.
Using a representative sample of corporate China, we find that
the costs of financing for SOEs are significantly lower than for
other companies after controlling for some fundamental factors
for profitability and individual firm characteristics. In addition, our
estimates show that if SOEs were to pay a market interest rate,
their existing profits would be entirely wiped out. Our findings
suggest that SOEs are still benefiting from credit subsidies, and
they are not yet subject to the market interest rates. In an envi-
ronment where credit rights are not fully respected, dividend pol-
icy, though important, should come second and not first.

1. Introduction

After many years of mounting losses, the latest data seem
to indicate that Chinese state-owned enterprises (SOEs)
are ªnally making proªts. A World Bank (2005) study

* We wish to thank Hans Genberg, He Dong, and Matthew Yiu

for helpful discussions and Angela Maria D’Uggento for offer-
ing her guidance to improve our sampling methodology. We
acknowledge the useful comments provided by an anonymous
referee, as well as by John Knight, May-Françoise Renard,

Asian Economic Papers 9:3

© 2010 The Earth Institute at Columbia University and the Massachusetts

Institute of Technology

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

claims that SOE proªt margins increased from 2.7 percent in 1999 to 5.7 percent in
2005, and growth in industrial proªts averaged 36 percent over the same period.
The study goes on to propose that the Chinese government should make an effort to
cash in its dividends, which SOEs reportedly “forget” to pay out to their main share-
holder. The World Bank study argues that this would be beneªcial to China’s public
ªnance and, more importantly, it would help rein in rapid investment growth in
China by imposing more discipline on SOE managers.

These results have surprised many because the general impression has been that the
majority of Chinese SOEs have low performance by international standards, as evi-
denced by the limited appreciation these ªrms obtain when listed in both the Hong
Kong and overseas ªnancial markets (Bai, Lu, and Tao 2006; Shan 2006). The World
Bank study also appears to have ignored the fact that SOEs have not been consistent
in honoring their obligations vis-à-vis bank debts, as SOE lending was perhaps the
main source of non-performing loans (NPLs) at state-owned commercial banks
(SOCBs) (Zhou 2004).1 Therefore, one may question whether the proªts are as high
as reported if the costs of capital (not only including paying dividends to sharehold-
ers but also market-based interest rates to creditors) are properly accounted for. In
addition, there are issues as to how much dividend the SOEs would have to pay to
the government and whether the dividend policy could be at the expense of the
health of the banking sector as long as SOCB creditor rights are not respected.
Finally, continued inferior creditor right protection could signiªcantly impair the
improvement of SOCB corporate governance deriving from their diversiªed owner-
ship, which was recently obtained with the stock exchange listings, thus leading to
repeated write-offs of bad loans as experienced in some other transition economies
should the economy experience a large downturn.

Indeed, before accepting the favorable interpretation put forward by the World
Bank study and discussing dividend policy, one has to ask whether these SOE
proªts are real. Speciªcally, two issues need to be addressed. First, even though
various analyses concur that SOE proªtability has improved, the assessment is not
always as rosy as in the World Bank study. OECD (2005) shows that SOE returns
are signiªcantly improving: For example, on average, in 2002–03, total factor

Changwen Zhao, and by several other participants at the 2010 Asian Economic Panel meet-
ing in Seoul. The views presented in the paper are those of the authors alone and do not rep-
resent those of the institutions with which they are afªliated.

1 Throughout the paper, for simplicity, we refer to the group consisting of the Industrial and

Commercial Bank of China, Bank of China, China Construction Bank, and Agricultural Bank
of China as the SOCBs. Though we are aware that three of these four banks have been listed
on the stock exchange (Agricultural Bank of China is about to be listed at the time of writ-
ing) and the state is no longer the single shareholder (even if it still holds the majority).

51

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

productivity increased by 5 percent, delivering a return on assets (ROA) of 10 per-
cent, twice as large as that in 1998–99. But the report is cautious to state that large
pockets of weak SOEs remain.2

Second, another aspect deserving special attention is how bank debts are treated in
calculating costs. It is well known that SOEs in China are quite reluctant to pay back
their loans to SOCBs. This is because the relationship between SOEs and SOCBs was
politically inºuenced and forbearance on debt has been the rule rather than the ex-
ception (Cull and Xu 2000, 2003; Brandt and Li 2003). These considerations raise
doubts on whether the proªts currently posted by China’s SOEs are as high as re-
ported or they could fade away once the generally accepted accounting principles
are used.

Indeed, the preferential treatment accorded by SOCBs to SOEs can be grounded nei-
ther on SOEs’ returns (which are noticeably lower than at private companies) nor
on SOEs being less leveraged. Thus, it seems that SOEs do not entirely respect credi-
tor rights.

The objective of this paper is thus twofold. First, we want to provide a careful as-
sessment of Chinese SOEs’ proªts by adjusting for realistic interest service outlays.
We ªrst estimate the interest service outlays at market rates, benchmarking them
also to Chinese private enterprises. We then use these estimates to impute what
would be the “realistic” costs of bank debts for SOEs and compute the revised ªg-
ures for proªts. The resulting adjusted proªts would provide a more credible assess-
ment of SOE performance. As we will show, in spite of this adjustment, we still ªnd
that there has been an improvement in SOE proªtability over recent years.

The second issue we address is a policy one. That is, whether China’s government
would be better served by encouraging SOCBs to improve their lending practices
rather than just cashing in SOE proªts. As it stands—even after the successful stock
exchange listings of China Construction Bank, Bank of China, and Industrial and

2 Speciªcally, the report ªnds: (1) Performance is best where the state controls the company

through a large minority stake, or where state ownership is intermediated (i.e., at indirectly
state-controlled ªrms, with state legal person controlling shareholders); and it is the worst at
directly state controlled companies. (2) Although private Chinese and listed SOEs compare
quite favorably (though there is a visible bias to listing best performing SOEs), an interna-
tional comparison of enterprise rates of ROA still ranks the Chinese median ªrm return the
lowest in a group of the worst performing OECD and non-OECD economies. (3) Two-thirds
of state held ªrms in the industrial sector earn less than a 5 percent rate of return on assets
prior to payment of interest and nearly 15 percent of state-controlled industrial companies
trade with negative equity funds. (4) Distressed companies now represent 7 percent of ªrms,
11 percent of workers, 23 percent of assets, and 22 percent of outstanding debts.

52

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Commercial Bank of China—the government is the main shareholder at both SOEs
and SOCBs. As such, the government might consider that stiffening credit policies
by SOCBs may be more effective than simply cashing in SOE dividends. Indeed, as
ªnance theory postulates, shareholders should be residual claimants on ªrms’
proªts only after all creditors are duely paid. In other words, SOCBs have a priority
claim on SOE proªts and the government could take this opportunity to encourage
SOCBs to improve their credit management vis-à-vis SOEs. More importantly, by re-
asserting creditor rights, this would further improve the credit culture and the mar-
ket economy in general in China.

The rest of the paper proceeds as follows. Section 2 recaps the relevant literature at
the heart of the debate on whether and to what extent Chinese SOEs have become
proªtable. Here, we also summarize the issue of policy-inºuenced loans to induce
SOCBs to keep lending on favorable terms to SOEs. Section 3 outlines the methodol-
ogy we employed to build a representative sample. Section 4 is devoted to the em-
pirical analysis by ªrst presenting some preliminary descriptive ªndings and then
performing our regression analysis to come up with accurate estimates of the loan
rate subsidy SOEs appear to enjoy. Section 5 computes the adjusted proªts for SOEs
and shows that, after making SOEs pay loan rates on par with otherwise equivalent
private enterprises, SOE proªts are entirely wiped out on the average of the refer-
ence period. Based on these calculations, Section 6 discusses policy implications
and concludes.

2. The landscape of the Chinese corporate sector and a literature review

Since the mid 1990s, the landscape of the Chinese corporate sector has experienced
signiªcant changes. From a database of ªrms with annual sales of 5 million yuan
maintained by the National Bureau of Statistics of China (NBS), we ªnd that the
share of private enterprises in the total number of enterprises has increased from
6.5 to 45 percent and the share in total value-added from 2.6 to 18 percent from 1998
to 2005. The SOE share and its share in total value-added decreased rapidly, respec-
tively, from 39.2 to 10 percent and from 57 to 37.6 percent. An OECD (2005) study
shows that, according to pre-tax returns on equity (ROE), private ªrms are generally
more proªtable than SOEs, although SOEs’ proªtability is also on the rise. In addi-
tion, private ªrms have a lower level of indebtedness (measured as a percent of their
assets) and a lower debt/equity ratio as compared to state ªrms. Nonetheless, debt/
equity ratios are decreasing for both private and state-owned enterprises.

Despite being more proªtable, private companies continue to face difªculties in
their access to bank credit. According to the same OECD study, about 41 percent of
private enterprises have no access to credit and 56 percent have no access to bank

53

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Table 1a. Proªtability of Chinese enterprises by ownership

All enterprises
State-owned enterprises
Collective enterprises
Private enterprises

Source: OECD, Economic Surveys, China.

ROA

1998

6.1
4.8
11.2
7.8

2003

12.2
10.2
16.5
15.0

ROE

1998

3.8
2.0
10.8
6.0

2003

12.2
10.2
16.5
14.4

credit. Smaller private enterprises are even more constrained than other ªrms.
For private ªrms, major hurdles are a lack of collateral and ownership discrimina-
tion. Even though their access to bank ªnancing is improving with a 67 percent
increase in lending between 1998 and 2003, private companies are still ªnan-
cially constrained.

Public ownership in the banking and industrial sectors appears to be one of the key
factors behind the fragility of Chinese banking. Two statistics are revealing: al-
though SOEs’ contribution to the Chinese GDP was around 25 percent, they re-
ceived about 65 percent of total loans. In addition, the ROA and ROE of private
companies are higher than those of state enterprises (see Table 1a).

Another aspect deserving special attention among the factors determining SOE
proªts is how bank debts are treated in calculating costs. It is well known that SOEs
in China are quite reluctant to pay back their loans to SOCBs. This is because the re-
lationship between SOEs and SOCBs was politically inºuenced and forbearance on
debt has been the rule rather than the exception (Cull and Xu 2000, 2003; Brandt and
Li 2003; Héricourt and Poncet 2009). These considerations raise doubts about
whether the proªts currently posted by China’s SOEs are as high as reported or if
the proªts could fade away once the generally accepted accounting principles are
used. A simple calculation using the data published by the OECD (2005) helps ex-
emplify this point. By taking the ratio of the interest outlays to debt outstanding, we
impute the implicit interest rate companies pay to creditors. Imputed interest rates
on debt are signiªcantly lower for SOEs with respect to private companies and the
difference does not disappear over the years (see Table 1b).

Indeed, such preferential treatment accorded by creditors to SOEs can be grounded
neither on SOEs’ returns (which, as seen, are noticeably lower than at private com-
panies) nor on SOEs being less leveraged (the debt gearing ratio measured as stock
of debt over value-added is visibly larger at SOEs and the difference is, if anything,
increasing; see Table 1b). Although requiring more careful study, this evidence

54

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Table 1b. Imputed interest rates on debt and gearing ratios of enterprises by ownership

Imputed interest rates on debt

SOEs (a)
Private enterprises (b)
Difference (a) (cid:2) (b)

Gearing ratios

SOEs (c)
Private enterprises (d)
Ratio (c) / (d)

1999

5.06
6.44
(cid:2)1.38

302.20
122.70
2.46

2000

4.68
5.83
(cid:2)1.15

252.20
108.00
2.34

2001

4.31
5.65
(cid:2)1.34

238.90
92.00
2.60

2002

4.26
5.49
(cid:2)1.23

222.90
80.10
2.78

2003

3.92
5.23
(cid:2)1.31

193.90
72.70
2.67

Source: Our calculations on data from OECD, Economic Surveys, China.

seems to be consistent with the hypothesis that SOEs do not entirely respect credi-
tors’ rights.

These descriptive analyses are also conªrmed by more rigorous statistical analyses.
Xiao (2006), using the NBS data set of over 20,000 large and medium-sized ªrms for
the period 1995–2002 and adjusting for ªrm characteristics and fundamentals, ªnds
that SOEs are still much more likely to generate bad debts for the banking system
than non-state enterprises despite the fact that non-performing debts in SOEs have
been falling since 2000. Bai, Lu, and Tao (2006) use the same data set but a different
sample period ranging from 1998–2003 to investigate whether privatization or own-
ership change brings about economic and social efªciency. They ªnd that ownership
reform helps increase economic efªciency in those reformed ªrms. Speciªcally, Bai
et al. attribute the reduction of agency costs, measured by the ratio of administrative
costs, to the improvement of economic efªciency. Using a different survey data set
with 12,400 ªrms in 120 Chinese cities conducted by the World Bank, Dollar and Wei
(2007) also ªnd that state-owned ªrms have low marginal returns to capital relative
to private and foreign ªrms. Such efªciency losses amount to 5 percent of GDP if
SOEs can improve ªnancial controls and corporate governances further.

Although these existing studies shed light on SOE performance, they do not address
the issue of whether the proªts of SOEs are real after other costs are accounted for,
especially interest rate costs. We focus on this issue in the following sections.

3. Data, sampling methodology, and descriptive statistics

Our data sample is obtained from the NBS database, which contains more than
280,000 industrial ªrms with annual sales above 500 million yuan. It is estimated
that the ªrms included in this census represent about 80 percent of all industrial
value-added activities among Chinese ªrms. For yearly data we use approximately
69 ªnancial indicators including, asset, liability, revenue of major activities, proªts,

55

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

value-added taxes, intermediate industrial input, cash ºows, debt payments, and
other indicators that allow us to conduct our analysis. Given that there are some ma-
jor discrepancies in certain ªnancial indicators for data before 2000, we start our
data sample from 2001 to avoid such problems. In addition, given that it was impos-
sible for us to obtain the whole database, we use a sampling methodology to con-
struct a representative sample to reºect the NBS database.

Our sample was constructed by following two methodological rules. First, we ex-
tracted a random component designed to make a closed sample of Chinese enter-
prises. Second, because of a large number of ªrm drop-outs, which resulted from
enterprises’ birth and disappearance and/or from merger and acquisition (M&A)
activity as well as statistics that were discontinued by the NBS, we superimposed
the closed sample component over an open sample component. The latter compo-
nent was randomly extracted.

The closed sample component was built according to the following considerations
and methodology. To respect the bounds represented by the necessity to minimize
costs and time, we determined the dimension of the sample (n) on the basis of the
ªnancial resources of the research/project and of the tolerable error, with a conª-
dence level of 95 percent. We obtained a sample composed of 5,497 units based on
the following formula:

(cid:2)

n

{[(

N

(cid:3)

2
z N
a
2
/
(cid:3)
(cid:5)
2
P
) / (
1
1

(cid:4)

P

)]

z

2
a

}

/

2

,

where n is the number selected for the sample size; z is a standardized variable with
mean 0 and variance 1; 1 (cid:2) (cid:3) is the degree of trust; N is the total number of units in
the population to be sampled; (cid:4) is the allowed error size; and P is the unknown pro-
portion, which we set at 0.5.

To select the statistical units, we used a stratiªed random sampling method that pro-
vides greater precision and gives a better representation of the original population
than a simple random sample of the same size. Moreover, providing greater preci-
sion, a stratiªed sample generally requires a smaller sample size, although this ad-
vantage is achieved at the cost of more administrative and operative efforts com-
pared to the simple random sample.

In this perspective, with reference to the 2001 data, we divided the population of
211,181 ªrms (N) into 14,250 strata, deriving from the combination of four stratiªca-
tion variables that we considered the most relevant for the aims of the research. The
stratiªcation variables are:

56

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

• Province (30 sub-strata);
• Ownership (5 sub-strata: SOE; cooperatives; private enterprises; enterprises with
capital from Hong Kong and/or Macau and/or Taiwan; foreign-owned enter-
prises);

• SITC Sectors (19 sub-strata);
• Size of employment (5 sub-strata: 0–99 employees; 100–299 employees; 300–499

employees; 500–999 employees; 1,000 employees and above).

On the basis of these stratiªcation variables, starting from the distribution of the
population of the ªrms (N), we deªned the sample design using a proportional to
size allocation technique. According to this method, the frequencies of the statistical
units in each stratum of the stratiªed sample are proportional to those of the strati-
ªed population. In other words, with proportional stratiªcation, the sample size of
each stratum is proportional to the population size of the stratum and this means
that each stratum has the same sampling fraction. This technique is based on the as-
sumption that selection costs and variances are about equal across strata.3

To overtake the practical problem of the proportional selection from the population
strata containing a low number of ªrms, we introduced a cut-off value that excludes
all the cells with a frequency less than 14 units (that means 0.008 percent of the pop-
ulation) from the selection. The allocation of the 5,497 units of the sample among the
strata is shown in Tables 5 and 6 hereafter.4 The ªnal sample (n) is formed by sum-
ming the random samples obtained from each stratum. Finally, because our research
question regards the speciªcity of SOEs, we oversampled SOEs within each stratum.

The open sample component was then added to the observations extracted to form
the closed sample. The superimposition of this additional component should also
help minimize our sampling error.

The composition of the total sample by ownership class is described in Appendix 1.
The second column reports the percentage shares in the a priori base closed sample
and the third column shows the shares in the a priori total sample, that is, after
oversampling SOEs and after superimposing the open sample component. Columns
(4) to (8) report the actual shares in the ex post total sample. It is possible to notice

3 The advantages of proportional stratiªcation are: (1) it provides equal or better precision

than a simple random sample of the same size; (2) the gains in precision are greatest when
values within strata are homogeneous; and (3) the gains in precision accrue to all survey
measures.

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

4 The number of observations within each stratum Nh is known, and N (cid:5) N1

(cid:6) N2

(cid:6) N3

(cid:6)

. . . (cid:6) NH 1

(cid:6) NH.

57

Asian Economic Papers

Honor Thy Creditors Beforan Thy Shareholders

that the ex post shares are reasonably close to the a priori ones. Only the SOEs are
slightly under represented. Finally, the size of the sample is on average near that of
the a priori desired number, however observations are somewhat under-sampled in
year 2004 and somewhat over-sampled in 2005.

4. Empirical analysis

Our empirical analysis hinges on estimating the costs of debt for SOEs and compar-
ing them to what we found for the other ownership classes, particularly for private
enterprises. We construct two different proxies to measure the costs of debt:

intrateit

(cid:5) intpayit / totdebtit,

intrate1it

(cid:5) ªncostit / totdebtit,

(1)

(2)

where intpayit is the interest payment for ªrm i in year t; totdebtit is the total debt for
ªrm i outstanding at the end of year t; and ªncostit is the total ªnancial costs for ªrm
i in year t.

Even though the proxy in equation (1) is the appropriate measure of the interest
rate, the proxy in equation (2) may be a better measure of the total costs of debt be-
cause it also includes non-interest costs. For our purposes, it makes sense to consid-
er both proxies.

Before moving to the econometric analysis, we present some descriptive evidence
on the two proxies and on other basic performance measures. According to our in-
terest, all of these measures are broken down by ownership class.

4.1 Descriptive statistics
The perception that SOEs pay lower rates is conªrmed at the descriptive level. This
is true irrespectively of whether we consider intrate or intrate1. The average data for
intrate are reported in Table 2a, while those for intrate1 are shown in Table 2b.

With respect to intrate, SOEs paid 133 basis points less than the total sample average
during the sample period from 2001–05. The SOE gap amounts to 265 basis points
with respect to cooperative enterprises and 198 basis points compared to private en-
terprises. However, intrate for SOEs does not seem to differ signiªcantly from two
other special classes of enterprises—those with ownership located in Hong Kong,
Macau, and Taiwan and those with ownership located outside of greater China.
Therefore, lower costs of debt for SOEs, especially with respect to private ªrms, is
mostly systematic across the years.

58

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Table 2a. Estimated interest rate by year and ownership—intrate

intpay / debt

SOE (tot)

Coop

Private

HK-Macau-Taiwan

Foreign capital

Total

2001
2002
2003
2004
2005

2.46
2.23
2.67
2.86
2.61

Memorandum items:

2001–05
2001–03
2004–05

2.55
2.46
2.72

4.94
4.65
5.38

10.46

5.20
4.96
10.46

4.84
4.64
4.61
3.81
4.57

4.53
4.69
4.31

Source: Authors’ calculations based on our own database.

2.89
2.81
2.59
2.17
2.29

2.57
2.77
2.25

2.98
2.61
2.28
1.92
2.38

2.47
2.62
2.22

4.13
3.92
3.93
3.31
3.93

3.88
3.99
3.70

Note: Interest rates are calculated as the ratio of interest payments in the year to total debts outstanding at the end of the year. We

excluded outlying ªrms with negative interest payments or with intrate

100%.

Table 2b. Estimated interest rate by year and ownership—intrate1

ªncost / debt

SOE (tot)

Coop

Private

HK-Macau-Taiwan

Foreign capital

Total

2001
2002
2003
2004
2005

2.55
2.66
2.73
2.67
2.51

Memorandum items:

2001–05
2001–03
2004–05

2.62
2.65
2.58

10.09
5.79
12.10

12.45

5.96
5.70
12.45

7.14
10.76
13.35
7.12
7.26

4.87
4.99
4.69

Source: Authors’ calculations based on our own database.

3.19
3.34
3.01
2.49
4.37

2.76
2.70
2.87

3.56
3.41
3.12
2.60
3.90

3.37
3.35
3.41

4.37
4.18
4.17
3.57
4.41

4.19
4.24
4.10

Note: Interest rates are calculated as the ratio of ªnance costs in the year to total debts outstanding at the end of the year.

We excluded outlying ªrms with negative ªnance costs or with intrate1

100%.

We reach similar results after examining intrate1. With respect to the entire period,
SOEs pay 157 basis points less than the average company, 225 basis points less than
private enterprises, 4 basis points less than Hong Kong, Macau, and Taiwan compa-
nies, and 75 basis points less than foreign capital ªrms. Note that the favorable gap
for SOEs does not reduce visibly over time.

The low costs of debt for SOEs seem to be neither justiªed on the grounds of better
proªtability nor on the basis of lower leverage, where both variables affect the prob-
ability of default—negatively for the former and positively for the latter. The basic
measure of proªtability we consider is ROA, given by the ratio of total proªts to to-
tal assets.

From Table 3a, we notice that average ROA increases by approximately 1 percent-
age point (from 6.26 percent in 2001–03 to 7.24 percent in 2004–05) while reaching
6.66 percent on average, a relatively low value by international standards. However,
proªtability varies noticeably across ownership classes from the highest levels

59

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Coop

Private

HK-Macau-Taiwan

Foreign capital

Total

Honor Thy Creditors Beforan Thy Shareholders

Table 3a. ROA by year and ownership

2001
2002
2003
2004
2005

SOE (tot)
(cid:2)1.09
(cid:2)0.98
(cid:2)1.07
(cid:2)0.33
(cid:2)2.50

Memorandum items:

2001–05
2001–03
2004–05

(cid:2)1.23
(cid:2)1.05
(cid:2)1.54

9.67
9.44
13.48

15.21

10.88
10.72
15.21

8.54
8.13
9.04
8.89
10.76

9.23
8.59
10.08

3.49
3.66
4.26
3.26
3.67

3.70
3.82
3.52

5.62
5.24
6.54
4.94
6.31

5.82
5.82
5.82

Source: Authors’ calculations based on our own database.

Note: ROA is calculated as the ratio of proªts to total assets at the end of the year.

Table 3b. Weighted ROA by aggregated ownership class

SOE (tot)

2001–05

0.92

Memorandum items:

2001–03
2004–05

0.77
1.14

Coop

4.59

4.50
7.46

Private

HK-Macau-Taiwan

Foreign capital

6.05

5.46
6.66

3.94

3.91
3.99

8.25

6.55
10.12

Source: Authors’ calculations based on our own database.

Note: ROA is calculated as the ratio of the sum of proªts within the class to the sum of total assets within the class.

6.05
5.86
6.80
6.34
7.78

6.66
6.26
7.24

Total

3.61

2.95
4.50

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

reached by private ªrms (9.23 percent and increasing between the two sub-periods)
and cooperative enterprises (10.88 percent), to the intermediate values for Hong
Kong-Macau-Taiwan companies (3.70 percent and slightly decreasing between the
two sub-periods) and foreign capital ªrms (5.82 percent and stable), to the minimum
of SOEs, which record persistently negative levels ((cid:2)1.23 percent for the entire pe-
riod, worsening from (cid:2)1.05 to (cid:2)1.54 percent across the two sub-periods). Neverthe-
less, judging the SOE sector performance on this per capita level would be mislead-
ing if, as it happens, the improvement in performance is achieved mostly by the
larger SOEs.

To be sure, we should remark that if we use a weighted average, the ROA for SOEs
is no longer negative: It is 0.92 percent over 2001–05 and, even though remaining
well below those for the other ownership classes, shows some improvement from
0.77 percent in 2001–03 to 1.14 percent in 2004–05 (Table 3b).

Table 3c reports the leverage ratio, deªned as the ratio of total debts over total liabil-
ities. The sample average suggests that leverage increases only slightly from 59.46 in
2001–03 to 59.93 percent in 2004–05 and averaged 59.65 percent over the entire pe-
riod, which is a relatively high value by international standards. In addition, lever-

60

Asian Economic Papers

Honor Thy Creditors Beforan Thy Shareholders

Table 3c. Leverage ratio by year and ownership

SOE (tot)

Coop

Private

HK-Macau-Taiwan

Foreign capital

2001
2002
2003
2004
2005

67.80
68.61
69.13
72.17
69.12

Memorandum items:

2001–05
2001–03
2004–05

68.49
67.68
69.64

62.95
62.12
60.13

52.38

60.86
61.18
52.38

62.25
61.05
58.64
61.45
60.05

60.25
60.20
60.30

55.72
55.91
52.47
54.76
57.17

55.13
54.49
56.16

53.92
51.84
52.59
54.54
51.54

52.59
52.63
52.53

Source: Authors’ calculations based on our own database.

Note: Leverage is calculated as the ratio of total debts to total liabilities both at the end of the year.

To come up with the average values reported in the table, we excluded some outlying ªrms with negative debts.

Total

61.01
60.18
58.38
61.34
59.54

59.65
59.46
59.93

Table 4a. Firm average asset size (billion CNY) by year and ownership class

SOE (tot)

Coop

Private

HK-Macau-Taiwan

Foreign capital

Total

2001
2002
2003
2004
2005

259.2
237.4
183.7
234.7
259.0

Memorandum items:

18.8
19.2
21.9

16.9

14.7
16.3
16.2
16.5
22.5

2001–05

234.5

20.6

17.7

Source: Authors’ calculations based on our own database.

60.4
62.2
57.1
53.9
67.9

61.0

82.7
83.1
86.4
78.5
133.8

96.0

67.1
63.5
54.1
63.6
69.9

63.8

age is systematically higher for SOEs and has increased over the years. Foreign-
funded ªrms have the lowest leverage at 52.59 percent, followed by the Hong Kong-
Macau-Taiwan companies at 55.13 percent, private ªrms at 60.25, cooperative enter-
prises at 60.86 percent, and SOEs at 68.49 percent.

Obviously, the low costs of debt for SOEs might be explained by other factors. For
instance, a major expected difference between the SOEs and private enterprises is
asset size, whereby creditors might grant lower borrowing rates to SOEs because
their large asset size can be utilized for collateral and makes them less likely to de-
fault. Indeed, asset size differs noticeably across ownership classes (Table 4a).
Typically, SOEs are more than twice as large as foreign-owned ªrms, almost four
times as large as enterprises receiving capital from Hong Kong-Macau-Taiwan, and
more than ten times as large as cooperatives or private ªrms.

The employment number may be another indicator. As shown in Table 4b, against
the overall average number of 208 per-ªrm employees, SOEs are twice the average
(453), whereas Hong Kong-Macau-Taiwan and foreign-funded companies are in the

61

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Table 4b. Firm average number of employees by year and ownership class

SOE (tot)

Coop

Private

HK-Macau-Taiwan

Foreign capital

Total

2001
2002
2003
2004
2005

473
476
477
321
495

Memorandum items:

2001–05

453

149
143
146

86

144

117
118
115
105
113

114

Source: Authors’ calculations based on our own database.

296
303
319
288
314

306

266
264
267
260
296

273

217
216
213
179
207

208

Table 4c. Intrate and intrate1 by size class (average number of employees, N)

N 17
[5%]

4.64
5.08

16 N
26
[10%]

4.85
4.70

25 N
48
[25%]

4.41
4.78

47 N
86
[50%]

4.29
4.68

85 N
171
[75%]

3.83
4.07

170 N
313
[90%]

3.35
3.71

312 N
493
[95%]

2.81
3.18

Intrate
Intrate1

Source: Authors’ calculations based on our own database.

Note: Interest rates are calculated as the ratio of interest payments (ªnance costs) in the year to total debts outstanding at the end of

the year. To come up with the average values reported in the table, we excluded some outlying ªrms with negative interest payments

(ªnance costs).

intermediate ranking with 306 and 273 employees, respectively. Cooperatives have
144 employees and private enterprises have 114 employees and are placed at the
bottom of the ladder.

Indeed, the data conªrm that the costs of debt are noticeably lower as ªrm size in-
creases. Table 4c reports the interest rates according to our two deªnitions. The drop
in the costs of debt is particularly visible as ªrm size moves beyond 85 employees,
which is the median value in the sample.

Another consideration is the industrial sector. SOEs are traditionally concentrated in
sectors that may require economies of scale as a natural monopoly. Specialization in
these sectors might also induce lenders to perceive lower probabilities of default for
SOEs. Indeed, as shown in Table 5, the degree of over-representation of SOEs is larg-
est in tap water production and supply and in electric power, steam, and hot water.
These two sectors comprise approximately one-third of the total SOEs in our sample
and it is worth noticing that the cost of debt in these two sectors—2.76 and 2.94 per-
cent, respectively—is far below the average (3.88 percent).

Finally, the costs of debt may also vary across provinces where, at times, the level of
economic development is low, the degree of privatization is small, the industrial

62

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Table 5. Distribution by sector and presence of SOEs

By sector

06-Coal mining and dressing
08-Ferrous metals mining and dressing
09-Nonferrous metals mining and dressing
07-Petroleum and natural gas extraction
10-Nonmetal minerals mining and dressing
11-Logging and transport of timber and bamboo
12-Fishing
13-Food processing
14-Food production
15-Beverage production
16-Tobacco processing
17-Textile industry
18-Garments and other ªber products
19-Leather, furs, down, and related products
20-Timber, bamboo, cane, palm ªber, and straw
21-Furniture manufacturing
22-Papermaking and paper products
23-Printing and record medium reproduction
24-Cultural, educational, and sports goods
25-Petroleum processing and coking
26-Raw chemical materials and chemicals
27-Medical and pharmaceutical products
28-Chemical ªber
29-Rubber products
30-Plastic products
31-Nonmetal mineral products
32-Smelting and pressing of ferrous metals
33-Smelting and pressing of nonferrous metals
34-Metal products
35-Ordinary machinery manufacturing
36-Special purposes equipment manufacturing
37-Transport equipment manufacturing
39-Electronic equipment
40-Electric equipment and machinery
41-Electronic and telecom equipment
42-Instruments, cultural, and ofªce machinery
43-Other manufacturing
44-Electric power, steam, and hot water
45-Gas production and supply
46-Tap water production and supply

Source: Authors’ calculations based on our own database.

A priori composition
of the total sample
% share

Ex post composition
of the total sample
% share

SOEs
% share

1.22
0.41
0.17
0.00
0.39
0.00

6.16
1.89
0.81
0.00
10.32
5.30
2.64
1.95
1.03
2.42
2.25
1.33
0.39
6.52
0.88
0.43
0.82
4.66
7.19
2.15
1.39
5.37
7.98
4.06
4.44
5.88
3.58
1.25
1.95
0.06
3.00
0.00
1.93

1.10
0.35
0.08
0.00
0.45
0.00
0.00
5.62
1.95
0.75
0.01
9.93
5.58
2.67
1.86
1.07
2.45
2.13
1.41
0.41
6.34
0.93
0.43
0.88
4.68
6.97
2.02
1.21
5.45
7.55
3.89
4.35
3.62
4.57
2.21
1.76
0.87
2.77
0.00
1.68

2.74
0.00
0.00
0.00
0.00
0.00
0.02
9.15
3.90
1.56
0.00
2.54
0.28
0.12
0.53
0.02
0.81
7.75
0.02
0.00
5.23
0.49
0.00
0.26
0.97
7.16
0.26
0.00
1.93
7.02
4.97
8.18
1.36
1.46
0.83
0.85
0.32
19.78
0.00
12.23

structure is highly concentrated in heavy or resource-oriented industries, and for-
eign presence is minimal. As shown in Table 6, SOEs are at least twice as repre-
sented with respect to the overall sample in 17 provinces, in decreasing order:
Xizang, Qinghai, Xinjiang, Shaanxi, Gansu, Heilongjiang, Guangxi, Guizhou, Jilin,
Shanxi, Beijing City, Yunnan, Jiangxi, Tianjin, Liaoning, Hunan, and Nei Mongol,
mostly in the western and northeastern heavy industry hinterland. It is important to
highlight that in 13 of these 17 provinces the average cost of debt is below—often
much below—the national average.

Although these simple statistical analyses may be revealing, an econometric frame-
work is still required to show whether this is indeed the case empirically.

63

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Table 6. Distribution by province and presence of SOEs

By province

Zhejiang
Guangdong
Jiangsu
Shandong
Shanghai City
Fujian
Liaoning
Henan
Hebei
Tianjin
Hunan
Beijing City
Sichuan
Hubei
Jiangxi
Shanxi
Anhui
Guangxi
Jilin
Heilongjiang
Guizhou
Shaanxi
Chongqing
Nei Mongol
Gansu
Yunnan
Xinjiang
Xizang
Ningxia
Qinghai

A priori composition
of the total sample
% share

Ex post composition
of the total sample
% share

SOEs
% share

16.40
15.10
17.00
8.60
5.80
4.50
4.80
3.60
2.80
2.50
2.40
2.40
2.00
1.70
1.50
1.10
1.20
1.00
1.10
0.90
0.80
0.70
0.50
0.50
0.50
0.40
0.20
0.10
0.10
0.00

16.58
16.47
16.47
8.81
5.99
4.75
4.33
3.44
2.97
2.34
2.31
2.20
1.86
1.60
1.23
1.22
1.16
0.95
0.95
0.83
0.78
0.56
0.52
0.48
0.43
0.43
0.19
0.07
0.07
0.03

2.52
7.09
1.64
3.98
2.09
1.95
12.24
5.42
4.75
6.98
4.67
7.25
1.85
2.88
3.78
4.99
0.83
4.53
3.90
4.02
3.31
3.15
0.04
0.97
2.11
1.36
1.06
0.49
0.00
0.16

Source: Authors’ calculations based on our own database.

Speciªcally, we need to control for proªtability, asset size, ªrm size, industrial sector,
and the location of each enterprise. We turn to the empirical framework in the fol-
lowing sub-section.

4.2 Econometric results
In line with the previous discussion, the regressions we estimate have the following
form:

yit

(cid:5) (cid:3) (cid:6) (cid:7)

1SOEit

(cid:6) (cid:7)

2COOPit

(cid:6) (cid:7)

(cid:7)

6PROVINCEit

3HKMTWit
(cid:6) (cid:7)
7Xit

(cid:6) (cid:7)
(cid:6) ε

4FORKit
it ,

(cid:6) (cid:7)

5SECTORit

(cid:6)

(3)

where the dependent variable yit will alternatively be intrateit or intrate1it; SOE,
COOP, HKMTW, and FORK are dummies taking value 1, respectively, for SOEs, co-
operatives, companies owned from Hong Kong-Macau-Taiwan, and enterprises
with ownership outside of greater China, and taking value 0 otherwise. Here the
omitted variable is PRIV, which is the dummy for private enterprises. SECTOR is a

64

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

vector of 0–1 sectoral dummies and PROVINCE is a vector of 0–1 locational dum-
mies. Here, several insigniªcant sectors (provinces) are omitted. Xit is a vector of
variables that control for economic fundamentals that may affect the costs of debt.
These variables include asset size or employment size, and history of the enter-
prise.5 Index i refers to the ªrm, and t runs from 2001 to 2005 for the estimation on
the entire sample, from 2001 to 2003 for the estimation on the ªrst sub-sample, and
from 2004 to 2005 for the estimation on the second sub-sample.

We ran three panel regressions—similarly for intrate and intrate1—with the ªrst one
estimated over the entire period 2001–05, the second focusing on the 2001–03 sub-
period, and the third estimated over the 2004–05 sub-period. All estimates were per-
formed using a generalized least squares panel speciªcation with random effects
and robust standard errors. The choice of the random effects estimation must be
justiªed because if we were to adopt it for a sample that should instead be estimated
via ªxed effects our estimates would be inconsistent. Our choice is motivated by
three considerations. First, the number of the observations in the sample for each
year is rather large (the minimum is 5,597 in 2004, and the maximum is 9,276 in
2005). Second, under the alternative ªxed effects speciªcation, regressors other than
the constant term were generally insigniªcant and the ªt of the model was really
poor. Third, the Hausman test often suggested that the ªxed effects model should be
rejected.6

Table 7 shows the results for intrate, and conªrms our expectations. Larger enter-
prises have lower costs of debt. In addition, with respect to ownership, even after
controlling for size, location, and productive specialization, SOEs, Hong Kong-
Macau-Taiwan ªrms, and foreign-owned companies still pay lower loan rates com-
pared to private companies (i.e., the omitted ownership dummy), and no signiªcant
difference is detected for the cooperatives. To be sure, the debt costs gap compared

5 We also attempted speciªcations by including proªtability (ROE and ROA) and leverage,

but the results turned out with the “wrong” sign, that is, negative on proªtability and posi-
tive on leverage. This would seem to imply a soft budget constraint whereby creditors are
enforcing higher payments on enterprises that can afford them. While postponing further
analysis to future work, for the moment, we decided to discard these performance variables.

6 As the values of the Hausman test reported in the tables will show, rejection was achieved at
rather comfortable conªdence levels for intrate regressions. This was not the case for the
intrate1 regressions where the Hausman test suggested rejection of the ªxed effects model at
a comfortable level for the 2004–05 sub-period, and for the total 2001–05 sample; on the con-
trary, according to the Hausman test, the ªxed effects speciªcation could not be rejected for
the 2001–03 sub-period. Nevertheless, we opted for the random effects model for the intrate1
regressions in light of the fact that the Hausman test can be misleading when the ªxed ef-
fects model has a poor ªt since the matrix may not be positively deªnite.

65

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Table 7. Random effects panel estimates for intrate

Explanatory variables

SOE
Coop
HK-Macau-Taiwan
Foeign Capital
Log(employees)
Constant

Number of obs
R-square between
R-square overall
Wald square
Hausman test(1)

Province dummies
Sector dummies

Entire sample 2001–05
(cid:2)0.02209 ((cid:2)14.41***)
(cid:2)0.00140 (0.33)
(cid:2)0.01041 ((cid:2)8.01***)
(cid:2)0.01391 ((cid:2)11.49***)
(cid:2)0.00407 ((cid:2)7.18***)
(cid:2)0.05815 (20.59***)

20861

0.0645
0.0609
705.17***

Sub-sample 2001–03
(cid:2)0.02423 ((cid:2)14.43***)
(cid:2)0.00120 (0.23)
(cid:2)0.01056 ((cid:2)6.02***)
(cid:2)0.01425 ((cid:2)9.25***)
(cid:2)0.00518 ((cid:2)6.95***)
(cid:2)0.06607 (17.47***)

12788

0.06
0.0563
525.28***

Sub-sample 2004–05
(cid:2)0.01880 ((cid:2)7.20***)
(cid:2)0.02838 (1.18)
(cid:2)0.01086 ((cid:2)7.02***)
(cid:2)0.01396 ((cid:2)8.23***)
(cid:2)0.00304 ((cid:2)4.11***)
(cid:2)0.04899 (13.58***)

8073

0.0803
0.0775
312.18***

25.43 [0.1135]

12.15 [0.7334]

25.74 [0.0409]

YES
YES

YES
YES

YES
YES

Source: Authors’ calculations based on our own database.

Note: The values reported in square brackets is the conªdence level at which the hypothesis that the ªxed effects model should be used

may be accepted. Thus, in this case, the hypothesis that the ªxed effects model should be used is accepted, respectively, at the 11%, 73%

and 4% level.

***Indicates that the hypothesis that the coefªcient is zero can be rejected at the 1 percent level of conªdence.

Table 8. Random effects panel estimates for intrate1

Explanatory variables

SOE
Coop
HK-Macau-Taiwan
Foeign Capital
Log(employees)
Constant

Number of obs
R-square between
R-square overall
Wald square
Hausman test(1)

Province dummies
Sector dummies

Entire sample 2001–05
(cid:2)0.02543 ((cid:2)13.79***)
(cid:2)0.00082 (0.19)
(cid:2)0.01071 ((cid:2)7.12***)
(cid:2)0.00878 ((cid:2)5.77***)
(cid:2)0.00404 ((cid:2)6.90***)
(cid:2)0.06097 (21.20***)

26591.00
0.06
0.05
635.46***

Sub-sample 2001–03
(cid:2)0.02556 ((cid:2)11.28***)
(cid:2)0.00249 (0.42)
(cid:2)0.01410 ((cid:2)8.17***)
(cid:2)0.00960 ((cid:2)5.41***)
(cid:2)0.00532 ((cid:2)7.23***)
(cid:2)0.06981 (19.03***)

16122.00
0.05
0.05
497.12***

Sub-sample 2004–05
(cid:2)0.02525 ((cid:2)10.31***)
(cid:2)0.03493 (1.18)
(cid:2)0.00650 ((cid:2)2.80***)
(cid:2)0.00682 ((cid:2)2.68***)
(cid:2)0.00311 ((cid:2)3.76***)
(cid:2)0.05265 (13.10***)

10469.00
0.07
0.06
290.62***

51.05 [0.0001]

21.40 [0.2090]

38.79 [0.0007]

YES
YES

YES
YES

YES
YES

Source: Authors’ calculations based on our own database.

Note: The values reported in square brackets is the conªdence level at which the hypothesis that the ªxed effects model should be used

may be accepted.

***Indicates that the hypothesis that the coefªcient is zero can be rejected at the 1 percent level of conªdence.

to private enterprises is largest for SOEs, intermediate for foreign-owned compa-
nies, and lowest for Hong Kong-Macau-Taiwan ªrms. The cost of debt for SOEs is
lower than for otherwise equivalent private companies by 221 basis points over the
entire period. While this gap decreases somewhat (54 basis points) between the ªrst
sub-period 2001–03 and the second sub-period 2004–05, it still runs at 188 basis
points in the later sub-period. Thus, the distortion coming from the credit market is
certainly not trivial.

The results for intrate1 are also consistent with our a priori (see abridged results in
Table 8). Firm size implies lower interest rates while SOEs, Hong Kong-Macau-

66

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Taiwan ªrms, and foreign-owned companies enjoy a lower cost of debt with respect
to private companies. Finally, here the gap compared to private enterprises is largest
for SOEs, intermediate for Hong Kong-Macau-Taiwan ªrms, and lowest for foreign-
owned companies. Interestingly, the results obtained for intrate1 in terms of the dy-
namics of the gap, which is on average 254 basis points, show no evidence of a re-
duction over time, which is something that was instead observed for the estimates
on intrate.

5. Revised SOE proªtability: An estimate

Revised SOE proªts were computed according to the following methodology. In
practice, we augmented the interest payment (ªnancial cost) variable by means of
the value of the coefªcients estimated in the regression. In turn, we considered two
types of corrections—the interest rate estimated for private enterprises, and the in-
terest rate estimated for foreign capital enterprises. Thus, for instance, according to
the private enterprise correction, SOEs’ interest payments (according to the intpay
method) are increased for the entire 2001–05 period by 86.5 percent because the esti-
mated coefªcient is (cid:2)0.02209, which is exactly 86.5 percent of SOEs’ actual cost of
debt (0.02552). In turn, referring to the foreign capital enterprise correction, SOEs’
interest payments (according to the intpay method) are increased for the entire
2001–05 period by 32.0 percent since the coefªcient estimated for the foreign capital
enterprises ((cid:2)0.01391) is subtracted from the coefªcient estimated for the SOEs
((cid:2)0.02209) thus delivering 0.00818, which is exactly 32.0 percent of SOEs’ actual cost
of debt. The same method is applied to the other sub-periods and also, mutatis mu-
tandis, to intrate1. By doing this, we are putting SOEs on the same par with private
companies or with foreign capital enterprises.

Next, we calculate the ratio of the additional interest payment (ªnancial cost)—what
SOEs should pay were they treated as private ªrms—to the observed SOEs’ total
proªts. The results of these calculations are reported in Table 9.

According to our calculations, SOEs’ proªts would have been entirely wiped out if
SOEs were made to pay the same interest rates as otherwise equivalent private en-
terprises. Over the entire sample period, the percentage of SOE proªts dented by the
correction to their cost of debt is 155.9 and 171.8 percent of proªts, respectively, de-
pending on whether we refer to intrate or intrate1. The only good news is the obser-
vation that the percentage drops signiªcantly moving from the ªrst sub-period
2001–03 to the second. Nevertheless, even in the second sub-period (2004–05),
92.5 percent of the proªts would be cancelled according to the intrate method and
even more following the intrate1 method.

67

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Table 9. Percentage of SOE proªts dented by applying their private enterprise interest rates

Private enterprise interest rates

Foreign capital enterprise interest rates

Year

2001–05

Intpay method
(Intrate)

155.9

Memorandum items:

2001–03
2004–05

221.7
92.5

Finco method
(Intrate1)

Intpay method
(Intrate)

Finco method
(Intrate1)

171.8

214.8
127.7

57.7

91.3
23.8

112.5

134.2
93.2

Note: Based on panel random effects estimates with robust standard errors.

If, instead of imputing private enterprise interest rates, we charge SOEs the interest
rates estimated for foreign capital enterprises, the drop in SOE proªts is still huge.
Over the entire 2001–05 period, the drop amounts to 57.7 and 112.5 percent accord-
ing to the intpay and ªnco methods, respectively.

6. Conclusion

This paper investigated whether the proªts of SOEs in China are real using a data
set that is representative of corporate China (i.e., the (cid:8)250,000 ªrms with annual
sales over 50 million yuan) for the period 2001 to 2005 by China’s NBS. Our empiri-
cal ªndings suggest that SOE proªts might be overstated because SOEs have histori-
cally beneªted from subsidized bank credit. Owing to political interference inducing
state-owned commercial banks (SOCBs) to lend to them, SOEs were hardly disci-
plined by lenders. Although SOE losses were the major source of SOCB NPLs, it is
common knowledge that even those SOEs repaying their loans do so on favorable
terms. Thus, subsided loans to SOEs contribute to a large extent to maintaining
SOCBs’ poor proªtability track record and, through that, cause repeated capital in-
jections in the SOCBs by the government.

Within our framework, the paper made three contributions. First, we constructed a
representative sample of corporate China where SOEs are oversampled purposely to
allow more precise identiªcation of their peculiarities compared to other Chinese
enterprises. Second, using that sample, we proved that, indeed, the cost of debt is
signiªcantly lower for SOEs, even after controlling for individual ªrm features.
Third, we estimated that should SOEs pay the same loan rates as otherwise equiva-
lent private enterprises, their additional interest outlays would be larger than SOE
proªts on average from 2001–05 and—even though decreasing relatively—the addi-
tional interest payment would still wipe out all SOE proªts from 2004–05, the most
recent years in our sample. And even charging SOEs the same interest rates esti-

68

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

mated for foreign capital enterprises—thus making the hardly tenable assumption
that the ability/willingness to honor debt obligations is the same for the two classes
of companies—the SOE proªts would at least halve if not vanish.

Accordingly, it seems that safeguarding creditors’ rights should be the utmost prior-
ity. This means inducing SOEs to pay market interest rates on (and to service scru-
pulously) their loans. Through that, SOCB performance would improve greatly—in
a way consistent with the new incentives after the stock exchange listing of three of
the four SOCBs—and the state would save further recapitalization. In addition, in
this case, SOE managers would undergo creditors’ discipline, to which the disci-
pline exerted by dividends could be added should proªts still remain positive after
adequate loan servicing.

Overall, our results suggest that strengthening creditors’ rights should be the pri-
mary step in the process to complete transition to a market economy in China by
bringing SOEs under more strict discipline. Dividend policy should come second
and not ªrst.

References

Bai, Chong-En, Jiangyong Lu, and Zhigang Tao. 2006. The Multitask Theory of State Enterprise
Reform: Empirical Evidence from China. American Economic Review 96 (2):353–357.

Brandt, Loren, and Hongbin Li. 2003. Bank Discrimination in Transition Economies: Ideology,
Information, or Incentives? Journal of Comparative Economics 31 (3):387–413.

Cull, Robert, and Lixin Colin Xu. 2000. Bureaucrats, State Banks, and the Efªciency of Credit
Allocation: The Experience of Chinese State-Owned Enterprises. Journal of Comparative Econom-
ics 28 (1):1–31.

———, and Lixin Colin Xu. 2003. Who Gets Credit? The Behavior of Bureaucrats and State
Banks in Allocating Credit to Chinese State-Owned Enterprises. Journal of Development Econom-
ics 71 (2):533–559.

Dollar, David, and Shang-Jin Wei. 2007. Das (Wasted) Kapital: Firm Ownership and Investment
Efªciency in China. NBER Working Paper No. 13103. Cambridge, MA: NBER.

Héricourt, Jérôme, and Sandra Poncet. 2009. FDI and Credit Constraints: Firm-Level Evidence
from China. Economic Systems 33 (1):1–21.

OECD. 2005. Economic Surveys: China. Paris: OECD.

Shan, Weijian. 2006. The World Bank’s China Delusions. Far Eastern Economic Review 169
(7):29–32.

World Bank. 2005. SOE Dividends: How Much and To Whom? Washington, DC: World Bank.

69

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

Honor Thy Creditors Beforan Thy Shareholders

Xiao, Geng. 2006. Nonperforming Debts in Chinese Enterprises: Patterns, Causes, and Implica-
tions for Banking Reform. Asian Economic Papers 4 (3):61–113.

Zhou, Xiaochuan. 2004. Some Issues Concerning the Reform of the State-Owned Commercial
Banks. Speech by the Governor of the People’s Bank of China at the IIF Spring Membership
Conference, 16 April, Shanghai.

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3

70

Asian Economic Papers

Honor Thy Creditors Beforan Thy Shareholders

e
r
a
h
s

%

5
0
0
2

e
r
a
h
s

%

4
0
0
2

e
r
a
h
s

%

3
0
0
2

e
r
a
h
s

%

2
0
0
2

e
r
a
h
s

%

1
0
0
2

e
l
p
m
a
s

l
a
t
o
T

e
l
p
m
a
s
d
e
s
o
l
c
e
s
a
B

e
r
a
h
s

%

e
r
a
h
S
%

e
l
p
m
a
s

l
a
t
o
t
e
h
t

f
o
n
o
i
t
i
s
o
p
m
o
c

t
s
o
p
x
E

e
l
p
m
a
s
e
h
t

f
o
n
o
i
t
i
s
o
p
m
o
c

i
r
o
i
r
p
A

.

4
1
1

.

2
2
6

.

1
3
1

.

1
3
1

2
0

.

0
0
1

.

8
5
1

.

8
9
5

.

1
2
1

3

.

2
1

0
0

.

0
0
1

8
.
3
1

6
.
7
5

8
.
3
1

0
.
3
1

8
.
1

0
0
1

4
.
4
1

0
.
6
5

2
.
4
1

1
.
3
1

3
.
2

0
0
1

1
.
5
1

8
.
4
5

6
.
4
1

0
.
3
1

6
.
2

0
0
1

6
7
2
9

,

7
9
5
5

,

0
9
7
,
7

5
6
1
,
7

4
1
8
,
6

2
.
6
1

0
.
9
5

0
.
2
1

0
.
2
1

8
.
0

0
0
1

0
0
5
,
7

1
.
9

2
.
4
6

1
.
3
1

1
.
3
1

5
.
0

0
0
1

0
0
0
,
5

)
0
4
2
o
t

0
0
2
m
o
r
f
(
n
a
w
i
a
T

,

u
a
c
a
M

,

g
n
o
K
g
n
o
H

)
1
5
1
(cid:6)
3
4
1
(cid:6)
1
4
1
(cid:6)
0
1
1
(
E
O
S

)
0
9
1
o
t

9
5
1
m
o
r
f
(

e
t
a
v
i
r
P

s
s
a
l
c
p
i
h
s
r
e
n
w
o
y
B

1

x
i
d
n
e
p
p
A

)
9
4
1
(cid:6)
2
4
1
(cid:6)
0
4
1
(cid:6)
0
3
1
(cid:6)
0
2
1
(

)
r
e
g
r
a
l

r
o
0
0
3
(
d
e
n
w
o
n
g
i
e
r
o
F

s
e
v
i
t
a
r
e
p
o
o
C

l
a
t
o
T

.
e
s
a
b
a
t
a
d
n
w
o

r
u
o
n
o

d
e
s
a
b

s
n
o
i
t
a
l
u
c
l
a
c

s
r
o
h
t
u
A

:
e
c
r
u
o
S

s
e
s
i
r
p
r
e
t
n
e

f
o

r
e
b
m
u
n

l
a
t
o
T

:
s
m
e
t
i

m
u
d
n
a
r
o
m
e
M

71

Asian Economic Papers

l

D
o
w
n
o
a
d
e
d

f
r
o
m
h

t
t

p

:
/
/

d
i
r
e
c
t
.

m

i
t
.

/

e
d
u
a
s
e
p
a
r
t
i
c
e

p
d

/

l

f
/

/

/

/

/

9
3
5
0
1
6
8
3
1
4
8
a
s
e
p
_
a
_
0
0
0
2
5
p
d

.

f

b
y
g
u
e
s
t

t

o
n
0
7
S
e
p
e
m
b
e
r
2
0
2
3
Download pdf