State-Owned Enterprises Going Public:
The Case of China*
Xiaozu Wang
School of Management
Fudan University
Lixin Colin Xu†
Development Research Group
The World Bank
&
Guanghua School of Management
Peking University
Tian Zhu
Division of Social Science
Hong Kong University of Science and Technology
Revised: December 2003
ABSTRACT
Public listing is a key reform measure for large state-owned enterprises (SOEs) in
China. We find evidence that public listing lowers state ownership significantly, lessens
firms’ reliance on debt finance, and allows firms to increase capital expenditure, at least
temporarily. We also find that ownership structure affects post-listing performance.
However, we find no statistical evidence of a positive effect of public listing on firms’
profitability. We suggest alternative interpretations of the last finding.
JEL Classification: P31, P27, G30
Keywords: state-owned enterprises, public listing, reform, China
___________________________
* We wish to thank three anonymous referees and a number of seminar participants for helpful
comments on the early versions of the paper. All errors remain ours. We are grateful to the Research
Grants Council of Hong Kong Special Administrative Region and the World Bank for financial
support. The paper’s findings, interpretations, and conclusions are entirely those of the authors and do
not necessarily reflect the views of the World Bank, its Executive Directors, or the countries they
represent.
†
Corresponding author. MC 3-300, Development Research Group, The World Bank, 1818 H Street,
N.W., Washington, DC 20433. Phone: (202) 473-4664. Fax: (202) 522-1155. Email:
[email protected].
1.
Introduction
Unlike most formerly socialist countries, China until recently avoided
privatizing state-owned enterprises (SOEs) and instead sought to reform them through
piecemeal measures, such as by increasing managers’ decision-making autonomy,
introducing financial incentives, and bringing in performance contracts between the
government and SOEs (Naughton, 1995; Shirley and Xu, 2001). These reform
measures were accompanied by improved productivity of SOEs during the 1980s
(Groves et al., 1994; Jefferson, Rawski, and Zheng, 1994; Zhuang and Xu, 1996; Li,
1997; Xu, 2000). However, the performance of Chinese state industry has since
deteriorated (Lardy, 1998). In the early 1990s the Chinese government began to shift
the focus of SOE reform to privatization of small SOEs and corporatization of larger
ones (Cao, Qian and Weingast, 1999; Lin and Zhu, 2001).
Public listing of SOEs in the domestic stock exchanges is a key measure of
corporatization. Indeed, the vast majority of China’s publicly listed companies are
formerly state-owned or state-controlled firms, mostly large and better-performing
ones.1 SOEs’ low efficiencies are often attributed to a lack of managerial autonomy,
soft budget constraints and the agency-incentive problem (Groves et al., 1994; Qian,
1996; Qian and Roland, 1996). In theory, public listing can potentially help separate
government from enterprises and hence increase enterprise autonomy and harden
1
Our data set, to be described later, does not contain information about the types of the share-issuing
firms. Based on our interviews with officials of the China Securities Regulatory Commission (CSRC),
about 75% of listed companies are formerly state-owned. Another 10% are formerly shareholding
companies with a significant portion of shares held by SOEs. Only less than 10% of listed companies
are formerly private-owned firms or foreign-invested firms, which in most cases had SOEs as their
joint venture partners.
1
budget constraints. It may improve managerial incentives if it results in a more clearly
defined structure of rights and responsibilities and the involvement of shareholders
with incentives and the ability to monitor managers. Public listing should also help to
raise capital for SOEs and thus reduce their traditionally high debt-to-asset ratios.
In this research, we explore the extent to which public listing has contributed
to the reform of SOEs, paying particular attention to its impact on firms’ operating
performance and financial structure. Using a panel of pre- and post-listing data of all
Chinese companies listed on the two domestic stock exchanges between 1994 and
2000, we find evidence that public listing lowers state ownership significantly, lessens
firms’ reliance on debt finance, and allows firms to increase capital expenditure, at
least temporarily. We also find that the ownership structure affects post-listing
performance. However, we find no statistical evidence of public listing exercising a
positive effect on firms’ profitability. Specifically, firms’ operating performances
after listing are significantly lower than their pre-listing level. We suggest alternative
interpretations of this result.
A number of recent papers have studied publicly-listed Chinese firms. Xu and
Wang (1998) study the impact of ownership concentration and the share of state
ownership on the performance of listed companies in China, but their study does not
deal with the issue of whether or not public listing itself improves company
performance. Chen, Firth and Kim (2000) use a sample of about 330 IPOs in China
between 1992 and 1995 to compare the differences in performance between A shares
and B shares (the former are issued to domestic investors, and the latter to foreign
investors). However, neither do they address the issue of how public listing affects
company performance. Aharony, Lee and Wong (2000) use the decline in
2
performance to demonstrate the existence of financial packaging in Chinese IPOs that
issue shares to foreigners.
In a paper that is more closely related to ours, Sun and Tong (2003) study the
effects of public listing on several measures of firm performance in China. Our study
differs from theirs in a number of respects. First, our study covers a longer time
period. Second, they do not control for the overall trend of the financial performance
in the country’s state-owned sector, and thus they cannot distinguish intrinsic listing
effects from the overall economic downturn in the 1990s for the state sector. Third,
we also look at the effects of ownership structure on firm performance. Fourth, they
do not explicitly control for the possibility of financial packaging. Finally, they draw
some of the conclusions on listing effects based on the levels (instead of ratios) of
profits and sales, an approach which we view as problematic.
Our study builds upon the empirical literature on the impact of public listing
or initial public offering (IPO) on firm performance (Roell, 1996). This literature
focuses on developed countries, particularly the United States, and finds that public
listing of privately-held companies tends to worsen company performance.
Specifically, Ritter (1991) finds that IPO firms underperform a set of comparable
firms matched by size and industry. Laughran and Ritter (1995) find that both IPOs
and seasoned equity offerings significantly underperform relative to non-issuing firms
for five years after the offering date. Jain and Kini (1994), Degeorge and Zeckhauser
(1993) and Mikkelson, Partch, and Shah (1997) find that the performance of IPO
firms—measured by return on assets (ROA) or return on sales (ROS)—declines in the
first few years following the offering but do not decline further afterwards.
One explanation for post-listing performance decline is managerial moral
hazard resulting from reduced ownership stakes by management after listing (Jain and
3
Kini, 1994; Holthausen and Larcker, 1996). Another explanation is that the pre-listing
performance may be exaggerated (Laughran and Ritter, 1995; Pagano, Panetta, and
Zingales, 1998). For example, offering firms may window-dress their accounting
figures prior to going public. They may also time the offerings to coincide with
periods of unusually good performance or favorable market valuations. Consequently,
the over-stated pre-IPO performance may result in a superficial decline in post-IPO
performance.
Our study is also related to the literature on share issue privatization, which
refers to using public listing as a way of divesting the government’s ownership in
SOEs (Megginson and Netter, 2001). Share issue privatization has been one of the
major forms of privatization around the world since the 1980s. In summarizing the
long-run performance of share issue privatization, Megginson and Netter (2001) state
that, “the average long-term, market-adjusted return earned by international investors
in share issue privatizations is economically and significantly positive.”
While public listing in developed countries either turns a privately-held
company into a more widely-held public company, or transforms an SOE into a
private-owned public company, public listing in China is largely used to corporatize
SOEs. China’s share issue corporatization aims to transform an SOE into a modernform corporation that features both state and non-state institutional shareholders in
addition to small individual shareholders. If the public listing of private firms worsens
company performance in developed economies and share issue privatization of SOEs
in these countries improves performance, it is an intriguing empirical question as to
whether public listing would improve or worsen firm performance in the intermediate
case of share issue corporatization and in the context of China’s transitional economy.
4
In the following section, we provide some background information on public
listings and the development of the stock market in China. Section 3 describes the
data and presents some preliminary results comparing the sample firms’ financial
outcomes and ownership structures before and after public listing. Main findings from
regression analyses are reported in Section 4. The last section concludes.
2.
Public Listings in China
China’s stock market was officially established in 1990 when eight firms first
went public in the Shanghai Stock Exchange (SHSE). In the following year, Shenzhen
Stock Exchange (SZSE) was also established. The following decade witnessed
phenomenal growth in China’s stock market, as outlined in Table 1.
(Insert Table 1 here)
At the end of 2000, 1088 firms were listed on the two exchanges, with a total
market capitalization close to RMB5 trillion (about US$0.6 trillion 2 ), or 54% of
China’s GDP. The stock market has also become an increasingly important means of
raising capital for China’s SOEs, resulting in more than RMB480 billion new equity
issuance during 2000 alone.
China’s publicly-listed companies are allowed to issue four types of shares.
The predominant type is A shares; these are listed in China, denominated in RMB and
their sales are restricted to domestic investors. B shares are also listed in China and
denominated in RMB, and until June 2001 their purchase was restricted to foreign
2
The current exchange rate is roughly US$1 = RMB8.2.
5
investors using foreign currency. The two other types of shares are H and N shares,
which are issued in Hong Kong and New York respectively by A-share or B-share
issuing firms. While most companies only issue A shares, the majority of B-share
issuing companies also issue A shares. By the end of 2000, of the total 114 B-share
issuing firms only 28 issued B-shares exclusively; the rest also issued A shares. All
the 19 H-share firms had also issued A-shares.
The shares of listed companies are typically divided into state, legal-person
and public shares.3 The first two categories of shares cannot be traded on the stock
exchanges, and their transfer requires special approval from the China Securities
Regulatory Commission (CSRC). Public shares are tradable shares issued to the
public and are normally held by small individual shareholders.
The distinction between state and legal person shareholders is often times
superficial. State shares are held by government bodies such as state asset
management agencies, or institutions authorized to hold shares on behalf of the state
such as a wholly state-owned investment company. Legal person shares are shares
held by any entity or institution with a legal person status, including an SOE or a
company controlled by an SOE. We do not have precise information about the
identity of legal person shareholders, but it is safe to say that state ownership, directly
or indirectly, accounts for a significant portion of all the legal person shares. Some
authors, however, suggest that the distinction between state ownership and legalperson ownership can be consequential (Tian, 2000; Berkman, Cole and Fu, 2002;
Sun and Tong, 2003). We thus will let the empirics tell us whether legal-person
ownership entails consequences different from state ownership.
3
A company can also issue employee-held shares, which normally account for less than 1% of the total
shares. They may become tradable three years after the IPO if approved by the regulator.
6
Before 2001, the question of whether a Chinese company could make an IPO
was determined largely by an administrative process rather than the market process
seen in developed economies. When an SOE wants to go public, it must seek
permission from the local government or/and its affiliated central government
ministries, which receive an IPO quota from the CSRC.4 Under such a quota system,
how many and which firms go public each year depends not only on the quality of the
firm and on macroeconomic conditions, but also on the availability and distribution of
the quota. All firms in our sample went public under the quota system.
3.
Data and Preliminary Findings
Data and Variables
The data for this study is a panel of accounting and ownership data of all
companies listed on the SHSE or SZSE.5 There are 1057 firms in our initial data set
covering all firms listed between 1991 and June 2000. Since there was a major
accounting reform in 1993, which made it difficult to compare a company’s
4
There are no explicit rules governing quota allocations. Information on how much quota is issued to
whom is hard to obtain. Based on our interviews with investment bankers and regulators in China,
quota may even be allocated to such organizations as the National Union of Women and the
Communist Youth League. In 2000, the government decided to abandon the quota system and let the
market determine which firms can go public. The first non-quota IPO appeared in 2001.
5
The data was purchased from Genius Information Technology, a Shenzhen-based financial
information service company in China. The data was corroborated for accuracy with the China Stock
Market and Accounting Research Database (CSMAR) produced by GTA Information Technology, a
company also based in Shenzhen.
7
accounting numbers before and after the reform, we opt to use firms that went public
in or after 1994. Firms that went public in 1994 were required to adjust their pre-1994
accounting numbers to be consistent with the new accounting rules. After dropping
missing values or invalid data entries, we have a sample of 793 firms for the period
from January 1994 to June 2000.
A novel feature of our data set is that it contains pre-listing information, which
allows us to compare companies’ pre- and post-listing performance.6 Another feature
of the data set is that it is free of survival bias that may cause problems in studying
listing effects on company performance. No firm in our data set ceased operations or
was de-listed after going public. Although China’s bankruptcy law was passed in
1986, listed companies can usually count on the government or state-owned banks to
bail them out of financial difficulties and hence avoid bankruptcy. Also no publicly
listed firms returned to private ownership in our sample period. Only in 2001 did we
observe the first incidence of de-listing.
In our regression analyses, we follow the existing literature in choosing our
dependent and explanatory variables. This allows us to highlight the similarities as
well as differences in the effects of public listing in China in comparison with
countries that have been previously examined in the literature. Definitions of variables
are listed in Table 2.
6
IPO firms are required by law to provide three years of audited accounting data prior to listing.
However, since the CSRC was established in 1992, two years after the first stock exchange was
established, and major disclosure rules were only issued in 1993 but were not immediately strictly
enforced, the disclosure standard was not consistent during the first half of 1990s. As a result, about
20% of firms in our sample did not produce complete three-year pre-listing data.
8
(Insert Table 2 here)
To minimize the possibility of a small number of outliers driving the results,
we follow other authors in the literature and Winsorize the data. Specifically, we reset
the value of a variable that is in the tail one percentile of the full sample to that of the
1st percentile and the 99th percentile respectively.
Financial Outcomes and Ownership Structures before and after Listing
We report in Table 3 the summary statistics of the financial outcomes and
ownership structures for both the full sample (column 3) and the sub-samples of the
pre-listing years (column 4) and post-listing years (column 5). These are calculated
using all observations that will be used in at least one of the subsequent regressions.
Note that the post-listing statistics include observations from the IPO years.
(Insert Table 3 here)
The average size of the listed firms, measured by either the book value of
assets (denoted as asset) or sales (sales), is quite large, with the average value of
assets at RMB1,179 million and average sales at RMB703 million. Public listing
significantly increases a firm’s assets: the post-listing average asset size is almost
double that of the pre-listing level. Sales also increase after listing, only less
dramatically than assets—by an average of 24% from the pre-listing level of RMB606
million to RMB752 million. Firms generally maintained their high level of sales
growth (salegrow) after public listing with the average annual growth rate of 0.167 or
16.7% after listing compared with 0.207 or 20.7% before listing.
9
Public listing apparently loosens the financial constraints faced by the Chinese
firms and helps them to lower their leverage ratio. The average debt-to-asset ratio
(debt) drops from 0.34 before listing to 0.28 after listing. Capital expenditure (Capex)
on average increases by more than 15% from RMB98 million before listing to
RMB114 million after listing.
A prominent feature of the Chinese listed firms is their high ownership
concentration in the hands of a few large shareholders. After public listing, the state
(state_stock) and large institutional shareholders (lperson_stock) on average each hold
slightly more than 30% of the total shares. For the full sample that combines pre- and
post-listing periods, the top five shareholders hold close to 60% of the total shares
(A5). The ownership concentration among the top five shareholders is also very
high—the Herfindahl concentration index of shareholding among these large
shareholders (Herfindahl_top5) is 0.647.7 Both measures of ownership concentration
drop after listing. More significantly, public listing has helped to transform the state
ownership of enterprises. The average share of state ownership goes down by almost
15 percentage points from the pre-listing level of 45% to the post-listing level of
30.6%.
However, the operating performance of listed firms as measured by return on
assets (ROA) and return on sales (ROS) deteriorates considerably after public listing.
The average ROA in the post-listing periods drops to almost a third of the level before
listing, from 0.153 or 15.3% to 0.057 or 5.7%. The average ROS drops after listing by
almost a half, from 0.191 to 0.101.
Since many factors, such as macroeconomic factors during the sample period,
could play a role in the decline of firm performance, in the next section we use
7
If the 60% of the total shares were distributed equally among the top five shareholders, the Herfindahl
index would be 0.2.
10
regression analyses to examine how operating performance and other financial
outcomes are affected by public listing once additional factors are controlled for.
4.
Regression Analysis
Methodology
The basic regression we use to measure the effects of going public is the
following:
yit = α 0 + δ 1 log(salesi ,t −1 ) + δ 2 log(debt i ,t −1 )
+ ∑s =0 γ s Lis + β i + α t + ε it .
6
(1)
Our dependent variables include a number of financial outcome measures. We
use ROA as the overall performance measure. Because IPO firms often experience a
rapid expansion in their asset base, and this alone can be responsible for the drop in
ROA, we also examine ROS, another conventional measure of operating
performance, to check the robustness of what we may find about the changes in ROA.
Since a primary objective of public listing is to raise equity capital as the external
source of investment for business expansion, we also include financial leverage
(debt), investment rate (ln_capex) and sales growth (salegrow) as measures of the
outcomes of public listing in our investigation.
On the right-hand side, Lis is the dummy variable that is 1 when firm i at year
t is in the sth year after going public and zero otherwise. Note that we use the subscript
s to denote the age of listing and the subscript t to denote the calendar year. Thus our
specification allows both listing age-specific effects and calendar year-specific
11
effects.8 Our specification also includes fixed effects to control for firm heterogeneity.
This allows us to interpret the listing effects on an outcome measure as the difference
between its levels at particular listing ages and the pre-listing level.9 We also control
for variables that are the “usual suspects” in explaining the outcomes regardless of
whether a firm is listed or not (as in Pagano, Panetta, and Zingales, 1998). We use the
variable log(lagged sales) to capture the size effect, i.e. larger firms may exert more
market power and therefore generate more profits, and log(lagged leverage ratio) to
control for financial structure and its informational content.10
It should be noted that the calendar-year effect in the above specification is
identified by the average effect for the listed firms at the particular calendar year net
of those explained by listing-age dummies and other explanatory variables. Our
sample period of 1994 to 2000 coincides with a period of massive deterioration of
financial performance of the state sector across the board. When the dependent
variable is ROA or ROS, the year dummies are intended to capture the effect of
economy-wide factors on firm performance. However, such an approach is less than
8
If all the firms went public in the same year, it would be impossible to distinguish the listing-age
effects and changes in macroeconomic trends. But our firms went public in different years, as shown in
Table 1.
9
Some may find it surprising that using fixed effects implies being able to use the pre-listing level as a
benchmark, since typically the operation of fixed effects is to subtract the firm-level mean from each
regression. But the inference is valid once one realizes that one can subtract an observation of any
period (say, the pre-listing period, s = -1) to get rid of the fixed effects and obtain a consistent estimate
of Equation (1).
10
We also tried adding more variables such as the share of intangibles in total assets and investment
rate (both lagged by one period) as additional explanatory variables, and we obtained qualitatively
similar results. However, in doing so we lost a significant number of observations.
12
satisfactory. To the extent that listed firms differ from typical SOEs, the calendar-year
effects may merely reflect macro effects common only to listed firms but not to a
random SOE. Thus a more satisfactory way of filtering out the macro shocks is, in
place of year dummies, to control directly for the average values of ROA and ROS for
the SOE sector. Since we are primarily interested in the listing effects relative to a
typical SOE, the national SOE average is a more proper benchmark against which
listed companies are compared.
We use data from the Chinese Statistical Yearbooks to calculate the average
ROA and ROS of all the industrial SOEs, denoted as ROA_national and
ROS_national respectively. Once the national average of firm performance is
controlled for, the listing-age effect for post-listing year s can be characterized as
( yit , s − y t ) − ( yi , t − s −1, −1 − y t − s −1 ) − δ ( X it − X i ,t − s −1 ) . In this we first obtain the difference
between firm performance and the national average in post-listing year s, subtract the
difference between firm performance and the national average in pre-listing year –1,
and then filter out the influence of other explanatory variables. This is similar to the
difference-in-difference approach that is often used in recent empirical applications in
the panel setup.
Since our benchmark is the performance of the pre-listing years, the sample
used for the regression consists only of firms with data for at least one pre-listing
year. If a firm has more than one year of pre-listing data, then the benchmark is the
average of the values for all the pre-listing years.
It is perhaps useful to explain here why we do not adopt the matching
approach to identify the listing effects. With such an approach, a matched sample
would be found for the sample of listed firms, and the listing effects would then be
computed as the before-after difference for the listed sample subtracting the before-
13
after difference for the matched sample. Matching is usually done through the closest
match based on size-industry category (e.g., Pagano, Panetta and Zingales, 1996 and
1998). Matching, however, poses serious data requirements. The researchers need to
have access to another, much larger, data set in which important characteristics of the
sample firms—most often performance, size, and industry—are close to those of the
listed firms. Poor matching would result in mis-specified test statistics and biased
estimates (Heckman, Ichimura and Todd, 1997 and 1998). It is difficult, if not
impossible, to find reasonably good matches for these listed companies, which are
overwhelmingly large firms. To our best knowledge, there is no Chinese data set
available that contains enough useful information on large enterprises between 1994
and 2000.11
11
When revising the paper for this journal, we examined a few potentially useful data sets but
concluded they were not good enough. For instance, we looked at a survey data set that covers
roughly 400 SOEs for the period between 1994 and 1999, which is partly sponsored by the Chinese
Academy of Social Science (CASS). Its predecessors, covering the periods from 1980-1989 and 19901994, have been used by previous researchers (Groves et al., 1994; Li, 1997; Xu, 2000; Shirley and Xu,
2001; Cull and Xu, 2000). However, the size of the firms in the CASS data set is a lot smaller than that
of listed firms: the asset size is on average only 1/7 of the listed sample. More importantly, the CASS
sample suffers from significant survival bias—the SOEs were those that had survived for at least 20
years! Not surprisingly, when we plot the average ROA for the sample, it has a significant rising trend,
in sharp contrast with the significant declining trend in the listed sample. Since we know that the
national average for the state sector shows a declining trend, the CASS sample is clearly not a good
benchmark for comparison with listed firms. We also looked at other samples for Chinese firms such as
the Chinese industrial censuses, and found that these have too little information for our purposes.
14
Effects of Public Listing
A primary objective of public listing is to raise capital for SOEs, which
normally rely on bank loans as their only source of external finance. Indeed, one of
the most-cited reasons for public listing in developed economies is that it loosens the
financial constraints faced by firms and facilitates business expansion (Roell, 1996;
Pagano, Panetta, and Zingales, 1998). To shed light on this point, we look at the
listing age patterns of the investment rate (measured by the level of capital
expenditure), the financial leverage and the growth of sales. Column (1) of Table 4
shows that the log of the level of capital expenditure increases significantly in the year
of listing, although one year later it goes back to the pre-listing level and then declines
somewhat several years after listing. Column (2) shows that public listing leads to a
significant reduction in the debt-asset ratio for a few years, after which the debt level
returns to the pre-listing level. Column (3) shows that listed firms are able to maintain
their high pre-listing sales growth rate through all the observed post-listing years.
These findings suggest that using public listing as a means to raise equity capital as an
external source of investment for SOEs has, at least in part, achieved its goal.
(Insert Table 4 here)
Public listing as a reform measure is motivated by and also aimed at stopping
the deterioration of financial performance in the state sector. The evidence, however,
shows that public listing does not improve SOEs’ bottom lines. On the contrary, SOE
listing is associated with a significant drop in operating performance measured by
ROA (Column 4 of Table 4). The listing-age effects are: -6.1% in the listing year, -
15
7.8% in post-listing year 1, -8.3% in post-listing year 2, -10.4% in year 3, -10.5% in
year 4, and -12.2 in years 5 and 6.12 In other words, the overall operating performance
of China’s listed firms is significantly lower than the pre-listing level, and
performance declines in the years following the listing.
The decline in performance cannot simply be attributed to the increase in the
size of assets. As total assets normally increase significantly after an IPO, operating
income scaled by assets has a downward bias. However, operating income scaled by
sales (i.e., ROS) also shows a similar decline after listing. Column (5) of Table 4
shows that the listing effects on ROS from the listing year to the 5th/6th year after
listing are, -2.4%, -6.3%, -8.8%, -10.7%, -13.7% and -13.1%, respectively. Note that
the negative effects of public listing on ROA and ROS are very precisely estimated.
The above finding suggests that performance decline after an IPO may be a
general pattern for firms going public—a finding which applies to Western firms as
well as to Chinese firms. In the Chinese case, a number of reasons could be
responsible for the deterioration of accounting profits. One reason, which also applies
to Western firms, can be found in the overstatement of the pre-listing performance
through timing the issuing or window-dressing accounting figures prior to listing (i.e.,
financial packaging). In a study of 81 listed Chinese firms that issued shares to
foreigners, Aharony, Lee and Wong (2000) suggest that the post-listing decline in
financial performance may be due to financial packaging before the IPO through, for
example, earnings management.13 According to these authors, a principal means of
12
We bundle the fifth and sixth year together because there are relatively few observations for each
year.
13
See Aharoni, Lee and Wong (2000) for details on the incentives for and techniques of financial
packaging in China.
16
earnings management in the Chinese context is using credit sales (i.e., accounts
receivable) to beef up earnings figures in the pre-listing years.
To examine this possibility, we construct a variable, ∆ARit , which is accounts
receivable over net cash sales in post-listing year t subtracting that in pre-listing year
–1. Column (1) of Table 5 regresses ∆ARit onto the listing dummies. Since ∆ARi ,t is
always zero for pre-listing year -1, the data for that year is not included in the fixedeffects regression, and the listing dummies start from post-listing year 1. Relative to
the pre-listing era, there appears to be an increasingly lower share of credit sales:
∆ARit becomes smaller over time, and the coefficients are close to being statistically
significant.
In light of evidence of the presence of financial packaging before listing, we
now examine to what extent the post-listing decline in financial performance is
attributable to financial packaging. In Columns (2) and (3) of Table 5, we show the
results of a re-run of the ROA and ROS regressions adding ∆ARit as an additional
explanatory variable to hold constant the effects of financial packaging. The
coefficients of the listing age dummies are almost identical to those in Table 4.
Therefore, we conclude that financial packaging through earnings management plays
only a minor role in explaining post-listing performance decline. The fact that ∆ARit
is statistically significant in the ROS regression (and later regressions in Table 6)
suggests that it does play a role.
(Insert Table 5 here)
17
As mentioned earlier, another common explanation for IPO’s performance
decline in the Western context is the increase in agency costs due to reduced
ownership stakes by management after public listing. In the Chinese case, public
listing reduces state ownership and generally increases managerial autonomy. It may
potentially reduce the cost of political control of firms but increase agency costs
(Qian, 1996). The net impact of these two opposing effects on performance is unclear.
However, due to a weak legal system, a strong case can be made for the expropriation
of interests of minority owners by the large shareholders who dominate the
management and operations of the listed firms. In most IPO cases, selected profitable
business units of an SOE are carved out and they form an independent entity for
public listing. The original SOE becomes the parent company and retains most, if not
all, unprofitable units and liabilities as well as the least productive workforce. This
arrangement sharply increases the operating pressure on the parent SOE, which may
use its power to divert revenues from the listed firms through, for example, related
transactions, leading to a decline in the latter’s profits. We do not have evidence for
the above hypothesis, though it is plausible. Future research is needed to shed light on
it. If expropriation by parent companies is indeed very serious, then the performance
decline of listed firms may not be attributed to the listing event per se, but to poor
corporate governance and the incomplete nature of the SOE reforms.
In any case, regardless of what causes firm performance to decline after
listing, it is apparent that public listing has failed to turn loss-making SOEs into
profit-making companies. This finding is also consistent with the possibility that the
government or the issuing SOEs simply use the stock market to raise money for SOEs
while having no intention of privatizing them, and the firms are still controlled by the
18
state given the fact that state owners have yet to relinquish their dominant positions in
the listed firms.
Ownership Structure and Post-listing Performance
Our discussions in the introduction and previous sections lead us to consider
the role played by ownership in the performance of listed companies. We focus on the
post-listing periods in this subsection for a number of reasons. First, the relationship
between ownership structure and firm performance in transitional economies is an
interesting research topic (e.g., Claessens and Djankov, 1999; Cull, Matesova and
Shirley, 2001). Second, pre-listing performance may be contaminated by financial
packaging effects. Lastly, a significant number of firms miss ownership information
in pre-listing years.
We observed earlier that public listing changes the ownership structure of a
firm and that, for most listed firms in China, the majority of shares is held by the state
and a number of legal persons. In what follows, we examine how ownership
concentration by top shareholders and the balance of power among them affect firm
performance.
Following Demsetz and Lehn (1985), we construct two measures of ownership
concentration: (1) the percentage of shares held by the top five shareholders (A5), and
(2) the Herfindahl concentration index of ownership among the top five shareholders
(Herfindahl_top5). It should be noted that the way we construct the Herfindahl index
is slightly different from the method used by other authors. Demsetz and Lehn (1985),
for example, use the Herfindahl index to capture concentration of ownership among
all shareholders. Under our construction, the Herfindahl index measures the
concentration of control power among the top five shareholders. A high Herfindahl
19
index implies that power is likely to be in the hands of the largest shareholder, while a
small Herfindahl index means that ownership is more evenly distributed and there is a
balance of power among the large shareholders.
The regression we run is:
y it = α 0 + α i + α t + γ X it + β 1 A5 it + β 2 Herfindahl _ top 5 it
+ β 3 tradable _ share it + β 4 lperson _ share it + β 5 ∆AR it + ε it
(2)
Here X represents the other control variables that we mentioned earlier. The default
for the type of share ownership is the state share. ∆ARit is controlled for so that any
carryover of financial packaging effects is held constant. The results are reported in
Table 6.
(Insert Table 6 here)
Legal person ownership shows no substantial difference from state ownership
in terms of its impact on performance. This result contrasts with the findings by Tian
(2000), Berkman, Cole and Fu (20002), and Sun and Tong (2003), who suggest that
legal-person ownership (mostly indirect state ownership) is somewhat superior to
state ownership. This result is also consistent with Xu, Zhu and Lin (2002), which
uses a national survey of firms that experienced ownership transformation in late
1990s and finds that legal-person ownership had similar effects to state ownership.
More interestingly, the percentage of tradable shares has a statistically significant
negative effect on performance. Since tradable shares are normally held by small
domestic individual shareholders, this finding is consistent with the notion that
dispersed ownership causes the free-rider problem in monitoring the management and
hence results in relatively poorer performance. The finding corroborates that of Lizal
20
and Svejnar (2002), which uses a panel of Czech firms between 1992 to 1998 and
finds less than impressive effects of domestic private ownership.
Ownership concentration also matters. A5 is positive for the ROA (but not the
ROS) regression. This is consistent with the notion that concentrated ownership
allows the internalization of costs of monitoring. More interestingly, the effect of
Herfindahl_top5 is negative and significant in both the ROA and the ROS
regressions. As we have already controlled for ownership by types of shareholders
and concentration by top five shareholders, this result does not merely reflect omitted
controls on state ownership and other ownership variables.14 Our interpretation of the
result is that an ownership structure that features more balances of power among top
owners introduces checks and balances in the control structure and hence reduces the
likelihood of a dominant shareholder maximizing his or her own private interest at the
expense of other shareholders (Bennedsen and Wolfenzon, 2000). Our test of the
effect of the balance of power on firm performance is, to our knowledge, the first in
the literature.
5.
Conclusion
In this paper, we use a panel of pre- and post-listing data of all publicly listed
companies in China between 1994 and 2000 to explore the effectiveness of public
listing as a means of reforming SOEs in China. We find that public listing has had
some intended impact on firms’ ownership structure and finance. However, listed
firms also experience a sharp deterioration in accounting profits, which may be
14
As a lower Herfindahl concentration index may be associated with a lower level of state ownership,
our result could be due to the negative impact of state ownership if it were not controlled for.
21
attributable to one or both of pre-listing window-dressing of accounting figures and
post-listing expropriation by the parent SOEs.
We also find some statistical relations between ownership and firm
performance. State and legal-person ownership are indistinguishable whereas
domestic private ownership is inferior in terms of its impact on performance. The
degree of ownership concentration by a few large shareholders is positively correlated
with operating performance, but a more balanced ownership structure among these
top shareholders is found to be good for performance. This latter finding suggests that
under a weak legal and regulatory system, having a few large shareholders on a
relatively equal footing may improve corporate governance and prevent misbehavior
by a dominant shareholder.
The empirical evidence we have does not allow us to make a conclusive
judgment on the overall success (or failure) of public listing as a means of reforming
SOEs. Given the short history of China’s stock market, it may be too early to make
such a judgment. As the stock market, legal systems and other market institutions in
China develop over time and as more private entrepreneurs accumulate sufficient
wealth to acquire more stakes and become large shareholders in listed firms, it is
conceivable that public listing can become a useful means in transforming SOEs and
improving their performance. It would be very interesting to see if corporatization in
general and public listing in particular is simply a transitional phenomenon or a viable
alternative to privatization in the long run.
22
References
Aharony, J., Chi-Wen Jevons Lee, and T.J. Wong. (2000). “Financial Packaging of IPO Firms
in China,” Journal of Accounting Research, 38(1), 103-26.
Bennedsen, M., and D. Wolfenzon (2000). “The Balance of Power in Closely Held
Corporations,” Journal of Financial Economics, 58, 113-39.
Berkman, Henk, R. Cole, and J. Fu (2002). “From State to State: Improving Corporate
Governance Where the Government Is the Controlling Block Holder,” Working
Paper, University of New South Wales.
Cao, Y., Y. Qian, and B. R. Weingast (1999). “From Federalism, Chinese Style to
Privatization, Chinese Style,” Economics of Transition, 7(1), 103-131.
Chen, G., M. Firth, and J. Kim (2000). “The Post-issue Market Performance of Initial Public
Offerings in China’s New Stock Markets,” Review of Quantitative Finance and
Accounting, 14(4), 319-339.
China Securities Regulatory Commission (1999). China Securities and Futures Statistics,
Beijing: China Finance and Economic Publishing House.
Claessens, S., and S. Djankov (1999). “Ownership Concentration and Corporate Performance
in the Czech Republic,” Journal of Comparative Economics, 27(3), 498-513.
Cull, R., J. Matesova, and M. Shirley (2001). “Ownership Structure and the Temptation to
Loot: Evidence from Privatized Firms in the Czech Republic,” World Bank Policy
Research Working Papers 2568, World Bank.
Cull, R., and L.C. Xu (2000). “Bureaucrats, State Banks, and the Efficiency of Credit
Allocation: The Experience of Chinese State-Owned Enterprises,” Journal of
Comparative Economics, 28, 1-31.
Degeorge, F., and R. Zeckhauser (1993). “The Reverse LBO Decision and Firm Performance:
Theory and Evidence,” Journal of Finance, 48(4), 1323-1348.
Demsetz, H., and K. Lehn (1985). “The Structure of Corporate Ownership: Causes and
Consequences,” Journal of Political Economy, 93(6), 1155-1177.
Groves, T., Y. Hong, J. McMillan, and B. Naughton (1994). “Autonomy and Incentives in
Chinese State Enterprises.” Quarterly Journal of Economics, 109(1), 183-209.
Heckman, J., H. Ichimura, and P. Todd (1997). “Matching as an Econometric Evaluation
Estimator: Evidence from Evaluating a Job Training Programme,” Review of
Economic Studies, 64(4), 605-54.
Heckman, J., H. Ichimura, and P. Todd (1998). “Matching As an Econometric Evaluation
Estimator,” Review of Economic Studies, 65(2), 261-294.
Holthausen, R. W., and D. F. Larcker (1996). “The Financial Performance of Reverse
Leveraged Buyouts,” Journal of Financial Economics, 42(3), 293-332.
Jain, B. A., and O. Kini (1994). “The Post-Issue Operating Performance of IPO Firms,”
Journal of Finance, 49(5), 1699-1726.
Jefferson, G. H., T. G. Rawski, and Y. Zheng (1996). “Chinese Industrial Productivity:
Trends, Measurement Issues, and Recent Developments,” Journal of Comparative
Economics, 23(2), 146-180.
Lardy, N. R. (1998). China’s Unfinished Economic Revolution. Washington, D.C.: Brookings
Institution.
23
Li, W. (1997). “The Impact of Economic Reform on the Performance of Chinese State
Enterprises, 1980-1989,” Journal of Political Economy, 105(5): 1080-1106.
Lin, Y., and T. Zhu (2001). “Ownership Restructuring in Chinese State Industry: An Analysis
of Evidence on Initial Organizational Changes,” China Quarterly, 0(166), 305-341.
Lizal, L., and J. Svejnar (2002). “Privatization Revisited: The Effects of Foreign and
Domestic Owners on Corporate Performance,” Working Paper, University of
Michigan.
Loughran, T., and J. R. Ritter (1995). “The New Issue Puzzle,” Journal of Finance, 50(1), 2351.
Megginson, W. L., and J. M. Netter (2001). “From State to Market: A Survey of Empirical
Studies on Privatization,” Journal of Economic Literature, 39(2), 321-389.
Mikkelson, W. H., M. M. Partch, and K. Shah (1997). “Ownership and Operating
Performance of Companies That Go Public,” Journal of Financial Economics, 44(3),
281-307.
Naughton, B. (1995). Growing Out of the Plan: Chinese Economic Reform, 1978-1993. New
York: Cambridge University Press.
Pagano, M., F. Panetta, and L. Zingales (1996). “The Stock Market as a Source of Capital:
Some Lessons from Initial Public Offering in Italy,” European Economic Review,
40(3-5), 1057-1069.
Pagano, M., F. Panetta, and L. Zingales (1998). “Why Do Companies Go Public? An
Empirical Analysis,” Journal of Finance, 53(1), 27-64.
Qian, Y. (1996). “Enterprise Reform in China: Agency Problems and Political Control,”
Economics of Transition, 4(2), 422-447.
Qian, Y., and G. Roland (1996). “The Soft Budget Constraint in China,” Japan and the World
Economy, 8(2), 207-223.
Ritter, J. (1991). “The Long-run Performance of Initial Public Offering,” Journal of Finance,
46(1), 3-27.
Roell, A. (1996). “The Decision to Go Public: An Overview,” European Economic Review,
40(3-5), 1071-1081
Shirley, M., and L.C. Xu (2001). “The Empirical Effects of Performance Contracts,” Journal
of Law, Economics, and Organization, 17(1), 168-200.
Sun, Q., and W. Tong (2003). “China Share Issue Privatization: The Extent of Its Success,”
Journal of Financial Economics, forthcoming.
Tian, G.L. (2000). “Government Shareholding and the Value of China’s Firms,” Working
Paper, London Business School.
Xu, L.C. (2000). “Control, Incentives, and Competition: The Impact of Reform in Chinese
State-Owned Enterprises,” Economics of Transition, 8(1), 151-173.
Xu, L.C., T. Zhu, and Y. Lin (2002). “Politician Control, Agency Problems, and Ownership
Reform: Evidence from China,” mimeo, the World Bank.
Xu, X., and Y. Wang (1999). “Ownership Structure and Corporate Governance in Chinese
Stock Companies,” China Economic Review, 10(1), 75-98.
Zhuang, J., and C. Xu (1996). “Profit-Sharing and Financial Performance in the Chinese State
Enterprises: Evidence from Panel Data,” Economics of Planning, 29(3), 205-222.
24
Table 1. Development of China’s Stock Market1
Year
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
Total Number of Listed Firms
10
14
53
183
291
323
530
745
851
949
1088
Capital Raised2 (RMB billion)
4.590
0.500
9.409
31.454
13.805
11.886
34.152
93.382
79.518
88.297 151.137
Market Capitalization3 (RMB billion)
N/A
N/A
104.813
353.101
369.061
347.428
984.238 1752.924 1950.564
2647.117 4809.094
Market Capitalization/GDP4 (%)
N/A
N/A
3.93
10.2
7.89
5.94
14.5
23.44
24.9
32.32
53.79
Number of Investors5 (million)
N/A
N/A
2.1665
7.777
10.590
12.425
23.072
33.333
39.107
44.820
58.011
Total Book Value of Assets (RMB billion)
N/A
N/A
48.1
182.1
330.9
429.5
635.2 966.058 1240.752 1610.736 2167.388
State Shares (as a % of total shares)
N/A
N/A
41.38
49.06
43.31
38.74
35.42
31.52
34.25
36.11
38.87
Legal Person Shares (as a % of total shares)
N/A
N/A
27.86
23.07
23.51
24.99
28.38
32.74
30.39
27.81
24.49
Note:
1
Source: China Securities Regulatory Commission, China Securities and Futures Statistical, China Finance and Economic Publishing House, Beijing, 1999, and
CSRC’s official web site http://www.csrc.gov.cn/CSRCSite/deptlistcom/stadata/stadata.htm. Some of the statistics were not kept by the CSRC until 1992.
2
Including both IPO and seasoned offerings of A and B shares.
3
By the end of 2002, the Chinese market had total market capitalization of US$462.9 billion, which was slightly higher than the US$456.3 billion (HK$3559
billion) total market capitalization of the main board of Hong Kong Stock Exchange.
4
According to the Flow of Funds Accounts published by the US Federal Reserve Board of Governors, the US stock market had a total market capitalization of
US$11833.9 billion at the end of 2002, or 113% of US GDP of the same year.
5
Including institutional and individual accounts.
25
Table 2. Definitions of Variables
Variable Name
asset
Definition
Book value of asset.
debt
Ratio of total debt to debt plus total shareholders’ equity. Calculated as
(short term debt + long term debt) / book value of assets.
ROA
Return on assets: operating income (OI) of year t (before depreciation,
amortization and extraordinary items) divided by book value of assets at
the end of year t.
ROS
Return on sales, calculated as the ratio of operating income to sales.
sales
Value of sales used to measure firm size and used as a proxy of market
power.
capex
Capital expenditure, defined as the change in net fixed assets from year t-1
to year t, plus depreciation in year t.
ln_capex
Natural log of Capex
salegrow
Growth rate of sales. Calculated as (salest - salest-1)/ salest-1.
state_stock
State-owned shares divided by total shares.
lperson_share
Legal person shares divided by total shares.
tradable_share
Tradable shares divided by total shares.
A5
Shares held by top five shareholders divided by total number of shares.
Herfindahl_top5
Herfindahl index of ownership concentration among top five shareholders.
Si2 , where Si is the ratio of shares held by the ith
Calculated as
∑
shareholder to total shares held by all top five shareholders.
∆AR
Credit sales/net cash sales in year t – credit sales/net cash sales in year
before the IPO. “Credit sales” is the change in accounts receivable from
the previous year. “Net cash sales” is revenue minus credit sales.
26
Table 3. Financial Outcomes and Ownership Structures
before and after Public Listing
All
Pre-listing
Post-listing
asset
Mean
1179.299
759.56603
1388.6824
(RMB million)
std dev
3094.8891
3116.9222
3062.7647
6511
2167
4344
Sales
# of obs
Mean
703.624
606.636
752.979
(RMB million)
std dev
1384.845
1361.643
1394.070
ROA
ROS
Debt
Capex
(RMB million)
Salegrow
A5
Herfindahl_top5
state_share
lperson_share
# of obs
6547
2208
4339
Mean
0.087
0.153
0.057
std dev
0.085
0.093
0.060
# of obs
6043
1879
4164
Mean
0.134
0.191
0.108
0.190
std dev
0.176
0.120
# of obs
6065
1901
4164
Mean
0.304
0.344
0.284
std dev
0.195
0.207
0.185
# of obs
6474
2149
4325
Mean
110.260
98.304
113.889
std dev
344.262
392.746
328.107
# of obs
4750
1106
3644
Mean
0.177
0.207
0.167
std dev
0.563
0.486
0.586
# of obs
5758
1437
4321
Mean
0.595
0.674
0.589
std dev
0.158
0.228
0.149
# of obs
4652
339
4313
Mean
0.647
0.656
0.647
std dev
0.242
0.254
0.240
# of obs
4750
409
4341
Mean
0.318
0.450
0.306
std dev
0.279
0.360
0.267
# of obs
4645
391
4254
Mean
0.310
0.353
0.306
std dev
0.275
0.344
0.268
# of obs
4644
391
4253
Note: The values of ROA, ROS, debt, salegrow, A5, Herfindahl_top5 are expressed as fractions, while asset,
sales and capex are in RMB million, adjusted by the GDP deflator (1995=100). An observation is included
in sample summary statistics if it is used in one of the subsequent regressions. The full sample is an
unbalanced panel in that new firms are added over time and a firm may re-enter the panel after
disappearance.
27
Table 4: Effects of Public Listing
(1)
(2)
(3)
(5)
(6)
ln_ capex
Debt
salegrow
ROA
ROS
0.308
-0.111
0.023
-0.061
-0.024
(2.39)**
(9.57)***
(0.57)
(19.37)***
(4.49)***
0.203
-0.084
0.047
-0.078
-0.063
(1.03)
(4.93)***
(0.76)
(21.11)***
(10.24)***
L2
-0.186
-0.049
0.071
-0.083
-0.088
(0.71)
(2.17)**
(0.87)
(17.69)***
(10.86)***
L3
-0.338
-0.025
0.090
-0.104
-0.107
(1.00)
(0.86)
(0.87)
(23.46)***
(13.84)***
L0
L1
L4
L5_6
-0.787
0.023
0.022
-0.105
-0.137
(1.81)*
(0.65)
(0.17)
(15.59)***
(11.79)***
-1.165
0.030
-0.042
-0.122
-0.131
(2.19)**
(0.67)
(0.26)
(11.88)***
(7.29)***
log (salest-1)
0.250
0.001
-0.529
0.015
0.035
(3.41)***
(0.27)
(27.46)***
(5.17)***
(7.06)***
log (debtt-1)
-0.013
0.000
0.002
0.000
(1.39)
(0.08)
(6.22)***
(0.78)
ROA_national
1.009
(8.11)***
ROS_national
0.447
(2.32)**
Constant
14.021
(7.32)***
(3.05)***
(25.36)***
(3.70)***
(5.51)***
Observations
2461
3679
3445
2468
2468
Number of Firms 505
657
634
572
572
R-squared
0.20
0.23
0.44
0.17
0.15
0.317
10.878
-0.206
-0.558
Note. *, **, *** represent statistical significance at the 10, 5, and 1 percent levels. The
coefficients for year dummies are not reported. Because lagged values are required, we are
essentially restricting ourselves to firms that have at least two years’ pre-listing data available.
28
Table 5: Financial Packaging and Firm Performance
(1)
(2)
(3)
∆AR
ROA
ROS
L0
-0.061
-0.025
(19.36)***
(4.61)***
L1
-0.019
-0.078
-0.063
(0.77)
(21.11)***
(10.19)***
L2
-0.046
-0.083
-0.088
(1.45)
(17.69)***
(10.90)***
L3
L4
L5_6
-0.064
-0.104
-0.108
(1.63)
(23.45)***
(14.04)***
-0.084
-0.106
-0.134
(1.62)
(15.58)***
(11.66)***
-0.106
-0.122
-0.132
(1.70)*
(11.88)***
(7.37)***
log (salest-1)
0.015
0.037
(5.14)***
(7.36)***
log (debtt-1)
0.002
0.001
(6.17)***
(1.18)
ROA_national
1.007
(8.10)***
ROS_national
0.483
(2.52)**
∆AR
Constant
-0.001
0.036
(0.31)
(4.70)***
0.006
-0.205
-0.588
(0.12)
(3.68)***
(5.82)***
Observations
3803
2468
2468
Number of Firms
667
572
572
R-squared
0.01
0.44
0.18
Note. *, **, *** represent statistical significance at the 10, 5, and 1 percent levels. The coefficients for
year dummies are not reported. ∆AR measures how much credit sales as a percentage of net cash sales
differ from that in the year right before IPO. By definition, ∆AR = 0 for the year right before the IPO, and
that year’s data are excluded from the analysis.
29
Table 6. Ownership Structures and Post-listing Performance
lperson_stock
tradable_stock
A5
herfindahl_top5
∆AR
log (salest-1)
log (debtt-1)
ROA_national
(1)
(2)
ROA
ROS
-0.000
-0.002
(0.03)
(0.11)
-0.133
-0.376
(7.21)***
(6.52)***
0.032
0.041
(2.05)**
(0.82)
-0.034
-0.085
(2.73)***
(2.17)**
0.011
0.060
(2.96)***
(5.37)***
0.009
0.049
(4.34)***
(7.73)***
-0.000
-0.002
(0.06)
(1.62)
1.135
(12.12)***
ROS_national
2.328
(8.19)***
Constant
-0.147
-1.036
(3.42)***
(7.57)***
Observations
2441
2441
Number of Firms
793
793
R-squared
0.12
0.11
Note: The sample includes only post-listing years. Unlike the previous analysis, we do not
require pre-listing data in this analysis, which is why more firms are included in this table.
30
abc
C:\x iaozu \EoT\PubL ist_resu bmission _final.doc
December 12, 2003 2:02 P M
31