Content uploaded by Weiying Zhang
Author content
All content in this area was uploaded by Weiying Zhang on May 13, 2019
Content may be subject to copyright.
The Road to Capitalism: Competition and
Institutional Change in China1
Shaomin Li and Shuhe Li
City University of Hong Kong, Kowloon, Hong Kong, China,
E-mail: efshuhe@cityu.edu.hk
and
Weiying Zhang
Peking University, Beijing, China
Received August 16, 1999; revised February 18, 2000
Li, Shaomin, Li, Shuhe, and Zhang, Weiying—The Road to Capitalism: Competition
and Institutional Change in China
We develop a theory of institutional change and apply it to analyze China’s transition
toward capitalism. We focus on how product market competition induces institutional
change through the interaction between bureaucrats and managers in regional
government-controlled economies. When cross-regional competition is sufficiently in-
tense, each region has to cut production costs. Given that the efforts of managers are not
verifiable, local governments may have to grant total or partial residual shares to the
managers. In general, intense product competition stimulates the rise of a private property
system. We submit our theory to a vigorous empirical test using China’s industrial census
data of more than 400,000 firms. The test supports strongly our postulation that cross-
regional competition is the driving force behind China’s transition toward capitalism. J.
Comp. Econ., June 2000, 28(2), pp. 269–292. City University of Hong Kong, Kowloon,
Hong Kong; and Peking University, Beijing, China. © 2000 Academic Press
Key Words: China; interregional competition; institutional change; privatization.
Journal of Economic Literature Classification Numbers: D23, H70, L13, P21.
1We thank Steven Chiu, Xiaokai Yang, and especially John P. Bonin and the two anonymous
referees for their valuable comments. Nancy Hearst provided excellent English editing. The authors
are listed alphabetically.
Journal of Comparative Economics 28, 269–292 (2000)
doi:10.1006/jcec.2000.1653, available online at http://www.idealibrary.com on
269 0147-5967/00 $35.00
Copyright © 2000 by Academic Press
All rights of reproduction in any form reserved.
1. INTRODUCTION
Institutional arrangements, particularly property rights, are central to incen-
tives and hence to economic performance. Under the condition that information
about individual attributes and actions is largely decentralized, most economists
agree that private ownership is the most high-powered incentive instrument. This
view is supported by the fact that more than twenty former socialist economies,
with about one-third of the world population, are trying to undergo the transition
to capitalism.2Yet our knowledge of institutional change is limited. How do
institutions change? What factors drive efficient institutional change? In partic-
ular, what are the driving forces behind the transition from government to private
ownership? These questions provide both unprecedented challenges as well as
opportunities for economists to study institutional change.
Among the transition economies, the Chinese case is particularly intriguing.
When Deng Xiaoping and his comrades began the reform in 1978, no one,
including Deng himself, expected to witness a nearly double-digit annual growth
rate in the subsequent two decades and the rise of a predominant non-state sector.
China’s phenomenal performance is cited by some economists as an example of
why privatization is not a necessary precondition for efficiency because the high
growth rate in China has occurred under the dominance of public ownership (e.g.,
Stiglitz, 1994). Some argue that the success of township and village enterprises
(TVEs), which are a form of collective ownership, challenges standard property
rights theory (Weitzman and Xu, 1994; Li, 1996). However, the Chinese expe-
rience is not consistent with these arguments. In the last two decades, especially
since the early 1990’s, both state-owned enterprises (SOEs) and TVEs have been
increasingly privatized and most newly established firms are private enterprises.
In 1978, nearly four-fifths of the total industrial output in China came from
SOEs. By 1997, the SOEs’ share had shrunk to slightly more than a quarter
(Statistical Survey of China, 1998, p. 99). The major players behind the rise of
a private ownership system are local governments at various levels (China
Reform Foundation, 1997; Cao et al., 1999).
Interestingly, the Chinese economic reform began with decentralization, rather
than with the development of a private ownership system, and with revitalization,
rather than privatization, of state firms. What are the driving forces behind the
unintended and accelerating rise of a private ownership system in China? What
motivates local governments to privatize the enterprises under their control and
to issue licenses to newly established private firms?
The work of North (1990) and Weingast (1995) provides some hints about
answers to these questions. North (1990) maintains that institutions are the rules
2“Capitalism,” according to Webster’s Collegiate Dictionary (ninth edition), is “an economic
system characterized by private or corporate ownership of capital goods, by investments that are
determined by private decision rather than by state control, and by prices, production and the
distribution of goods that are determined mainly by competition in a free market.”
LI, LI, AND ZHANG270
of the game, while organizations are the players, and competition among orga-
nizations is the key to institutional change. Weingast (1995), and Qian and
Weingast (1997) propose that market-preserving federalism provides a good
political foundation for economic development. They argue that cross-regional
competition played a central role not only in the rise of England’s economic
power in the eighteenth century and that of the United States in the nineteenth
century but also in the rise of the Chinese economy during the last two decades.
More recently, Cao et al. (1999) argue that federalism, Chinese style, has induced
privatization, Chinese style. However, one issue these authors do not address, at
least not formally, is how cross-regional competition stimulates the rise of a
private ownership system. It is not clear whether it is in the interest of local
bureaucrats to privatize. Although privatization generally makes the pie that they
share with firm managers and others bigger, it also decreases the relative shares
of the local bureaucrats.
We develop a theory of institutional change for transition economies by
characterizing it as the rise of a private ownership system. We then apply our
theory to explain China’s road to capitalism. We argue that the rise of the private
ownership system consists of two essential components, the privatization of
existing SOEs and collective-owned enterprises (COEs) and the establishment of
new private firms.
For ease of exposition, we focus on how cross-regional competition in the
product market triggers privatization of SOEs and COEs. However, the same
logic can be applied to analyze how cross-regional competition induces the
establishment of new private enterprises. In our paper, firm ownership is defined
by residual claimancy.3Privatization is the process of shifting residual claims
from the government to managers. In our model, there are two local governments
and two enterprises or firms. The enterprises were formerly owned by the central
government. At the initial stage of reform, the central government gave the
enterprises to the local governments, with each local government owning one
enterprise. After the localization, the central government still maintains the
authority to set tax rates as well as to retain a share of the tax revenue, but the
local government obtains the residual claim on after-tax profits and also has the
right to decide whether or not to shift residual claims to management. In other
words, the local government has the autonomy to decide whether or not to
privatize. To simplify our analysis, we assume that the manager has all the
control rights over the firm’s business, except for the rights of taxation and
privatization. We further assume that the manager’s residual claim rights are well
3Traditionally, ownership is defined by residual rights. Grossman and Hart (1986) define owner-
ship as control rights over assets. Economists recognize that both residual claims and control rights
are indispensable to ownership. Here we omit control rights not because they are irrelevant but for
technical tractability. Nevertheless, we conjecture that our results apply to control rights as well.
CHINA’S ROAD TO CAPITALISM 271
preserved in privatized firms. Thus, when the manager holds all residual claim
rights to the firm, he is the de facto owner of the firm, enjoying both residual
control rights and residual claim rights.
The two firms play a Bertrand–Nash price game in markets with differentiated
products. The production cost is determined by the manager’s non-verifiable
effort. The local government is concerned with its own total revenue, i.e., the sum
of its share of tax revenues and any profit remittance, which depends on its
market share and the profit margin. We show that, when competition is suffi-
ciently intense in the product market, the local government will be induced to
shift the residual claims to the manager. The reasoning for this is inductive. As
the product market becomes more competitive, the market share, and therefore
the profits, is more sensitive to production costs. In order to maintain a minimum
market share for survival, the manager must be motivated to work harder. Given
that verification of the manager’s effort is impossible, privatization is the only
effective means by which the local government can motivate the manager. In
contrast, if the central government sets the after-tax residual share or if two local
governments collude perfectly to maximize their joint revenue, public ownership
may prevail. We find that efficiency generally improves as a consequence of
privatization.
Our theory offers an explanation for the ongoing process of privatization and
the establishment of new private firms in China. We show that this process is a
consequence of cross-regional competition, which has followed the decentrali-
zation policy introduced at the early stage of reform. More generally, the
relevance of our theory extends beyond China and former socialist economies.
According to a World Bank report, the output of SOEs accounts for a large share
of GDP in many countries, including not only transition economies but also
developing economies and even industrial economies (World Bank, 1995).
However, across countries and across time, SOEs are poor performers. In the past
decade, privatization of SOEs has taken place not only in socialist and develop-
ing economies but also in developed economies. We conjecture that the inten-
sifying cross-country competition resulting from globalization has been, and will
continue to be, one of the most fundamental driving forces behind the worldwide
movement toward privatization and the transition to capitalism.
2. THE MODEL: COMPETITION AND PRIVATIZATION
Consider a multi-stage game of two firms and two local governments. The two
firms were originally owned by the central government. At stage 0, the central
government delegated ownership to local governments; firm iis controlled by
local government iand run directly by manager ifor i1, 2. The central
government still maintains authority to set the profit tax rate (1
), where 0
1. The local government’s share of tax revenue is
, where 0
1. These
LI, LI, AND ZHANG272
parameters are the same for both regions. At stage 1, given (1
) and
, local
government idetermines the manager’s after-tax profit retention rate

i僆[0, 1]
in institutional competition; that is, it decides whether or not to privatize firm i.
At stage 2, given (1
) and
,

1and

2, the two managers make unverifiable
effort choices in cost-reduction competition. At stage 3, the two firms compete
with each other in the product market by setting prices.
Since our model is timeless, the residual is best interpreted as the present value
of all future residual flows when the theory is applied. Consequently, privatiza-
tion should be understood as a permanent transfer of residual claims from the
government to private hands. A short-term contract between the local govern-
ment and management, such as the contract management responsibility system
(jingying chengbao zerenzhi) practiced in China, can be seen as partial privat-
ization of state-owned enterprises; i.e.,

i1. A complete sell-out of an SOE
is full privatization; i.e.,

i1.
We model stages 1 through 3, assuming that the central government’s initial
decision to decentralize and to set the tax rate and the local government’s share
of tax revenue are determined exogenously, although the tax parameters can
affect privatization as will be seen later. For this purpose, consider a spatial
model. Suppose that the two firms are located at the two ends of a Hotelling’s
linear city having a length of 1. They produce an identical consumer product.
There is a continuum of consumers evenly located along the linear city and each
consumer buys one unit of good and incurs a unit transportation cost t. A natural
interpretation of tis that it is a parameter of the degree of competition in the
product market; a lower trepresents a higher degree of competition. We assume
that tsummarizes all the factors affecting competition, such as transportation
costs, trade barriers, enforcement, and other costs. Some of these factors are
under the control of the central government, and others are under the control of
local governments or are determined by technology. Since we focus on how com-
petition triggers privatization, we assume that these factors are given exogenously.
Let
ibe the profit, pibe the price, cibe the unit cost, and xibe the market
share of firm i. Then
i(pici)xi. We assume that the cost function of firm
iis cic0ai, where c0is an intrinsic cost that is the same for both firms,
and aiis manager i’s work effort, where 0 ⱕaiⱕc0.4Assume that price setting
is costless and, if the manager is indifferent between two prices, i.e., they yield
the same net utility, he will choose the one that results in larger profit. We assume
that managers are risk-neutral and that manager i’s disutility of effort is
ai
2/2,
where
0. The objective functions to be maximized, of the central govern-
ment, local government i, and manager i, are (1
)(1
)(
1
2),
(1
)
i(1

i)
i, and

i
i
ai
2/2.
We treat the cost and disutility functions as common knowledge, but assume
that managerial efforts and costs cannot be verified by a third party. Partly as a
4See also Hart (1983) for a similar formulation.
CHINA’S ROAD TO CAPITALISM 273
result of this, the managerial contract is based solely on profits.5Given that total
demand and total transportation costs are both fixed, social welfare can be measured
by the reverse of the sum of total production costs and total disutility. Thus, the
first-best symmetric allocation aFB is the solution to the following problem:
min c0⫺a⫹
a2
s.t. 0 ⱕaⱕc0.
We have for solutions ai
FB aFB 1/(2
)ifc01/(2
) and ai
FB aFB
c0if c0ⱕ1/(2
) for i1, 2.
We now search for symmetric subgame perfect equilibria by backward induction.
In the third stage, given the residual shares, production costs, and pj, manager i’s
problem is to choose pito maximize after-tax retained profits. That is,
max

i
i
s.t. piⱖ0.
By using a standard analysis of Hotelling’s linear city model, in a Nash equi-
librium of the price game, we have for ijand 兩a1a2兩3t,
pi⫽t⫹cj
3⫹2ci
3,
xi⫽1
2t
冉
t⫹cj
3⫺ci
3
冊
,
i⫽1
2t
冉
t⫹cj
3⫺ci
3
冊
2. (1)
For firm ito survive, xi0. The inequality 兩a1a2兩3tis the simultaneous
survival condition. Otherwise, both market share and profits will be zero.
In the second stage, given the residual shares and aj, manager i’s problem is
to choose his action aito maximize his utility:
max

i
2t
冉
max
再
0, t⫹ai
3⫺aj
3
冎冊
2
⫺
ai
2
2
s.t. 0 ⱕaiⱕc0.
5We emphasize a particular feature of the spatial model in this paper. At the symmetric equilib-
rium of the spatial model, both the market share and profits of each firm are determined uniquely by
the transportation cost tand an independent variable of managerial effort ai. Hence, we can isolate
the effect of competition from the usual incentive considerations. As the two local governments
collude perfectly and competition disappears, the agency problem does not matter in our model. In
contrast, in a standard agency model, the principal needs to design incentive-compatible payment
schemes even in the absence of competition.
LI, LI, AND ZHANG274
For 9
t
,6we have the following reaction functions for ij:
ai⫽

i
3t⫺aj
9
t⫺

i
. (2)
From (2), the efforts of the two managers are strategic substitutes. In addition,
given manager j’s effort aj, manager i’s effort increases monotonously with his
residual share

iwith ai0 for

i0.
For interior equilibrium solutions of the action game, we have for ij
ai⫽

i
9
t⫺2

j
3
9
t⫺

i
⫺

j
,
xi⫽9
t⫺2

j
29
t⫺

i
⫺

j
,
i⫽t9
t⫺2

j
2
29
t⫺

i
⫺

j
2(3)
so long as the parameters are such that actions are interior solutions.
A necessary condition for interior solutions to exist is 9
t2
. Differenti-
ating xiwith respect to

iand

j, when 9
t2
, we have
⭸xi
⭸

i⫽
9
t⫺2

j
29
t⫺

i
⫺

j
2⬎0
⭸xi
⭸

j⫽
9
t⫺2

i
29
t⫺

i
⫺

j
2⬍0. (4)
That is, firm i’s market share increases with the residual share of its own
manager, and it decreases with the residual share of its rival’s manager. This is
the fundamental reason that the two local governments compete for privatiza-
tion.7
In the first stage, given

j, local government i’s problem is to choose

ito
maximize its total revenue, subject to manager i’s participation constraint,
6This is a sufficient second-order condition.
7However, the partial effect of competition on effort, given shares, depends on the relative sizes
of the residual shares. It is easy to show that one’s effort increases with the degree of competition if
and only if one’s residual share is greater than the residual share of one’s rivals. At symmetric
equilibrium, the efforts are independent of the degree of competition and depend only on the residual
shares. For a general model in which the effect of competition on effort is ambiguous, see Schmidt
(1997). In our model, although competition has an ambiguous partial effect on effort, strategic
interactions between the two governments imply that, as competition becomes more intense, each
local government has to grant a higher residual share to its manager to induce more effort because
effort becomes more valuable. Thus the total effect of competition on effort is, in general, positive
because it induces a change in the residual shares.
CHINA’S ROAD TO CAPITALISM 275
max
1⫺
⫹1⫺

i
9
t⫺

1
⫺

2
2t9
t⫺2

j
2
2
s.t. 0 ⱕ

iⱕ1,

i
t9
t⫺2

j
2
29
t⫺

1
⫺

2
2⫺

i
9
t⫺2

j
2
23
9
t⫺

1
⫺

2
2ⱖ0.
The participation constraint is equivalent to 9
tⱖ

i
.8Note that local
government i’s revenue consists of two components,
(1
)
i, which is the
tax revenue, and (1

i)
i, which is the profit revenue. The trade-off facing
local government iis that a higher

igenerates a higher profit
ibut decreases
its relative share. We refer to the first effect as the incentive effect and to the
second effect as the distribution effect.
Write the first part of the above objective function as gi(

i,

j)(
(1
)(1

i)
)/(9
t

1

2
)2, for ij. Differentiating with respect
to

i, we have
⭸gi
⭸

i⫽2
⫹1⫺
⫺9
t
⫺

i
2⫹

j
2
9
t⫺

1
⫺

2
3. (5)
From (5), it is clear that there is no interior symmetric equilibrium solution to the
residual share game that is derived from the action game with interior solutions.9
Let t*2(
(1
)
)/(9
). For 9
t2
, the participation constraint is
satisfied, and it is easy to check that for all

i僆[0, 1],
⭸gi
⭸

i

i,0⫽2
⫹1⫺
⫺9
t
⫺

i
2
9
t⫺

i
3⬍0ift⬎t*,
⭸gi
⭸

i

i,1⫽2
⫹1⫺
⫺9
t
⫺

i
2⫹
2
9
t⫺

i
⫺
3ⱖ0iftⱕt*.
Collecting our results, we have the following proposition.
8The reader may be curious about this constraint since it implies that, given other factors, the
smaller the residual share for the manager is, the more likely it is to be met. The reason for this is
that the above inequality includes the manager’s incentive compatibility constraint.
9The only exceptional case is tt*. In this rare case, after-tax residual shares do not matter as
long as they are symmetric across regions. Thus, there is a discontinuity introduced when

iis
reduced from a positive value to zero.
LI, LI, AND ZHANG276
P
ROPOSITION
1. If t t*, (

*
1,

*
2,a*
1,a*
2,p*
1,p*
2), where

*
1

*
20, a*
1
a*
20, and p*
1p*
2tc0,is a subgame perfect equilibrium. If (2
/9
)
tⱕt*and c0(
/3
), (

*
1,

*
2,a*
1,a*
2,p*
1,p*
2), where

*
1

*
21, a*
1
a*
2
/(3
), and p*
1p*
2tc0(
/3
), is a subgame perfect
equilibrium.
Proposition 1 says that, if competition is sufficiently intense, i.e., if 2
/9
tⱕt*, the local government will be induced to privatize its firm. The intuition
is that, under sufficiently intense competition, the incentive effect dominates the
distribution effect. The incentive effect is the marginal effect of the change in the
residual share on total profit, which, ignoring the common denominator and other
irrelevant items, is given by I2
(
(1
)(1

i)
)0. The
distribution effect is the marginal effect of the change in the residual share on the
local government relative share for fixed total profit and is given by S
(9
t

i

j
)0. If tt*, IS, for the symmetric case, and
therefore the local government prefers to retain all the profits even at the expense
of the manager’s incentives. On the other hand, if tt* and 9
t2
,IS,
for the symmetric case, and therefore the local government prefers to retain tax
revenue only, because the revenue increase resulting from a bigger profit share
will not offset the tax revenue loss.
The underlying reason for this result is that the sensitivity of market share xi
to

idepends positively on the intensity of competition.10 To see this, differen-
tiate Eq. (4) with respect to t,
⭸2xi
⭸t⭸

i⫽9
9
t⫺

i
⫺

j
⫺29
t⫺2

j
29
t⫺

i
⫺

j
3⬍0,
⭸2xi
⭸t⭸

j⫽9
9
t⫺

i
⫺

j
⫺29
t⫺2

i
29
t⫺

i
⫺

j
3⬎0 (6)
for

i

jand 9
t2
. That is, the marginal effects of both one’s own and
one’s rival’s incentives on one’s own market share decrease as competition
becomes less intense. Recall that the effect of one’s rival’s residual on one’s own
market share is negative. Since the local government’s revenue depends directly
on its market share, pressure to motivate its manager to cut costs is stronger when
competition is more intense and weaker when it is less intense.
Note that the critical level t* is positively correlated with
; hence, the larger
is, other things being equal, the more private ownership is likely to prevail.
Consequently, the larger the share of total tax revenue allocated to the local
government is, the larger the private sector is. The intuition is that the higher the
tax share to the local government, given the tax rate, the more likely it is that the
10 This is consistent with a more general result that shows that the market share is more sensitive
to production costs as competition becomes more intense (Hay and Liu, 1997).
CHINA’S ROAD TO CAPITALISM 277
incentive effect will dominate the distribution effect. Both the critical level t*
and the equilibrium action a* are negatively correlated with the tax rate 1
.
Thus, a lower tax rate may promote a higher degree of private ownership and
induce more efficient action. However, as the tax rate approaches zero, the local
government will have little incentive to privatize since it will get almost no
tax revenue and also almost no profit revenue if the share of the manager is
close to 1.11
Obviously, equilibrium efforts are smaller than the first-best actions. Never-
theless, when competition is sufficiently intense, efficiency can be improved by
promoting privatization. Recall that welfare can be measured by the reverse of
c0a
a2. In Proposition 1, the privatization case produces greater welfare
than the non-privatization case.12
Next, we show that for
9
t2
, there is no symmetric equilibrium to
the residual sharing game when the actions are interior solutions to (3). Note that
there is no interior equilibrium solution to the residual sharing game. We thus
need to examine only the corner solutions to the residual sharing game. First,

*
1

*
20 is not an equilibrium since (⭸gi/⭸

i)(0, 0) 0. Second,

*
1

*
21 is not an equilibrium since (⭸gi/⭸

i)(1, 1) 0. Thus no interior
solutions of the action game can be supported by an equilibrium for (
/9
)
t(2
/9
).
Now we check whether corner solutions of the action game can be supported
as an equilibrium for (
/9
)tt*. First we show that a1a20 cannot
be supported as an equilibrium. Note that a1a20 can be treated as a
limiting case in (3) and (5) with

1and

2approaching zero and that a1a2
0 can only be supported by

1

20 in equilibrium. From the perspective of
local governments, granting positive shares to managers to support zero efforts
is inconsistent with government revenue maximization. When (
/9
)tt*,
(5) implies that

1

20 is not an equilibrium for the residual sharing game,
and hence a1a20 cannot be supported as an equilibrium.
For (
/9
)tt*, we now show that a1a2c0can be supported as
an equilibrium when c0ⱕmin{(
/3
), 3(t*t)}. Given ajc0and

i, the
best response of manager iis ai

i
(3tc0)/(9
t

i
), provided that this
value lies between 0 and c0. Government imaximizes ((
(1
)(1

i)
)/(9
t

i
)2)(t(9
t3
c0)2/2) subject to manager i’s participation
constraint. It can be easily checked that the sign of the derivative of government
i’s objective function with respect to

iis the same as t*t(

i
/9
). Note
that aiⱕc0is equivalent to

iⱕ(3
c0/
). Given c0ⱕmin{(
/3
), 3(t*
t)}, we have t*t(

i
/9
)ⱖ0if

iⱕ(3
c0/
). Thus the best response
11 In fact, the condition in the second part of Proposition 1 does not hold when
1.
12 Note that (c0(
/3
)(
2/9
)) c0.
LI, LI, AND ZHANG278
of government iis

i(3
c0/
).13 It follows that manager i’s best response
becomes aic0.14 Collecting our results, we have the following proposition.
P
ROPOSITION
2. If (
/9
)tⱕt*and c0ⱕmin{(
/3
), 3(t*t)}, (

*
1,

*
2,a*
1,a*
2,p*
1,p*
2), where

*
1

*
2(3
c0/
), a*
1a*
2c0,and p*
1
p*
2t,is a subgame perfect equilibrium.
From Proposition 2, we observe that, when product market competition is
sufficiently intense, managers are motivated to cut all possible costs in their
actions, and as a result the first-best can be achieved. The reader may find it
surprising that, when competition intensifies sufficiently, even without full pri-
vatization, managers have full incentives to reduce costs. There are two reasons
for this result. First, the profit function, and consequently the objective function
of the manager under the conditions in Proposition 2, exhibits increasing returns
to effort. Thus, a small increase in the residual share of the manager may induce
a significant cost reduction. Second, if the potential for cost reduction is limited,
local governments may not need to grant total after-tax residual shares to
motivate the manager since partial residuals will suffice to induce the manager to
cut costs fully.
If 9
t
, Eq. (2) does not hold in general because the second-order
condition is not satisfied. Hence, we need to look for corner solutions to the
action game. Let ki9
t/

i
. Equivalently, manager imaximizes the
following objective function by choosing ai,
3t⫹ai⫺aj2⫺kiai2⫽1⫹kiai⫺aj⫺3t1⫺kiai⫺aj⫺3t,
provided that aiⱖaj3t. Manager i’s objective function becomes non-
positive when aiaj3tsince, in this case, the market share of region iis
zero. We claim that for c03t,ajc0, and

i
9
t, we have a*
ic0.
Note that under the specified conditions, aj3t0, 1 ki0, and the
above function is negative if (aj3t)/(1 ki)ai(aj3t)/(1 ki),
and it is non-positive if 0 ⱕaiⱕ(aj3t)/(1 ki). Thus, it is maximized at
a*
ic0provided that c0ⱖ(c03t)/(1 ki).
Now we are ready to show that a1a2c0can be supported as an
equilibrium for 3tc0ⱕ
t/
. Given ajc0, the participation constraint
of manager ibecomes

i
2t
冉
t⫹ai⫺c0
3
冊
2
⫺
ai
2
2ⱖ0,
which implies

i
9
t. As long as the participation constraint is satisfied,
manager ichooses aic0provided that c0ⱖ(c03t)/(1 ki). To maximize
13 This suffices to motivate manager ito cut costs to the minimum level, which is zero.
14 Note that c0ⱕ(
/3
) and
9
timply c03t. Hence the participation constraint,
(

i
t/2) (
c0
2/2) 0, is met for

i(3
c0/
).
CHINA’S ROAD TO CAPITALISM 279
revenue, local government isets the after-tax residual profit share such that the
participation constraint is binding; that is,

i(
c0
2/
t). Note that c0(c0
3t)/(1 ki) and

i僆[(9
t/
), 1]. Granting a residual share higher than
(
c0
2/
t) cannot induce any further cost cutting. The only other alternative for
government iis to choose

i0, which results in ai0, xi0, and
i
0. To summarize, we have the following proposition.
P
ROPOSITION
3. If 3tc0ⱕ
t/
,(

*
1,

*
2,a*
1,a*
2,p*
1,p*
2), where

*
1

*
2(
c0
2)/(
t), a*
1a*
2c0,and p*
1p*
21, is a subgame perfect
equilibrium.
The equilibrium actions in Proposition 3 are first-best actions.15 Note that the
conditions in Proposition 3 imply 9
t
. Thus, when competition is suffi-
ciently intense, the first-best action is achievable. Again, as in Proposition 2, full
residual transfer to management is not necessary due to the fact that the potential
for full cost reduction is limited. In both Propositions 2 and 3,

*
iincreases as
decreases; thus the degree of privatization will be higher when the tax rate is
higher. If the local government is able to collect more tax effectively, it need not
or better not rely on profit remittance as much as when tax revenue is lower.
Let us use a numerical example to illustrate Proposition 3. Suppose t1,
0.64,
0.01. Then for c0僆(3, 8], Proposition 2 holds. Take c06, we
have

*
1

*
2(9/16). Now suppose tbecomes 0.64. We then have

*
1

*
2
(225/256). In both Propositions 2 and 3,

*
iis positively correlated with c0. Thus,
a higher intrinsic cost requires that a larger residual share go to the manager to
induce full cost cutting. Although a higher intrinsic cost may imply a higher
degree of private ownership, it may not affect efficiency.
For competition to work effectively, regional governments cannot erect trade
barriers, but they must have the autonomy to make other economic decisions to
respond to competitive pressures. Thus far we have assumed that regional
governments have full autonomy to set after-tax residual shares. When the
autonomy of regional governments is restricted by the central government, the
privatization process may be slowed down. Indeed, in the following proposition,
we show that, if all the rights of setting after-tax residual shares are in the hands
of the central government, public ownership will prevail.
Suppose that the central government, rather than local governments, sets the
after-tax residual shares and treats the two regions equally. Equal treatment
implies that

1

2and hence a1a2. It follows that the total revenue of the
government, including all levels and all regions, becomes (1
(1

i)
)t,16 which is maximized at

i0. Thus, we have the following proposi-
tion.
15 This is because c0ⱕ(
t/
)(
/3
)(1/2
) and, hence, aFB c0.
16 Note that profits are independent of the symmetric actions of the managers.
LI, LI, AND ZHANG280
P
ROPOSITION
4. If the central government sets the after-tax residual shares and
treats the two regions equally, then the maximization of total government revenue
results in

*
1

*
20 and, accordingly, a*
1a*
20, p*
1p*
2tc0.
The outcome from centralization of decision making is equivalent to that from
joint maximization by two local governments. If the two regions collude per-
fectly, they achieve jointly the same result as found in Proposition 4. The
intuition is simple. From (3), at the symmetric equilibrium, both market share xi
and profits
iare independent of actions and ownership structure so that xi1
2,
it/2. By colluding at

1

20, each local government can collect full
profits equal to
t/2. However, such collusion cannot be an equilibrium in
general. In particular, when tt*, we have (⭸gi/⭸

i)(

i,0)0 for
sufficiently small

ifrom (5). That is, given that government jchooses

j0,
the best response for government iis to set

i0. Each local government is
tempted to cheat on its rival with respect to privatization until equilibrium is
restored; i.e.,

*
1

*
21 for 2
/9
tt*. In other words, when
competition in the product market is sufficiently intense, two local governments
have to compete with each other regarding the privatization of their firms.
Summarizing the propositions, when the degree of competition is sufficiently
low, public ownership will prevail because the distribution effect dominates the
incentive effect. When the degree of competition is sufficiently high, full or
partial private ownership will occur, depending on whether the intrinsic cost is
high or low. When the intrinsic cost is sufficiently high, full private ownership is
needed to induce the manager to cut costs. When the intrinsic cost is sufficiently
low, partial private ownership will be sufficient to induce full cost-cutting.
Second, when the profit tax rate is sufficiently high, private ownership will be
less likely to occur, because there is little incentive effect through the manager’s
after-tax profit, and hence, the distribution effect will be dominant.17 In the
extreme case of a 100% profit tax rate, the manager will get zero after-tax profit;
thus there will be no incentive effect and public ownership will prevail. A lower
tax rate may promote private ownership because it increases the incentive effect
through the manager’s after-tax profit. However, too low a tax rate may reduce
the local government’s incentive to privatize.18 The intuition is that if the local
government is unable to collect more tax effectively, it is better for it to rely more
on profit remittance. In fact, as the tax rate approaches zero, the local government
will lose incentive to privatize fully, since it will get nearly zero tax revenue and
almost zero profit revenue if the share to the manager is close to 1. Thus, the
effect of the tax rate on privatization is not linear; similar to the Laffer curve, it
is inverse U-shaped.
Third, a larger share of tax revenue to the local government promotes private
17 Recall that the critical t* is negatively correlated with the tax rate (1
).
18 Recall that

*
iis positively correlated with (1
).
CHINA’S ROAD TO CAPITALISM 281
ownership, because it increases the incentive effect through the local govern-
ment’s tax revenue.19 However, a sufficiently high tax share, even 100%, does
not guarantee privatization, which depends also on competition intensity and tax
rate. On the other hand, a zero share of tax revenue to the local government
erodes its incentive to privatize fully. When the central government cannot
collect taxes effectively, the share of tax revenue to the local government may
rise and the effective tax rate may become lower, as in the case of China. While
the first effect promotes privatization, the second has an ambiguous effect on
privatization; thus the overall effect is indeterminate.
Thus far we have interpreted a larger

ias a higher degree of privatization of
existing SOEs, which is one of the two elements contributing to the rise of a
private ownership system. The other component is competition for the establish-
ment of new firms. By similar reasoning, more intense product competition can
result in the setting up of more new private enterprises and fewer new SOEs.
Rather than assuming that the two local governments consider to what extent
they want to privatize their existing SOEs, we assume that they consider what
types of new firms, i.e., private vs state-owned, they want to approve. Then the
results of our model apply when we ignore the funding issue for the new firm as
we do in our model. In reality, new start-up firms require funding. The local
government must finance new start-up SOEs from its own budget. Consequently,
the local government has even less incentive to approve the establishment of new
SOEs if the funding issue of a firm is taken into account. Therefore, our analysis
suggests that more intense interregional competition induces a faster growth of
the private ownership system in a region by facilitating both privatization and the
establishment of new private firms.
3. APPLICATION: THE CHINESE EXPERIENCE
In this section, we apply our theory to explain China’s transition to capitalism.
We then conduct a formal empirical test using China’s industrial census data.
3.1. China’s Road to Capitalism
In the past two decades, and particularly since the early 1990’s, privatization
in China has been accelerating. However, this has not been a deliberate policy of
the central government. Our theory sheds light on this unintended rise of a private
ownership system.
The authority of the central government began to expand in 1949; it dominated
local governments in almost all major economic decision making until the late
1970’s. Partly as a result of such a high degree of centralization, public owner-
ship was predominant from the mid-1950’s to the early 1980’s. The fiscal budget
was centralized and the majority of SOEs were under central government control.
19 Recall that t* is positively correlated with
.
LI, LI, AND ZHANG282
The economic reform launched in 1978 can be characterized as an evolutionary
process of reassigning the economy’s residual claims and control rights from the
central to the local government and from the government to firm managers.
China’s decentralization policy has had two major components. The first is a
fiscal revenue-sharing system (caizheng baogan) between adjacent levels of
governments, under which lower-level governments have an obligation to hand
over a fixed amount or a fixed proportion of their revenues to the superior
government, while keeping for themselves the remaining revenues. This fiscal
decentralization was accompanied by the second major component, a delegation
of state enterprises to local governments (qiye xiafang). By 1996, state-owned
industrial enterprises controlled by the central government accounted for only
1.6% of the total number of firms at or above the township level and 17.8% of
the total industrial output (China Industrial Census, 1996).
The decentralization policy granted local government officials great autonomy
over their economies, including the autonomy to set prices, to make investment
with self-raised funds, and, more importantly, the autonomy to restructure their
firms and issue licenses to newly established firms. Overall, decentralization has
delimited better the property rights between governments at different levels, such
that the government at each level becomes the real residual claimant and
controller of its own public economy. Thus, each region acts as a conglomerate
or as a holding company. This system boosted the local governments’ incentives
to make profits, and, more importantly, it forced local governments to compete
with one another, thus contributing to the marketization of the entire economy.
Although local governments may still use some planning mechanisms to control
their enterprises, they can conduct business with other regions only through a
bargaining process, since no one region has authority over the others. The
relationships among provinces, municipalities, counties, townships, and villages
are more or less marketized.
At the early stage of decentralization, many local governments attempted to
protect their enterprises from competition with other regions by erecting trade
barriers. However, as the size of each local economy became smaller and the
number of local economies increased at lower government levels,20 the erection
of trade barriers by a local government became more costly and, hence, compe-
tition became more intense. Protectionism often failed because efficiency gains
from specialization and exchange exceeded significantly the net benefits of
erecting trade barriers as both informal and formal arrangements emerged to
capture the gains. Since the late 1980’s, local governments have begun to sign
treaties pledging to protect one another’s enterprises as their own (Yang, 1989;
Clarke, 1996). In 1993, the central government enacted the “Law of Anti-
20 China’s administrative divisions are as follows. At the provincial level, there are thirty-one
jurisdictions, many of which are larger than medium-sized countries. At the county level, there are
more than two thousand jurisdictions. At the township level, there are more than 50,000 jurisdictions.
CHINA’S ROAD TO CAPITALISM 283
improper Competition,” prohibiting local governments from erecting trade bar-
riers.
All government bureaucrats seek rents and are reluctant to give up their power.
However, without a monopoly, rents can be guaranteed only by improving the
efficiency of their firms. As our model predicts, because the competitiveness of
each local economy depends on its cost effectiveness relative to that of its rivals,
competition eventually forces local governments to grant more residual shares to
their enterprises and, finally, to privatize.
Under the planning system, SOEs remitted all of their earnings to the govern-
ment, so that there was no distinction between taxes and profits. In 1984, SOEs
began to remit taxes and profits to governments separately. A few years later, the
contract management responsibility system emerged, under which local govern-
ments at various levels decide how after-tax profits will be shared between the
government and managers. This contracting system can be viewed as a form of
partial privatization; it has boosted greatly managerial incentives in state enter-
prises (Zhang, 1997). However, it has also induced pervasive short-term behav-
ior, which is harmful to long-run competitiveness. In particular, intensified
competition has made SOEs less and less profitable compared to the non-state
sector, so that local governments have become burdened with the increased
losses of their SOEs. Thus, various forms of privatization began to emerge and
to accelerate in the 1990’s (China Reform Foundation, 1997; Wang and Xu,
1996).
Although our model characterizes central–local government relations, the
same logic can be applied to any two adjacent levels of governments.21 However,
competitive pressures and their impact on privatization vary at different levels of
government. The effectiveness of cross-regional competition depends on the
following factors: the leverage of regional governments to protect their products
and to erect trade barriers, the degree of autonomy of regional governments to
make decisions in reaction to competitive pressures, the number of competitors,
and the costs of contract enforcement and transportation of their products. In
general, the lower the level of jurisdiction, the greater the effectiveness of
competition from the above-mentioned factors and, hence, the higher the degree
of privatization for the following reasons.
First, a lower-level government has less leverage, i.e., administrative and legal
means, to protect its enterprises and to erect trade barriers. Second, a lower-level
government and its firms tend to have more autonomy. Typically, firms con-
trolled by lower-level governments produce fewer strategic products, such as
arms, and thus, they are less subject to central planning and political control.
Third, a lower-level government and its firms face more homogeneous compet-
itors. Enterprises controlled by the same level of government tend to produce
21 Unlike at the central government level, which is a monopoly, regional competition can induce
decentralization at local levels.
LI, LI, AND ZHANG284
products of similar complexity and technology. Finally, firms controlled by
lower-level governments produce more simple, standardized products, such as
basic consumption goods, and hence, enforcement of contracts is easier.
County-level governments are the lowest-level jurisdictions that can issue
licenses to firms. Licenses to individual business and private firms are issued
mostly by the administration for commerce and industry of the county-level
governments. Due to intense cross-county competition, most counties have
permitted or encouraged the establishment of new private firms in the reform era,
especially since the early 1990’s. The share of the private sector expanded
quickly as both privatization and the establishment of new private firms accel-
erated during the early to mid-1990’s. However, the share varied significantly
from county to county due to different intensities of competition; hence, we have
a good opportunity to test our theory.
3.2. An Empirical Test
We now conduct an empirical test of our theory using newly available Chinese
industrial census data from 1993 to 1995. The censuses are conducted by the
State Statistical Bureau of China and cover all manufacturing firms subordinate
to the township government or above. The number of firms included varies
between 400,000 and 500,000. In 1995, the output of these firms accounted for
91% of the total industrial output of all firms with independent accounting
systems (Chinese Industrial Census, 1995; China Statistical Yearbook, 1996, p.
414). Firms not included in the census are the very small, often family-run,
workshops. Thus, the firms included cover basically the entire population of
China’s manufacturing industry. The data set contains ownership type, level of
government control, geographic location, revenue, and other performance and
demographic variables. In China, virtually all firms are subordinate to (lishu)
governmental organizations at different levels.22 There are five levels of govern-
ment control in terms of lishu, central, provincial, municipality or prefecture,
county, and township. The firms subordinate to a government may be state-,
collective-, or even privately owned. In general, SOEs are subordinate to county
governments or above; COEs are subordinate to county-level governments and
above or to township governments and village committees (TVEs); private
enterprises are generally subordinate to village committees, to township and
county governments, or to some semi-official business associations.
For the test, we use the regression equation
Bit ⫽h0⫹h1Ti,tk⫹h2Zi,tk⫹h3Bj,tk⫹h4Bi,tk⫹
⑀
it,
22 “Firm A lishu government X” means that A is under the jurisdiction of X, and the latter is
responsible for administering the former. If A is a public firm, X has the authority to make essentially
all decisions for A. In other words, X is similar to a board of directors of A.
CHINA’S ROAD TO CAPITALISM 285
where Bit is the share of the private sector in region iat year t,Ti,tkis the index
of transportation in region iat year tk,Zi,tkis the degree of autonomy of
region iat year tk,Bj,tkcaptures the neighborhood effect (ji), and Bi,tk
is the self-effect. The transportation index and the degree of autonomy are
proxies for the intensity of interregional competition, which is not observable
directly. The initial private sector share serves as a control variable. If our theory
holds, we should observe all the explanatory variables exerting a positive effect
on Bit. Higher Bj,tkand Ti,tkimply stronger competition from private firms in
the neighboring regions. A higher Bi,tkimplies stronger competition from
private firms within the region. A higher Zi,tkmeans that region ihas more
autonomy, namely, more freedom to restructure its SOEs and to establish new
private firms. Thus, it is more responsive to cross-regional competition. Our
theory predicts that all these factors contribute to a higher private-sector share by
facilitating privatization and the establishment of new private firms through
intensifying competition.
In our test, Bit is measured by the proportion of revenues contributed by the
private sector in a county in 1995. The Ti,tkis measured by two variables at the
provincial level; one is the ratio of the length of the coastal line to the land area,
and the other is the ratio of the length of railways to the land area in 1992. The
Zi,tkis measured by the ratio of the revenue of SOEs subordinate to county or
lower-level governments to the total revenue of all SOEs in a county in 1994.23
The Bj,tk, is measured by the proportion of revenue of the private sector in the
neighboring counties, excluding the county under evaluation, in 1993. Neigh-
boring counties are all counties in the same prefecture as county i. There are 334
prefectures and 2,134 counties in China, with an average of 6.4 counties per
prefecture (China Statistical Yearbook, 1996, p. 3). Here, Bi,tkis the proportion
of revenue of the private sector in the county under evaluation, i, in 1993. The
unit of analysis is the county. We aggregate information from all firms in 1993
(446,265), 1994 (485,052), and 1995 (450,233) at the county level and calculate
indices of privatization and autonomy for all counties based on the industrial
censuses. Then, we estimate the regression equation. Given the comprehensive
coverage of our data, this is the most rigorous test we can design.
In one test, the private sector is broadly measured by the non-state sector,
including privately owned firms, foreign-invested firms, domestic joint ventures,
joint stock firms, and collectively owned firms.24 The correlation coefficient
between Bi,tkand Bj,tkis 0.471. In order to avoid possible multicollinearity
between Bi,tkand Bj,tk, we test three models. Model A uses both Bi,tkand
Bj,tk; Model B uses the share of revenue of the private sector in the entire
23 The data for Zi,tkin 1993 are not available.
24 Studies show that a large part, some 20 to 50% in various regions, of the collectively owned
firms are in fact de facto privately owned firms, especially since the 1990’s (Li, 1998). Thus, firms
in the non-state sector can be seen as either fully private or partially private firms.
LI, LI, AND ZHANG286
prefecture, including both the county under evaluation and the neighboring
counties; Model C uses only Bj,tk, the neighborhood effect.
Table 1 summarizes the regression results. All three models perform equally
well. All the independent variables in each model influence the dependent
variable in the hypothesized direction and are highly significant. According to
Model A, a 1% higher private share in the neighboring counties triggers about a
0.2 to 0.24% higher private share in a county two years later (Tables 1 and 2).
In order to test for the robustness and consistency of our theory, we use a
different, narrowly defined, measure of the private sector, which excludes all
SOEs, all COEs in 1993, and the COEs at the county level and above in 1995.25
The degree of local autonomy is measured by the ratio of the revenue of SOEs
and COEs subordinate to township-level governments to the revenue of all SOEs
and COEs in 1994. The results, shown in Table 2, are very similar to those of
Table 1 with R-squares ranging from 48 to 52%. The test results of different
specifications and measures all provide very strong support for our theory.
One of the limitations of our model is that we ignore the effect of regional
specialization. We assume different regions produce identical products, which
are differentiated only by the transportation costs for consumers. In reality, there
is a pattern of regional specialization in China, with many northern regions
specializing in capital goods, southern regions in consumer goods, and other
areas in raw materials and energy. In reaction to competition, a region may
differentiate its products through specialization, thus reducing or delaying the
effect of competition on privatization. That is, if there were no regional special-
ization, the degree of privatization would be higher. Thus, the effect of compe-
tition on privatization might be underestimated in our empirical investigation.
Summarizing the discussions and tests in this section, we conclude that our
theoretical predictions are quite consistent with the observed characteristics in
China. The economies of the coastal regions are more privatized than those
inland because the former enjoy not only lower transportation costs, which
facilitate cross-regional competition, but also, and more importantly, greater
autonomy.26 Similarly, SOEs in northeastern and southwestern China tend to be
less privatized, since there are few private enterprises in neighboring regions,
making competition less intense. These factors have contributed to the widening
of the income gap between the coastal and inland regions in the reform era.
Sectors with simple or standard contracts are more privatized than sectors with
complex or specific contracts since the former involve lower enforcement costs
and hence face stronger competition. The former include labor-intensive indus-
tries, such as textiles and consumer electronics. The latter generally include
25 For the 1993 data, we cannot distinguish the government level of the COEs; thus, we exclude
all COEs from the private sector.
26 In the 1980’s, the central government granted much more autonomy to the coastal regions than
it did to the inland regions.
CHINA’S ROAD TO CAPITALISM 287
TABLE 1
Regression Results of Privatization Determinants (Broad Definition of Private Sector) (Dependent Variable, Proportion of Revenue of the Private Sector
in a County in 1995; Number of Cases 2,002 (All Counties in China))
Models: A B C
Estimates
R-square 36.1%; 28.3% 24.7%
Adjusted R-square 35.9%; 28.1% 24.5%
prob F0.0001 0.0001 0.0001
ABC
Independent variables Unstandardized Standardized (prob 兩T兩) Unstandardized Standardized (prob 兩T兩) Unstandardized Standardized (prob 兩T兩)
Intercept 0.151 0.000 (0.0001) 0.171 0.000 (0.0001) 0.189 0.000 (0.0001)
1. Coastal line/land (provincial level) 2.174 0.070 (0.0001) 2.340 0.075 (0.0001) 4.199 0.135 (0.0001)
2. Railway/land (provincial level) in 1992 1.766 0.092 (0.0020) 2.152 0.113 (0.0019) 2.226 0.117 (0.0001)
3. Local autonomy in 1994 0.156 0.243 (0.0001) 0.173 0.271 (0.0001) 0.182 0.285 (0.0001)
4. Proportion of revenue of the private sector
in the neighboring counties in 1993 0.195 0.149 (0.0001) 0.390 0.298 (0.0001)
5. Proportion of revenue of the private sector
in the county in 1993 0.371 0.394 (0.0001)
6. Proportion of revenue of the private sector
in the prefecture in 1993 (including the
county under evaluation and the
neighboring counties) 0.521 0.386 (0.0001)
Sources. For variables 3 to 6, China Industrial Census (1993–1995); for variables 1 and 2, China Statistical Yearbook (1993, p. 515); and Atlas of the People’s
Republic of China (1989).
Note. The private sector includes privately owned firms, foreign-invested firms, domestic joint ventures, joint stock firms, and collectively owned firms. Local
autonomy is measured by the ratio of the revenue of SOEs subordinate to county or lower-level governments to the total revenue of all SOEs in the county.
LI, LI, AND ZHANG288
TABLE 2
Regression Results of Privatization Determinants (Narrow Definition of Private Sector) (Dependent Variable, Proportion of Revenues by the Private Sector
in a County in 1995; Number of Cases 2,002 (All Counties in China))
Models: A B C
Estimates
R-square 52.0%; 49.3% 48.2%
Adjusted R-square 51.9%; 49.2% 48.1%
prob F0.0001 0.0001 0.0001
ABC
Independent variables Unstandardized Standardized (prob 兩T兩) Unstandardized Standardized (prob 兩T兩) Unstandardized Standardized (prob 兩T兩)
Intercept 0.125 0.000 (0.0001) 0.124 0.000 (0.0001) 0.127 0.000 (0.0001)
1. Coastal line/land (provincial level) 2.676 0.092 (0.0001) 3.397 0.117 (0.0001) 3.869 0.133 (0.0001)
2. Railway/land (provincial level) in 1992 0.720 0.039 (0.0132) 0.898 0.049 (0.0026) 0.895 0.049 (0.0030)
3. Local autonomy in 1994 0.612 0.580 (0.0001) 0.605 0.573 (0.0001) 0.604 0.572 (0.0001)
4. Proportion of revenue of the private sector
in the neighboring counties in 1993 0.242 0.116 (0.0001) 0.467 0.223 (0.0001)
5. Proportion of revenue of the private sector
in the county in 1993 0.422 0.236 (0.0001)
6. Proportion of revenue of the private sector
in the prefecture in 1993 (including the
county being evaluated and the
neighboring counties) 0.538 0.250 (0.0001)
Sources. For variables 3 to 6, China Industrial Census (1993–1995); for variables 1 and 2, China Statistical Yearbook (1993, p. 515); and Atlas of the People’s
Republic of China (1989).
Note. The private sector includes all firms except (1) all SOEs and (2) all collectively owned firms in 1993 and the collectively owned firms at the county
level and above in 1995. Local autonomy is measured by the ratio of the revenue of SOEs and collectively owned firms subordinate to township-level
governments to the revenue of all SOEs and collectively owned firms in 1994.
CHINA’S ROAD TO CAPITALISM 289
capital-intensive and contract-intensive industries, such as machine tools, bank-
ing, and insurance (Li and Zhou, 1999). For instance, in 1985 the output of SOEs
accounted for 17 and 64% of the total output of the garment and machine tools
industries, respectively. By 1995, the SOEs’ share in the garment industry had
shrunk to 6.9%, while the SOEs’ share in the machine tools industry was still
40% (China Statistical Yearbook, 1986, pp. 242–43, 1996, pp. 414, 418). The
privatization of TVEs has proceeded more quickly than that of SOEs because
TVEs operate in more competitive markets and their township or village gov-
ernments have no leverage to protect them.
Although our model focuses on product market competition, similar analyses
can be applied to other markets, such as capital and labor markets. In reality, the
direct motivation for privatization by local governments is often related to the
financial health of their enterprises (Wang and Xu, 1996). This is a precise
consequence of competition in various markets. Competition erodes the monop-
oly profits that SOEs enjoyed formerly. In many regions, local governments are
unable to get funds from capital markets and to subsidize massive losses, so that
they have to privatize those firms that incur losses. Another important reason for
privatization is that SOEs and collective enterprises cannot keep good managerial
teams and skilled workers because the more efficient private and foreign joint
ventures are able to offer much higher salaries (Liu, 1995). For the same reasons,
almost all new firms established by local governments in recent years have been
private firms.
4. CONCLUSIONS
In this paper, we develop a theory of institutional change in the context of a
transition economy. We show that the rise of a private ownership system occurs
as a consequence of cross-regional competition. Initially the Chinese economic
reforms were not intended to privatize state and collective enterprises. However,
the decentralization policy triggered privatization eventually and the establish-
ment of new private firms through cross-regional competition. In 1997, the
Chinese Communist Party’s Fifteenth Party Congress promoted the formation of
joint-stock systems and various other organizational forms to bail out the vast
majority of failing SOEs; this move is widely viewed as a covert act of
privatization. Formal open privatization in China has thus far not been adopted
as a central government policy for ideological reasons. However, competition is
far more powerful than ideology. Regardless of whether or not the central
government will draw up a blueprint for full privatization, both our theory and
reality show that the privatization process will continue to accelerate with its own
logic and vigor. China has reached a point of no return on the road to capitalism.
The Chinese experience demonstrates that the invisible hand is not only
powerful in allocating resources; it is also powerful in creating institutions. Once
decentralization begins, market competition may precipitate a self-enforcing
LI, LI, AND ZHANG290
development of a private ownership system. The newly founded and privatized
firms intensify, in turn, market competition. This is the major lesson that other
transition and emerging economies may draw from China’s experience.
Nevertheless, the emergence of a private ownership system requires a sound
legal system to protect property rights. In particular, de facto ownership by
managers must eventually become de jure private ownership. Commercial laws
are also needed to enforce contracts between enterprises. Although China enacted
the General Principles of Civil Law in 1986, the Law of Civil Litigation in 1991,
and the Contract Law in 1999, and many other laws since these, there are two
major problems in its current commercial law system. First, there are no clear and
detailed rules to protect private property. To facilitate efficient private invest-
ments, detailed civil codes and procedures are needed to protect private property
under different contingencies. Second, cross-regional commercial disputes are
settled in local courts that are virtually controlled by local governments, in that
the local governments provide the courts with both financial and personnel
resources. To mitigate local protectionism and to facilitate interregional compe-
tition, local courts must become independent of local government control or
major cross-regional commercial disputes must be settled by higher-level courts,
whose jurisdiction is common to the regions.
We are only beginning to understand the driving forces behind institutional
change, in general, and the rise of private ownership systems, in particular. Much
work remains to be done. Although we have demonstrated how decentralization
can promote the development of a private ownership system through competi-
tion, we do not yet have a theory to explain what drives decentralization, nor do
we have a theory to explain how de facto property rights evolve into de jure
property rights. Exploration of these topics will enhance our understanding of
institutional change. This study provides a foundation on which to base such
endeavors.
REFERENCES
Atlas of the People’s Republic of China. Beijing: Foreign Languages Press, 1989.
Cao, Yuanzheng, Qian, Yingyi, and Weingast, Barry R., “From Federalism, Chinese Style, to
Privatization, Chinese Style.” Econ. Transition 7, 1:103–131, 1999.
China Reform Foundation, Xianshi de xuanze: Guoyou xiao qiye gaige shijian de chubu zongjie
(Realistic Choice: Preliminary Summary of the Practice of Reforming Small State-Owned
Enterprises). Shanghai: Shanghai yuandong chubanshe (Shanghai Far East Press), 1997.
Clarke, Donald C., “The Creation of a Legal Structure for Market Institutions in China,” In John
McMillan and Barry Naughton, Eds., Reforming Asian Socialism: The Growth of Market
Institutions. Ann Arbor, MI: Univ. of Michigan Press, 1996.
Grossman, Sanford J., and Hart, Oliver D., “Costs and Benefits of Ownership: A Theory of Vertical
and Lateral Integration.” J. Polit. Econ. 94, 4:691–719, Aug. 1986.
Hart, Oliver D., “The Market Mechanism as an Incentive Scheme.” Bell J. Econ. 14, 2:366–382,
Autumn 1983.
Hay, Donald A., and Liu, Guy S., “The Efficiency of Firms: What Difference Does Competition
Make?” Econ. J. 107, 442:597–617, May 1997.
CHINA’S ROAD TO CAPITALISM 291
Li, David D., “A Theory of Ambiguous Property Rights in Transition Economies: The Case of the
Chinese Non-state Sector,” J. Comp. Econ. 23, 1:1–19, Aug. 1996.
Li, Shaomin, “Success in China’s Industrial Market: An Institutional and Environmental Approach.”
J. Int. Marketing 6, 1:56–80, 1998.
Li, Shaomin, and Zhou, Dongsheng, “Institutional Change and Ownership Choice in China.” Paper
Presented at the Annual Meeting of the Academy of International Business (AIB), Charleston,
SC, Nov. 20–21, 1999. Univ. of Hawaii at Manoa: AIB. [ISSN 0047-2506]
Liu, Shiding, “Shunde qiye zichan suoyouquan zhuti jiegou de biange” (Changes in Enterprise
Ownership Structure in Shunde Municipality). Gaige (Reform Journal) 6:28–35, June 1995.
North, Douglass C., Institutions, Institutional Change, and Economic Performance. Cambridge:
Cambridge Univ. Press, 1990.
People’s Republic of China, State Statistical Bureau. China Industrial Census. Beijing: State
Statistical Publishing House, 1993–1996.
People’s Republic of China, State Statistical Bureau. China Statistical Yearbook. Beijing: State
Statistical Publishing House, 1986–1996. [Available at http://www.chinagate.com/]
People’s Republic of China, State Statistical Bureau. Statistical Survey of China. Beijing: State
Statistical Publishing House, 1998.
Qian, Yingyi, and Weingast, Barry R., “Federalism as a Commitment to Preserving Market Incen-
tives.” J. Econ. Persp. 11, 4:83–92, Fall 1997.
Schmidt, Klaus M., “Managerial Incentives and Product Market Competition.” Rev. Econ. Stud. 64,
2:191–213, April 1997.
Stiglitz, Joseph E., Whither Socialism? Cambridge, MA: MIT Press, 1994.
Wang, Yanzhong, and Xu, Heping, Zhucheng qiye gaige tantao (Studies of Zhucheng’s Enterprise
Reform). Beijing: Jingji guanli chubanshe (Economics and Management Press), 1996.
Weingast, Barry R., “The Economic Role of Political Institutions: Market-Preserving Federalism and
Economic Development.” J. Law Econ. Organ. 11, 1:1–31, April 1995.
Weitzman, Martin, and Xu, Chenggang, “Chinese Township and Village Enterprises as Vaguely
Defined Cooperatives.” J. Comp. Econ. 18, 2:121–145, Apr. 1994.
World Bank, Bureaucrats in Business: The Economics and Politics of Government Ownership.
Oxford/New York: Oxford Univ. Press, 1995.
Yang, Jisheng, “East-West Dialogue in China.” Liaowang zhoukan (Outlook Weekly, overseas ed.) 9:
5–7, Feb. 27, 1989.
Zhang, Weiying, “Decision Rights, Residual Claim and Performance: A Theory of How the Chinese
State Enterprise Reform Works.” China Econ. Rev. 8, 1:67–82, Spring 1997.
LI, LI, AND ZHANG292
