Causes of the Soft Budget Constraint: Evidence on Three Explanations

Causes of the Soft Budget Constraint: Evidence on Three Explanations

JOURNAL OF COMPARATIVE ECONOMICS ARTICLE NO. 26, 104–116 (1998) JE971507 Causes of the Soft Budget Constraint: Evidence on Three Explanations1 Davi...

100KB Sizes 11 Downloads 125 Views

JOURNAL OF COMPARATIVE ECONOMICS ARTICLE NO.

26, 104–116 (1998)

JE971507

Causes of the Soft Budget Constraint: Evidence on Three Explanations1 David D. Li and Minsong Liang University of Michigan, Ann Arbor, Michigan 48109-1220 Received August 1, 1997; revised December 1, 1997

Li, David D., and Liang, Minsong—Causes of the Soft Budget Constraint: Evidence of Three Explanations This paper conducts empirical evaluations of three recent theories of the soft budget constraint (SBC) using data from China’s enterprise reform period. For enterprises with financial losses, we find that employment of nonproduction workers, investment with below-average rate of return, and distribution of excessive amounts of bonuses have all contributed significantly to the losses. In addition, there is no evidence that these factors decrease in response to financial losses. These findings support all three theories identifying political influence, creditor’s lack of information and commitment, and insider control as causes of the SBC. J. Comp. Econom., March 1998, 26(1), pp. 104–116. University of Michigan, Ann Arbor, Michigan 48109-1220. q 1998 Academic Press

Key Words: the soft budget constraint; political control; agency costs; insider control; reform; transition. Journal of Economic Literature Classification Numbers: P31, P51.

1. INTRODUCTION Many formal economic theories exploring the causes of the soft budget constraint (SBC) have appeared in the literature recently. The SBC refers to the expectation of refinancing or bail-out of managers with unprofitable investment projects. The concept of the SBC was first introduced by Kornai (1980); it has been widely documented and recognized as a major source

1

We thank Eric Maskin, Andrei Shleifer, and Bernard Yeung for helpful discussions. We are grateful to Chong-en Bai, John Bonin, and Yijiang Wang for comments and suggestions on earlier drafts of the paper. Financial support from the William Davidson Institute at the University of Michigan Business School is gratefully acknowledged. 0147-5967/98 $25.00 Copyright q 1998 by Academic Press All rights of reproduction in any form reserved.

AID

JCE 1507

/ 6w13$$$201

104

04-07-98 18:11:02

cea

EVIDENCE ON THE SOFT BUDGET CONSTRAINT

105

of inefficiency in various economic systems, especially the former socialist economies during transition. The first theory claims that the SBC is caused by politicians’ influence on enterprises’ behavior.2 Shleifer and Vishny (1994) argue that politicians have noneconomic objectives, e.g., politicians are interested in maximizing their political base by increasing enterprises’ employment or output. Therefore, when politicians have direct control over enterprises or when politicians can bargain effectively with the enterprise, the profitability of the enterprise will be reduced to accommodate politicians’ noneconomic objectives. In return, politicians will provide subsidies to compensate the enterprise for any losses.3 The second theory bases the SBC on creditors’ lack of information and lack of commitment not to refinance bad projects. As a result, poor investment decisions are made. Dewatripont and Maskin (1995) argue that large creditors, having less information than managers, are prone to approve managers’ projects that are ex ante unprofitable, i.e., taking the initial investment into account. For large creditors, it is worthwhile and feasible to refinance a project after the initial investment is sunk. Small creditors would not have the liquidity to continue these projects and would be more likely to terminate them. Exploring the informational explanation, Bai and Wang (1995) show that the creditors who rely on managers to screen projects may force the managers to refinance some ex post unprofitable projects. By continuing such bad projects, the creditors pass on losses to managers and, therefore, the managers are more careful in screening projects. The third theory emphasizes insiders’, or managers’, control rights as a cause of the SBC (Li, 1997). Suppose that insiders are effectively the decision makers of the enterprise even though they are not entitled to 100% of its liquidation value. For example, managers borrow from creditors but still hold key control rights. Suppose further that insiders enjoy significant benefits associated with control. Facing financial losses, insiders may well object to the liquidation of the operation to avoid loss of their control benefit. Consequently, many bankrupt enterprises survive because of insider control. However, this outcome may well be ex post socially efficient if the insiders’ control benefit are taken into account. Nevertheless, it is ex ante inefficient because insiders will be careless and abuse their control rights to their own benefit. Furthermore, rational creditors may anticipate such a result and refuse to lend to insider-controlled enterprises. 2 Kornai’s (1980 and 1992) own explanation of the SBC is based on bureaucratic paternalism; i.e., in a bureaucratic hierarchy, the superior has to patronize and protect the subordinate. The enterprise manager, as the subordinate, takes advantage of this paternalism and repeatedly makes financial losses. We have not yet seen further theoretical development following this insightful point, although the Shleifer and Vishny (1994) theory is close to this view in spirit. 3 The exact amount of subsidies depends on the relative bargaining power and thus may or may not cover fully the losses.

AID

JCE 1507

/ 6w13$$$201

04-07-98 18:11:02

cea

106

LI AND LIANG

Little empirical research has been done to evaluate these theories of the SBC. The purpose of this paper is to examine empirical evidence concerning these proposed theories. We examine two groups of evidence on these theories, using a data set from Chinese state enterprises during the reform era of 1980 to 1994. The first group of evidence is about the extent to which the factors identified by these theories contribute to enterprises’ financial losses. We use the enterprises’ employment of nonproduction workers, investment with belowaverage rate of return, and distribution of bonuses in excess of these regulated by the government as proxies for the factors identified by these theories, respectively. It is found that if all of the nonproduction workers were eliminated, on average, about 38% of the financial losses could be avoided. It is also found that about 55% of the unprofitable enterprises had a rate of return on investment below the industry average. These inefficient investment projects accounted for 126% of the total losses. In other words, if the return rate had been increased to the industry average, these enterprises would have been profitable and the profit is 26% of the absolute value of the losses. Moreover, the effect of insider control was evidenced in that about 81% of enterprises with financial losses distributed excessive bonuses. Among such enterprises, the amount of excessive bonus expenditure accounted for 39% of the enterprises’ financial losses. The second group of evidence is on whether the enterprise responds to financial losses by reducing any of the three factors that cause the losses initially. All three theories imply that this is not the case. Our regressions indicate that there was no significant reduction in nonproduction workers, in investment, and in excessive bonuses after financial losses. In particular, the regression results show that unlike in the case of profitable enterprises, in an enterprise with financial losses, investment is not sensitive to the rate of return on previous investments. Although more detailed and elaborate empirical investigations are necessary, the preliminary findings in the paper indicate that all the factors identified by the three theories are important causes of the SBC. This implies that the SBC is a complex economic problem with multiple causes and that there are no simple and easy solutions to the problem. The data set we use comes from two large surveys of Chinese state-owned enterprises. The first covers 1980 to 1989 and the second 1990 to 1994. Each survey consists of about 1300 enterprises and about 681 enterprises had valid replies in both surveys. Enterprises were sampled almost evenly from four provinces in China: Jiangsu in the coastal region, Jilin in the northeast, Shanxi from the north, and Sichuan in the southwest. The sample’s coverage of industries, 39 in total, is representative of China’s overall industrial structure, although large and medium enterprises are overrepresented.4 4

The first survey was jointly conducted by scholars from the Chinese Academy of Social Sciences (CASS), University of California at San Diego, University of Michigan, and University

AID

JCE 1507

/ 6w13$$$202

04-07-98 18:11:02

cea

EVIDENCE ON THE SOFT BUDGET CONSTRAINT

107

In the next two sections, we discuss the empirical methodology. Section 4 presents the empirical findings and the last section provides a short conclusion. 2. IDENTIFYING CASES OF FINANCIAL DIFFICULTY (LOSSES) AND BAIL-OUT A key component of the SBC phenomenon is financial losses or financial difficulty, which we now define specifically as a situation where the enterprise’s cash flow is not sufficient to meet its outstanding financial obligations, such as wage payments and interest payments. Financial difficulty or losses is a necessary condition for the SBC according to Kornai’s original definition, although it is difficult to identify empirically financial losses. The enterprise’s reported profit is a natural starting point. In our data, this information is highly unreliable because Chinese accounting standards are not uniform and enterprise insiders have an incentive to manipulate this information. Thus, we need to resort to more rudimentary accounting categories. In order to identify cases of financial loss, we use an index of basic cash flow, defined as sales revenue minus total operational expenses. Sales revenue does not include newly borrowed investment funds, which may be bail-out funds. Expenses include the cost of raw materials, depreciation, wages, fringe benefits, and bonuses. Our index of basic cash flow does not deduct interest payments, tax payments, or the enterprise’s promised tax remittance to the government. Therefore, when the basic cash flow is negative, the enterprise is in serious financial difficulty. We exclude all financial commitments to outsiders in our definition of basic cash flow due to data limitations based on the inability to separate ex ante financial commitment from ex post payments. In addition, large disparities in outstanding cash flow commitment to outsiders exist across enterprises. In the early years of economic reform, most enterprises remitted almost all of their profits to the government. Later on, incentive contracts were signed between the enterprises and the government so that enterprises retained some of their profits. In recent years, corporate taxes at varying rates were applied to enterprises. Furthermore, enterprises began using bank loans, rather than noninterest bearing budgetary appropriation, to finance their investment. All of these changes were implemented differently across enterprises so that it is difficult to calculate accurately the outstanding cash flow commitment to outsiders for each enterprise in a given year. Therefore, we settle for a narrow and restrictive definition of basic cash flow. Facing a negative basic cash flow, an enterprise must secure funds from

of Oxford. See Dong and Tang (1995) for a summary of the project. The second survey is a follow-up of the first one with some changes in the design of the questionnaire.

AID

JCE 1507

/ 6w13$$$202

04-07-98 18:11:02

cea

108

LI AND LIANG

some source. In addition to any tax and interest arrears, the enterprise requires external funds or it must use its own reserve funds to meet this deficiency. Bail-out, i.e., external help, is another key component of the SBC. In studying the SBC, we need to rule out those cases in which the enterprise uses internal funds to finance the deficit in basic cash flow. In China, there are two sources of internal funds, working capital and accumulated retained profit. We eliminate the enterprises for which these funds are used to self-finance the deficit in the basic cash flow. A potential objection to our operational definition of financial difficulty and bail-out is that a negative basic cash flow may not indicate poor economic performance because, among many things, the product market price that an enterprise faces may have been distorted. Our first response is that price distortions generate poor economic signals to enterprises and lead to inefficiency. However, the SBC is a different type of inefficiency and it exists in addition to that caused by the distorted signals. Whether these two kinds of inefficiencies cancel each other out in reality is a fascinating empirical question and is beyond the scope of this paper. The second answer to this objection is that price distortions have been steadily declining during China’s reform, especially after 1992, when the two price tracks, planned and market, were merged. However, as will be shown later, the sources of the SBC are not reduced during the recent years. 3. THE DESIGN OF EMPIRICAL TESTS A relevant theory of the SBC must satisfy two empirical criteria. First, the factor proposed by the theory as the cause of the SBC must be a significant source for the enterprise’s financial difficulty. In other words, if the proposed factor were absent ex ante, the enterprise’s financial difficulty would be significantly reduced or even eliminated. For the sake of convenience, let us call this the ex ante criterion. Second, once financial difficulty emerges and bailout happens, the proposed cause of the SBC is not significantly corrected in response to the events. Let us call this the ex post criterion. Both criteria are important. The first one is clearly a necessary condition for empirical relevancy of the theory. The second one is indispensable because otherwise the proposed theory would be inconsistent in that the proposed cause of the SBC should have been corrected ex ante by rational managers. Take the example of the political-influence theory of the SBC. It argues that the government pushes for surplus employment (or other inefficient policies) ex ante. Then losses are incurred and subsidies are provided to bail out the enterprise. This is an outcome of bargaining between the government and the managers. The subsidies, according to the theory, are meant to compensate the enterprise for keeping its surplus employment. Thus, when losses arise, no one is supposed to reduce excess employment.

AID

JCE 1507

/ 6w13$$$202

04-07-98 18:11:02

cea

EVIDENCE ON THE SOFT BUDGET CONSTRAINT

109

The information/commitment theory also implies the second criterion. Managers propose a bad investment project, according to the theory, expecting that later on, when low profitability is revealed, creditors have to make additional investment to make up for the initial bad investment. Therefore, an initial bad investment leads to further bad investments. As a result, the event of financial loss does not reduce investments. Similarly, the insider control theory must also pass the second criterion. It predicts that managerial abuses, i.e., an ex ante behavior such as proposing bad projects or stealing enterprise assets, lead to financial losses. Then, managers use their insider control power to block liquidation and to obtain bail-out funds. Managerial control is not reduced ex post as a consequence of financial losses, otherwise, expecting this, managers would exercise self-control and reduce managerial abuses initially. To examine the political-influence theory, we focus on the issue of surplus employment. In China, employment is certainly a concern for government officials. Before the economic reform, state enterprises were the major employers. During the reform, maintaining social stability by preventing massive layoffs has been a political goal of the Chinese government. In the surveys we use, the category of nonproduction workers (fei sheng chan yi xian gong ren) is a natural proxy for excessive employment. Employees in a Chinese state enterprise are divided into the following categories: productive workers, nonproduction workers, managers, engineers, sales personnel, and others. Nonproduction workers are the closest proxy for workers with zero marginal product, although not all of them are necessarily unproductive. They are mostly unskilled, hired to do nonessential jobs, and, in the majority of cases, forced upon the enterprise by governments. In the surveys, an overwhelming proportion of enterprise managers complained about a severe surplus of these workers. Our index of the political-influence theory is defined as rlloss Å

lex 1 w , ÉpÉ

(1)

where lex is the reported number of nonproduction workers in an enterprise, w is the average basic wage in the same enterprise, and p is basic cash flow.5 Intuitively, rlloss measures the proportion of the enterprise’s financial losses that could have been recovered or avoided if all nonproduction workers were eliminated. We also calculate this for profitable enterprises, where the index measures the extra proportion of profits that could be increased if the nonproduction workers were eliminated.

5

Bonuses and perquisites are not included in the basic wage. Otherwise, the index would be higher.

AID

JCE 1507

/ 6w13$$$202

04-07-98 18:11:02

cea

110

LI AND LIANG

Similarly, our index for the information/commitment theory is constructed by considering the contribution of bad investments made during the previous 3 years on financial losses. Specifically, the index is defined as rkloss Å

(rV 0 r) 1 I , ÉpÉ

(2)

where r is the enterprise’s rate of return on 3-year investment, which includes the current year’s investment; rV is the industry average of r; I is the enterprise’s sum of 3-year investments; and p is the basic cash flow. To calculate rV , we divide the sample into three industry groups, mining, heavy manufacturing, and light manufacturing, and compute the average rate of return for each industry. The index rkloss measures the proportion of losses that would be recovered if the enterprise’s past investments were not below the average level. If rkloss is negative, it measures the contribution of the enterprise’s above-average investments to its profit. As for the insider control theory, a good proxy for the extent of insider control is the amount of bonus and fringe benefits that an enterprise issues to employees. In fact, a simple version of the insider control theory is that insiders overconsume enterprises’ cash-flow so that financial difficulty arises but insiders can resist liquidation of the enterprise. During the reform in China, each enterprise is allowed to retain a certain percentage of profit. However, not all retained profit is supposed to be allocated to bonus and fringe benefits for employees. Part of the retained profit should be reinvested. By government regulation, bonuses should not exceed 25% of the base wage bill, or, about 3 months’ wage. Thus, bonuses over the 25% ceiling provide a natural index of insider control. Our index for the insider control theory is constructed as rbloss Å

b 0 0.25 1 w , ÉpÉ

(3)

where b is total bonus, w is the basic wage, and p is the net cash flow. The coefficient 0.25 is the norm stipulated by the government for bonus payments as a percentage of the basic wages. The index rbloss indicates to what extent financial losses would be avoided if enterprise insiders had not issued excessive amounts of bonuses. Notice that the value of each of the three indices, rlloss, rkloss, and rbloss, is not restricted to the interval [0, 1] and that the sum of these three may not be equal to 1. Suppose we find rlloss ú 1, this means that by laying off all of the nonproduction workers, the enterprise can become profitable. This is a plausible scenario if either there is a large number of such workers or the financial loss is relatively small. If the sum of these three indices is bigger than 1, this indicates that correcting all three factors at the same time would

AID

JCE 1507

/ 6w13$$$203

04-07-98 18:11:02

cea

EVIDENCE ON THE SOFT BUDGET CONSTRAINT

111

make the enterprise profitable. Of course, the situation where the indices fall outside the interval [0, 1] will become less likely, if we include tax obligations, promised interest payments, etc., in the definition of the basic cash flow. However, as explained above, there is no reliable data to do this. Thus we need to be careful in interpreting the results. We adopt a uniform approach to test the ex post criterion. The idea is to test whether, after financial difficulty and bail-out, the enterprise reduces the factor that causes the financial difficulty. Specifically, for each of the theories, we run yi,t/1 Å ct / cj / a 1 SBCit / b1xit∗SBCit / b2xit∗(1 0 SBCit) / eit ,

(4)

where yi,t/1 is the growth rate between year t and year t / 1 of the factor that is theorized to cause financial losses. For the political-influence theory, yi,t/1 is the growth rate of nonproduction workers. For the other two theories, it is the growth rate of investment over the average of the previous 3 years and that of excessive bonuses, respectively. The term ct is a year-specific intercept, which is indexed by year dummies, and cj is the industry specific intercept indexed by three industry dummies (mining industries, heavy manufacturing, and light manufacturing). If SBCit Å 1, enterprise i in year t is in financial difficulty and is bailed out. The term xit is one the three indices, rllossit , rkllossit , and rblossit , depending on which theory is tested. For the information/commitment theory, we also include the rate of return (ROR) on previous 3-year investment. The ex post criterion implies that the coefficients a and b1 are close to 0. In contrast, in cases of a hard budget constraint, both of these coefficients are expected to be significantly negative since managers have to reduce the factors that cause the financial difficulty. 4. ESTIMATION RESULTS Table 1 summarizes the cases of financial difficulty and bail-out. There were no cases in which the enterprise used internal funds to cover fully the financial loss. In other words, all cases of financial difficulties were bailed out. There seem to be two factors responsible for this outcome. First, since it does not include obligations to pay taxes, interests, and profit remittances, our definition of basic cash flow is conservative. When it is negative, the enterprise is in severe financial trouble and it is difficult for the enterprise to use internal funds to cover this deficit. Second, Chinese state enterprises keep small amounts of internal funds since most profits are either distributed to the government or to the employees as bonuses. The percentage of enterprises in financial difficulty in Table 1 is considerably lower than most national statistics of financial difficulty. The most important reason is sample selection, since better-performing enterprises are

AID

JCE 1507

/ 6w13$$$203

04-07-98 18:11:02

cea

112

LI AND LIANG TABLE 1 Cases of Financial Losses and Bail-Out

Year

Cases of financial losses and bail-out

% in Total cases

1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994

95 142 104 89 76 66 94 61 65 137 102 132 158 156 262

14.0 20.9 15.3 13.1 11.2 9.7 13.8 9.0 9.5 20.1 15.0 19.4 23.2 22.9 38.5

Note. Financial losses refers to cases of negative basic cash flow, which is defined as sales revenue minus total operational expenses. We found no cases in which the enterprise used internal funds to cover fully its negative cash flow. That is, the set of cases of financial difficulty is the set of cases of bail-out.

more likely to be chosen and are more likely to respond to surveys. In addition, our definition of financial difficulty is rather strict. Table 2 lists the summary statistics of rlloss, rkloss, and rbloss for enter-

TABLE 2 Contributions to Financial Losses by Various Factors

Factors (Indices)

Percentage of relevant cases

Conditional mean

SD

Nonproduction employment (rlloss) Bad investments (rkloss) Excessive bonuses (rbloss)

100 54.9 81.4

38.0 125.6 38.8

62.3 121.2 65.6

Note. The index rlloss is the percentage of the financial losses due to hiring nonproduction workers; rkloss is that due to investments with below-average rates of return; and rbloss is that due to distributing bonuses over 25% of basic wage. Relevant cases means cases with rlloss ú 0, rkloss ú 0, and rbloss ú 0, respectively. Conditional mean is the sample mean for the relevant cases.

AID

JCE 1507

/ 6w13$$$203

04-07-98 18:11:02

cea

EVIDENCE ON THE SOFT BUDGET CONSTRAINT

113

prises with financial losses. The results on rlloss show that all such enterprises had nonproduction workers and, on average, 38% of the loss could have been recovered by eliminating all nonproduction workers. This result shows that nonproduction employment is a prominent factor contributing to financial losses. The results on rkloss are less straightforward. Table 2 indicates that about 55% of the enterprises in financial difficulty had below-average rates of return. Of those enterprises with below-average return rate, on average, 126% of the financial losses could have been recovered if the previous 3-year investments had performed at the industry average level. In other words, these enterprises would be profitable with a profit being equal to 26% of the absolute value of the losses. For the 45% of the cases in which the rate of return was higher than industry average, we found that their average 3-year investment was less than one-half of the average investment in the other cases. Most likely, these smaller investments were for short-term and minor investment projects for which the rate of return was likely to be volatile. Given that models of Dewatripont and Maskin (1995) and Bai and Wang (1995) apply to large and long-term investment projects, these cases may not be relevant. Thus, the results on rkloss do not seem to contradict directly the information/commitment theories. Table 2 also shows that 81% of enterprises in financial difficulties paid bonuses above the government-regulated amounts. For those enterprises, the amount of excessive bonuses was about 39% of the total negative cash flow. These results provide a intuitive sense of the relevancy of the insider control theory. Turning to the second criterion, Table 3 lists a group of regressions examining whether the enterprise decreases its employment of nonproduction workers after the event of financial difficulty. The table shows that the event of financial difficulty per se did not have any significant effect on nonproduction employment. However, the event of financial losses coupled with a high rlloss would prompt a slight reduction in nonproduction employment. For example, regression 3 in the table shows that for enterprises with SBC Å 1, if rlloss Å 100 (one standard deviation above its mean), there would be a 2% reduction in such employment next year. Overall, these results seem to provide some evidence for the political-influence theory of the SBC. Table 4 provides tests of the information/commitment theories of the SBC. It shows that the event of financial difficulty made the enterprise’s investments insensitive to the rate of return on previous years’ investments. In profitable enterprises, investments were positively dependent on the rate of return on previous investments. To the extent that rates of return on investments during the past 3 years is a reasonable predictor of future rate of return, these

AID

JCE 1507

/ 6w13$$$203

04-07-98 18:11:02

cea

114

LI AND LIANG TABLE 3 Changes in Nonproduction Employment after Financial Losses Occur OLS regression with dependent variable: Growth rate of nonproduction employment in year t / 1; No. of observation: 3948

SBC RLLOSS 1 SBC RLLOSS 1 (1-SBC) R2

Regression 1

Regression 2

Regression 3

00.0147 (01.17)

00.00759 (00.531) 00.0208 (01.34) 00.175 (02.07**) 0.0132

00.0254 (01.78*) 00.0169 (02.01**) 0.0130

0.0116

Note. Also included in the regressions are year dummies (1981–1994) and industry dummies (for heavy and light industries). All independent variables are in year t. SBC Å 1 means that the enterprise is in financial difficulty; SBC Å 0, otherwise. Heteroskedastic-consistent t statistics are in parentheses. *, **, and *** indicate statistical significance at the 10, 5, and 1% level, respectively.

results indicate that the amount of post-financial-difficulty investment was not dependent on future rate of return. The table also shows weak evidence that investment will be reduced very slightly if rkloss is very high. According

TABLE 4 Investments after Financial Losses Occur OLS regression with dependent variable: Growth rate of investment in year t / 1 over the average of previous 3 years; No. of observation: 3948

SBC ROR 1 SBC ROR 1 (1-SBC) RKLOSS 1 SBC RKLOSS 1 (1-SBC) R2

Regression 1

Regression 2

00.0637 (01.50)

00.0501 (01.14) 0.0328 (0.748) 0.148 (8.93***)

0.0130

Regression 3

0.0348 (0.623) 0.0986 (4.76***) 00.00634 (01.96*) 00.0729 (03.99***) 0.0372

0.0306

Note. Also included in the regressions are year dummies (1984–1994) and industry dummies (for heavy and light industries). ROR is the rate of return on 3-year investments. If RKLOSS õ 0, the enterprise’s rate of return on investment is higher than the industry average. SBC Å 1 means that the enterprise is in financial difficulty; SBC Å 0, otherwise. All independent variables are in year t. Heteroskedastic-consistent t statistics are in parentheses, *, **, and *** indicate statistical significance at the 10, 5, and 1% level, respectively.

AID

JCE 1507

/ 6w13$$$203

04-07-98 18:11:02

cea

EVIDENCE ON THE SOFT BUDGET CONSTRAINT

115

TABLE 5 Changes in Excessive Bonuses after Financial Losses Occur OLS regression with dependent variable: Growth rate of (bonus 0 0.25 1 basic wage) in year t / 1; No. of observation: 3948

SBC RBLOSS 1 SBC RBLOSS 1 (1-SBC) R2

Regression 1

Regression 2

Regression 3

03.10 (00.400)

03.93 (00.446) 01.05 (00.100) 09.42 (00.776) 0.0012

03.13 (0.330) 08.47 (00.709) 0.0011

0.0011

Note. Also included in the regressions are year dummies (1981–1994) and industry dummies (for mining and light industries). SBC Å 1 means that the enterprise is in financial difficulty; SBC Å 0, otherwise. If RBLOSS õ 0, bonus õ 0.25 1 basic wage. All independent variables are in year t. Heteroskedastic-consistent t statistics are in parentheses.

to the estimates, if rkloss is 100%, then investment will be reduced by only 0.6%. Again, it seems that the findings in Table 4 are broadly consistent with the information/commitment theories. Table 5 concerns the insider control theory of the SBC and the evidence is clear. There is no sign that the enterprise reduced the amount of extrabonus in response to both the event of financial losses (SBC Å 1) and the percentage of losses due to previous year’s extra-bonus expenditure (rbloss). These regression results are supportive of the insider control theory of the SBC. 5. CONCLUDING REMARKS Using a large data set of Chinese state enterprises during the reform era, we provide empirical evaluations of three theories of the SBC. These theories characterize the SBC as being caused by political influence, creditors’ lack of information and commitment, and insiders’ control, respectively. Our empirical findings show that each of those factors contributed significantly to enterprises’ financial losses. Moreover, consistent with the theories, enterprises did not take measures to reduce significantly these factors that cause financial losses after being bailed out. Overall, the findings suggest that the SBC is a complex problem with multiple causes and can be resolved only with comprehensive reform measures. REFERENCES Bai, Chongen, and Wang, Yijiang, A Theory of the Soft Budget Constraint, Department of Economics Working Paper. Boston: Boston College, 1995.

AID

JCE 1507

/ 6w13$$$203

04-07-98 18:11:02

cea

116

LI AND LIANG

Dewatripont, Mathias, and Maskin, Eric, ‘‘Credit and Efficiency in Centralized and Decentralized Economies.’’ Rev. Econom. Stud. 62, 4:541–555, Oct. 1995. Dong, Fureng, and Tang, Zhongkun, ZhongGuo GuoYou QiYe ZhiDu BianGe YanJiu. Beijing: RenMin ChuBan She, 1995. [Studies of Institutional Changes of China’s State Enterprises. Beijing, China: The People’s Publisher, 1995] Kornai, Ja´nos, The Socialist System: The Political Economy of Communism. Princeton, NJ: Princeton Univ. Press, 1992. Kornai, Ja´nos, Economics of Shortage. Amsterdam, Holland: North-Holland, 1980. Li, David D., Insider Control and the Soft Budget Constraint: A Simple Theory, Department of Economics Working Paper. Ann Arbor, MI: University of Michigan, 1997. Shleifer, Andrei, and Vishny, Robert, ‘‘Politicians and Firms.’’ Quart. J. Econom. 109, 4:995– 1025, Nov. 1994.

AID

JCE 1507

/ 6w13$$$204

04-07-98 18:11:02

cea