- Email: [email protected]
Policy burden, privatization and soft budget constraint
Journal of Comparative Economics 36 (2008) 90–102 www.elsevier.com/locate/jce
Policy burden, privatization and soft budget constraint Justin Yifu Lin a , Zhiyun Li b,∗ a China Center for Economic Research, Peking University, Beijing 100871, China b MF003, Manufacture des Tabacs, MPSE, Toulouse Université des Sciences Sociales, 31000, Toulouse, France
Received 17 February 2007; revised 11 November 2007 Available online 8 December 2007
Lin, Justin Yifu, and Li, Zhiyun—Policy burden, privatization and soft budget constraint We propose a new cause for the pervasive syndromes of soft budget constraint (SBC) in socialist and transition economies, that is, the policy burdens on enterprises result in the SBC problems. The policy burdens induce low effort input of firm manager and thus the low efficiency of production. And with the policy burdens, increasing market competition will make the SBC syndromes arise more likely. Privatization will not necessarily harden the budget constraint of the enterprise. On the contrary, when a SOE still bears the policy burdens, privatization will only aggravate the SBC problems. Because in this case, a private enterprise will demand more ex post subsidies from the government, than a SOE under the same condition. Our results help to explain many stylized facts in transition and socialist economies. Journal of Comparative Economics 36 (1) (2008) 90–102. China Center for Economic Research, Peking University, Beijing 100871, China; MF003, Manufacture des Tabacs, MPSE, Toulouse Université des Sciences Sociales, 31000, Toulouse, France. © 2007 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved. JEL classification: L32; L33; P31 Keywords: Policy burden; Market competition; Soft budget constraint; Privatization
1. Introduction
A well-known phenomenon in socialist and transition economies is that, when state-owned enterprises (SOEs) incur losses, the State normally cannot restrain from bailing them out by providing additional subsidies or credits, which is known as the soft budget constraint (SBC) problem. Soft budget constraint, a term coined by Kornai (1980), is connected to various problems in socialist and transition economies, such as shortage and low efficiency of SOEs. And due to the negative consequences of SBC, hardening the budget constraints of enterprises has been a principal objective of the economic reforms in transition economies. However, the reforms of Eastern European economies (EEEs) in the 1990s have proved to be unsuccessful in eradicating the lasting SBC syndromes, even after the SOEs had been massively privatized in many of those countries (World Bank, 1996, 2002).
* Corresponding author. Fax: 86 10 6275 1474.
E-mail addresses: [email protected] (J.Y. Lin), [email protected] (Z. Li). 0147-5967/$ – see front matter © 2007 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jce.2007.11.001
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
91
As for the SBC syndromes in socialist and transition economies, there has been a huge amount of literature, most focusing on the consequences of SBC, such as moral hazard or adverse selection problems of firm managers.1 But about the institutional causes for the formation of SBC, there still exist debates, and little detailed work has been done on these topics. In this paper, we propose a new hypothesis on the causes of SBC, that is, the policy burdens on enterprises results in SBC problems. And privatization will not necessarily harden the enterprises’ budget constraints, if they continue to undertake the State’s policy burdens; rather, privatization will probably exacerbate the SBC syndromes under certain conditions. The results of our model provide important implications for understanding many stylized facts in socialist and transition economies, as well as in fledged market economies. Before expatiating on our theory, we first provide a short review of the literature. As Kornai et al. (2003) summarize, understanding the SBC syndromes entails bearing in mind a complex chain of causality, which is depicted in a schematic form in Fig. 1. Block (1) represents the political and social factors that generate the motives behind the formation of SBC. Block (2) represents the motives that create the SBC syndromes, such as the motivation of the State or creditor to refinance loss-making enterprises. Finally, block (3) represents the consequences of SBCs.
Fig. 1. The SBC syndrome: the chain of causality.
Many studies on SBCs focus on the relationship between block (2) and block (3), as mentioned above, and to some degree they have reached a consensus on how SBCs can influence the working of an economy. Dewatripont and Maskin (1995) is a seminal paper in modeling this relationship, from which a large amount of related literature has been developed. In their paper, they formulate the SBC in the context of dynamic commitment inconsistency and effectively capture the main ideas of Kornai (1980) on the SBC syndromes. In their two-period dynamic game, they state that the state could have incentives to refinance an inefficient and uncompleted project, because the marginal benefit of refinancing exceeds the marginal cost of abandoning it, which means refinancing is an ex post efficient decision for the support organization. But little detailed work has been done on the relationship between block (1) and block (2), such as the root causes for the SBC syndromes. Also, one important and unanswered question is why socialist economies are more vulnerable to SBC than fledged market economies. Or, in other words, what are the institutional factors that contribute to the pervasive SBC syndromes in those economies? Kornai (1980) attributes the causes of SBC primarily to political constraints—that is, to the paternalism of socialist governments. Dewatripont and Maskin (1995), however, have shown that paternalism is neither a necessary nor a sufficient condition for SBC. They suggest that the highly centralized system of socialist economies is the primary cause of SBC, because in a decentralized system, such as in market economies, the transaction cost of refinancing will be so high that refinancing is ex post inefficient, and will thus harden budget constraints. Qian and Roland (1998) propose a similar idea: they suggest that China’s more decentralized fiscal system increases competition among different regions, and thus the opportunity costs of refinancing bad projects are very high, which could partly explain China’s success in economic transitions.2 Another influential theory on SBC is that the public ownership of socialist economies is the cause for their pervasive SBC syndromes. In the model of Li (1992), public ownership means that refinancing decisions are made jointly by 1 And a comprehensive and excellent survey of the literature can be found in Kornai et al. (2003). There is also a symposium on the SBC in Anon (1998). 2 The validity of their statement depends on the tightness of the financial system. In effect, under existing financial arrangements, banks are owned by the central government. For economic development in their region, local governments can assist local enterprises to borrow excessively from state banks to finance their investment projects. If the state banks accumulate a large amount of non-performing loans due to many of those projects turning bad, the central government is obliged to rescue the banks. Therefore, under a decentralized fiscal system and a centralized banking sector, local governments can treat the funds in the state banks as a common resource, resulting in a situation resembling the tragedy of common.
92
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
the government and the enterprise, which could contribute to the SBC problem. From another angle, Schmidt (1996) states that privatization makes inside information inaccessible for the government, which makes ex post punishment credible and thus hardens the firm’s budget constraint. The massive privatization of SOEs in the EEEs did not, however, eradicate the SBC syndromes at all; on the contrary, the state provided even more subsidies to those privatized enterprises, which suggests that public ownership is not necessarily the cause of the SBC syndromes. Segal (1998) proposes that the monopolistic status of an enterprise could cause SBC: because the monopoly cannot capture the entire social surplus from its production, the government might have incentives to subsidize an unprofitable enterprise in order to support its production and thus guarantee social surplus. The theory is not, however, convincing, because it is only a partial equilibrium model and, in general, monopoly will incur efficiency losses as a whole in the economy. Boycko et al. (1996) suggest that politicians might pay subsidies to enterprises to induce them to retain excess labor, which could benefit the politicians in elections. There are, however, no dynamic factors in their model, and therefore no problems of commitment, which may constrain the extension of their analysis. Röller and Zhang (2005) study a similar problem in a dynamic context, and find that requiring the firm to provide social goods could result in SBC problems, and increasing competition in the private goods market could make SBC problems more likely. In this paper, we provide a new explanation for the pervasive SBCs in socialist and transition economies. We argue that the policy burdens imposed by the State on state-owned and other enterprises are the root cause of the SBC syndromes, and only by eliminating the policy burdens can the State effectively harden the budget constraints of the enterprises. When the SOEs continue to undertake the policy burdens, privatization will not eradicate SBC problems; rather, this will exacerbate them because it increases the cost of state subsidization. Our theory shares some features in common with Boycko et al. (1996) and Röller and Zhang (2005), but there are at least two important differences. First, they emphasize just the social policy burden on the enterprises. In this paper, we introduce a new, possibly more important, strategic policy burden on the enterprises, due to the government’s comparative advantage-defying (CAD) strategy, which is a key feature of the enterprises in socialist and many developing countries since World War II. Second and more importantly, we study the relationship between privatization and SBCs. We find that when a privatized firm continues to undertake the policy burdens, privatization will only aggravate the SBC syndromes. This can effectively explain the persistent SBC syndromes in the EEEs after massive privatization of the SOEs. This result also provides important implications for the choice of transitional paths, from a centrally planned economy to a decentralized market system. Finally, we study the impacts of market competition on SBC problems, and find a result similar to Röller and Zhang’s (2005)—that is, increasing competition makes it more likely for SBC problems to arise when the policy burdens exist. But a further implication of our model is that, in a less competitive market where the SOEs can make positive profits and there are no ex post subsidies, the State also implicitly support the SOEs by restricting market competition. One point worthy of attention is that our theory also helps to explain SBC problems in fledged market economies, since the existence of policy burdens is not a feature unique to socialist and transition economies. It is also common for enterprises in market economies to undertake governments’ policy burdens, such as maintaining redundant workers or implementing special government policies. In those cases, when the enterprises incur losses, the government is obliged to support them subsequently with additional subsidies, which is an SBC problem. The rest of this paper is organized as follows: Section 2 provides a background of the formation of policy burdens in socialist and developing countries, and explains our theory of the SBC syndromes. Section 3 develops a model that studies the relationship between policy burdens and the SBC of SOEs. Section 4 investigates the effects of privatization on SBC problems. Section 5 discusses some policy implications of our model, and Section 6 is a short conclusion. 2. The main ideas A distinct feature of socialist and many developing economies is their extreme bias towards the development of heavy industries. However, the endowments of those economies are commonly characterized by their lack of physical capital, and relative abundance in labor supply. And the heavy industry-biased strategy is obviously not consistent with their comparative advantages. To implement CAD strategies, the State needs to impose its policy burdens on the enterprises—that is, the strategic policy burden and the social policy burden. The former stems from the fact that the enterprises are forced to enter CAD industries or adopt CAD technologies, both of which will render them non-viable in a free, competitive market. The social policy burden refers to keeping redundant workers, providing retirement
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
93
pensions and other social services, such as schooling and medical care. Since the capital-intensive industries cannot provide enough job opportunities, in order to solve employment problems and maintain social stability, the State needs the SOEs to retain excess workers, which results in a social policy burden. These two kinds of policy burdens are therefore interrelated. With these policy burdens, the enterprises are not profitable in a free, competitive market. The term viability is defined as follows: if, without any external subsidies or protections, a normally managed firm is expected to earn a socially acceptable profit in a free, open and competitive market, the firm is viable; otherwise, it is nonviable (Lin, 2003). In a transition economy, the allocation of social resources becomes more and more dependent on market mechanism, and it will thus be more difficult for SOEs with the policy burdens to be profitable. To keep these SOEs implementing the policy burdens, the State needs to guarantee their survival in market competition, which can be achieved in two ways: one is to subsidize them implicitly by restricting competition and protecting the market powers of the SOEs, as was done before the transition; and the second is to support them explicitly by ex post subsidies or credits, in the case of loss making. We argue that privatization will not necessarily eradicate the SBC syndromes. On the contrary, it will only aggravate the syndromes if the privatized enterprise continues to bear the policy burdens. In our paper, privatization implies a combination of two changes: the turnover of control rights and an increasing share of cash flows to private shareholders. Either of these changes will increase the bargaining power of the private firm in demanding subsidies from the government. To impose a policy burden, it is more expensive for the government to allocate it to a private enterprise than to a SOE, since the former will demand more subsidies from the government to implement that policy burden. 3. Policy burdens, SOEs and the SBC syndromes 3.1. Competition and profit: a preliminary result As a preparation for later analysis, we first consider a Nash–Cournot equilibrium in a free-entry market, where the SOEs do not bear policy burdens. We assume there are N SOEs in the market, and they produce the same product. The demand for the firms’ product is given by a strictly decreasing inverse demand function, P (Q), where Q is the total supply of this product. We assume the cost function is the same for all SOEs, which is C(q) = K + e(q), where K is capital input, q is the output of a firm and e(q) refers to the effort of the firm manager. And for any q 0, we assume e (q) > 0, e (q) > 0 and e(0) = 0. We assume that N¯ is the long-term Nash–Cournot equilibrium number of firms in the free-entry market, and the profit of firm i is defined as π(qi ) = P (Q) · qi − e(qi ) − K, where Qi = Q−i + qi (Q−i is the total output of the other N − 1 firms). Given the number of firms, N N¯ , we achieve the following conditions of Nash–Cournot equilibrium: (1) profit maximization: qi ∈ arg max p(Q−i + qi ) · qi − K − e(qi ), (2) output market equilibrium: p(Q) = p(Q−i + qi ), (3) participant constraint: w − e(qi ) = 0, i = 1, 2, . . . , N where w is the payment to the SOE managers, whose reservation pay-offs are set at 0. According to Novshek (1980), there exists a solution to the problem above, which is denoted as {q ∗ (N ), w ∗ (N )} in our model. We get the equilibrium profit π ∗ (N ) = π(q ∗ (N )) and the payment to the SOE manager w ∗ (N ) = e(q ∗ (N )).3 If we further assume p + p · q < 0, which is always satisfied if marginal revenue is falling at a rate steeper than the slope of the demand curve, we get the following lemma. Lemma 1. If N1 < N2 N¯ , then, at the Nash–Cournot equilibrium in a free-entry market, π ∗ (N1 ) > π ∗ (N2 ) and q ∗ (N1 ) > q ∗ (N2 ). 3 For simplicity, we denote this as q ∗ , π ∗ and w ∗ in the following sections.
94
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
Lemma 1 tells us that when the number of firms in the market increases, the equilibrium profit and equilibrium output of each firm will decrease. Or, in other words, increasing competition reduces the profit of each firm. For the following sections, we always assume that the exogenously given number of SOEs is smaller than N¯ —that is: Assumption 1. N N¯ is always satisfied. We will later use the result of Lemma 1 to analyze the effects of market competition on the SBC syndromes. 3.2. Policy burdens and the SBCs of SOEs In the real world, we often see SOEs undertaking various policy burdens, such as providing social goods or retaining redundant workers. In this subsection, we introduce the policy burdens into a dynamic game between the government and the SOEs, and investigate the impacts of the policy burdens on the SBC syndromes. To focus on the relationship between the policy burdens and the SBC outcomes, we need to make some simplification to our model setting. We will study the behavior of just one SOE, which undertakes the policy burdens imposed by the government. We assume that the other N − 1 firms do not bear the policy burdens, and all behave as in the former game of Nash–Cournot competition, each producing at the efficient level of q ∗ . Naturally, the introduction of the policy burdens makes it more difficult to evaluate the performance of an SOE manager. For example, when a SOE retains redundant workers or adopts CAD technologies according to government requirement, in the case of loss making, the government would not be able to distinguish between the losses resulted from the policy burdens and those from the less effort inputs of the SOE managers. On the contrary, if the firm does not bear the policy burdens and competes with other firms under the same conditions, the relative profitability of the firm can be a sufficient indicator to reflect the efficiency of the SOE. Taking into account this information difference after the introduction of the policy burdens, we make the following assumption. Assumption 2. When an enterprise bears the policy burdens, its output level, q, and thus the effort level, e(q), are not verifiable. Unlike in the former case, the payment to the SOE manager is now no longer contingent on output q, since that information is not verifiable. We here use L to denote the policy burden on the SOE, which can be the cost of retaining redundant workers and/or that of adopting a CAD technology, and let π(q) denote the profit of the SOE when it does not bear the policy burden. Under the assumptions above, we have π(q) = P (q + (N − 1)q ∗ ) · q − e(q) − K. The profit of the SOE undertaking the policy burden is Π(q) = π(q) − L. Let B(L) represent the benefit of the policy burden to the government, and we assume: Assumption 3. B(L) > L. Assumption 3 implies that the policy burden always generates additional benefit to the government, compared with the cost of the SOE’s forgone profit.4 For example, redundant employment could engender public support for the government, or the establishment of CAD industries could contribute to the government’s political credit. We assume that the government pays the SOE manager a fixed payment of w ∗ ; the pay-off function of the SOE manager is thus US = w ∗ − e(q) and the government’s payoff (without providing subsidies) is UG = B(L) + Π = B(L) + π(q) − L. Here we consider the following dynamic game between the government and the SOE manager. 4 This assumption is similar to the assumption of Boycko et al. (1996) that the politician cares less about the forgone SOE profit to the government, resulting from maintaining redundant employees.
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
95
Game 1. Stage 1. The government decides whether to have the SOE undertake the policy burden. Stage 2. The SOE manager observes the government’s decision and chooses his effort input, e(q). Stage 3. Production is completed. The government observes the final profit of the firm and decides whether to subsidize the SOE. Before solving the game, we still need to specify some features of the behavior of the government and the SOE manager, to complete the description of Game 1. First, we assume that the government subsidizes the SOE only when it incurs losses—that is, Π = π(q) − L < 0—and the amount of subsidy is S = |π(q) − L|, equal to the absolute value of loss. The government’s subsidy cost is θ · S, with the assumption that Assumption 4. 0 < θ < 1. This assumption suggests that the government pays just a part of the total cost of the subsidy.5 Second, we assume that at Stage 3, if the SOE incurs losses (Π < 0) and the government does not provide a subsidy, the SOE will go bankrupt, and the payment to the SOE manager and the pay-off of the government will both be 0. Finally, we assume that, with the same pay-off level, the government always prefers the SOE to take the policy burden. Solving Game 1, we get the following results. Proposition 1. The sub-game perfect equilibrium outcomes of Game 1 are given by (1) If π ∗ < B(L), the SOE manager chooses an output of qL = {q | π(q) = L + θ1 · (π ∗ − B(L))}, and the firm incurs losses. The government imposes the policy burden, L, on the SOE and provides a subsidy of θ1 [B(L) − π ∗ ]. (2) If π ∗ B(L), the SOE manager chooses an output of qH = {q | π(q) = L + (π ∗ − B(L))}, and the firm does not incur losses. The government imposes the policy burden, L, on the SOE and provides no subsidy. Proof. All proofs are provided in Appendix A.
2
According to Proposition 1, the relative magnitudes of π ∗ and B(L) decide the equilibrium outcomes of the game. We refer to part (1) of Proposition 1 as the SBC outcome, because the government explicitly provides an ex post subsidy to the loss-making SOE. Part (2) can also be considered a case of implicit subsidy, in the sense that the government implicitly subsidizes the SOE by protecting its market powers.6 Proposition 1 reveals the impacts of the policy burden on the production decisions of the SOE, and the formation of the SBC syndromes. First, when the SOE undertakes the policy burden, the output levels are always different from the optimal level, q ∗ , which implies an efficiency loss in SOE production. Low SOE efficiency is commonly observed all over the world, and most researches attribute it to the public ownership of SOEs. This is not necessarily true, according to our analysis; or at least, it is not the only cause for the poor performance of SOEs. The result of Proposition 1 reveals that the low efficiency of SOEs may result from implementing the policy burden imposed by the government.7 Second, an SOE with a policy burden is non-viable in a free, competitive market, due to the additional cost of L. Proposition 1 suggests that the government can help the survival of SOEs in two ways: one is to subsidize them explicitly when they incur losses (part (1)); the other is to support them implicitly by restricting market competition and therefore protecting their market powers (part (2)). 5 The government can transfer a part of its subsidy cost by an inflation tax on the general public. In the case of China, local governments can also
compel state banks to grant loans to their local SOEs, which results in huge amounts of non-performing loans that need to be absorbed by the state banks. Eventually, the central government is required to inject funds to write off the state banks’ non-performing loans. 6 A high value of π ∗ implies that the SOE has market power. By restricting market competition, the government implicitly supports the firm at the cost of efficiency loss of the overall economy. For example, China’s large-sized SOEs make huge profits, because they are monopolies in their industries, and their monopoly statuses are protected by the government. 7 This result is similar to the explanation in Boycko et al. (1996)—that public enterprises are inefficient because they address the objectives of the politicians—but our result is reached in a dynamic context.
96
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
Fig. 2. Policy burden and low effort input.
There are several corollaries that can be derived from Proposition 1, which are summarized below. Corollary 1. When the SOE bears the policy burden, its output levels are always lower than the optimal level—that is, qL , qH < q ∗ . According to Corollary 1, the policy burden results in lower output levels, and thus lower levels of effort input by the SOE manager, which reflects an incentive problem resulting from the policy burden. The welfare loss is depicted in Fig. 2. Interestingly, when L < π ∗ < B(L), the SOE with the policy burden can also make positive profit, for example just by choosing to produce q ∗ . But the manager has no incentive to produce that much because he expects that the government will subsidize the SOE subsequently in the case of loss making, and producing less will bring him a higher pay-off. This case clearly demonstrates the incentive problems of the SBC. In socialist and transition economies, we often see that resources are allocated through administrative systems, and the SOE managers are normally subject to various managerial restrictions imposed by the government, such as output quotas and price ceilings. Corollary 1 also implies that a restriction on managerial autonomy can potentially enhance the production efficiency of the SOEs. Corollary 2. When the SOE bears the policy burden, an increase of N makes the SOE more vulnerable to losses, and results in more subsidy, S, to the SOE in the case of loss making. Corollary 2 is an intuitive but quite important result. When the number of firms (N ) in the market increases, the profit of each firm will decrease (Lemma 1), and an SOE with a policy burden will thus be more vulnerable to losses. In addition, when π ∗ decreases and the SOE incurs losses, the government will provide more subsidies to the firm (part (1) of Proposition 1). The relationship between the number of firms in the market and government subsidy is depicted in Fig. 3. Corollary 2 helps to explain the transitions of China’s state sectors in the past three decades. Since the beginning of China’s economic reforms in 1978, more and more private firms have entered the markets with low entrance barriers, and SOEs—especially those in more competitive markets—have incurred huge losses. The rapidly increased subsidies to the loss-making SOEs have become a heavy burden on the government. At the end of 1990s, the central government decided to abandon small and medium-size SOEs, most of which were in competitive markets, and kept only the large SOEs that were monopolies in their markets and contributed nearly all the profits of the state sectors.8 8 In 2001, China’s 50 largest SOEs accounted for 73.6 percent of the total profits of all SOEs in China (Zhang and Lu, 2003).
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
97
Fig. 3. Competition and the size of subsidy.
Corollary 3. When the SOE bears the policy burden and incurs losses, a smaller θ will result in more subsidies from the government and less output of the SOE. Corollary 3 says that when the government’s unit cost of subsidization becomes lower (a smaller θ ), the output level of the SOE will be lower, and the government will provide more subsidies to a loss-making SOE. A higher θ clearly represents a tougher monetary and fiscal policy, which makes subsidies more costly for the government. An implication of Corollary 3 is that a tougher monetary and fiscal policy can increase the efficiency of the SOEs, and reduce the amount of the subsidies. This result also suggests that under loose fiscal or financial policies, decentralization can result in softer budget constraints for SOEs, especially when local governments are able to intervene in the operations of the State’s local financial institutions. Corollary 4. When the SOE bears the policy burden, a larger B(L) will result in less output from the SOE. Corollary 4 renders a similar idea as Corollary 3, but its validity is more general in the sense that it does not require the SOE to incur losses. The intuition behind this result is that, if the government receives more benefits from the policy burden, the SOE manager can also share those benefits and increase his pay-off by just reducing his effort input. 4. Privatization, policy burdens and SBCs A well-known theory states that public ownership of SOEs is the root cause for SBC problems in socialist and transition economies; however, the experience of SOE reforms in the EEEs has proved that this theory is not necessarily true. In many of those economies, the SOEs were massively privatized, but this did not harden the budget constraints of the privatized firms. On the contrary, it softened their budget constraints, and governments provided even more subsidies to the privatized SOEs than before (World Bank, 1996, 2002). We argue that when SOEs continue to undertake the policy burdens, privatization will not harden the budget constraints of the privatized SOEs; rather, they will exacerbate their SBC syndromes. In this section, within a framework similar to that in Section 3, we investigate the relationship between policy burdens, privatization and SBC problems, and propose a new explanation for the persistent SBC syndromes in transition economies. According to Grossman and Hart (1986), privatization implies a combination of two changes in the context of our model. The first is the turnover of control rights from the government to the enterprise manager, so that the manager can decide whether or not to accept the policy burden assigned by the government. The second is an increased share of cash flows to the manager9 and, correspondingly, a decreased share to the government. For simplicity and without 9 Here, the manager is assumed to be a representative of private shareholders of the enterprise.
98
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
loss of generality, we assume that the government owns no share in the privatized enterprise. Now we consider the following game between the government and the enterprise manager, when the enterprise has been privatized. Game 2. Stage 1. The government decides whether to assign the policy burden, L, to the enterprise. Stage 2. The enterprise manager observes the government’s action and decides whether or not to accept the policy burden, and then chooses his effort level, e(q). Stage 3. Production is completed. If the privatized enterprise bears the policy burden, the manager will require a subsidy of S from the government. Stage 4. The government decides whether to provide the subsidy of S to the enterprise. Game 2 is similar structurally to Game 1, yet privatization has changed the control right of the firm and the reservation pay-off for the manager. First, the manager now has the control right of the privatized firm and is thus able to reject the policy burden assigned by the government. Second, the manager’s reservation pay-off changes from 0 to π ∗ , and that of the government changes from π ∗ to 0. Finally, when the enterprise undertakes the policy burden, the manager will require a subsidy of S from the government, and the government will decide whether or not to accept it. Other settings and assumptions remain the same as in Game 1. Solving Game 2, we get the following proposition. Proposition 2. The sub-game perfect equilibrium outcomes of Game 2 are given by (1) If π ∗ < θ1 B(L), the government will assign the policy burden, L, to the enterprise, and the enterprise manager will accept it and choose to produce q = {q | π(q) = min(L, π ∗ )}. The subsidy provided by the government is S = θ1 B(L). (2) If π ∗ θ1 B(L), the government will assign the policy burden, L, to the enterprise, but the enterprise manager will reject it and choose to produce q ∗ = {q | π(q) = π ∗ }. From Proposition 2 we see that when the competitive profit is relatively low (π ∗ < B(L)/θ ), the private enterprise will accept the policy burden assigned by the government, and receive an ex post subsidy of B(L)/θ , which is known as an SBC problem. In this case, when L < π ∗ , the policy burden will also result in inefficiency of production in the private enterprise, so the policy burden may also reduce the production efficiency of a private firm. On the other hand, if the competitive profit is relatively high (π ∗ B(L)/θ ), the enterprise will reject the policy burden proposed by the government and produce at the efficient output level, q ∗ , which generates a higher pay-off to the manager. The equilibrium outcomes of Game 2 differ from those of Game 1 in several ways, such as the threshold of SBC outcomes, the outputs of the enterprises and the amount of government subsidization. These differences all result from the change of ownership of the enterprise. The manager of the privatized enterprise—with a higher reservation pay-off—can achieve his maximum utility not only by choosing the output level, but by deciding whether or not to accept the policy burden proposed by the government. Table 1 compares the SBC outcomes of Game 1 (the SOE) with Game 2 (the private enterprise), which provides important implications for the impact of privatization on the SBC syndromes. The first important information in Table 1 is that, when enterprises undertake the same policy burden, L, all other things being equal, it is easier for a private enterprise (PE) than for an SOE to experience SBC problems. For instance, Table 1 Policy burden and the SBCs of the SOE and the private enterprise (PE)
1. Condition of resulting SBC 2. Amount of subsidy 3. Output of enterprise
Proposition 1 (SOE)
Proposition 2 (PE)
π ∗ < B(L) 1 ∗ θ [B(L) − π ]
π ∗ < θ1 B(L)
qL = {q | π(q) = L − θ1 (B(L) − π ∗ )}
1 θ B(L)
q = {q | π(q) = min(L, π ∗ )}
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
99
when B(L) π ∗ < B(L)/θ , the SOE will not have the SBC problem, but the PE will, according to Proposition 2. This result can be summarized in the following corollary. Corollary 5. When an SOE and a PE undertake the policy burden, L, all other things being equal, it is easier for the PE to experience SBC problems. Second, it is interesting that in the case of SBC outcomes, the subsidy to the PE, B(L)/θ , is more than that to the SOE, (B(L) − π ∗ )/θ . This result implies that, to have the enterprises bear the same level of policy burden, the government needs to pay a higher level of subsidy to the PE. The intuition behind this result is that privatization increases the reservation pay-off to the firm manager and endows him with the right to reject the policy burden assigned by the government; as a result, the manager will require more subsidies when he is ready to undertake the policy burden. This result has important policy implications and helps to explain why, after massive privatization in the EEEs, subsidies to privatized SOEs have increased. We reiterate this result formally as follows. Corollary 6. When an SOE and a PE undertake the policy burden, L and π ∗ < B(L), all other things being equal, the government needs to provide more subsidies to the PE. From Table 1, we also see that in the SBC outcomes, the PE is more efficient than the SOE in the sense that it produces more, since min(L, π ∗ ) > L − (B(L) − π ∗ )/θ . In addition, in the case of the non-SBC outcomes of Game 1 and Game 2, the PE is also more efficient than the SOE, because it produces at the efficient output level of q ∗ , but in this case the PE does not undertake the policy burden. We can easily reach the following corollary. Corollary 7. When an SOE and a PE undertake the policy burden, L and π ∗ < B(L), all other things being equal, the PE is more efficient in production than the SOE. 5. Discussions and applications Our models in the above two sections are an extension of the work of Lin and Tan (1999), Lin et al. (1996, 1998, 2000). In this section, we will first provide a short discussion of the key assumptions of our models, and then apply the major results of our analysis to explain some stylized facts of the SBC syndromes in socialist and transition economies. Furthermore, the results of our analysis provide important implications for how to harden the budget constraints of enterprises in transition economies, as well as in market economies. There are two pivotal assumptions in our models. The first is that the policy burden on the enterprise brings additional benefits to the government, which is implied by Assumption 3, B(L) > L, in our models. In general, the government cares not only about economic, but political issues, such as nation building, national security, social stability and public support for the government. Therefore, the policy burden on enterprises could bring additional political benefits to the government. The other key assumption is that the government can sustain only part of the cost of subsidizing enterprises, which is represented by Assumption 4, 0 < θ < 1, in our model. For example, in transition economies, it is normal for local governments to intervene in the operation of state banks. Local governments naturally prefer to use bank credits rather than their own fiscal revenue to subsidize local enterprises; also the central government can transfer the cost to the public through an inflation tax. If either of the above two assumptions is true, the government will have an incentive to impose the policy burden on enterprises, whether they are SOEs or private enterprises. Next we apply the results of our model to explain some stylized facts in socialist and transition economies. First, we know that in socialist economies, the government restricts the control rights of SOE managers through such measures as price ceilings or production quotas. Our model suggests that when enterprises undertake policy burdens, an ex ante expectation of ex post subsidies could induce the incentive problem for the SOE manager—that is, a lower level of effort input. According to our analysis, a restriction of managerial autonomy could reduce efficiency loss resulting from less effort input. Second, an interesting phenomenon in China’s SOE reforms is that governments originally subsidized the SOEs with fiscal revenue, then with bank credits, and recently with the public sale of SOE shares. According to the setting of our model, subsidizing with fiscal revenue implies that the government needs to pay nearly all the cost of subsidization,
100
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
while by bank credit and public sale, it pays only partial cost of subsidization. So through the latter two methods, the government is able to provide a larger amount of subsidy to the loss-making SOEs. Most importantly, our model also proposes that privatization will not necessarily eliminate SBCs in transition economies (Proposition 2). On the contrary, when the SOEs continue to undertake the policy burdens, privatization can result in even more subsidies to enterprises (Corollary 6). This important result helps us to better understand the persistent SBC problems in many transition economies—such as Russia—after SOEs have been massively privatized. In the literature, researchers have reached the consensus that hardening the budget constraints of enterprises is a necessary condition for the success of economic reform in transition economies, but there are still debates about how to do this effectively. We argue that the policy burden on the enterprises is the root cause of SBCs, and only by eliminating the policy burden is it possible to effectively harden the budget constraints. To eliminate the policy burden on the enterprises, we need to reduce the government’s incentives for their imposition; and, according to our analysis, this can be achieved in two ways. One is to reduce the value of B(L)—the benefits that the policy burden can bring to the government. For example, market-oriented reforms will result in a price system that more correctly reflects the shadow prices of different production factors, and thus will increase the costs of implementing CAD development strategies, which helps to remove the strategic policy burden on enterprises. Furthermore, an effective social security system can help the unemployed and maintain social stability, therefore reducing the value of the social policy burden to the government. The other method is to adopt tight fiscal and financial policies, so as to harden the government’s financial constraints. For example, we need to establish effective auditing systems and protect the independent operation of the state banks. Under such a policy, governments—especially local governments—will necessarily pay a higher unit cost for subsidizing enterprises. With the implementation of these two kinds of measures, governments will have low incentives to impose policy burdens, which will help to harden the budget constraints of enterprises. As noted in Lin (2003), the distortions in the factor and product markets, the lack of job opportunities and weak financial discipline could, however, all be endogenous to the government’s CAD strategy. The elimination of SBC symptoms will be successful only if the government gives up the CAD strategy. 6. Conclusion The SBC syndromes have been persistent and pervasive phenomena in socialist and transition economies, and even in market economies. In the literature, researchers attribute the formation of SBCs to various institutional factors, such as paternalism, monopolization and ownership structures. These theories cannot explain, however, the persistent SBC syndromes in transition economies after SOEs have been massively privatized. In this paper, we argue that the policy burden on enterprises is the root cause of the SBC syndromes, and privatization will not necessarily harden the budget constraints of enterprises. We present a model of the policy burden, privatization and the SBC that helps to explain many stylized facts in socialist and transition economies. We find that, in the case of SOEs, the policy burden will induce less effort input from the SOE manager, and therefore low production efficiency. When market competition increases and competitive profit decreases to certain levels, the policy burden on SOEs will result in SBC problems. In the case of SBC, a looser fiscal or financial policy will result in more subsidies to SOEs and lower output levels. Privatization will not necessarily harden the budget constraints of enterprises. On the contrary, when a SOE continues to bear the policy burden, privatization will aggravate SBC problems, because in this case, a private enterprise will demand more ex post subsidies from the government than an SOE under the same condition. Furthermore, when a SOE continues to bear the policy burden, privatization makes it more likely for the privatized SOE to encounter SBC problems. Finally, in the case of SBC outcomes, policy burdens can also reduce the production efficiency of private enterprises, if the policy burden is not too large. The results of our models help to explain the persistent SBC syndromes in transition economies after the massive privatization of their SOEs. Our analysis provides important policy implications for how to effectively harden the budget constraints of the enterprises in economic reforms. Our results also help to explain many other stylized facts in transition economies.
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
101
Acknowledgments The authors would like to thank participants at seminars at Peking University and Toulouse University of Social Sciences. The authors are grateful to two anonymous referees and Editor Daniel Berkowitz for helpful comments and suggestions. Of course, we are responsible for any remaining errors. Appendix A
Proof of Lemma 1. The first-order condition of Nash–Cournot market equilibrium is p(Q) + p (Q) · q = e (q), Q = N · q. If we analyze comparative statistics, we will get dQ 0 p + p q p − e = 1 −N dq q · dN dQ dq dπ dq = p q · +p· − e · < 0. dN dN dN dN
∴
dQ > 0, dN
dq < 0, dN
2
Proof of Proposition 1. Game 1 is easily solvable by backward induction. Stage 3: If Π = π(q) − L 0, the government does not subsidize the SOE, and the pay-offs to the government and the SOE manager are, respectively, UG = B(L) + π(q) − L and US = w ∗ − e(q). If Π = π(q) − L < 0 and the government subsidizes the SOE, the pay-offs to the government and the SOE manager are, respectively, UG = B(L) − θ · S and US = w ∗ − e(q). If Π = π(q) − L < 0 and the government does not subsidize the SOE, the pay-offs to the government and the SOE manager are, respectively, 0 and −e(q). Stage 2: If Π = π(q) − L 0, the SOE manager solves maxq : US = w ∗ − e(q), s.t. B(L) + π(q) − L π ∗ and we will get qH = {q | π(q) = L + (π ∗ − B(L))}. If Π = π − L < 0 and the government subsidizes the SOE, the SOE manager will choose maxq : US = w ∗ − e(q) s.t. B(L) − θ · S π ∗ , S = |π(q) − L|. If Π = π − L < 0 and the government does not subsidize the SOE, the SOE manager will choose q = 0. Stage 1: Given the strategy of the manager, the government has the SOE bear the policy burden, L. 2 Proof of Corollary. From the proof of Lemma 1, we know that π(q) = P (q + (N − 1)q ∗ ) · q − K − e(q) is increasing in q for q q ∗ . Since B(L) > L, θ ∈ (0, 1), from Proposition 1, we have: (1) When π ∗ B(L), π(qH ) = π ∗ + L − B(L) < π ∗ , ∴ qH < q ∗ . (2) When π ∗ < B(L), π ∗ − π(qL ) = π ∗ − L + θ1 [B(L) − π ∗ ] ( θ1 − 1)[B(L) − π ∗ ] > 0, ∴ qL < q ∗ .
2
Proof of Corollaries 2, 3 and 4. These results are obvious from Lemma 1, Proposition 1 and Corollary 1.
2
Proof of Proposition 2. The proof is similar to that of Proposition 1, and the only thing to be pointed out is that, when Π = π − L < 0 and the government provides a subsidy, the private enterprise manager will solve the optimization problem of maxq,S : UP = π(q) − L + S, s.t. B(L) − θ · S 0, π(q) < L and UP π ∗ . 2 References Anon, 1998. Symposium on soft budget constraint and transition economies. Journal of Comparative Economics 26 (1), 9–116. Boycko, M., Sheleifer, A., Vishny, R., 1996. A theory of privatization. Economic Journal 106, 309–319. Dewatripont, M., Maskin, E., 1995. Credit and efficiency in centralized and decentralized economies. Review of Economic Studies 62 (4), 541–555. Grossman, S., Hart, O., 1986. The costs and benefits of ownership: A theory of vertical and lateral integration. Journal of Political Economy 94 (4), 691–719. Kornai, J., 1980. Economics of Shortage. Amsterdam, North-Holland. Kornai, J., Maskin, E., Roland, G., 2003. Understanding the soft budget constraint. Journal of Economic Literature 41 (4), 1095–1136. Li, D., 1992. Public ownership as the cause of a soft budget constraint. Mimeo, Harvard University.
102
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
Lin, J.Y., 2003. Development strategy and economic convergence. Economic Development and Cultural Change 51 (2), 277–308. Lin, J.Y., Tan, G., 1999. Policy burdens, accountability and soft budget constraint. American Economic Review 89 (2), 426–431. Lin, J.Y., Cai, F., Li, Z., 1996. China’s Miracle: Development Strategies and Economic Reform. Chinese Univ. of Hong Kong Press. Lin, J.Y., Cai, F., Li, Z., 1998. Competition, policy burdens, and state-owned enterprise reform. American Economic Review 88 (2), 422–427. Lin, J.Y., Cai, F., Li, Z., 2000. Perfect Information and the SOE Reforms in China. Chinese Univ. of Hong Kong Press. Novshek, W., 1980. Cournot equilibrium with free entry. Review of Economic Studies XLVII, 473–486. Qian, Y., Roland, G., 1998. Federalism and soft budget constraint. American Economic Review 88 (5), 1143–1162. Röller, L., Zhang, Z., 2005. Bundling and social and private goods and the soft budget constraints. Journal of Comparative Economics 33 (1), 47–58. Schmidt, K., 1996. The cost and benefit of privatization: An incomplete contract approach. Journal of Law, Economics and Organization 12, 1–24. Segal, I.R., 1998. Monopoly and soft budget constraint. RAND Journal of Economics 29 (3), 596–609. World Bank, 1996. World Development Report 1996: From Plan to Market. Oxford Univ. Press, Oxford. World Bank, 2002. Transition: The First Ten Years, Analysis and Lessons for Eastern Europe and the Former Soviet Union. World Bank, Washington, DC. Zhang, Z., Lu, Y., 2003. Facilitating the SOE reforms. Economics and Finance Forum 1 (in Chinese).
Policy burden, privatization and soft budget constraint Justin Yifu Lin a , Zhiyun Li b,∗ a China Center for Economic Research, Peking University, Beijing 100871, China b MF003, Manufacture des Tabacs, MPSE, Toulouse Université des Sciences Sociales, 31000, Toulouse, France
Received 17 February 2007; revised 11 November 2007 Available online 8 December 2007
Lin, Justin Yifu, and Li, Zhiyun—Policy burden, privatization and soft budget constraint We propose a new cause for the pervasive syndromes of soft budget constraint (SBC) in socialist and transition economies, that is, the policy burdens on enterprises result in the SBC problems. The policy burdens induce low effort input of firm manager and thus the low efficiency of production. And with the policy burdens, increasing market competition will make the SBC syndromes arise more likely. Privatization will not necessarily harden the budget constraint of the enterprise. On the contrary, when a SOE still bears the policy burdens, privatization will only aggravate the SBC problems. Because in this case, a private enterprise will demand more ex post subsidies from the government, than a SOE under the same condition. Our results help to explain many stylized facts in transition and socialist economies. Journal of Comparative Economics 36 (1) (2008) 90–102. China Center for Economic Research, Peking University, Beijing 100871, China; MF003, Manufacture des Tabacs, MPSE, Toulouse Université des Sciences Sociales, 31000, Toulouse, France. © 2007 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved. JEL classification: L32; L33; P31 Keywords: Policy burden; Market competition; Soft budget constraint; Privatization
1. Introduction
A well-known phenomenon in socialist and transition economies is that, when state-owned enterprises (SOEs) incur losses, the State normally cannot restrain from bailing them out by providing additional subsidies or credits, which is known as the soft budget constraint (SBC) problem. Soft budget constraint, a term coined by Kornai (1980), is connected to various problems in socialist and transition economies, such as shortage and low efficiency of SOEs. And due to the negative consequences of SBC, hardening the budget constraints of enterprises has been a principal objective of the economic reforms in transition economies. However, the reforms of Eastern European economies (EEEs) in the 1990s have proved to be unsuccessful in eradicating the lasting SBC syndromes, even after the SOEs had been massively privatized in many of those countries (World Bank, 1996, 2002).
* Corresponding author. Fax: 86 10 6275 1474.
E-mail addresses: [email protected] (J.Y. Lin), [email protected] (Z. Li). 0147-5967/$ – see front matter © 2007 Association for Comparative Economic Studies. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jce.2007.11.001
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
91
As for the SBC syndromes in socialist and transition economies, there has been a huge amount of literature, most focusing on the consequences of SBC, such as moral hazard or adverse selection problems of firm managers.1 But about the institutional causes for the formation of SBC, there still exist debates, and little detailed work has been done on these topics. In this paper, we propose a new hypothesis on the causes of SBC, that is, the policy burdens on enterprises results in SBC problems. And privatization will not necessarily harden the enterprises’ budget constraints, if they continue to undertake the State’s policy burdens; rather, privatization will probably exacerbate the SBC syndromes under certain conditions. The results of our model provide important implications for understanding many stylized facts in socialist and transition economies, as well as in fledged market economies. Before expatiating on our theory, we first provide a short review of the literature. As Kornai et al. (2003) summarize, understanding the SBC syndromes entails bearing in mind a complex chain of causality, which is depicted in a schematic form in Fig. 1. Block (1) represents the political and social factors that generate the motives behind the formation of SBC. Block (2) represents the motives that create the SBC syndromes, such as the motivation of the State or creditor to refinance loss-making enterprises. Finally, block (3) represents the consequences of SBCs.
Fig. 1. The SBC syndrome: the chain of causality.
Many studies on SBCs focus on the relationship between block (2) and block (3), as mentioned above, and to some degree they have reached a consensus on how SBCs can influence the working of an economy. Dewatripont and Maskin (1995) is a seminal paper in modeling this relationship, from which a large amount of related literature has been developed. In their paper, they formulate the SBC in the context of dynamic commitment inconsistency and effectively capture the main ideas of Kornai (1980) on the SBC syndromes. In their two-period dynamic game, they state that the state could have incentives to refinance an inefficient and uncompleted project, because the marginal benefit of refinancing exceeds the marginal cost of abandoning it, which means refinancing is an ex post efficient decision for the support organization. But little detailed work has been done on the relationship between block (1) and block (2), such as the root causes for the SBC syndromes. Also, one important and unanswered question is why socialist economies are more vulnerable to SBC than fledged market economies. Or, in other words, what are the institutional factors that contribute to the pervasive SBC syndromes in those economies? Kornai (1980) attributes the causes of SBC primarily to political constraints—that is, to the paternalism of socialist governments. Dewatripont and Maskin (1995), however, have shown that paternalism is neither a necessary nor a sufficient condition for SBC. They suggest that the highly centralized system of socialist economies is the primary cause of SBC, because in a decentralized system, such as in market economies, the transaction cost of refinancing will be so high that refinancing is ex post inefficient, and will thus harden budget constraints. Qian and Roland (1998) propose a similar idea: they suggest that China’s more decentralized fiscal system increases competition among different regions, and thus the opportunity costs of refinancing bad projects are very high, which could partly explain China’s success in economic transitions.2 Another influential theory on SBC is that the public ownership of socialist economies is the cause for their pervasive SBC syndromes. In the model of Li (1992), public ownership means that refinancing decisions are made jointly by 1 And a comprehensive and excellent survey of the literature can be found in Kornai et al. (2003). There is also a symposium on the SBC in Anon (1998). 2 The validity of their statement depends on the tightness of the financial system. In effect, under existing financial arrangements, banks are owned by the central government. For economic development in their region, local governments can assist local enterprises to borrow excessively from state banks to finance their investment projects. If the state banks accumulate a large amount of non-performing loans due to many of those projects turning bad, the central government is obliged to rescue the banks. Therefore, under a decentralized fiscal system and a centralized banking sector, local governments can treat the funds in the state banks as a common resource, resulting in a situation resembling the tragedy of common.
92
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
the government and the enterprise, which could contribute to the SBC problem. From another angle, Schmidt (1996) states that privatization makes inside information inaccessible for the government, which makes ex post punishment credible and thus hardens the firm’s budget constraint. The massive privatization of SOEs in the EEEs did not, however, eradicate the SBC syndromes at all; on the contrary, the state provided even more subsidies to those privatized enterprises, which suggests that public ownership is not necessarily the cause of the SBC syndromes. Segal (1998) proposes that the monopolistic status of an enterprise could cause SBC: because the monopoly cannot capture the entire social surplus from its production, the government might have incentives to subsidize an unprofitable enterprise in order to support its production and thus guarantee social surplus. The theory is not, however, convincing, because it is only a partial equilibrium model and, in general, monopoly will incur efficiency losses as a whole in the economy. Boycko et al. (1996) suggest that politicians might pay subsidies to enterprises to induce them to retain excess labor, which could benefit the politicians in elections. There are, however, no dynamic factors in their model, and therefore no problems of commitment, which may constrain the extension of their analysis. Röller and Zhang (2005) study a similar problem in a dynamic context, and find that requiring the firm to provide social goods could result in SBC problems, and increasing competition in the private goods market could make SBC problems more likely. In this paper, we provide a new explanation for the pervasive SBCs in socialist and transition economies. We argue that the policy burdens imposed by the State on state-owned and other enterprises are the root cause of the SBC syndromes, and only by eliminating the policy burdens can the State effectively harden the budget constraints of the enterprises. When the SOEs continue to undertake the policy burdens, privatization will not eradicate SBC problems; rather, this will exacerbate them because it increases the cost of state subsidization. Our theory shares some features in common with Boycko et al. (1996) and Röller and Zhang (2005), but there are at least two important differences. First, they emphasize just the social policy burden on the enterprises. In this paper, we introduce a new, possibly more important, strategic policy burden on the enterprises, due to the government’s comparative advantage-defying (CAD) strategy, which is a key feature of the enterprises in socialist and many developing countries since World War II. Second and more importantly, we study the relationship between privatization and SBCs. We find that when a privatized firm continues to undertake the policy burdens, privatization will only aggravate the SBC syndromes. This can effectively explain the persistent SBC syndromes in the EEEs after massive privatization of the SOEs. This result also provides important implications for the choice of transitional paths, from a centrally planned economy to a decentralized market system. Finally, we study the impacts of market competition on SBC problems, and find a result similar to Röller and Zhang’s (2005)—that is, increasing competition makes it more likely for SBC problems to arise when the policy burdens exist. But a further implication of our model is that, in a less competitive market where the SOEs can make positive profits and there are no ex post subsidies, the State also implicitly support the SOEs by restricting market competition. One point worthy of attention is that our theory also helps to explain SBC problems in fledged market economies, since the existence of policy burdens is not a feature unique to socialist and transition economies. It is also common for enterprises in market economies to undertake governments’ policy burdens, such as maintaining redundant workers or implementing special government policies. In those cases, when the enterprises incur losses, the government is obliged to support them subsequently with additional subsidies, which is an SBC problem. The rest of this paper is organized as follows: Section 2 provides a background of the formation of policy burdens in socialist and developing countries, and explains our theory of the SBC syndromes. Section 3 develops a model that studies the relationship between policy burdens and the SBC of SOEs. Section 4 investigates the effects of privatization on SBC problems. Section 5 discusses some policy implications of our model, and Section 6 is a short conclusion. 2. The main ideas A distinct feature of socialist and many developing economies is their extreme bias towards the development of heavy industries. However, the endowments of those economies are commonly characterized by their lack of physical capital, and relative abundance in labor supply. And the heavy industry-biased strategy is obviously not consistent with their comparative advantages. To implement CAD strategies, the State needs to impose its policy burdens on the enterprises—that is, the strategic policy burden and the social policy burden. The former stems from the fact that the enterprises are forced to enter CAD industries or adopt CAD technologies, both of which will render them non-viable in a free, competitive market. The social policy burden refers to keeping redundant workers, providing retirement
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
93
pensions and other social services, such as schooling and medical care. Since the capital-intensive industries cannot provide enough job opportunities, in order to solve employment problems and maintain social stability, the State needs the SOEs to retain excess workers, which results in a social policy burden. These two kinds of policy burdens are therefore interrelated. With these policy burdens, the enterprises are not profitable in a free, competitive market. The term viability is defined as follows: if, without any external subsidies or protections, a normally managed firm is expected to earn a socially acceptable profit in a free, open and competitive market, the firm is viable; otherwise, it is nonviable (Lin, 2003). In a transition economy, the allocation of social resources becomes more and more dependent on market mechanism, and it will thus be more difficult for SOEs with the policy burdens to be profitable. To keep these SOEs implementing the policy burdens, the State needs to guarantee their survival in market competition, which can be achieved in two ways: one is to subsidize them implicitly by restricting competition and protecting the market powers of the SOEs, as was done before the transition; and the second is to support them explicitly by ex post subsidies or credits, in the case of loss making. We argue that privatization will not necessarily eradicate the SBC syndromes. On the contrary, it will only aggravate the syndromes if the privatized enterprise continues to bear the policy burdens. In our paper, privatization implies a combination of two changes: the turnover of control rights and an increasing share of cash flows to private shareholders. Either of these changes will increase the bargaining power of the private firm in demanding subsidies from the government. To impose a policy burden, it is more expensive for the government to allocate it to a private enterprise than to a SOE, since the former will demand more subsidies from the government to implement that policy burden. 3. Policy burdens, SOEs and the SBC syndromes 3.1. Competition and profit: a preliminary result As a preparation for later analysis, we first consider a Nash–Cournot equilibrium in a free-entry market, where the SOEs do not bear policy burdens. We assume there are N SOEs in the market, and they produce the same product. The demand for the firms’ product is given by a strictly decreasing inverse demand function, P (Q), where Q is the total supply of this product. We assume the cost function is the same for all SOEs, which is C(q) = K + e(q), where K is capital input, q is the output of a firm and e(q) refers to the effort of the firm manager. And for any q 0, we assume e (q) > 0, e (q) > 0 and e(0) = 0. We assume that N¯ is the long-term Nash–Cournot equilibrium number of firms in the free-entry market, and the profit of firm i is defined as π(qi ) = P (Q) · qi − e(qi ) − K, where Qi = Q−i + qi (Q−i is the total output of the other N − 1 firms). Given the number of firms, N N¯ , we achieve the following conditions of Nash–Cournot equilibrium: (1) profit maximization: qi ∈ arg max p(Q−i + qi ) · qi − K − e(qi ), (2) output market equilibrium: p(Q) = p(Q−i + qi ), (3) participant constraint: w − e(qi ) = 0, i = 1, 2, . . . , N where w is the payment to the SOE managers, whose reservation pay-offs are set at 0. According to Novshek (1980), there exists a solution to the problem above, which is denoted as {q ∗ (N ), w ∗ (N )} in our model. We get the equilibrium profit π ∗ (N ) = π(q ∗ (N )) and the payment to the SOE manager w ∗ (N ) = e(q ∗ (N )).3 If we further assume p + p · q < 0, which is always satisfied if marginal revenue is falling at a rate steeper than the slope of the demand curve, we get the following lemma. Lemma 1. If N1 < N2 N¯ , then, at the Nash–Cournot equilibrium in a free-entry market, π ∗ (N1 ) > π ∗ (N2 ) and q ∗ (N1 ) > q ∗ (N2 ). 3 For simplicity, we denote this as q ∗ , π ∗ and w ∗ in the following sections.
94
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
Lemma 1 tells us that when the number of firms in the market increases, the equilibrium profit and equilibrium output of each firm will decrease. Or, in other words, increasing competition reduces the profit of each firm. For the following sections, we always assume that the exogenously given number of SOEs is smaller than N¯ —that is: Assumption 1. N N¯ is always satisfied. We will later use the result of Lemma 1 to analyze the effects of market competition on the SBC syndromes. 3.2. Policy burdens and the SBCs of SOEs In the real world, we often see SOEs undertaking various policy burdens, such as providing social goods or retaining redundant workers. In this subsection, we introduce the policy burdens into a dynamic game between the government and the SOEs, and investigate the impacts of the policy burdens on the SBC syndromes. To focus on the relationship between the policy burdens and the SBC outcomes, we need to make some simplification to our model setting. We will study the behavior of just one SOE, which undertakes the policy burdens imposed by the government. We assume that the other N − 1 firms do not bear the policy burdens, and all behave as in the former game of Nash–Cournot competition, each producing at the efficient level of q ∗ . Naturally, the introduction of the policy burdens makes it more difficult to evaluate the performance of an SOE manager. For example, when a SOE retains redundant workers or adopts CAD technologies according to government requirement, in the case of loss making, the government would not be able to distinguish between the losses resulted from the policy burdens and those from the less effort inputs of the SOE managers. On the contrary, if the firm does not bear the policy burdens and competes with other firms under the same conditions, the relative profitability of the firm can be a sufficient indicator to reflect the efficiency of the SOE. Taking into account this information difference after the introduction of the policy burdens, we make the following assumption. Assumption 2. When an enterprise bears the policy burdens, its output level, q, and thus the effort level, e(q), are not verifiable. Unlike in the former case, the payment to the SOE manager is now no longer contingent on output q, since that information is not verifiable. We here use L to denote the policy burden on the SOE, which can be the cost of retaining redundant workers and/or that of adopting a CAD technology, and let π(q) denote the profit of the SOE when it does not bear the policy burden. Under the assumptions above, we have π(q) = P (q + (N − 1)q ∗ ) · q − e(q) − K. The profit of the SOE undertaking the policy burden is Π(q) = π(q) − L. Let B(L) represent the benefit of the policy burden to the government, and we assume: Assumption 3. B(L) > L. Assumption 3 implies that the policy burden always generates additional benefit to the government, compared with the cost of the SOE’s forgone profit.4 For example, redundant employment could engender public support for the government, or the establishment of CAD industries could contribute to the government’s political credit. We assume that the government pays the SOE manager a fixed payment of w ∗ ; the pay-off function of the SOE manager is thus US = w ∗ − e(q) and the government’s payoff (without providing subsidies) is UG = B(L) + Π = B(L) + π(q) − L. Here we consider the following dynamic game between the government and the SOE manager. 4 This assumption is similar to the assumption of Boycko et al. (1996) that the politician cares less about the forgone SOE profit to the government, resulting from maintaining redundant employees.
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
95
Game 1. Stage 1. The government decides whether to have the SOE undertake the policy burden. Stage 2. The SOE manager observes the government’s decision and chooses his effort input, e(q). Stage 3. Production is completed. The government observes the final profit of the firm and decides whether to subsidize the SOE. Before solving the game, we still need to specify some features of the behavior of the government and the SOE manager, to complete the description of Game 1. First, we assume that the government subsidizes the SOE only when it incurs losses—that is, Π = π(q) − L < 0—and the amount of subsidy is S = |π(q) − L|, equal to the absolute value of loss. The government’s subsidy cost is θ · S, with the assumption that Assumption 4. 0 < θ < 1. This assumption suggests that the government pays just a part of the total cost of the subsidy.5 Second, we assume that at Stage 3, if the SOE incurs losses (Π < 0) and the government does not provide a subsidy, the SOE will go bankrupt, and the payment to the SOE manager and the pay-off of the government will both be 0. Finally, we assume that, with the same pay-off level, the government always prefers the SOE to take the policy burden. Solving Game 1, we get the following results. Proposition 1. The sub-game perfect equilibrium outcomes of Game 1 are given by (1) If π ∗ < B(L), the SOE manager chooses an output of qL = {q | π(q) = L + θ1 · (π ∗ − B(L))}, and the firm incurs losses. The government imposes the policy burden, L, on the SOE and provides a subsidy of θ1 [B(L) − π ∗ ]. (2) If π ∗ B(L), the SOE manager chooses an output of qH = {q | π(q) = L + (π ∗ − B(L))}, and the firm does not incur losses. The government imposes the policy burden, L, on the SOE and provides no subsidy. Proof. All proofs are provided in Appendix A.
2
According to Proposition 1, the relative magnitudes of π ∗ and B(L) decide the equilibrium outcomes of the game. We refer to part (1) of Proposition 1 as the SBC outcome, because the government explicitly provides an ex post subsidy to the loss-making SOE. Part (2) can also be considered a case of implicit subsidy, in the sense that the government implicitly subsidizes the SOE by protecting its market powers.6 Proposition 1 reveals the impacts of the policy burden on the production decisions of the SOE, and the formation of the SBC syndromes. First, when the SOE undertakes the policy burden, the output levels are always different from the optimal level, q ∗ , which implies an efficiency loss in SOE production. Low SOE efficiency is commonly observed all over the world, and most researches attribute it to the public ownership of SOEs. This is not necessarily true, according to our analysis; or at least, it is not the only cause for the poor performance of SOEs. The result of Proposition 1 reveals that the low efficiency of SOEs may result from implementing the policy burden imposed by the government.7 Second, an SOE with a policy burden is non-viable in a free, competitive market, due to the additional cost of L. Proposition 1 suggests that the government can help the survival of SOEs in two ways: one is to subsidize them explicitly when they incur losses (part (1)); the other is to support them implicitly by restricting market competition and therefore protecting their market powers (part (2)). 5 The government can transfer a part of its subsidy cost by an inflation tax on the general public. In the case of China, local governments can also
compel state banks to grant loans to their local SOEs, which results in huge amounts of non-performing loans that need to be absorbed by the state banks. Eventually, the central government is required to inject funds to write off the state banks’ non-performing loans. 6 A high value of π ∗ implies that the SOE has market power. By restricting market competition, the government implicitly supports the firm at the cost of efficiency loss of the overall economy. For example, China’s large-sized SOEs make huge profits, because they are monopolies in their industries, and their monopoly statuses are protected by the government. 7 This result is similar to the explanation in Boycko et al. (1996)—that public enterprises are inefficient because they address the objectives of the politicians—but our result is reached in a dynamic context.
96
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
Fig. 2. Policy burden and low effort input.
There are several corollaries that can be derived from Proposition 1, which are summarized below. Corollary 1. When the SOE bears the policy burden, its output levels are always lower than the optimal level—that is, qL , qH < q ∗ . According to Corollary 1, the policy burden results in lower output levels, and thus lower levels of effort input by the SOE manager, which reflects an incentive problem resulting from the policy burden. The welfare loss is depicted in Fig. 2. Interestingly, when L < π ∗ < B(L), the SOE with the policy burden can also make positive profit, for example just by choosing to produce q ∗ . But the manager has no incentive to produce that much because he expects that the government will subsidize the SOE subsequently in the case of loss making, and producing less will bring him a higher pay-off. This case clearly demonstrates the incentive problems of the SBC. In socialist and transition economies, we often see that resources are allocated through administrative systems, and the SOE managers are normally subject to various managerial restrictions imposed by the government, such as output quotas and price ceilings. Corollary 1 also implies that a restriction on managerial autonomy can potentially enhance the production efficiency of the SOEs. Corollary 2. When the SOE bears the policy burden, an increase of N makes the SOE more vulnerable to losses, and results in more subsidy, S, to the SOE in the case of loss making. Corollary 2 is an intuitive but quite important result. When the number of firms (N ) in the market increases, the profit of each firm will decrease (Lemma 1), and an SOE with a policy burden will thus be more vulnerable to losses. In addition, when π ∗ decreases and the SOE incurs losses, the government will provide more subsidies to the firm (part (1) of Proposition 1). The relationship between the number of firms in the market and government subsidy is depicted in Fig. 3. Corollary 2 helps to explain the transitions of China’s state sectors in the past three decades. Since the beginning of China’s economic reforms in 1978, more and more private firms have entered the markets with low entrance barriers, and SOEs—especially those in more competitive markets—have incurred huge losses. The rapidly increased subsidies to the loss-making SOEs have become a heavy burden on the government. At the end of 1990s, the central government decided to abandon small and medium-size SOEs, most of which were in competitive markets, and kept only the large SOEs that were monopolies in their markets and contributed nearly all the profits of the state sectors.8 8 In 2001, China’s 50 largest SOEs accounted for 73.6 percent of the total profits of all SOEs in China (Zhang and Lu, 2003).
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
97
Fig. 3. Competition and the size of subsidy.
Corollary 3. When the SOE bears the policy burden and incurs losses, a smaller θ will result in more subsidies from the government and less output of the SOE. Corollary 3 says that when the government’s unit cost of subsidization becomes lower (a smaller θ ), the output level of the SOE will be lower, and the government will provide more subsidies to a loss-making SOE. A higher θ clearly represents a tougher monetary and fiscal policy, which makes subsidies more costly for the government. An implication of Corollary 3 is that a tougher monetary and fiscal policy can increase the efficiency of the SOEs, and reduce the amount of the subsidies. This result also suggests that under loose fiscal or financial policies, decentralization can result in softer budget constraints for SOEs, especially when local governments are able to intervene in the operations of the State’s local financial institutions. Corollary 4. When the SOE bears the policy burden, a larger B(L) will result in less output from the SOE. Corollary 4 renders a similar idea as Corollary 3, but its validity is more general in the sense that it does not require the SOE to incur losses. The intuition behind this result is that, if the government receives more benefits from the policy burden, the SOE manager can also share those benefits and increase his pay-off by just reducing his effort input. 4. Privatization, policy burdens and SBCs A well-known theory states that public ownership of SOEs is the root cause for SBC problems in socialist and transition economies; however, the experience of SOE reforms in the EEEs has proved that this theory is not necessarily true. In many of those economies, the SOEs were massively privatized, but this did not harden the budget constraints of the privatized firms. On the contrary, it softened their budget constraints, and governments provided even more subsidies to the privatized SOEs than before (World Bank, 1996, 2002). We argue that when SOEs continue to undertake the policy burdens, privatization will not harden the budget constraints of the privatized SOEs; rather, they will exacerbate their SBC syndromes. In this section, within a framework similar to that in Section 3, we investigate the relationship between policy burdens, privatization and SBC problems, and propose a new explanation for the persistent SBC syndromes in transition economies. According to Grossman and Hart (1986), privatization implies a combination of two changes in the context of our model. The first is the turnover of control rights from the government to the enterprise manager, so that the manager can decide whether or not to accept the policy burden assigned by the government. The second is an increased share of cash flows to the manager9 and, correspondingly, a decreased share to the government. For simplicity and without 9 Here, the manager is assumed to be a representative of private shareholders of the enterprise.
98
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
loss of generality, we assume that the government owns no share in the privatized enterprise. Now we consider the following game between the government and the enterprise manager, when the enterprise has been privatized. Game 2. Stage 1. The government decides whether to assign the policy burden, L, to the enterprise. Stage 2. The enterprise manager observes the government’s action and decides whether or not to accept the policy burden, and then chooses his effort level, e(q). Stage 3. Production is completed. If the privatized enterprise bears the policy burden, the manager will require a subsidy of S from the government. Stage 4. The government decides whether to provide the subsidy of S to the enterprise. Game 2 is similar structurally to Game 1, yet privatization has changed the control right of the firm and the reservation pay-off for the manager. First, the manager now has the control right of the privatized firm and is thus able to reject the policy burden assigned by the government. Second, the manager’s reservation pay-off changes from 0 to π ∗ , and that of the government changes from π ∗ to 0. Finally, when the enterprise undertakes the policy burden, the manager will require a subsidy of S from the government, and the government will decide whether or not to accept it. Other settings and assumptions remain the same as in Game 1. Solving Game 2, we get the following proposition. Proposition 2. The sub-game perfect equilibrium outcomes of Game 2 are given by (1) If π ∗ < θ1 B(L), the government will assign the policy burden, L, to the enterprise, and the enterprise manager will accept it and choose to produce q = {q | π(q) = min(L, π ∗ )}. The subsidy provided by the government is S = θ1 B(L). (2) If π ∗ θ1 B(L), the government will assign the policy burden, L, to the enterprise, but the enterprise manager will reject it and choose to produce q ∗ = {q | π(q) = π ∗ }. From Proposition 2 we see that when the competitive profit is relatively low (π ∗ < B(L)/θ ), the private enterprise will accept the policy burden assigned by the government, and receive an ex post subsidy of B(L)/θ , which is known as an SBC problem. In this case, when L < π ∗ , the policy burden will also result in inefficiency of production in the private enterprise, so the policy burden may also reduce the production efficiency of a private firm. On the other hand, if the competitive profit is relatively high (π ∗ B(L)/θ ), the enterprise will reject the policy burden proposed by the government and produce at the efficient output level, q ∗ , which generates a higher pay-off to the manager. The equilibrium outcomes of Game 2 differ from those of Game 1 in several ways, such as the threshold of SBC outcomes, the outputs of the enterprises and the amount of government subsidization. These differences all result from the change of ownership of the enterprise. The manager of the privatized enterprise—with a higher reservation pay-off—can achieve his maximum utility not only by choosing the output level, but by deciding whether or not to accept the policy burden proposed by the government. Table 1 compares the SBC outcomes of Game 1 (the SOE) with Game 2 (the private enterprise), which provides important implications for the impact of privatization on the SBC syndromes. The first important information in Table 1 is that, when enterprises undertake the same policy burden, L, all other things being equal, it is easier for a private enterprise (PE) than for an SOE to experience SBC problems. For instance, Table 1 Policy burden and the SBCs of the SOE and the private enterprise (PE)
1. Condition of resulting SBC 2. Amount of subsidy 3. Output of enterprise
Proposition 1 (SOE)
Proposition 2 (PE)
π ∗ < B(L) 1 ∗ θ [B(L) − π ]
π ∗ < θ1 B(L)
qL = {q | π(q) = L − θ1 (B(L) − π ∗ )}
1 θ B(L)
q = {q | π(q) = min(L, π ∗ )}
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
99
when B(L) π ∗ < B(L)/θ , the SOE will not have the SBC problem, but the PE will, according to Proposition 2. This result can be summarized in the following corollary. Corollary 5. When an SOE and a PE undertake the policy burden, L, all other things being equal, it is easier for the PE to experience SBC problems. Second, it is interesting that in the case of SBC outcomes, the subsidy to the PE, B(L)/θ , is more than that to the SOE, (B(L) − π ∗ )/θ . This result implies that, to have the enterprises bear the same level of policy burden, the government needs to pay a higher level of subsidy to the PE. The intuition behind this result is that privatization increases the reservation pay-off to the firm manager and endows him with the right to reject the policy burden assigned by the government; as a result, the manager will require more subsidies when he is ready to undertake the policy burden. This result has important policy implications and helps to explain why, after massive privatization in the EEEs, subsidies to privatized SOEs have increased. We reiterate this result formally as follows. Corollary 6. When an SOE and a PE undertake the policy burden, L and π ∗ < B(L), all other things being equal, the government needs to provide more subsidies to the PE. From Table 1, we also see that in the SBC outcomes, the PE is more efficient than the SOE in the sense that it produces more, since min(L, π ∗ ) > L − (B(L) − π ∗ )/θ . In addition, in the case of the non-SBC outcomes of Game 1 and Game 2, the PE is also more efficient than the SOE, because it produces at the efficient output level of q ∗ , but in this case the PE does not undertake the policy burden. We can easily reach the following corollary. Corollary 7. When an SOE and a PE undertake the policy burden, L and π ∗ < B(L), all other things being equal, the PE is more efficient in production than the SOE. 5. Discussions and applications Our models in the above two sections are an extension of the work of Lin and Tan (1999), Lin et al. (1996, 1998, 2000). In this section, we will first provide a short discussion of the key assumptions of our models, and then apply the major results of our analysis to explain some stylized facts of the SBC syndromes in socialist and transition economies. Furthermore, the results of our analysis provide important implications for how to harden the budget constraints of enterprises in transition economies, as well as in market economies. There are two pivotal assumptions in our models. The first is that the policy burden on the enterprise brings additional benefits to the government, which is implied by Assumption 3, B(L) > L, in our models. In general, the government cares not only about economic, but political issues, such as nation building, national security, social stability and public support for the government. Therefore, the policy burden on enterprises could bring additional political benefits to the government. The other key assumption is that the government can sustain only part of the cost of subsidizing enterprises, which is represented by Assumption 4, 0 < θ < 1, in our model. For example, in transition economies, it is normal for local governments to intervene in the operation of state banks. Local governments naturally prefer to use bank credits rather than their own fiscal revenue to subsidize local enterprises; also the central government can transfer the cost to the public through an inflation tax. If either of the above two assumptions is true, the government will have an incentive to impose the policy burden on enterprises, whether they are SOEs or private enterprises. Next we apply the results of our model to explain some stylized facts in socialist and transition economies. First, we know that in socialist economies, the government restricts the control rights of SOE managers through such measures as price ceilings or production quotas. Our model suggests that when enterprises undertake policy burdens, an ex ante expectation of ex post subsidies could induce the incentive problem for the SOE manager—that is, a lower level of effort input. According to our analysis, a restriction of managerial autonomy could reduce efficiency loss resulting from less effort input. Second, an interesting phenomenon in China’s SOE reforms is that governments originally subsidized the SOEs with fiscal revenue, then with bank credits, and recently with the public sale of SOE shares. According to the setting of our model, subsidizing with fiscal revenue implies that the government needs to pay nearly all the cost of subsidization,
100
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
while by bank credit and public sale, it pays only partial cost of subsidization. So through the latter two methods, the government is able to provide a larger amount of subsidy to the loss-making SOEs. Most importantly, our model also proposes that privatization will not necessarily eliminate SBCs in transition economies (Proposition 2). On the contrary, when the SOEs continue to undertake the policy burdens, privatization can result in even more subsidies to enterprises (Corollary 6). This important result helps us to better understand the persistent SBC problems in many transition economies—such as Russia—after SOEs have been massively privatized. In the literature, researchers have reached the consensus that hardening the budget constraints of enterprises is a necessary condition for the success of economic reform in transition economies, but there are still debates about how to do this effectively. We argue that the policy burden on the enterprises is the root cause of SBCs, and only by eliminating the policy burden is it possible to effectively harden the budget constraints. To eliminate the policy burden on the enterprises, we need to reduce the government’s incentives for their imposition; and, according to our analysis, this can be achieved in two ways. One is to reduce the value of B(L)—the benefits that the policy burden can bring to the government. For example, market-oriented reforms will result in a price system that more correctly reflects the shadow prices of different production factors, and thus will increase the costs of implementing CAD development strategies, which helps to remove the strategic policy burden on enterprises. Furthermore, an effective social security system can help the unemployed and maintain social stability, therefore reducing the value of the social policy burden to the government. The other method is to adopt tight fiscal and financial policies, so as to harden the government’s financial constraints. For example, we need to establish effective auditing systems and protect the independent operation of the state banks. Under such a policy, governments—especially local governments—will necessarily pay a higher unit cost for subsidizing enterprises. With the implementation of these two kinds of measures, governments will have low incentives to impose policy burdens, which will help to harden the budget constraints of enterprises. As noted in Lin (2003), the distortions in the factor and product markets, the lack of job opportunities and weak financial discipline could, however, all be endogenous to the government’s CAD strategy. The elimination of SBC symptoms will be successful only if the government gives up the CAD strategy. 6. Conclusion The SBC syndromes have been persistent and pervasive phenomena in socialist and transition economies, and even in market economies. In the literature, researchers attribute the formation of SBCs to various institutional factors, such as paternalism, monopolization and ownership structures. These theories cannot explain, however, the persistent SBC syndromes in transition economies after SOEs have been massively privatized. In this paper, we argue that the policy burden on enterprises is the root cause of the SBC syndromes, and privatization will not necessarily harden the budget constraints of enterprises. We present a model of the policy burden, privatization and the SBC that helps to explain many stylized facts in socialist and transition economies. We find that, in the case of SOEs, the policy burden will induce less effort input from the SOE manager, and therefore low production efficiency. When market competition increases and competitive profit decreases to certain levels, the policy burden on SOEs will result in SBC problems. In the case of SBC, a looser fiscal or financial policy will result in more subsidies to SOEs and lower output levels. Privatization will not necessarily harden the budget constraints of enterprises. On the contrary, when a SOE continues to bear the policy burden, privatization will aggravate SBC problems, because in this case, a private enterprise will demand more ex post subsidies from the government than an SOE under the same condition. Furthermore, when a SOE continues to bear the policy burden, privatization makes it more likely for the privatized SOE to encounter SBC problems. Finally, in the case of SBC outcomes, policy burdens can also reduce the production efficiency of private enterprises, if the policy burden is not too large. The results of our models help to explain the persistent SBC syndromes in transition economies after the massive privatization of their SOEs. Our analysis provides important policy implications for how to effectively harden the budget constraints of the enterprises in economic reforms. Our results also help to explain many other stylized facts in transition economies.
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
101
Acknowledgments The authors would like to thank participants at seminars at Peking University and Toulouse University of Social Sciences. The authors are grateful to two anonymous referees and Editor Daniel Berkowitz for helpful comments and suggestions. Of course, we are responsible for any remaining errors. Appendix A
Proof of Lemma 1. The first-order condition of Nash–Cournot market equilibrium is p(Q) + p (Q) · q = e (q), Q = N · q. If we analyze comparative statistics, we will get dQ 0 p + p q p − e = 1 −N dq q · dN dQ dq dπ dq = p q · +p· − e · < 0. dN dN dN dN
∴
dQ > 0, dN
dq < 0, dN
2
Proof of Proposition 1. Game 1 is easily solvable by backward induction. Stage 3: If Π = π(q) − L 0, the government does not subsidize the SOE, and the pay-offs to the government and the SOE manager are, respectively, UG = B(L) + π(q) − L and US = w ∗ − e(q). If Π = π(q) − L < 0 and the government subsidizes the SOE, the pay-offs to the government and the SOE manager are, respectively, UG = B(L) − θ · S and US = w ∗ − e(q). If Π = π(q) − L < 0 and the government does not subsidize the SOE, the pay-offs to the government and the SOE manager are, respectively, 0 and −e(q). Stage 2: If Π = π(q) − L 0, the SOE manager solves maxq : US = w ∗ − e(q), s.t. B(L) + π(q) − L π ∗ and we will get qH = {q | π(q) = L + (π ∗ − B(L))}. If Π = π − L < 0 and the government subsidizes the SOE, the SOE manager will choose maxq : US = w ∗ − e(q) s.t. B(L) − θ · S π ∗ , S = |π(q) − L|. If Π = π − L < 0 and the government does not subsidize the SOE, the SOE manager will choose q = 0. Stage 1: Given the strategy of the manager, the government has the SOE bear the policy burden, L. 2 Proof of Corollary. From the proof of Lemma 1, we know that π(q) = P (q + (N − 1)q ∗ ) · q − K − e(q) is increasing in q for q q ∗ . Since B(L) > L, θ ∈ (0, 1), from Proposition 1, we have: (1) When π ∗ B(L), π(qH ) = π ∗ + L − B(L) < π ∗ , ∴ qH < q ∗ . (2) When π ∗ < B(L), π ∗ − π(qL ) = π ∗ − L + θ1 [B(L) − π ∗ ] ( θ1 − 1)[B(L) − π ∗ ] > 0, ∴ qL < q ∗ .
2
Proof of Corollaries 2, 3 and 4. These results are obvious from Lemma 1, Proposition 1 and Corollary 1.
2
Proof of Proposition 2. The proof is similar to that of Proposition 1, and the only thing to be pointed out is that, when Π = π − L < 0 and the government provides a subsidy, the private enterprise manager will solve the optimization problem of maxq,S : UP = π(q) − L + S, s.t. B(L) − θ · S 0, π(q) < L and UP π ∗ . 2 References Anon, 1998. Symposium on soft budget constraint and transition economies. Journal of Comparative Economics 26 (1), 9–116. Boycko, M., Sheleifer, A., Vishny, R., 1996. A theory of privatization. Economic Journal 106, 309–319. Dewatripont, M., Maskin, E., 1995. Credit and efficiency in centralized and decentralized economies. Review of Economic Studies 62 (4), 541–555. Grossman, S., Hart, O., 1986. The costs and benefits of ownership: A theory of vertical and lateral integration. Journal of Political Economy 94 (4), 691–719. Kornai, J., 1980. Economics of Shortage. Amsterdam, North-Holland. Kornai, J., Maskin, E., Roland, G., 2003. Understanding the soft budget constraint. Journal of Economic Literature 41 (4), 1095–1136. Li, D., 1992. Public ownership as the cause of a soft budget constraint. Mimeo, Harvard University.
102
J.Y. Lin, Z. Li / Journal of Comparative Economics 36 (2008) 90–102
Lin, J.Y., 2003. Development strategy and economic convergence. Economic Development and Cultural Change 51 (2), 277–308. Lin, J.Y., Tan, G., 1999. Policy burdens, accountability and soft budget constraint. American Economic Review 89 (2), 426–431. Lin, J.Y., Cai, F., Li, Z., 1996. China’s Miracle: Development Strategies and Economic Reform. Chinese Univ. of Hong Kong Press. Lin, J.Y., Cai, F., Li, Z., 1998. Competition, policy burdens, and state-owned enterprise reform. American Economic Review 88 (2), 422–427. Lin, J.Y., Cai, F., Li, Z., 2000. Perfect Information and the SOE Reforms in China. Chinese Univ. of Hong Kong Press. Novshek, W., 1980. Cournot equilibrium with free entry. Review of Economic Studies XLVII, 473–486. Qian, Y., Roland, G., 1998. Federalism and soft budget constraint. American Economic Review 88 (5), 1143–1162. Röller, L., Zhang, Z., 2005. Bundling and social and private goods and the soft budget constraints. Journal of Comparative Economics 33 (1), 47–58. Schmidt, K., 1996. The cost and benefit of privatization: An incomplete contract approach. Journal of Law, Economics and Organization 12, 1–24. Segal, I.R., 1998. Monopoly and soft budget constraint. RAND Journal of Economics 29 (3), 596–609. World Bank, 1996. World Development Report 1996: From Plan to Market. Oxford Univ. Press, Oxford. World Bank, 2002. Transition: The First Ten Years, Analysis and Lessons for Eastern Europe and the Former Soviet Union. World Bank, Washington, DC. Zhang, Z., Lu, Y., 2003. Facilitating the SOE reforms. Economics and Finance Forum 1 (in Chinese).