Chapter

11 Credit Market Imperfections and Output Response in Previously Centrally Planned Economies

Editor(s):
Timothy Lane, D. Folkerts-Landau, and Gerard Caprio
Published Date:
June 1994
Share
  • ShareShare
Show Summary Details
Author(s)
Guillermo A. Calvo and Fabrizio Coricelli1 

Introduction

The reform programs launched in Central and Eastern Europe and in the former Soviet Union have been accompanied by a sharp, and partly unexpected, fall in output (Chart 1). Moreover, three years into the reform experience, output remains rather flat.

Chart 1.Real GDP and GDP Indices

(Percentage changes over previous year)

Abstracting from the possibility—not so remote (Berg and Sachs, 1992)—that the contraction of output was simply a statistical artifact, two main types of explanation for this phenomenon have been advanced. One set of explanations relies on standard static models of aggregate demand and aggregate supply and ascribes the fall in output to exogenous shocks. Prominent among them, the external trade shock (the demise of the Council for Mutual Economic Assistance (CMEA)) and the fall in household demand. Another—possibly complementary—set of explanations emphasizes the more systemic nature of the collapse of output. Specifically, the decline in output may have resulted from a change in regime consisting in a shift from a system of organization and coordination of production and exchanges based on central intervention or government tutelage, to a decentralized system. Within this broad view we focus on the effects of credit policy in a system in which the role of credit markets has changed radically. In fact, most previously centrally planned economies began their reforms lacking credit markets, namely, lacking the information, trust, institutions, and rules, which, either codified in laws or in common practice, define credit markets. In the pre-reform regime banks mainly played the role of accounting firms. The information capital necessary for the functioning of credit markets and the same banking skills were absent in the pre-reform regime. Moreover, private credit markets did not exist in highly centralized economies. Besides legal restrictions, the lack of credible bankruptcy procedures was a major obstacle for the development of such private markets. Countries—like Hungary and Poland—that had undertaken partial reforms and introduced some degree of decentralization in economic decisions developed a primitive system of private credit markets—interenterprise credit—and a skeleton of a commercial banking system. However, even in these countries the nature of credit markets was peculiar because of the presence of de facto full insurance from the central bank and the lack of a system of enforceable private contracts.

The reforms of 1990–91 denoted a sudden break with the past. Credit policy was sharply tightened. The automatic extension of credit by the central bank was terminated. The full insurance guaranteed by the central bank as a lender of last resort was eliminated. The financial condition of enterprises suddenly became a determinant of their output and wage payments. The central channels of organization and coordination of exchanges disappeared. Firms had to carry out their transactions in the market, at prices that they freely set. Money became a fundamental instrument for transactions also within the enterprise sector, in contrast with the old regime where the only relevant monetary exchanges were associated with the household sector.

This paper explores whether such conditions in credit markets may help explain the output performance observed in the initial stages of reform in previously centrally planned economies. The paper extends our previous work, which was based on a simple dynamic framework of firms constrained by liquidity. It examines the response of credit markets and enterprise liquidity to the initial shock and presents a simple framework for analyzing both the impact effect of the credit contraction and the dynamic effects. The main innovation of the paper is the discussion of the circular process of exchanges within the firm sector and the role of money and interenterprise credit in such a process.

Impact Effect of the Credit Contraction

A central element of the reform programs in previously centrally planned economies has been the liberalization of prices and the increase of key administrative prices like energy. As a result, the price level jumped at the beginning of the reform programs (Table 1). With the declared objective of avoiding that the initial price jump would translate into persistent inflation, credit policy was sharply tightened at the beginning of the programs, as shown in the credit targets displayed in Table 1. Hungary stands out as the exception, as the initial “price shock” was much softer, consistent with a more gradual approach to price liberalization.

Table 1.Credit and Money in the Enterprise Sector After Reforms(Real stocks deflated by producer prices; in percent)
Credit to EnterprisesEnterprise

Money
Inflation
ActualTargetActualTarget
Bulgaria
(1991.I/1990IV)−75.1−61.0−71.7277.0167.0
(1991.IV/1990.IV)−67.0−61.0338.0234.0
Former Czechoslovakia
(1991.I/1990.IV)−28.3−11.4−34.439.525.0
(1991.IV/1990.IV)−18.4−8.10.758.730.0
Hungary1
(1990.I/1989.IV)−12.0−6.915.0
(1990.IV/1989.IV)−7.0−7.024.830.020.0
Poland
(1990.I/1989.IV)−49.0−10.0−38.2129.475.0
(1990.IV/1989.IV)13.025.0−15.0193.094.0
Romania
(1991.I/1990.IV)−16.031.0
(1991.IV/1990.IV)−39.0−45.0227.0120.0
Source: Authors’ calculations on data provided by the authorities of the various countries.

Quarterly data on Hungary are affected by seasonality. Gérard Belanger has pointed out to us that seasonal factors account for a difference of 6-8 percent between enterprise deposits in the fourth and first quarters (with the latter being lower). The annual change reported may thus be more relevant.

Source: Authors’ calculations on data provided by the authorities of the various countries.

Quarterly data on Hungary are affected by seasonality. Gérard Belanger has pointed out to us that seasonal factors account for a difference of 6-8 percent between enterprise deposits in the fourth and first quarters (with the latter being lower). The annual change reported may thus be more relevant.

In countries like Bulgaria, Romania, and Russia that started the reforms with a perceived sizable “monetary overhang,” the combination of stable nominal credit/money and the price jump was supposed to eliminate the overhang. However, ex post, the association between the initial fall in real credit/money aggregates and ex ante estimates of monetary overhang was rather weak, as suggested by the sharp contraction in real credit even in countries like the former Czechoslovakia and Poland (Table 1).

Therefore, irrespective of their initial conditions, all countries—except Hungary—experienced a sharp fall in real credit to enterprises and in enterprise real monetary balances. Could such a contraction in real credit account, at least partially, for the decline in output that occurred in the initial stages of reform programs? In previous papers we have argued that the “credit view” may explain part of the initial contraction in output, and it may also help explain some apparent puzzles, such as the behavior of wages. In this section we summarize the main elements of a simple conceptual framework underpinning the role of credit markets in the output decline in previously centrally planned economies, focusing on the impact effects of a credit contraction. The model tries to capture some key institutional aspects of these economies, relating both to the underdevelopment of credit markets and to the structure of control of state enterprises based on a powerful role for workers.

Simple Model of Liquidity-Constrained Firms

To focus on the role of credit, we assume the economy is populated by competitive, identical firms that produce a final good using intermediate inputs and labor. Both final and intermediate goods are tradable. Thus, with constant international prices, their domestic price is determined by the exchange rate. To simplify the analysis further, we assume a system of fixed exchange rates. Consequently, prices are exogenous. To produce final goods, firms must have a stock of inputs as production begins, and inputs must be paid in cash. The liquidity constraint is captured by assuming that firms face a liquidity-in-advance constraint:

where x stands for inputs of intermediate goods and m is real cash balances in terms of inputs. To account for the powerful role of the workers in state-owned enterprises, we assume that firms are managed by a workers’ council. The council cares about the welfare of all the (identical) workers belonging to the coalition attached to the firm. Since our focus is on the initial impact of a liquidity squeeze, we assume that the size of the coalition, or membership, is fixed. Therefore, the council maximizes the expected utility of the representative member.2

To rule out dynamics of the labor market that would derail us from the main focus of the paper, we assume that there are no unemployment benefits. As a result, the firm employs all its members, and thus employment becomes a fixed factor (its size is normalized to be equal to one). The production function can thus be written as f(x), and f is assumed to satisfy the usual neoclassical conditions. Thus, if firms are liquidity constrained, condition (1) holds with equality, and the output of final goods can be written as f(m). Letting ω denote total labor income, and assuming an exogenous constant exchange rate, firms can accumulate liquidity over time according to:

where p stands for the price of output in terms of inputs. As noted before, we assume that both goods are tradable and thus the relative price is exogenous.

Condition (2) highlights the fundamental role of wages in the liquidity accumulation of firms.

Let us consider the case in which households have perfect access to international capital markets. This justifies taking the (constant) world interest rate as the relevant discount factor for households. Thus, the workers’ council chooses the path of labor income that maximizes the present discounted value of labor income, that is,

subject to the liquidity accumulation equation (2) and the initial stock of liquidity m0. The linear specification of the workers’ council objective function implies that the above problem is a “bang-bang” type solution (Shell, 1967). Let be such that it satisfies the following condition:

If m0 = m, it is optimal to set mt = m0 for all t. However, if m0 < m , and thus the economy is in a liquidity crunch situation, then the optimal choice for the workers is to transfer to the firm an amount of money equivalent to mm0. Thus, the firm would borrow from its workers and circumvent the liquidity constraint. This situation can be defined as one in which credit markets work perfectly, as liquidity is transferred from the sector that has access to international credit markets—the workers-households—to the sector that is excluded from them—the firms. By contrast, credit market segmentation can be captured in a simple manner by assuming that wages cannot fall below a well-defined lower bound.

With credit market segmentation, if the firm operates initially in the liquidity-crunch region, the optimal choice would be to set wages at the minimum acceptable level until real monetary balances have attained their optimal steady-state level. The initial liquidity crunch may be determined either by a devaluation of the exchange rate—keeping the relative price p unchanged—or by a contraction of nominal balances. The liquidity crunch inevitably determines a contraction of output on impact. This takes place even when the relative price of inputs does not increase and real wages decline, that is, there is no “supply shock” in the conventional meaning.

In sum, when facing a liquidity crunch firms “borrow from their workers” to achieve, at the maximum speed, the optimal level of real monetary balances. However, the segmentation of credit markets implies that output declines on impact.

This simple model explains the puzzling behavior of wages during the early stages of reform, namely, that wages were set well below the ceilings established by the programs. Indeed, such a phenomenon is particularly puzzling, given the powerful role of workers in the decision making of enterprises.

There is a simple economic intuition behind these results. When the level of liquidity is below full capacity, the following inequality holds:3pf’ (m)–1 > r. The inequality states that the marginal return to liquidity used to purchase inputs pf’ (m)–1 is greater than the marginal return to liquidity left in the hands of the workers r. Thus, by lending to firms, workers get a rate of return larger than r.

The version of the model discussed above adopts a series of simplifying assumptions that do not crucially affect the main results. Before pointing to some natural extensions of the model, it is worth noting that the above model focused on the case of fixed exchange rates and tradable inputs and output. With flexible exchange rates it is less obvious how the real stock of liquidity can fall below full capacity. Indeed, the contraction of nominal liquidity would lead to an appreciation of the exchange rate, which could in turn maintain the real stock of liquidity at its full capacity. To generate the same result discussed in the text, the model has to be modified in one of two ways: (1) assuming some form of price-setting rigidity and/or (2) assuming an asymmetric distribution of liquidity before the program started. In both cases a contraction in liquidity would lead to output effects. Rigidities in price setting may arise because of the presence of a significant share of input prices that is administered, or because of anachronistic cost-plus pricing in sectors not exposed to foreign competition. The asymmetric distribution of liquidity implies that the increase in the price level will determine liquidity constraints for liquidity-poor firms, whereas liquidity-rich firms would still be characterized by excess liquidity. To avoid the output decline, a reallocation of liquidity across firms would be necessary. Thus, the effects of a credit crunch are likely to be at work also with flexible exchange rates, as long as there is segmentation in credit markets.

The main results of the model do not depend on the specific form of credit market segmentation, which was introduced by superimposing a lower bound on wages. In a richer—and perhaps more realistic—model the limits on enterprise borrowing from workers can arise from the fact that workers-households may face liquidity constraints as well. In addition, the main thrust of the above model carries over to a more realistic setting in which firms’ liquidity can be altered by bank credit (see Calvo and Coricelli (1992) for such extensions).4

Finally, a more realistic model would incorporate inventories and interenterprise arrears. Indeed, the cash constraint may be loosened by the presence of these two elements. However, both the downloading of input-inventories and the accumulation of arrears in payments to suppliers, although effective at the level of the single firm, may have negative aggregate effects on output as they create difficulties for firms that produce inventories and for suppliers of inputs. Nevertheless, the introduction of inventories and arrears would not alter qualitatively the basic result relating to the impact effect on output of a liquidity contraction. Whether the use of inventories and the accumulation of arrears may eliminate the liquidity crunch is an empirical question. The definition of liquidity of the firm is modified, but not the nature of the constraint (1). Nevertheless, as discussed in the next section, the presence of interenterprise arrears may have important dynamic effects.

The model implies that output converges toward its full-capacity level as firms accumulate liquidity over time. The actual experience of previously centrally planned economies seems to indicate that the speed of recovery is rather low. Even in countries like Poland that began to grow during 1992, the level of output remains depressed (Chart 2). There are several possible channels of low-output persistence, some of which are explored in the next section. The simple model discussed above obviously neglects the presence of exogenous shocks, like the demise of the CMEA, which may depress output. Although empirically relevant, these shocks do not alter the validity of the theoretical model. In fact, the CMEA shock itself could be seen as a case of trade implosion deriving from the destruction of a system of centralized credit and payment. In our model this can be characterized as a tightening of the liquidity constraint determined by a contraction of foreign credit.

Chart 2.Industrial Output Indices

Moreover, there are other channels accounting for a slower recovery that can be easily incorporated in the simple model exposed above. They relate for instance to the horizon of the firms, endogenous inflation, and its effects on money demand. These modifications, while altering the quantitative implications of the model in terms of speed of adjustment, do not change the main thrust of the simple model.

In contrast with these “incremental” changes of the simple model, we discuss in the next section the possibility of a qualitative change in the properties of the dynamic adjustment implied by the simple model. In particular, we develop a model with multiple equilibria, with each equilibrium distinguished by different levels of output and liquidity. In this type of model the issue is not simply the speed of adjustment, as countries may get stuck in the initial state of depressed output.

Low-Output Persistence

This section presents three extensions of the basic model discussed above, which help to explain low-output persistence.

Short Enterprise Horizons

A salient characteristic of most transformation processes is the willingness to restructure and privatize a large portion of the originally socialized sector. In general, this type of expectation should have an effect on the firms’ incentives. Thus, for example, if the firm were to be privatized at time T, the sum in expression (3) would have to run from 0 to T, which lowers incentives for liquidity accumulation and depresses output. More interestingly, if privatization time, T, is unknown, and the probability of its occurring in the “next instant,” if it has not occurred before, is δ (a constant), then one could argue that the optimal strategy for the enterprise is to maximize a sum like (3), where now the discount factor is (1 − δ) / (1 + r) (which corresponds to Yaari (1965) in discrete time). Consequently, the steady-state first-order condition (4) now becomes

Therefore, under privatization, risk output will be permanently lower than the social optimum (where condition (4) holds), even if the system converges quickly to its steady state.

The insight from this model is straightforward. Either privatize quickly, or make a credible announcement that privatization is unlikely to happen.5

Endogenous Inflation

Our simple model in the previous section assumes fixed exchange rates. However, it is easy to extend it to cases in which, for instance, the rate of devaluation is constant. Let the rate of devaluation be denoted by e. Then one can prove that steady-state condition (4) becomes

Given exogeneity of relative prices, the domestic rate of inflation is also equal to . Thus, condition (4”) implies that the higher the rate of inflation, the smaller is the demand for cash by enterprises (a perfectly conventional result). Besides, since output is given by f(m), it follows that higher inflation is associated with smaller steady-state output.

Let us now consider an economy that depends on seigniorage to finance (part of) its budget deficit. In such a case the rate of expansion of cash is a function of the price level: the higher the latter, the higher will be the former. As is well known (see, for example, Calvo (1992)), such money supply endogeneity may give rise to equilibria with different inflation rates. In other words, the same set of “fundamentals” may be associated with low or with high inflation, depending on the state of expectations. Since, as noted above, the level of output is inversely related to inflation, this extension of the model illustrates the possibility that the economy locks into a “bad” equilibrium with low output and high inflation.

There is more than one way in which an economy could lock into a bad equilibrium. One possibility is that, before the start of the transformation program, inflation was high. Thus, it would be perfectly normal for people to expect that it will take a while for inflation to decrease—possibly forcing the monetary authority to validate expectations, undermining the program’s credibility, and lengthening even more the period of high inflation/low output.

Another possibility, which is of even greater relevance for previously centrally planned economies, is that the bad equilibrium is a consequence of the initial liquidity crunch. The latter increases nominal interest rates, which could, in turn, be interpreted by individuals as a signal of higher future inflation. Higher inflationary expectations lower the demand for money, which, as argued above, may induce the monetary authority to validate the high-inflation equilibrium.

The solution implied by the model is deceivingly simple: just push the economy to the high-liquidity equilibrium. However, effective implementation of such a solution is not easy. For example, a natural candidate would seem to be a one-step increase in money supply. However, for this policy to be effective, it is necessary for people to believe that it will result in lower inflation—an unlikely outcome given that the public observes money supply increasing. In fact, people may infer that the central bank is conducting “business as usual” by once again heavily relying on the inflation tax. Therefore, the demand for money may not increase—it may actually decrease—making high steady-state inflation even harder to overcome. In addition, in the period that the policy is implemented, the inflation rate may exhibit a very large spike (because money supply suffers a sizable jump, while money demand is constant or declines)—making a mockery of the stabilization plan.

Alternatively, the government could launch a propaganda campaign aimed at lowering inflation expectations. This campaign could announce, say, that the currency will be pegged to the U.S. dollar. However, this policy may backfire. First, it will unravel quickly if the public is not persuaded about the effectiveness of the stabilization program. The demand for money will remain low, which will force the government to resort to a large infusion of money to finance the budget deficit—thus eventually leading to a balance of payments crisis. Second, if to avoid a balance of payments crisis the exchange rate is allowed to float, the low-inflation/high-output equilibrium can be achieved only if prices and nominal wages fall by, possibly, a substantial amount. Otherwise, without the above-mentioned dangerous one-step increase in nominal money, real monetary balances will not be able to increase toward the “good” equilibrium. For this to work out, however, public sector prices and wages must also fall by, in principle, the same proportional rate. In fact, to enhance the credibility of the shift toward the good equilibrium, public sector prices and wages should take the lead. If, as is unlikely, there are no frictions and all prices and wages fall by the necessary amount, success would be granted. However, if private sector prices and wages fall by less than is necessary, the relative position of government will be impaired. The fiscal deficit is likely to increase, the best workers in the public sector will be attracted to the private sector, and those that remain may shirk more and be more willing to go on strike. All of which may push the economy back to the bad equilibrium.6

Interenterprise Arrears

A salient characteristic of transformation programs has been the rapid development of interenterprise credit channels (see Begg and Portes (1992), Calvo and Coricelli (1992), Ickes and Ryterman (1992), Clifton and Khan (1993), Daianu (1993), and Rostowski (1993)). However, interenterprise credit is large also in market economies (see Rostowski (1993) and Begg and Portes (1992)). Therefore, the growth of interenterprise credit in previously centrally planned economies should, in principle, be viewed as a welcome development. Unfortunately, however, the rapid growth of that market has been accompanied by the emergence and, occasionally, the ballooning of interenterprise arrears, that is, involuntary interenterprise credit. The latter is in principle undesirable, since it very much occupies the place that shirking and stealing take in regular market economies.

To capture interenterprise arrears, let us assume that firms can acquire their inputs by paying with cash or by falling into arrears. Let θ indicate the share that is, in equilibrium, paid in cash. Taking the extreme case in which arrears are expected never to be paid back, the competitive nominal price of inputs and outputs (recall that inputs and outputs are fully tradable and their foreign currency price is unity whereas, in the simplest model, the exchange rate was set equal to unity) must be 1/θ.

To put a natural break on arrears, we will assume that they are costly, and that the present discounted value of the cost (in terms of input) of additional arrears is proportional to the latter, where the factor of proportionality is denoted by κ ξ1. Hence, focusing exclusively on firms that sell to other firms, the analog of cash accumulation equation (2), where now m is expressed in terms of input becomes

New arrears per unit of time in terms of output (or input) are equal to x − m, which rationalizes the third term in equation (5). The first term, in turn, is net revenue in terms of input bought in the free market at home. To prove it, notice that nominal gross revenue is θPf(m) where, as indicated above, P = 1/θ = nominal price of output and input in the free market at home. Thus, θ remains a factor in equation (5) because it is the relative price of output in the firm per unit of time to real cash balances in terms of input bought at home at time t, mt. The presence of θ in equation (5) has nothing to do with the relative price of output with respect to input since, by previous considerations, the latter is always equal to unity. Finally, the factor θtt+1 accounts for the change in the price of input between periods t and t + 1.

Given the linearity of the model, solutions will be, as a general rule, of the bang-bang variety. Therefore, to ensure well-defined solutions in which arrears take place but repudiation is not total, we will assume that θ cannot fall below some θ, such that 1 ξ θ ξ 0. Notice that, by equation (5), optimal xt will be chosen to maximize the right-hand side of that equation. Thus, at an interior maximum, we have

Moreover, the no-accumulation-of-arrears corner solution obtains if

whereas the maximum-accumulation-of-arrears corner solution holds if

Let us now turn our attention to optimality conditions for ω To simplify, we will focus exclusively on steady states, where θt = θ = constant through time. As in the previous sections, we will assume that the bounds on ω are wide enough for optimal ω to be interior to its feasibility region. If 0 θη μ 1, then increasing the wage bill by 1 more unit at time t implies, by equation (5), θ less units of real monetary balances at the beginning of period t + 1, which, if used to pay input in period t + 1, saves kθ in cost of arrears. Thus, the wage bill in period t + 1 could increase by k units. In view of expression (3), if the perturbation is taken at the optimal point, a necessary condition for an interior optimum will be k = 1 + r—a highly unlikely situation. Therefore, as expected, solutions will in general be at θ = 1 or θ = θ.

Consider the case in which θ = 1, that is, no accumulation of arrears. This case is similar to that in the previous section. Therefore, at optimum, f’(m) = 1 + r, which, combined with equation (7), implies that k ξ 1 + r.

Finally, if θ = θ then, by equations (3) and (5), at optimum,

When the latter is combined with inequality (8), it implies that k ≤ 1 + r.

Consequently, we have shown that no arrears will take place if k > 1 + r, whereas maximum accumulation of arrears will occur if the inequality is reversed.

Therefore, disregarding the possibility of an interior solution, the economy could exhibit a minimum liquidity equilibrium (that is, θ = θ), MLE, and a no-arrears equilibrium (that is, θ = 1), NAE, which corresponds to the equilibrium discussed in the previous section. Notice that, because when θ = 1, we have f’(x) = 1 + r, equation (9) implies that, as expected, output is lower at an MLE than at an NAE.

Given k and r (1 + r ≠ k), one of the two equilibria will emerge. However, these variables are, to some extent, endogenous. First, as noted above, the effective interest rate (that is, r + δ) is affected by the firm’s horizon. If the latter is short, the effective interest rate will be large, and a bad MLE is likely to materialize. Second, if as above, the model allows for inflation (nominal interest rate i ξ r), then one can show that all the above conditions hold with i substituting for r. Thus, when inflation increases beyond the critical point at which 1 + i = k, the economy will suddenly shift from the NAE to the MLE. This reinforces the possibility of a bad equilibrium taking hold, because the shift from an NAE to an MLE represents a catastrophic decline in the demand for money (which comes on top of the decline emphasized at the beginning of this section).

In addition, the marginal direct cost of arrears, k, is likely to be affected by the amount of arrears. The larger the latter is, the smaller may be the marginal direct cost of arrears, k. Thus, an initial liquidity crunch, for example, may force firms into arrears. As the latter accumulate, firms realize that they are not the only ones in trouble and, therefore, that penalties cannot be as severe as when just a few firms fall into financial difficulties. As a result, κ may decline so much that the bad MLE takes hold. In other words, an initial credit crunch could generate a situation in which interenterprise arrears accumulate without bound and output is permanently lower than in the no-arrears situation.

Obviously, much of the previous policy discussion applies to the present case. The present model, however, allows us to address the question of whether the clearing of arrears could be an effective policy to drive the economy to the good NAE. The model suggests two reasons for being skeptical. First, firms’ horizons could be very short and, thus, κ would have to be very large to induce firms to avoid falling into arrears. Besides, large ks may be hard to implement because bankruptcy is either nonexistent or bankruptcy procedures are exceedingly slow. Second, κ may depend not only on past arrears but also on expected arrears. Thus, the MLE could be quickly regenerated if firms expect that the previous arrears situation will re-emerge in the future.

Our skepticism is supported by experience in countries like Romania, where the elimination of arrears has quickly been followed by a buildup of new ones. Therefore, our discussion suggests that the solution in such cases must be found in massive privatization and effective bankruptcy regulations.

A Closer Look at the Evidence

The above discussion has suggested that (1) a contraction in real liquidity of enterprises is likely to lead on impact to a fall in output; (2) real wages tend to decline on impact. The speed of recovery of wages depends on the characteristics of the workers’ objective function, on the workers’ horizon, and on the timing of wage payments.7 (3) The initial contraction in output may be sustained by a short horizon of the firms, by a fall in money demand, and by a “demonetization” of enterprise transactions, associated with the blossoming of interenterprise arrears.

The model in previous sections implies a relatively homogeneous impact effect of the credit contraction. By contrast, the dynamic adjustment is likely to be highly heterogeneous. Indeed, even starting from similar initial conditions, countries may settle down on very different equilibria. The channels affecting convergence toward a specific equilibrium are diverse and are likely to play different roles across countries and over time within countries.

Limitations on data availability severely constrain the empirical test of these implications. In particular, we will not attempt to carry out an in-depth analysis of the dynamic adjustment in the different countries. Nevertheless, there are a host of stylized facts and more detailed evidence on some of the countries that lend support to the simple models developed in the previous sections. These stylized facts depict countries that seem to share the initial contractionary effects of credit tightening, whereas their dynamic adjustment after the initial shock is significantly differentiated. Although the experience with stabilization is too fresh to permit clear distinction of the successes from the failures, there seems to be a marked differentiation between one group of countries—the former Czechoslovakia, Hungary, and Poland—which, despite large initial costs, appears to be on a path toward the good equilibrium, and another group of countries—Romania and Russia—which has not yet succeeded in stabilizing inflation nor shown signs of recovery of economic activity. Available information, however, does not allow us to analyze in detail the dynamic adjustment of the various countries.

The following sections provide evidence that the difficulties of stabilizing these economies could have partly resulted from an excessive initial tightening of liquidity conditions. Once the economy has settled into an equilibrium with low liquidity and low output, relaxing monetary policy tends to be ineffective to reactivate production. Thus, monetary expansion tends to be reflected in higher inflation, giving rise to the stop-go cycle observed in countries like Russia and Romania. As discussed below, the same nature of the response of the economy to the initial liquidity contraction tends to exert pressure for a relaxation of policies.

In addition to factors related to credit markets, uncertainty on property rights and on the timing and characteristics of changes in ownership and control of enterprises plays a fundamental role in affecting the supply response during the transition.

Initial Credit Crunch

Table 1 shows that—except for Hungary—all Central and East European countries were characterized by a sharp drop in real credit to enterprises at the beginning of their reform programs. This drop was partly planned, as indicated by the credit targets. However, the actual decline was much larger than planned, especially for the former Czechoslovakia and Poland. The larger-than-expected decline in real credit was determined by a larger-than-anticipated jump in the price level. With Poland as an outlier, there seems to be a correlation between the pre-liberalization ratio of broad money to GDP—a very rough proxy for ex ante monetary overhang—and the size of the initial price level jump (Chart 3). However, the price jump has invariably been higher than anticipated, and the size of the initial adjustment seems to have been tightly connected with the increase in administered prices and the devaluation of the exchange rate.

Chart 3.Initial Price Level Jump

1 Inflation during the first month after price liberalization (in percent).

Accordingly, the behavior of the price level seems to have reflected some rigidity in the price system, either because of the large initial devaluation of the exchange rate followed by a fixed peg—as in Poland and the former Czechoslovakia—or because of the large increase in administered prices. In fact, in some countries both factors played a role (see also the discussion in Bruno (1992)). Thus, the assumption of an initial exogenous contraction in real bank credit appears plausible.8

For Poland, data availability—especially cross-section data—permits a more detailed empirical analysis. At a macroeconomic level the association between the contraction in credit and the drop in output is striking. Indeed, the fall in output accompanying the credit crunch was sudden, concentrated in the first two months of 1990. The empirical analysis of the output-credit link cannot be based on a time series analysis, as January 1990 coincided with a change in regime for the Polish economy. However, a closer inquiry into the role of credit factors in the initial collapse of output could be based on cross-section information.

The cross-section analysis may shed some light on the relation between the initial structure of credit markets inherited from the past and the sectoral behavior in response to the policy changes of January 1990. In particular, the first model developed above implies that the initial distribution of liquidity across sectors is a key determinant of the output behavior following the credit crunch. Indeed, with credit market segmentation, the effect of a credit contraction is likely to have heterogeneous effects across sectors. Specifically, sectors that are more dependent on outside sources of finance, in particular bank credit, are likely to be harder hit, as they cannot easily replace bank credit with alternative sources of financing. An empirical test of our view would in principle require information on the initial liquidity position of firms/sectors, and on the possible sources of liquidity for firms, in particular the stocks of input inventories, whose movements alter the liquidity constraint of firms at a point in time. Unfortunately, we do not have data on the sectoral distribution of liquidity. We are thus forced to use bank credit as a proxy for enterprise liquidity.

Notwithstanding these caveats, the negative association between ex ante credit dependence and the change in output at the beginning of 1990 shown in regression (2) in Table 2 is suggestive of the importance of the inherited credit market structure on output behavior.9 Of course, one would like to go beyond this simple correlation between ex ante credit dependence and initial output performance and try to assess the quantitative impact of the change in the liquidity conditions of enterprises on output. As shown in Table 2 a simple ordinary least-squares (OLS) regression on changes in output and changes in credit indicates a very weak statistical relation between credit and output across sectors. However, one might have expected to find a weak correlation between output and credit, given that the change in bank credit may be a poor proxy of the change in liquidity of enterprises. Indeed, the output effect of the credit contraction is mediated by the adjustment of other variables—for instance, inventories—affecting the liquidity of firms, in a way that makes the change in output at the firm/sector level only slightly related to the change in credit to that firm/ sector.

Table 2.Poland: Regressions on Output and Credit, 1990.1/1989.IV1(Dependent variable: change in real output)
OLSTSLS2
(1)(2)(3)
Constant−0.24−0.19−0.20
(12.89)(6.70)(6.10)
Change in credit30.020.20
(0.71)(2.10)
“Credit dependence”4−0.27
(2.45)
R20.010.07
Observations858585

Figures in parentheses are t-statistics.

Two-stage least-squares. Instruments: constant and credit dependence as defined above.

Log-difference.

Ratio of bank credit to total costs at the end of 1989.

Figures in parentheses are t-statistics.

Two-stage least-squares. Instruments: constant and credit dependence as defined above.

Log-difference.

Ratio of bank credit to total costs at the end of 1989.

This phenomenon may arise, for instance, if we assume that reducing the stock of inventories gives rise to adjustment costs, resulting in output losses, which are not very sensitive to the size of the inventory adjustment. In the limit, these adjustment costs may be fixed, thus totally independent of the size of the change in inventories. In this case, the output loss associated with an inventory reduction induced by a credit crunch is going to be evenly spread across sectors, and thus uncorrelated with the size of the sectoral contraction in credit. At the macroeconomic level, one would still observe a correlation between credit contraction and output decline. However, at the microeconomic level, the direct association between credit and output is lost. Therefore, even without resorting to the presence of price rigidities, the credit crunch could have large output effects through the reduction in inventories.10 In addition, if one assumes price rigidities, the output decline would be determined by the reduction of the demand for inventory goods by firms that deplete their stocks of input inventories (see discussion in Calvo and Coricelli (1992)). Even in this case the cross-section regression would likely generate no significant impact of credit on output, as the effect of credit on output depends also on the distribution of inventories across firms.11

The above discussion suggests that the role of financial constraints can be better detected by analyzing the behavior of inventories in response to the contraction in bank credit. An extended version of our model incorporating inventories would have clear cross-section implications, as firms/sectors suffering a more severe credit squeeze should display a sharper decline in inventories (see discussion in Calvo and Coricelli (1992)).

Table 3 contains the results of several regressions on the behavior of input inventories in the first quarter of 1990 for 85 branches of Polish industry. The macroeconomic importance of the behavior of inventories can be appreciated by noting that during the first quarter of 1990 the stock of input inventories in our sample dropped in real terms by about 30 percent. We focus on input inventories as they are the relevant aggregate for the credit view exposed above. Nevertheless, it is worth noting that credit factors have a large and significant effect on the behavior of finished goods inventories (see also Berg and Blanchard (1992)). We concentrate on nominal values, since a relevant price deflator for input inventories is not available, both for inventories and credit. However, most of the results are robust to the use of real measures of both inventories and credit, and to different price deflators.

Table 3.Poland: Regressions on Inventories and Credit, 1990.1/1989.IV1(Dependent variable: change in input inventories)
OLSTSLS2
(1)(2)(3)(4)
Constant0.640.400.860.36
(16.68)(7.97)(20.19)(5.78)
Change in credit30.130.080.18
(3.25)(5.98)(2.40)
Change in sales30.380.33
(2.26)(4.44)
“Credit dependence”4−0.27
(3.34)
R20.110.380.11

Sample: 85 observations; figures in parentheses are t-statistics.

Instruments: constant, change in sales and credit dependence as defined above.

Log-difference.

Ratio of bank credit to material costs at the end of 1989.

Sample: 85 observations; figures in parentheses are t-statistics.

Instruments: constant, change in sales and credit dependence as defined above.

Log-difference.

Ratio of bank credit to material costs at the end of 1989.

The first regression in Table 3 displays the result of a simple bivariate regression, including only the change in bank credit as an explanatory variable. The coefficient on the credit variable is significant at the 2 percent level, and the point estimate implies that on average about 15 percent of the change in inventories is explained by the change in credit. In the second regression we control for other factors that may affect the change in inventories by introducing the change of sales as an additional explanatory variable. Although slightly reduced, the coefficient on the credit variable remains significant at the 3 percent level.

Moreover, we ran simple OLS regressions using an ex ante measure of credit dependence instead of the change in credit. The sign of the credit dependence variable is negative as expected and significant at the 1 percent level. This result confirms the importance of the ex ante credit dependence on the sectoral behavior, suggesting the empirical relevance of credit market segmentation. In addition, the use of an ex ante measure of credit dependence is immune from problems of endogeneity of the credit variable.

To further account for the possibility of endogeneity of the credit variable we also carry out a two-stage least-squares (TSLS) regression, using the credit dependence index as an instrument for the credit variable. The results of the TSLS estimation essentially confirm those of the OLS regressions.

In sum, all specifications yield a significant and quantitatively important effect of the credit variable on inventory behavior. These results are suggestive of the importance of the credit channel at the beginning of the Polish reform program, although the inventory equations cannot help establish the quantitative effect on output of the credit contraction. Nevertheless, the above analysis suggests that a large proportion of the output behavior that in the output-credit regression is captured by the constant can in fact be associated with the credit channel through the output effect of the inventory contraction. Moreover, the fact that the ex ante credit dependence proves to be consistently significant both in inventory and output equations is suggestive of the important role of credit market factors in the behavior of real variables in the initial stages of reforms in Poland. This is somehow a remarkable result, if one takes into account the degree of noise present in a system like Poland in 1990 undergoing a radical change in economic regime.

Another way, consistent with the first model above, of establishing the importance of the credit view is to analyze the behavior of wages at the outset of the stabilization programs. In the same sectoral sample used for the output and inventory regressions for Poland in 1990, we find that the change in credit has a positive, and statistically significant, impact on the sectoral change in the wage bill (Table 4).12

More generally, the phenomenon of “borrowing from workers” predicted by the first model above seems to have been relevant not only in Poland but also in Bulgaria, the former Czechoslovakia, and Romania at the outset of reforms, as shown by increases in wages below the ceilings imposed in the stabilization programs (see Calvo and Coricelli (1993)).

Table 4.Poland: Regressions on Wages and Credit, 1990.1/1989.IV1(Dependent variable: change in nominal wage bill)
OLS Regressions
(1)(2)
Constant86678890
(4.84)(5.48)
Change in credit20.100.05
(8.07)(3.07)
Change In sales20.01
(4.35)
R20.010.07
Observations8585

Figures in parentheses are t-statistics.

First difference.

Figures in parentheses are t-statistics.

First difference.

Dynamic Response: Inflation, Firms’ Horizon, and Interenterprise Arrears

As shown in Chart 2, despite some signs of recovery in the former Czechoslovakia, Hungary, and especially Poland, three years into the reform programs economic activity remains rather depressed, particularly in industry. The first model discussed above, based on a “representative firm,” implied that the optimal behavior of a firm that maximizes the welfare of its workers would lead to a relatively fast recovery toward steady-state output. It was also stressed that uncertainty on the timing of privatization, by shortening the horizon of the workers (that is, increasing the “effective” discount rate from r to r + δ), could hinder the output recovery. This factor likely played an important role in most previously centrally planned economies, perhaps with the exception of the former Czechoslovakia and Hungary. In Poland, Romania, and Russia, after a short-lived initial period of wage moderation, real producer wages increased significantly.

In Poland, for instance, although during the first months of the program wages were set well below the wage ceilings, by the end of 1990 they overshot the ceilings and stayed above them for most of 1991. Despite falling output and growing unemployment, the share of enterprise value added appropriated by labor increased sharply.13 As a result, profitability fell over time, reducing the source of self-financing for enterprises. This seems to support the view of a shortening of the horizon of firms. Interestingly, in 1992 there was a switch in behavior. Apparently, the fear of continued wage pressure and decapitalization of firms was averted. A possible explanation for the change in wage behavior may be associated with a lengthening of firms’ horizons owing to a reduction of privatization risk (δ in the model of the first part of the previous section). Indeed, the continued delay of the mass privatization program may have signaled a generalized delay in privatization. In addition, and perhaps more important, programs of enterprise restructuring have emphasized the voluntary character of privatization—privatization becoming increasingly a “negotiated” process—and have enhanced the role of the insiders, both managers and workers, in the privatization process. This is likely to have reduced the risk of privatization, either by postponing the eventual date of privatization or by providing the workers a stake in the process. Both factors should induce a lengthening of the workers’ horizon.14

As discussed above, another important channel hindering the recovery of output is related to the persistence of inflation, and the consequent decline in money demand.

Inflation following stabilization varied sharply across countries. The former Czechoslovakia, Hungary, and Poland have been characterized by relatively low rates of inflation in the first two-three years after reforms. By contrast, Romania and Russia have failed to reduce inflation significantly after the initial price level jump. In fact, in Russia inflation has not abated more than one year after price liberalization and is hovering near hyperinflationary levels (Chart 4).

Chart 4.Inflation

1 The arrows indicate the date of price liberalization. In Romania price liberalization took place in stages.

High-inflation countries have also displayed a continuous decline in real monetary balances, while low-inflation countries did not experience such a process of sustained “demonetization.” Chart 5 shows the behavior of money velocity and identifies a clear differentiation between the behavior of Romania and Russia on one side and the former Czechoslovakia, Hungary, and Poland on the other. Velocity jumped in the aftermath of price liberalization in Bulgaria, Poland, Romania, and Russia, while it remained stable in the former Czechoslovakia and in Hungary. In the second year of their reforms, velocity declined significantly in Poland and, to a lesser extent, in the former Czechoslovakia and Hungary, indicating a process of “remonetization.” By contrast, velocity continued to increase in Romania, signaling a continued fall in money demand.

Chart 5.Money Velocities1

1 Velocity is measured as GDP/average broad money (M2).

2 1992 data are up to June.

Even when appended to privatization risk and endogenous money demand, the model of the “representative firm” provides an incomplete account of the dynamic adjustment of countries in transition. Indeed, independent of the incentives at the level of the individual firm, there could be problems in coordinating the recovery to the full-employment output in a system with interdependent firms. As shown in the previous section, when firms interact with each other the good equilibrium with output recovering its full-employment level is only one of the possible equilibria. Bad equilibria with low output and low liquidity can arise. A main force determining this multiplicity of equilibria is associated with interenterprise arrears.

Arrears have grown in some countries from zero to amounts larger than overall bank credit or broad money (Table 5). Moreover, the heterogeneity of the behavior of arrears across countries may offer an important clue for understanding the different macroeconomic situations characterizing the various countries. Interestingly, the countries that we identified above as unsuccessful in stabilizing their economies are those displaying an explosive behavior of arrears.15

Table 5.Interenterprise Arrears1(Ratio of interenterprise arrears to bank credit)
PolandRomaniaHungaryFormer

Czechoslovakia
Russia
1989.IV1.60.20.0
1990.I1.70.10.0
1990.II1.20.20.0
1990.III0.90.20.1
1990. IV1.00.10.20.1
1991.I0.90.50.20.1
1991.II0.90.70.20.2
1991.III0.81.10.20.2
1991.IV1.01.90.20.20.0
1992.I0.90.220.9
1992.II0.91.12.3
1992.III0.90.83
1992.IV

For Poland, the figures refer to interenterprise credit, whereas for the other countries they refer only to arrears, that is. overdue credits.

At the beginning of the year arrears were cleared through the global compensation scheme.

The figure is for July. In July bank credit was injected into the system to clear the arrears.

For Poland, the figures refer to interenterprise credit, whereas for the other countries they refer only to arrears, that is. overdue credits.

At the beginning of the year arrears were cleared through the global compensation scheme.

The figure is for July. In July bank credit was injected into the system to clear the arrears.

Interenterprise Arrears

In the area of interenterprise arrears country experiences were remarkably heterogeneous (Daianu, 1993; Rostowski, 1993). Specifically, arrears have literally exploded in Romania and Russia, whereas they have increased much less in Hungary and Poland. Moreover, after operations to clean up the arrears, both in Romania and Russia, arrears have rapidly grown again (Chart 6).16

Chart 6.Interenterprise Arrears

(PPI deflated)

In the model on interenterprise arrears in the previous section the main determinant of arrears was summarized by the parameter k, a proxy for the marginal cost of running into arrears for the individual firm. It was also suggested that such a cost is a decreasing function of the aggregate size of arrears. If firms cannot coordinate ex ante their behavior, there can be multiple equilibria with different levels of arrears.

If firms attach a positive probability to the government validation of the arrears through money creation, the likelihood of the high arrears equilibrium obviously increases. The cases of Romania and Russia clearly illustrate such a phenomenon of self-fulfilling prophecy. Indeed, in both countries the government response to the explosion of arrears has been a generalized cleanup effected through injection of bank credit. Not surprisingly, arrears have grown rapidly soon after these cleanup operations were implemented, confirming the important role of the expectation of a bailout for the growth of arrears. In particular, as this expectation is probably linked to the size of aggregate arrears, the latter may be one of the channels through which k becomes a decreasing function of aggregate arrears. The time-series behavior of arrears in Poland offers additional evidence on the importance of credibility of the stabilization program in affecting the accumulation of arrears. Indeed, at the beginning of the program, when credibility was probably high, interenterprise credit fell together with bank credit. As the credibility of the program weakened starting in the second half of 1990, interenterprise arrears began to move in the opposite direction to bank credit. The correlation coefficient between interenterprise credit and bank credit, or enterprise money, is indeed negative after the first half of 1990. The change in the behavior of interenterprise arrears mirrors the change in wage behavior discussed above.

In addition to the issue of credibility of the no-bailout stance, institutional factors may affect the perceived individual cost of running into arrears. Ultimately, these institutional factors relate to unclear property rights and the consequent lack of credible bankruptcy threat, which in turn rules out the possibility of enforcing private contracts.

Finally, the possibility for suppliers to switch to different customers is another important factor in affecting the expected cost of running into arrears. Given the highly concentrated structure of domestic markets and the rigid network of relations imposed by the central plan, in the initial stages of reform the flexibility of supplier-customer relations is likely to be associated with opportunities for exports. Hungary and Poland stand out as the countries with a more rapid growth of exports to market economies. This may have implied a high k for these economies, which in turn can account for the moderate increase in interenterprise arrears.

In sum, the multiple equilibrium model of the previous section helps characterize the different experiences across countries as different equilibria, largely associated with different perceived microeconomic costs of running into arrears (the parameter k). However, the model also illustrates certain key features of interenterprise arrears. Specifically, it generates a network, or chain, of arrears in a system in which all firms are viable—have nonnegative profits. This condition is important because it permits a view of interenterprise arrears as a form of stable equilibrium. The presence of loss-making firms would imply that in the chain of arrears a group of firms is subsidized by other firms, an unlikely sustainable phenomenon. Of course we do not deny that arrears reflect also the presence of loss-making firms. However, we claim that this is not the dominant feature of arrears. The mechanics of the chain of arrears, its features, and potential inefficiencies are illustrated graphically in Figure A1 in the annex.

Figure A1.Interenterprise Arrears: A Circular System with Viable Firms

The empirical relevance of the view of arrears as an equilibrium network can be analyzed for Romania, for which micro data on all state-owned industrial enterprises are available for 1992. These data permit analyzing (1) the degree of “circularity” of the process, namely, how important are the debts of firms that have a roughly balanced debt-credit (payables and receivables) position; and (2) the role of loss-making firms in the network of arrears. Finally, we report some suggestive evidence from a small-scale enterprise survey that provides qualitative answers on the features and determinants of arrears in Romania.

One simple measure of the degree of circularity is given by the ratio of total net arrears, defined as the sum of net debt positions of net debtor firms, to total gross arrears. This ratio has hovered between 25 and 30 percent during 1991 and 1992, indicating that most arrears are a substitute for enterprise liquidity. Perhaps even more relevant is the number of firms involved in the network of arrears and the number of firms with significant net debt positions. For instance, in June 1992, out of 1,692 state industrial firms, 1,455 displayed arrears in payments to suppliers, while most net arrears were concentrated in a handful of firms.17 Therefore, the vast majority of Romanian firms have both debts and credits of similar magnitudes, clearly illustrating the phenomenon of chain of arrears, where net arrears are not highly significant.

Regarding the importance of loss-making firms, only 200 firms, or 12 percent of the total number of firms, are loss making, and they do not absorb a disproportionately large share of arrears—in relation to their sales, for instance. Their share in total gross arrears is about 15 percent, while their share in total sales is about 10 percent (September 1992 data). Although loss-making firms are important recipients of net arrears, they do not play a crucial role in the whole chain of arrears.

In addition, results from a small-scale survey (of 30 firms) carried out at the end of 1992 by the World Bank seem to confirm the above discussion. In assessing the main causes of arrears in payments to suppliers, the most important reason is perceived to be the arrears in payments by their own customers. This seems consistent with the concept of a self-sustaining network of interenterprise arrears. The survey also shows that in the sample only between 5 and 25 percent of firms’ purchases are effected in cash. Firms revealed that in the hierarchy for the use of cash, wages are the first priority, whereas payments to suppliers are the last, coming after wages, banks, and taxes. This seems to be consistent with a system in which transactions among enterprises take place with very little use of cash, while cash is mainly used for wage payments. Finally, the survey indicated that at the end of 1992—one year after the cleanup of arrears—almost 80 percent of firms expected a new government bailout.

The same features of arrears just described may also help explain the pressures for government intervention. As noted in Calvo and Frenkel (1991), the interlocking relations induced by arrears hinder the possibility of an efficient selection among good and bad enterprises, and it creates the risk of a domino effect with good firms being dragged into trouble by firms in difficulties. Therefore, it is the far-reaching ramification of this network that seems to have triggered government intervention in countries like Romania and Russia.

Another important implication of the model of arrears in the previous section is that with arrears domestic prices tend to be higher than world market prices and that the price level is directly proportional to the share of arrears in sales. The model provides a possible explanation for the puzzling price jump in producer prices in Russia in 1992, which was much higher than the increase in consumer prices. Moreover, within producer prices, the increase in prices of inputs for production was much higher than the increase in prices of consumer goods. This phenomenon, indeed, could reflect the lack of cash payments for intermediate products, in contrast with the dominance of cash payments for consumer goods.18

Concluding Remarks

The paper supports the conventional view that structural reforms are essential to improve allocation and utilization of resources in previously centrally planned economies. Firms have to be given incentives to operate under long planning horizons, and bankruptcy regulations have to become effective. Unfortunately, however, these reasonable objectives could take an inordinate amount of time. Privatization is a lengthy process, and for bankruptcy regulation to become effective legal courts must not be hampered by other offsetting regulations and the legal profession must learn to navigate in the new waters. In the meantime, thus, palliatives and extra caution in choosing macropolicies is required.

In that respect, the approach adopted in the paper suggests that credit markets should be carefully monitored. A key characteristic of previously centrally planned economies at the beginning of their economic transformation programs is the lack of well-developed credit markets. In particular, the economy strongly depends on the existing, largely official, banking system. Therefore, for example, a cut in bank credit is tantamount to a cut in total credit, implying that a stabilization program aimed at sharply lowering inflation may excessively reduce the amount of credit available to enterprises and, thus, cause an unduly large fall in output. The paper has further argued that the initial credit contraction could put the economy on a path in which output is permanently lower than its potential. Once the economy is locked into the “bad” low-output equilibrium, however, credit expansion may just result in higher inflation, not higher output. Consequently, this fundamental asymmetry would call for ensuring that the credit market is not unduly strained during the first stages of an economic transformation program. Specifically, the vulnerability of the credit market and its dependence on official sources strongly suggests that an effective anti-inflation policy should be accompanied by measures that prevent an unduly large credit contraction when the transformation program starts.

In that connection, it is essential to deactivate the price/wage “engine.” As argued in the paper, short horizons and low default costs are conducive to low-output equilibria. Thus, until privatization and bankruptcy regulations become effective, it would be necessary to give further incentives to enterprises to try to offset those distortions. One possibility that has been experimented with in several recent programs is a ceiling on wages. The effectiveness of this measure, however, is highly dependent on the government coming to a binding agreement with labor (like in the Mexican Pacto)—not a very likely precondition in previously centrally planned economies. Therefore, wage ceilings may have to be supplemented by other measures that give further incentives to enterprises to stay below the ceiling. For example, credit itself may be a function of whether a firm complies with the wage ceiling regulation.19 Moreover, managers’ salaries could be subject to cuts, or managers themselves could be subject to dismissal, if their firms fail to comply.

The agenda for future research is still long and crowded. First, we need to know more about the actual mechanics of credit markets in previously centrally planned economies, which can only be obtained through thorough data gathering. In this respect, the above models could serve as a guide on what to look for. For example, the models suggest that private and public firms must behave differently about liquidity accumulation, since private firms likely display a lower rate of effective time preference. Second, we should explore models that incorporate oligopolistic and key political economy aspects of previously centrally planned economies. Third, a more systematic testing of models should be carried out.

Comment

Rudiger Dornbusch

With the devastating decline in output of transition economies to be explained, the Calvo-Coricelli paper offers a welcome hypothesis and even evidence of a channel so far largely disregarded. Their claim that the compression of real working capital must account for some of this output decline is altogether plausible—if working capital is productive, then, other things being equal, a sharp reduction in the real value of these assets must have real effects. How much of an effect real credit contraction has, and in what exact ways it influences output, is another issue, although even here the authors make progress by offering empirical evidence.

The emphasis on credit as an important aspect of the financial transmission mechanism is entirely fashionable.1 In U.S. macroeconomics the question of money versus credit continues to be hotly debated. Interestingly, the working capital perspective has not been an important part of that discussion. Rather the link has been made between financing and investment rather than between finance and production. It would be interesting therefore to investigate further whether the authors’ working capital hypothesis could be identified in the context of advanced economies. Whatever the findings might be, Calvo and Coricelli are right to emphasize that the price shock of Eastern Europe offers a natural experiment. The price liberalization, without commensurate expansion of money and credit aggregates, leads to an exogenous real contraction. In segmented credit markets in which firms cannot replenish their finance fully by borrowing, real credit contraction limits their ability to finance inputs and hence production. In principle we therefore can unscramble how money and credit work. The more segmented and imperfect credit markets are, the more strongly their effect should come to the forefront.

In the model discussed here there is a lag between the application of inputs and the resulting output. Let a and b be the unit labor and material coefficients of a competitive firm, P, W, M, and i the price, wage, materials cost, and nominal interest rate. With competitive capital markets and a one-period input-output lag for materials and for a share of wages, discounted price would equal marginal (average) cost. Output at the level of the firm is indeterminate.

Tight credit in this situation of perfect capital markets means higher real interest rates. Accordingly the relation between future prices and current costs will change. But there are no implications for output. To look at output, the demand response to changes in the relative prices of credit-intensive goods would have to come into play.

But with a credit constraint Kt one moves away from a pricing equation to a limitation of output. The amount of output Q that a firm can produce, supposing the expected price is right, becomes constrained by cash flow.

or, letting lowercase variables represent measurements in wage units (m = M/W) so that one has

A price rise in period t will raise current revenue but it will also reduce the purchasing power of revenue and working capital in terms of current wages and materials. A rise in nominal interest rates, given credit, further intensifies the financing constraint. The framework immediately suggests that not running up interest arrears or postponing the payment of wages are effective ways of lessening the bite of the financing constraint. Equally apparent is the manager’s call for increased credit as a means of sustaining production.

The Calvo-Coricelli model does not offer a sharp distinction between the level of credit in the economy and the distribution of that credit among forms. In segmented credit markets, that distinction is all important. With an uneven distribution, some forms might be altogether unconstrained whereas for others the lack of credit becomes the effective limit to production. The manner in which credit allocations are decided would therefore have a first-order impact on production. Likewise, it would be useful to make a distinction among firms in their credit intensity. Production with long gestation periods would be more vulnerable than industries where production and sales more nearly coincide. Building a tanker, for example, is not the same as fixing shows.

Similarly, the role of inventories deserves attention. In the short run firms can liquidate inventories to obtain cash for production. That is an important consideration because, in fixed-price, random supply, planned economies, inventories at each stage of production were high. Accordingly, the immediate credit squeeze may not have been so important because of the possibility of liquidating inventories. Of course, firms in the aggregate cannot acquire an increase in their credit position except as a result of an economy-wide increase in credit or a reduction in households’ balances. Most of the inventory liquidation therefore might result, together with arrears, in a redistribution of credit among firms and industries.

Two further points deserve attention. In the decontrol experience one would expect strong immediate effects and more dampened persistent ones. Capital markets are segregated, but there are always means to lessen credit dependence and force implicit borrowing in any number of ways, including nonpayment of credit or interest. All these channels will tend to come into play over time and hence lessen the role of tight credit.

In concluding, one would like to see the role of shocks to working capital in settings other than transformation economies. Such possibilities do exist. For example, in 1990 Peru increased oil prices thirtyfold, leading to a vast blip in the price level and a corresponding reduction in real money and credit. Output instantly dropped sharply. Other examples no doubt exist. In hyperinflation experiences of the 1920s, as reported by Bresciani-Turroni (1937), there is discussion of Kapitalaufzehrung as a result of extreme inflation. One can think of this as the reduction in the real value of working capital—a reduction not reflecting voluntary increases in the velocity of credit, but an involuntary reduction in the purchasing power of working capital. The Calvo-Coricelli hypothesis may thus have interesting implications for the interpretation of very different experiences, including the output drop in the context of extreme, unrepressed inflation.

ANNEX

The Mechanics of the Chain of Arrears

Figure A1 represents a system of three interconnected firms, labeled F1, F2, and F3, which are forced to conduct “cash” transactions with households. Solid lines connecting firms indicate flows of goods or services, while dotted lines indicate flows of “cash.” Numbers directly above the lines indicate values (“peso” sums). Therefore, in this system F3 sells final goods to households valued in 20 “pesos” in exchange for the same amount in “cash.” F1 is the other firm dealing with households. It hires labor for “cash”—the wage bill being 10 “pesos.” In addition, the production process requires F2 to supply inputs to F3, valued at 120 “pesos,” F3 to Fl, valued at 100 “pesos,” and F1 to F2, valued at 110 “pesos.” However, contrary to the firms-households trades, the “cash” counterpart of interenterprise trade is significantly less than the value of goods supplied. Thus, F2 pays nothing to F1, and F3 pays only 10 “pesos” to F2 on a transaction valued at 120 “pesos.” More interestingly, F3 sells inputs to F1 by 100 “pesos” and, in addition, lends 10 “pesos” in “cash” to F1. This allows F1 to hire workers.

The network depicted in Figure A1 could have come to life after the initial price/wage jump. The jump gave rise to a “cash” shortage, and thus “cash” ended up being exclusively utilized for transactions between firms and households, with interenterprise transactions financed by falling into arrears. However, no firm is a net borrower—no firm is implicitly subsidized by borrowing more than it could repay if its debtors did not default. This is so because each firm increases its arrears by 110 “pesos” each period but, at the same time, its customers also do so by the same amount. The system is feasible and could, in principle, continue operating indefinitely. However, gross arrears grow without limit. Notice that the system as a whole uses less cash and, thus, conventional statistics will exhibit a fall in real liquidity. Arrears may be a poor substitute for “cash” when firms are interconnected as in Figure A1. The equilibrium there, for example, depends on F3 being willing to lend 10 “pesos” to F1, even though the latter will not be able to repay the loan, and even though the loan is not directly connected with the goods sold to F1. In contrast, if F3 keeps the 10 “pesos” for itself, F1 will not be able to hire labor, which implies that there will be no inputs for F2. F2 output will collapse, and F3 will not be able to produce. Thus output everywhere grinds to a complete halt. Leaving aside the externalities stressed in the model of interenterprise arrears, these arrears are not per se detrimental to output. In fact, their presence may have cushioned the impact of the initial liquidity crunch. However, the major drawback of interenterprise arrears is that they slow the process of adjustment. In the case portrayed in Figure A1, for example, input prices are independent of “cash” supply. The same real equilibrium could be attained if F3 billed F1 for 200, F1 billed F2 for 220, and F2 billed F3 for 220 “pesos.” Thus, interenterprise trades without the discipline of “cash” payments result in weak price signals. Another implication of Figure A1 is that once the network of interenterprise arrears is established, the health of a given firm becomes intimately dependent on the health of the system as a whole, which again interferes with adjustment (see Calvo and Frenkel (1991)).

References

    BeggDavid andRichardPortesEnterprise Debt and Economic Transformation: Financial Restructuring in Central and Eastern Europe Centre for Economic Policy Research Discussion Paper No. 695 (1992).

    • Search Google Scholar
    • Export Citation

    BergAndrew andOlivierJean Blanchard“Stabilization and Transition: Poland 1990-1991,”International Monetary Fund Seminar Series No. 1992-07 (April1992).

    • Search Google Scholar
    • Export Citation

    BergAndrew andJeffreySachs“Structural Adjustment and International Trade in Eastern Europe: The Case of Poland,”Economic PolicyNo. 14 (April1992) pp. 11773.

    • Search Google Scholar
    • Export Citation

    BrunoMichael“Stabilization and Reform in Eastern Europe: A Preliminary Evaluation,”IMF Working Paper WP 92/30 (May1992).

    CalvoCuillermoA.“Are High Interest Rates Effective for Stopping High Inflation? Some Skeptical Notes,”World Bank Economic ReviewVol. 6January1992) pp. 5569.

    • Search Google Scholar
    • Export Citation

    CalvoCuillermoA. andFabrizioCoricelli“Stabilizing a Previously Centrally Planned Economy: Poland 1990,”Economic PolicyNo. 14 (April1992) pp. 176226.

    • Search Google Scholar
    • Export Citation

    CalvoCuillermoA.“Output Collapse in Eastern Europe: The Role of Credit,”Staff PapersInternational Monetary FundVol. 40 (March1993) pp. 3252.

    • Search Google Scholar
    • Export Citation

    CalvoGuillennoA. andJacobA. Frenkel“From Centrally Planned to Market Economies: The Road from CPE to PCPE,”Staff PapersInternational Monetary FundVol. 38 (June1991) pp. 26899.

    • Search Google Scholar
    • Export Citation

    CliftonEricV. andMohsinS. Khan“Interenterprise Arrears in Transforming Economies: The Case of Romania,”Staff PapersInternational Monetary FundVol. 40 (September1993) pp. 68096.

    • Search Google Scholar
    • Export Citation

    CoricelliFabrizio andTimothyD. Lane“Wage Controls During the Transition from Central Planning to a Market Economy,”World Bank Research ObserverVol. 8 (July1993) pp. 195210.

    • Search Google Scholar
    • Export Citation

    DaianuDaniel“Arrears in Post-Command Economies: Thoughts from a Romanian Perspective” (unpublished; 1993).

    FrydmanRoman andAndrzejRapaczynski“Privatization in Eastern Europe: Is the State Withering Away?”Finance and DevelopmentVol. 30 (June1993) pp. 1013.

    • Search Google Scholar
    • Export Citation

    IckesBarryW. andRandiRyterman“The Interenterprise Arrears Crisis in Russia,”Post-Soviet AffairsVol. 8 (October-December1992) pp. 33161.

    • Search Google Scholar
    • Export Citation

    LiptonDavid andJeffreyD. Sachs“Remaining Steps to Achieve a Market-Based Monetary System” (unpublished; 1992).

    RostowskiJacek“The Interenterprise Debt Explosion in the Former Soviet Union: Causes, Consequences, Cures,”Communist Economies and Economic TransformationVol. 5No. 2 (1993) pp. 13159.

    • Search Google Scholar
    • Export Citation

    ShellKarl“Optimal Programs of Capital Accumulation for an Economy in Which There Is Exogenous Technical Change,”in Essays on the Theory of Optimal Economic Growthed. by KarlShell (Cambridge, Massachusetts: MIT Press1967).

    • Search Google Scholar
    • Export Citation

    WardBenjamin“The Firm in Illyria: Market Syndicalism,”American Economic ReviewVol. 48 (September1958) pp. 56689.

    YaariMenahemE.“Uncertain Lifetime, Life Insurance, and the Theory of the Consumer,”Review of Economic StudiesVol. 32 (April1965) pp. 13750.

    • Search Google Scholar
    • Export Citation

    BernankeBen“Credit in the Macroeconomy,”Federal Reserve Bank of New York Quarterly ReviewVol. 18 (Spring1993) pp. 5070.

    Bresciani-TurroniCostantinoThe Economics of Inflation: A Study of Currency Depreciation in Post-War Germany (London: Allen & Unwin1937).

    • Search Google Scholar
    • Export Citation
1The authors would like to thank Willem Buiter, Rudiger Dornbusch, and Hugo Hopenhayn for very useful comments.
2The outcome of the choice of the workers’ council is optimal for the coalition of workers and resembles the outcome of efficient contracts in capitalist firms, with the difference being that in the case we discuss, the income-sharing rule is trivial, as all the rents of the firm are appropriated by the workers. This setup contrasts sharply with the classical version of the labor-managed firm, which assumes that firms maximize income per worker (Ward, 1958). That model implies the firm will fire workers even when there is no unemployment compensation.
3As a consequence of the assumption of strict concavity of the production function.
4In (he case of binding credit ceilings, the model is basically unchanged. When credit ceilings are not binding, the behavior of the stock of credit is demand-determined and depends on the path of interest rates on bank loans. However, in the relevant case in which interest rates are initially high and are expected to decline over time, the main results of the simple cash-in-advance model still hold.
5A more complete analysis will, of course, have to take into account the inefficiencies associated with quick privatization as well as those associated with keeping enterprises in government hands.
6Another interesting case of endogenous inflation is when prices are set at time t before knowing money supply at f. Under those conditions, a government that is averse to unemployment may end up validating more than one inflation level. A solution for this problem is sometimes sought in incomes policies. For a discussion of wage policy in the context of Poland, see Calvo and Coricelli (1992).
7In the linear specification of the workers’ utility function discussed above, wages will stay at their minimum level to maximize the speed of recovery of the steady-state level of output. Calvo and Coricelli (1992) show that with a concave utility function, real wages grow along the path to the steady state. Finally, it can be shown that if wages enter the liquidity in advance constraint of the firm, optimal paths exist along which wages grow at the same rate as liquidity, and output remains constant.
8An additional, indirect indication that the credit contraction at the beginning of the reform programs was not determined by an exogenous decline in the financing needs of enterprises is suggested by the fact that interenterprise credit (or arrears) increased in relation to bank credit in every country (see discussion below).
9Credit dependence is measured by the ratio of bank credit to total costs before the implementation of the stabilization program in the last quarter of 1989.
10An analogous phenomenon would apply to interenterprise arrears, as discussed above. Indeed, a credit contraction would lead to an increase in arrears, which produce output losses independently of the size of the arrears at the individual firm. Unfortunately, we do not possess at this time sufficient information to carry out an econometric analysis of interenterprise arrears.
11For instance, firms rich in inventory will have the opportunity of replacing a large proportion of bank credit and maintain output, while firms with poor inventory would suffer output losses even if they suffered, relative to the other firms, a smaller proportional decline in bank credit.
12However, the results are highly sensitive to the specification of the regression. In particular, in a log-linear specification the coefficient on credit was not significant. Nevertheless, the linear specification arises naturally from the budget constraint of the firm.
13In industry, the share of the wage bill in gross value added increased from about 18 percent at the beginning of 1990 to 45 percent at the end of 1991.
14This pattern seems to have been the one prevailing in Hungary since 1990 (Coricelli and Lane, 1993). Moreover, the importance of privatization risk in affecting wage behavior seems to be confirmed by the experience of the former Czechoslovakia. Indeed, in the former Czechoslovakia the initial drop in real wages has not been followed by any significant wage pressure. The very rapid process of privatization and the direct stake given to every citizen, through the purchase of vouchers, may have been important factors in determining wage moderation. For a discussion of privatization and the role of insiders in Eastern Europe, see Frydman and Rapaczynski (1993).
15The former Czechoslovakia is somehow an outlier. Indeed, arrears grew very rapidly in a context of low inflation. This suggests—contrary to what is argued in Rostowski (1993)—that credibility of the stabilization program is not the only factor affecting interenterprise arrears. Nevertheless, in relation to bank credit, arrears in the former Czechoslovakia remain much lower than in Romania and Russia.
16In Romania with the so-called global compensation scheme at the end of 1991 (see Clifton and Khan (1993)), and in Russia in July 1992.
17Indeed, 80 percent of net arrears were concentrated in 100 firms, accounting for 27 percent of sales, while the largest 10 debtors accounted for 50 percent of net arrears.
18Lipton and Sachs (1992) have argued that one of the main reasons for high producer prices in Russia was associated with the lack of use of cash for enterprises’ purchases. Their explanation is different from ours. They argue that there was an excessive supply of credit—of noncash rubles—for enterprise transactions.
19Alternatively, enterprises could be taxed on wages in excess of the ceiling. However, Poland in 1990 suggests that this is not a very effective policy (see Calvo and Coricelli (1992)), because when workers are set to “cannibalize” a firm, they may not really much care about its net worth, since they are not the owners. In contrast, a credit crunch has immediate consequences.
1See Bemanke (1993) for a fine survey.

    Other Resources Citing This Publication