1. The first year of the program ^3/

In January 1990, the Polish Government initiated a comprehensive program whose goal was both to stabilize a rapidly deteriorating macroeconomic situation and to begin to lay the groundwork for the establishment of a market economy. As far as stabilization measures were concerned, the budget balance was programmed to shift from a deficit position equivalent to 7 percent of GDP in 1989 to approximate balance in 1990. This was to be achieved through strict control of government expenditure on goods and services, large reductions in subsidies, and improvements in revenue collection. Monetary policy was programmed to be relatively tight in the early part of the year, but allowance was made for an expected remonetization in later months as inflation came down. Interest rates were increased sharply and were targeted to remain positive in real terms, in contrast to the sharply negative real rates in 1989. The exchange rate was devalued and then fixed at a rate of 9,500 zlotys to the dollar, a level that was believed would hold for a period of three months or so. ^4/ Another crucial element of the program was the introduction of a tax-based incomes policy, which penalized enterprises strongly for wage increases above a norm that was a specified fraction of the monthly increase in consumer prices. For the year as a whole, the incomes policy was designed to effect a reduction in real wages of about 30 percent.

As far as liberalization measures were cohcerned, food subsidies had already been eliminated at the end of 1989, while in January 1990 most remaining controls on the prices of goods and services were removed. ^5/ As for the prices that continued to be subject to-administrative oversight, the most important were those of coal and electricity which, notwithstanding very sharp increases (of between 300 and 600 percent) at the beginning of the program, remained significantly below world levels. Other measures included a significant liberalization of foreign trade, and currency convertibility for transactions in goods and nonfactor services. Finally, the government indicated its intention to develop plans for privatizing state enterprises; and initiated plans for breaking up monopolies; clarifying bankruptcy legislation; and modernizing the banking system.

The program was successful in many areas. Inflation, after a (larger-than-expected) surge in the early months of 1990, was reduced to about 3 percent per month by midyear. The incomes policy was successful in holding down the growth in wages and, in fact, wage growth in the first half of the year was lower than allowed for under the terms of the policy, with the consequent drop in real wages being greater than programmed. The budgetary situation also showed greater improvement than anticipated, aided by a much improved financial position of enterprises and banks. ^6/ The budget swung into sharp surplus in the first half of 1990 (equivalent to about 4 percent of annual GDP), compared to an anticipated position of small deficit. Again contrary to expectations, convertible-currency exports surged and imports fell sharply, with both the current account surplus and increase in reserves being much larger than programmed. ^7/ Consistent with the balance of payments performance, monetary policy proved to be somewhat tighter than envisaged at the outset of the program, with both money and credit to nongovernment falling sharply in real terms. Finally, in addition to the program’s unexpected financial performance, significant progress was also made on the microeconomic side, particularly as regards the elimination of shortages, privatization of small firms, and to some extent demonopolization.

While the financial performance of the program—as measured by the turnaround in the budget and the balance of payments—exceeded expectations, developments in the real economy—particularly the behavior of output—were cause for serious concern. In January 1990, sold production of socialized industry fell by about 30 percent, and remained relatively flat until the middle of the year. ^8/ Although the decline in real GDP is likely to have been much smaller, owing to a relatively better performance of other sectors of the economy such as services, construction, agriculture, and private industry, developments in these sectors would have been insufficient to alter the conclusion that at the beginning of 1990, Poland entered a depression of historical proportions. The unemployment rate, which had been effectively zero at the end of 1989, rose to 3 percent by midyear.

In part as a result of these developments, fiscal, monetary, and incomes policies were all eased in the second part of the year. ^9/ To some extent, developments in the second half of the year reflected these changes, with the monthly inflation rate rising from a low of 1.8 percent in August to an average 5.5 percent monthly in the last quarter, ^10/ and output of socialized industry recovering 4 percent from its level in the first half. ^11/, ^12/ Despite the recovery in the second half of 1990, production of socialized industry (for the year as a whole), was down 25 percent on the level of 1989, and real GDP was down by about 12 percent. Finally, as far as the program of systemic change was concerned, especially privatization of the large state enterprises, development of competitive labor and housing markets, banking system reform, and reform of the tax system, the bulk of the work remained to be carried out in subsequent years.

2. The second year of the program

Reflecting both the easing of financial policies in the latter part of 1990 and large administered price increases at the beginning of 1991, inflation rose to nearly 13 percent in January, with the “underlying” rate estimated to have been in excess of 4 percent. To get inflation back under control, monetary policy was tightened at the beginning of 1991 by increasing the NBP’s refinance rate; incomes policy was firmly targeted to reduce the level of real wages which had risen sharply in the latter part of 1990; and calls for a devaluation of the zloty were resisted despite the very sharp real appreciation that had taken place in the latter part of 1990. ^13/ The discretionary elements of fiscal policy were programmed to remain tight in 1991, although a significant worsening of the budget (to a position of small deficit) was envisaged. This reflected mainly a reduction in corporate income tax revenues, itself the result of many factors, including: (i) the nonrecurrence of capital gains on inventories and enterprise foreign currency deposits, which would contribute to a reduction in state enterprise profits; and (ii) the narrowing of interest rate spreads which would reduce bank profits. On the expenditure side, significant increases in appropriations for unemployment compensation and other safety nets would also serve to increase the deficit.

On the microeconomic side, the program for 1991 embodied a commitment to reduce state ownership in the economy by about 15 percent. This was to be achieved both through a program of mass privatization involving the free issue of vouchers to the public, ^14/ and by traditional methods (public offerings, private placements, and joint ventures). There was also a renewed commitment to make bankruptcy a reality for nonviable enterprises in the state sector. Systemic change was to continue on other fronts as well, including strengthening competition policy, liberalizing foreign investment legislation, financial sector reform, and further liberalization of those prices remaining under administrative control (energy, transportation, and alcohol). Based on this program, the authorities believed that output would begin to recover in the second half of the year and that, relative to 1990, GDP would rise by 3-4 percent in real terms, reflecting a continued strong expansion of an increasingly important private sector. This estimate was consistent with a substantial decline (about 60 percent in volume terms) in exports to the CMEA area, and with the shift of CMEA trade to world market pricing which represented a considerable deterioration in Poland’s terms of trade (on the order of 15-20 percent).

In the event, performance in 1991 was much worse than foreseen. While industrial output was expected to be flat in the first half of the year, followed by a recovery in the second half, in fact, sold production of industry (which from 1991 includes the larger private enterprises), actually fell by over 20 percent over the first six months of 1991, and displayed virtually no recovery in the second half. ^15/ For the year as a whole, industrial production is estimated to have been about 12 percent below the 1990 level (which itself was 25 percent below the 1989 level), while GDP in 1991 is believed to have declined by about 7 percent relative to 1990, reflecting the resilience of some other sectors—notably agriculture, private trade, and construction—in the face of the continuing deep depression in industry.

The decline in activity in 1991, and particularly its extent, came as a much greater surprise than the decline in 1990, which many believe to have been a necessary accompaniment to the stabilization-cum-reform measures that were enacted in that year. While a full explanation of the decline is still lacking and is considered in more detail below, among its main proximate causes were clearly a greater-than-foreseen decline in exports to the CMEA area in the wake of disarray in the U.S.S.R., ^16/ and the loss of much of the former GDR market following German unification; and a largely unanticipated shift in domestic demand toward imports. Several factors may have contributed to the latter, including: (i) the continued erosion of Poland’s competitiveness, and a substantial increase in the purchasing power of wages measured in foreign currency terms, consequent on the maintenance of a fixed exchange rate; (ii) an increase in demand for foreign inventory goods (following the considerable destocking of 1990), particularly for new private firms; and (iii) a lagged response to the liberalization of the trade and payments system that had taken place in 1990.

Weaker-than-expected activity had other ramifications. Enterprise profitability fell sharply at the beginning of the year (by about 50 percent) and continued to fall for much of the year. Apart from the decline in sales and the reversal of transitory accounting profits referred to above, the decline in profitability during 1991 may also have been related to a significant increase in unit labor costs—to which both the decline in productivity and a sharp increase in real product wages would have contributed—and to an increase in energy costs associated with the reform of CMEA trading arrangements. ^17/

The declines in output and enterprise profitability had a profound impact on the budget, where revenues were nearly 5 percent of GDP lower than originally envisaged and 11 percent of GDP lower than in 1990. ^18/ Despite some very sharp expenditure cuts, the budget deficit for 1991 reached about 6 1/2 percent of GDP, or 6 percent more than projected under the program. To finance the deficit, credit to government increased sharply and, despite a much smaller than envisaged expansion in credit to nongovernment, the result was considerable overshooting of the program’s inflation target—the through-the-year rate was 60 percent instead of 36 percent as originally envisaged—and a marked deterioration in the external accounts. ^19/

Finally, as concerns the structural reforms, there was a significant scaling back of the mass privatization program, in part because of the recognition that the viability of many firms originally included in the list had been overestimated, but also because of a genuine shift in the direction of policy, towards more active involvement of the state in restructuring the large enterprises (industrial policy), and a reduced priority accorded to transferring ownership to the private sector. With the program of ownership change being pushed back, the anticipated supply response to improved market incentives would be delayed further.

3. The anatomy of the recession

This section looks in detail at the main features of the depression that Polish industry entered at the beginning of 1990. The reasons for focusing on industry, rather than on GDP as a whole, are first that the depression was really concentrated in the industrial sector, so that if we want to explain why GDP declined so much, we should direct our attention to industry, rather than agriculture—which was not much affected by the changes initiated in 1990—or private trade, which was booming for most of the period since 1990. Second, most of the explanations that have been advanced for the decline in output—such as the credit-crunch view, the deficient-demand view, and others—seem mostly relevant for the industrial sector, rather than at the whole of aggregate supply. Finally, one of the issues that we will wish to investigate below is whether, looking at a disaggregated level, there is any significant difference in the behavior of the various subsectors of industry, as would be the case in an economy undergoing structural change, or whether all sectors were similarly affected, as might be the case if a single common shock were driving output.

Chart 2 plots monthly data on output, employment, productivity, and real product wages for the industrial sector for the two-year period beginning in December 1989, i.e., the month prior to the initiation of the Polish stabilization program.

View Full Size

Developments in Total Industry

Citation: IMF Working Papers 1992, 086; 10.5089/9781451850796.001.A001

• Download Figure
• Download figure as PowerPoint slide

1. Structural versus macroeconomic factors

There is no doubt that the process of economic reform in the PCPEs must generate a reallocation of productive resources on a massive scale, and that such a process must already have begun as price and trade liberalization have exposed enterprises to domestic and foreign competition, and enterprise budget constraints have been “hardened” as automatic bank financing has dried up. But the true extent of structural change in the productive core of the Polish economy remains a matter of debate. Although public enterprises are no longer mere executors of the central plan in this transitional period, it is safe to say that they have not yet become fully profit-maximizing entities in the same way that their private counterparts are in a market economy. ^29/

But why would structural change cause a fall in output rather than merely a shift in the composition of output? One reason is that an asymmetric response may be expected between those sectors that become unprofitable and must contract, and those that find increased opportunities for profitable expansion. On the one hand, enterprises might be forced to immediately stop loss-making production by limitations on their financial resources or by a complete disappearance of demand; on the other hand, a sizable expansion of production might require more time, particularly if it required significant increases in specialized factors of production. Moreover, public enterprises are unlikely to engage in substantial fixed investments in the highly uncertain environment of the transition phase, in which ownership changes are contemplated and the potential new owners may wish to significantly alter business strategy.

In order to gauge the extent to which output developments have reflected a process of structural change rather than a response to some common macroeconomic disturbances, we apply the technique of principal components to sectoral industrial production and employment data. Principal components is a technique that finds (mutually orthogonal). linear combinations of a set of variables accounting for the largest proportion of the variability (sum of the variances) of those series. The first principal component accounts for the highest fraction of the variability of the series, the second for the second highest, and so forth. ^30/ Naturally, there are as many principal components as there are series.

As far as the data on industrial production are concerned, the results in Table 1 show that the first few principal components account for a smaller proportion of the variability of the series in the Polish case than in the case of the benchmark country (the United States). The results are based on a sample of industrial production data from about twenty subsectors of industry in the case of Poland. The interpretation of the figures is that the higher is the fraction of the variability of the change in industrial production that is accounted for by each principal component, the smaller is the extent of structural change that has taken place within the industrial sector. In the limit, if one hundred percent of the variability of the change in output were explained by a single factor, output of all sectors would be proportional, and this would be interpreted as indicating that a single macroeconomic variable were responsible for most of the output decline and that virtually no structural change had taken place. ³¹ To have a benchmark for comparison, we use data on production of subsectors of industry in the United States over the same period (1990-91), with a roughly similar level of disaggregation. ^32/

Table 1.

Poland: Cumulative Fraction of Variance Explained by Principal Components

Table 1.

Poland: Cumulative Fraction of Variance Explained by Principal Components

	Log Differences of Output
Principal component	Poland	U.S.A.
1	0.51135	0.74052
2	0.65373	0.84525
3	0.72740	0.90481
4	0.77971	0.93052
5	0.82843	0.95339
6	0.87247	0.96773
7	0.90860	0.97680
8	0.92673	0.98337
	Log Differences of Employment
Principal component	Poland	U.S.A.
1	0.36421	0.30797
2	0.48550	0.47682
3	0.58305	0.58091
4	0.65851	0.67464
5	0.72038	0.74665
6	0.77353	0.80331
7	0.82040	0.85184
8	0.86051	0.89047

View Table

Table 1.

Poland: Cumulative Fraction of Variance Explained by Principal Components

	Log Differences of Output
Principal component	Poland	U.S.A.
1	0.51135	0.74052
2	0.65373	0.84525
3	0.72740	0.90481
4	0.77971	0.93052
5	0.82843	0.95339
6	0.87247	0.96773
7	0.90860	0.97680
8	0.92673	0.98337
	Log Differences of Employment
Principal component	Poland	U.S.A.
1	0.36421	0.30797
2	0.48550	0.47682
3	0.58305	0.58091
4	0.65851	0.67464
5	0.72038	0.74665
6	0.77353	0.80331
7	0.82040	0.85184
8	0.86051	0.89047

Thus, the results of the first panel of Table 1 indicate that the most important common macroeconomic factors are less capable of accounting for the behavior of output across the different industrial subsectors in Poland than in the United States. ^33/ In isolation, this result would suggest that within the industrial sector, more structural change has taken place in Poland than takes place in the course of a (hopefully) standard business cycle in the benchmark country. Given the very large relative price changes that the Polish economy was subjected to, this result is at one level perhaps not very surprising. However, if one looks for additional supporting evidence for the structural change hypothesis, the case for structural change becomes harder to make. For example, if resources were really being shifted towards sectors with lower comparative costs, one would expect this to be reflected in the employment data. Unfortunately, as the second panel in Table 1 shows, these data do not suggest that employment across the various subsectors has been particularly variable, as one would expect if structural change were really taking place. ^34/ Moreover, results of a multi-country study that attempted to decompose output developments into country-specific (that is, macroeconomic) and sector-specific factors found that the former was capable of explaining virtually all of the variability of output in all of the PCPEs considered (including Poland), thereby again casting some doubt on the structural change hypothesis. ^35/ Finally, our attempts to correlate output changes with available measures of comparative costs do not suggest that the direction of resource shifts has been towards activities in which Poland has some comparative advantage.

This last finding is reported in Table 2, which examines the extent to which output changes can be predicted on the basis of estimates of comparative advantage. For this purpose we make use of two studies that compute domestic resource costs (DRCs) for Poland prior to price and trade liberalization. ^36/ Domestic resource costs are defined as the ratio of value added at domestic prices to value added at international prices (that is, valuing products and inputs at their estimated domestic-currency equivalent to the world price). The lower the DRC, the more competitive a sector will be after price and trade liberalization because the price of its output will rise more than the cost of its inputs.

Table 2.

Poland: Output and Employment Changes and Comparative Advantage ^1/

^1/Gross section regressions of overall change in output and prices on a constant and: (1) Inverse of domestic resource cost as calculated by de la Calle (1990) and (2) inverse of domestic resource cost as computed by Hughes and Hare (1992) Table 2. Figures in parenthesis correspond to t-statistics (in absolute value).

Table 2.

Poland: Output and Employment Changes and Comparative Advantage ^1/

		DRC (1)	DRC (2)
Output changes		-0.17 (1.50)	-0.19 (0.77)
	R2	0.13	0.04
	DW	2.35	1.95
Employment changes		-0.11 (1.43)	-0.22 (0.16)
	R2	0.12	0.11
	DW	2.04	1.71

^1/Gross section regressions of overall change in output and prices on a constant and: (1) Inverse of domestic resource cost as calculated by de la Calle (1990) and (2) inverse of domestic resource cost as computed by Hughes and Hare (1992) Table 2. Figures in parenthesis correspond to t-statistics (in absolute value).

View Table

Table 2.

Poland: Output and Employment Changes and Comparative Advantage ^1/

		DRC (1)	DRC (2)
Output changes		-0.17 (1.50)	-0.19 (0.77)
	R2	0.13	0.04
	DW	2.35	1.95
Employment changes		-0.11 (1.43)	-0.22 (0.16)
	R2	0.12	0.11
	DW	2.04	1.71

^1/Gross section regressions of overall change in output and prices on a constant and: (1) Inverse of domestic resource cost as calculated by de la Calle (1990) and (2) inverse of domestic resource cost as computed by Hughes and Hare (1992) Table 2. Figures in parenthesis correspond to t-statistics (in absolute value).

It should be noted at the outset that the estimates of DRCs are highly tentative because of the large number of judgmental assumptions necessary to value products at world prices, including problems arising from the existence of nontraded goods, substantial quality differences, and peculiar exchange rate arrangements in the CMEA area. ^37/ At a more fundamental level, DRCs are based on the assumption of a fixed coefficient technology and do not consider the possibility of different elasticities of substitution across sectors. As is well known, energy was generally the most underpriced input in many of the PCPEs. This implies that, according to the DRC methodology, sectors with high energy intensity will appear to be the most uncompetitive. However, it might be that these sectors employed highly energy-intensive technologies because they had the highest elasticities of substitution between energy and other inputs (labor and capital). If this were the case, the DRCs would be misleading as far as providing a ranking of comparative advantage. Notwithstanding all these caveats, the DRC estimates are the only available measures of comparative advantage, and for this reason they cannot be overlooked.

To determine the extent to which output changes are correlated with this measure of comparative advantage, we ran regressions of the cumulative output change since the beginning of the reform program on (transformed) measures of DRC. The transformation is necessary because DRCs are not a monotonic measure: They are negative for those sectors that produce negative value added at world prices; when the DRC is positive, a higher DRC indicates a less competitive sector. To overcome this non-monotonicity, we applied the transformation suggested by Hare and Hughes (1992): for positive DRCs, we use the inverse of the DRC while for negative DRCs we use the ratio of value added at international prices to the domestic price.

The results presented in Table 2 do not provide much support for the structural change hypothesis since there is little evidence on the basis of the regressions that output is moving in the direction dictated by the available estimates of comparative advantage. ^38/ As explained above, however, the construction of the DRC measures is subject to numerous pitfalls and the results may partly reflect the imperfections associated with this measure of competitiveness.

2. Supply and demand shifts

The issue of whether demand shocks or supply shocks have been predominant in accounting for the output decline is an important element in our understanding of developments in the early stages of reform in Poland. While it seems obvious that both types of shock have been present, it is not clear which has exerted the largest influence. The tight financial policies pursued in order to ensure domestic stabilization and a satisfactory balance of payments position on the one hand, and the exogenous component of the drop in export demand from other former CMEA countries on the other, certainly had a restraining effect on aggregate demand. But the increase in a number of administered prices—notably for energy—and liquidity restrictions affecting enterprises is also likely to have had a negative impact on aggregate supply. ^39/

It is important to note that the demand-shock versus supply-shock question is not entirely the same as the structural-change versus macroeconomic-recession question. Although business cycle fluctuations have traditionally been associated with aggregate demand shocks, and supply shocks have frequently been taken to indicate a reallocation of resources, this interpretation may not always be valid, particularly in the circumstances of Poland. For example, a drop in demand may be the result of increased import competition, which reflects only structural change. Or a contraction in supply may be caused by a credit squeeze on enterprises, clearly a macroeconomic shock.

a. Price-output correlations

The correlation between price and output changes is, in principle, a good summary indicator of the predominance of supply or demand shifts. In any given market, a positive correlation between price and output changes would indicate that demand shifts were relatively more important than supply shifts. Similarly, the sign of the correlation between price and output changes in a cross-section of industrial sectors would indicate whether demand or supply shifts have been predominant in the economy.

The (cross-sectional) quarterly price-output correlations suggest that demand shocks were relatively more important in the first halves of both 1990 and 1991 (when the correlations are positive), while supply shocks predominated in the second halves of these two years (Table 3). One interpretation may be that fiscal, monetary, and incomes policies were all very tight in the first half of 1990, but were eased later in that year, and, to some extent, this pattern was repeated again in 1991. An alternative interpretation, and one that receives some support from the regression results reported in the next subsection, is that although supply shocks (in the form of large administered energy price increases) were indeed prevalent at the beginning of both 1990 and 1991 (and would, other things being equal, result in a negative correlation between price and output, contrary to what is actually observed in the data), the relative inelasticity of demand for various industrial goods implies that the impact of such supply shocks on the correlation between price and output will be relatively small. ^40/

Table 3.

Poland: Correlation Between Quarterly Price and Output Changes

Table 3.

Poland: Correlation Between Quarterly Price and Output Changes

	Industrial Output Growth	Industrial Price Increase	Y-P Correlation
1990:1	-0.27137	0.80140	0.08567
1990:2	-0.01364	0.05810	0.20954
1990:3	0.04082	0.07252	-0.56320
1990:4	0.10263	0.10966	-0.29421
1991:1	-0.16723	0.16657	0.30667
1991:2	-0.12610	0.03838	0.08451
1991:3	-0.02291	0.06366	-0.08214
1991:4	0.06180	0.05220	-0.15669

View Table

Table 3.

Poland: Correlation Between Quarterly Price and Output Changes

	Industrial Output Growth	Industrial Price Increase	Y-P Correlation
1990:1	-0.27137	0.80140	0.08567
1990:2	-0.01364	0.05810	0.20954
1990:3	0.04082	0.07252	-0.56320
1990:4	0.10263	0.10966	-0.29421
1991:1	-0.16723	0.16657	0.30667
1991:2	-0.12610	0.03838	0.08451
1991:3	-0.02291	0.06366	-0.08214
1991:4	0.06180	0.05220	-0.15669

b. A simple supply and demand model

Price-output correlations are an imperfect indicator of the extent of supply and demand shifts. For example, as mentioned previously, the sign and magnitude of such correlations are influenced not only by the size of the shocks affecting demand and supply, but also by the price elasticities of the underlying supply and demand functions. To better assess the relative importance of supply and demand shocks, estimation of a simple structural supply-demand model for each of the industrial subsectors would seem to be advisable.

In the contemplated framework, demand in sector i is given by

$\begin{array}{cc}\left(1\right)& \log {\mathrm{D}}_{\mathrm{t}}^{\mathrm{i}}\end{array}={\alpha }_0^{\mathrm{i}}+{\alpha }_0^{\mathrm{i}}\log {\mathrm{P}}_{\mathrm{t}}^{\mathrm{i}}+{\alpha }_2^{\mathrm{i}}\log {\mathrm{Y}}_{\mathrm{t}}$

where Pⁱ denotes a (relative) price index for sector i and Y is a scale variable that proxies aggregate spending (on a monthly basis). ^41/ Supply in sector i is given by

$\begin{array}{cc}\left(2\right)& \log \end{array}{\mathrm{S}}_{\mathrm{t}}^{\mathrm{i}}={\beta }_0^{\mathrm{i}}+{\beta }_1^{\mathrm{i}}\log {\mathrm{P}}_{\mathrm{t}}^{\mathrm{i}}+{\beta }_2^{\mathrm{i}}\log {\mathrm{w}}_{\mathrm{t}}^{\mathrm{i}}+{\beta }_3^{\mathrm{i}}\log {\mathrm{P}}_{\mathrm{t}}^{\mathrm{E}}+{\beta }_4^{\mathrm{i}}\log {\mathrm{C}}_{\mathrm{t}-1}$

where wⁱ is the real wage, P^E is an index of the price of energy input and C is the real stock of bank credit to enterprises. That is, supply is a function of the price of the product and of the cost of the relevant inputs (labor and energy). All of these prices and the stock of credit are deflated by the overall industrial price index. We also incorporate the stock of bank credits to enterprises in real terms in order to proxy for the possible effect of an overall liquidity constraint that might have been relevant during this period (as argued by Calvo and Corricelli (1992)). ^42/

Equations (1) and (2) were estimated using three stage least squares—with the price of the product treated as an endogenous variable—for seven industrial subsectors. As can be seen from Table 4, the results are encouraging, since the vast majority of the coefficients have their theoretically-predicted signs. ^43/ However, the relative price variable in the demand function and the (lagged) product wage variable in the supply function do not perform very well. As far as the former is concerned, the price elasticity of demand is usually estimated to be fairly small, with most of the estimated elasticities falling between 0.1 and 0.3. There are two possible interpretations of this result. One is that the demand functions are in fact highly inelastic. The other interpretation is that the demand equation is not well identified by the set of instrumental variables used in the estimation. As far as the finding that product wages do not seem to have played a substantial role over the estimation period is concerned, this may partly reflect the fact that firms did not believe that the very low real wages at the beginning of the program would be sustained. In this case, real product wages would underestimate the effective cost of labor over the relevant time horizon of firms and, therefore, firms would have been less likely to expand output on the basis of the very low wages at the beginning of the program, particularly if employment changes carried significant costs of adjustment.

Table 4.

Poland: Demand and Supply Estimation Results ^1/

^1/Estimation is by Three Stage Least Squares; sample period is 1989:12-1991:12; numbers in parenthesis below estimated coefficients are t-statistics (in absolute value).

Table 4.

Poland: Demand and Supply Estimation Results ^1/

	Metallurgy	Electro-Engineering	Chemical	Wood & Paper	Light Industry	Food	Mineral Industry
DEMAND
Constant	-1.42 (1.42)	-3.82 (4.87)	-1.15 (4.54)	0.13 (0.23)	-3.49 (6.06)	5.94 (36.41)	-0.45 (0.38)
Relative price	0.04 (0.29)	-0.09 (0.57)	-0.09 (2.23)	-0.42 (3.39)	-0.34 (1.88)	-0.07 (0.69)	-0.16 (1.01)
Aggregate spending	1.29 (5.83)	1.74 (9.95)	1.15 (20.48)	0.90 (7.06)	1.66 (13.37)	0.52 (7.96)	0.91 (4.07)
R-squared	0.60	0.81	0.95	0.74	0.88	0.78	0.43
Durbin-Watson	0.63	0.87	1.54	0.74	1.43	1.66	0.73
SUPPLY
Constant	1.28 (1.08)	3.74 (2.58)	-5.74 (1.13)	-3.36 (0.54)	2.60 (3.85)	3.94 (12.34)	-28.67 (0.16)
Relative price	1.84 (4.00)	4.53 (2.41)	9.51 (1.91)	7.53 (1.21)	4.25 (4.45)	1.75 (3.70)	24.90 (0.19)
Real credit (-1)	0.78 (2.50)	1.70 (1.31)	4.56 (1.85)	4.14 (1.16)	1.12 (2.30)	0.03 (0.17)	9.90 (0.18)
Coal price	-1.10 (4.78)	-2.37 (2.49)	-5.37 (1.98)	-3.61 (1.29)	-1.37 (4.75)	-0.51 (2.68)	12.90 (0.19)
Real wage (-1)	-0.16 (1.10)	-1.85 (1.36)	3.23 (1.71)	-3.02 (1.07)	-0.61 (1.04)	0.2 (1.19)	-3.51 (0.18)
R-squared	0.61	0.48	0.12	0.06	0.41	0.57	0.20
Durbin-Watson	0.83	0.49	0.77	1.26	0.66	1.11	0.57

^1/Estimation is by Three Stage Least Squares; sample period is 1989:12-1991:12; numbers in parenthesis below estimated coefficients are t-statistics (in absolute value).

View Table

Table 4.

Poland: Demand and Supply Estimation Results ^1/

	Metallurgy	Electro-Engineering	Chemical	Wood & Paper	Light Industry	Food	Mineral Industry
DEMAND
Constant	-1.42 (1.42)	-3.82 (4.87)	-1.15 (4.54)	0.13 (0.23)	-3.49 (6.06)	5.94 (36.41)	-0.45 (0.38)
Relative price	0.04 (0.29)	-0.09 (0.57)	-0.09 (2.23)	-0.42 (3.39)	-0.34 (1.88)	-0.07 (0.69)	-0.16 (1.01)
Aggregate spending	1.29 (5.83)	1.74 (9.95)	1.15 (20.48)	0.90 (7.06)	1.66 (13.37)	0.52 (7.96)	0.91 (4.07)
R-squared	0.60	0.81	0.95	0.74	0.88	0.78	0.43
Durbin-Watson	0.63	0.87	1.54	0.74	1.43	1.66	0.73
SUPPLY
Constant	1.28 (1.08)	3.74 (2.58)	-5.74 (1.13)	-3.36 (0.54)	2.60 (3.85)	3.94 (12.34)	-28.67 (0.16)
Relative price	1.84 (4.00)	4.53 (2.41)	9.51 (1.91)	7.53 (1.21)	4.25 (4.45)	1.75 (3.70)	24.90 (0.19)
Real credit (-1)	0.78 (2.50)	1.70 (1.31)	4.56 (1.85)	4.14 (1.16)	1.12 (2.30)	0.03 (0.17)	9.90 (0.18)
Coal price	-1.10 (4.78)	-2.37 (2.49)	-5.37 (1.98)	-3.61 (1.29)	-1.37 (4.75)	-0.51 (2.68)	12.90 (0.19)
Real wage (-1)	-0.16 (1.10)	-1.85 (1.36)	3.23 (1.71)	-3.02 (1.07)	-0.61 (1.04)	0.2 (1.19)	-3.51 (0.18)
R-squared	0.61	0.48	0.12	0.06	0.41	0.57	0.20
Durbin-Watson	0.83	0.49	0.77	1.26	0.66	1.11	0.57

^1/Estimation is by Three Stage Least Squares; sample period is 1989:12-1991:12; numbers in parenthesis below estimated coefficients are t-statistics (in absolute value).

As mentioned previously, and notwithstanding these two shortcomings of the estimation, the model produces on the whole reasonable parameter estimates. As far as the supply regression is concerned, the results suggest that energy price increases were a significant determinant of output developments in five of the seven sectors, while the real credit variable turns out to be significant in three of seven cases.

Using the above estimates, it is possible to decompose the total decline in output into a fraction caused by shifts in variables that affect the supply function and a fraction caused by shifts in variables that affect the demand function. Since the fit of the regressions is not perfect, there is a remainder, which is explained by the random disturbances, and which cannot in general be allocated because the disturbances to the demand and supply equations are not independent. The output change in sector i caused by changes in the exogenous demand-side variables can be computed as

$\begin{array}{cc}\left(3\right)& \mathrm{\Delta log}{\mathrm{Q}}_{\mathrm{DS}}^{\mathrm{i}}=[{\beta }_1^{\mathrm{i}}/({\beta }_1^{\mathrm{i}}-{\alpha }_1^{\mathrm{i}})]\left({\alpha }_2^{\mathrm{i}}\mathrm{\Delta log}\mathrm{Y}\right),\end{array}$

where the operator Δ indicates the cumulative change in the variable. Similarly, the change due to supply-side shifts is

$\begin{array}{cc}\left(4\right)& \mathrm{\Delta log}{\mathrm{Q}}_{\mathrm{SS}}^{\mathrm{i}}=[{\alpha }_1^{\mathrm{i}}/({\alpha }_1^{\mathrm{i}}-{\beta }_1^{\mathrm{i}})]\end{array}[{\beta }_2^{\mathrm{i}}\mathrm{\Delta log}{\mathrm{w}}^{\mathrm{i}}+{\beta }_3^{\mathrm{i}}\mathrm{\Delta log}{\mathrm{P}}^{\mathrm{E}}+{\beta }_4^{\mathrm{i}}\mathrm{\Delta log}\mathrm{C}]\mathrm{.}$

Both the horizontal shifts in supply and demand functions (that is, evaluated at a constant price) as well as the equilibrium change in output due to those shifts (as given in equations (3) and (4)) were computed. We used the median value of the estimated coefficients and we estimated the shifts for the first quarter of the reform program in Poland (first quarter of 1990), for the first half of 1990, and for the full years 1990 and 1991 (relative to levels at the end of 1989 and 1990, respectively). ^44/ The results indicate very large supply shifts in the first quarter of 1990, but more moderate supply shifts over the rest of the period. Because our median estimate of the demand function is essentially completely inelastic, shifts in supply generate increases in price but little decline in output as can be seen from the column identifying the change in output due to supply-side shifts in Table 5. However, as mentioned above, the inelasticity of demand functions may not be a robust finding.

Table 5.

Poland: Decomposition of Output Decline

Table 5.

Poland: Decomposition of Output Decline

	Total change	Supply shift	Supply-side change	Demand shift	Demand-side change	Error terms
1990Q1	-0.247	-0.846	-0.039	-0.251	-0.246	0.040
1990H1	-0.263	-0.736	-9.028	-0.269	-0.264	0.030
1990	-0.199	-0.771	-0.031	-0.197	-0.193	0.024
1991	-0.185	-0.303	-0.008	-0.217	-0.213	0.044

View Table

Table 5.

Poland: Decomposition of Output Decline

	Total change	Supply shift	Supply-side change	Demand shift	Demand-side change	Error terms
1990Q1	-0.247	-0.846	-0.039	-0.251	-0.246	0.040
1990H1	-0.263	-0.736	-9.028	-0.269	-0.264	0.030
1990	-0.199	-0.771	-0.031	-0.197	-0.193	0.024
1991	-0.185	-0.303	-0.008	-0.217	-0.213	0.044

It should be noted that, since the demand and supply shifts are not independent, there is no clear way of fully decomposing their effects. This is because, for example, an exogenous change in the price of energy may have some effect on aggregate demand, and thus would show up as both a demand-side and supply-side shock. However, to the extent that the dependence between demand and supply shocks is characterized by a longer lag structure, the decomposition that we perform would still give reasonable results for short periods.