\UPS RIOS Policy R"arch 'WORKING PAPERS Transiton and Macro-Adjustment | Policy Research Department The World Bank August 1993 WPS 1165 Social Costs of the Transition to Capitalism Poland, 1990-91 Branko Milanovic Contrary to e&pectations, Poland's stabilization program en- tailed unexpectedly high social costs. Unemployment reached 12 percent by the end of 1991, and real incomes fell 40 percent. The poverty gap rose from an estimated 1.4 percent of GDP to 4.8 percent. Policy Research Working Papesdisserninatethe findingsofworkin progress and encouragetheexchangeofideat amongBankstaffand all other interested in avelopment issues. Thesepapers, distributed by the Research Advisory Staff, carry the nlames of the authors,reflect only their views,andshould be used and cited accordingly.The findings, interpretations,andconclusions arethe authors'own.Theyshould not be attnbuted to the WorMd Bank, its Board of Directors, its management, or any of its member countries. Policy Research Transition and Macro-Adjustrnent WPS 1165 This paper - a product of the Transition and Macro-Adjustment Division, lolicy Research Department - is part of a larger effort in the department to analyze income distribution and poverty in transition economies. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Rebecca Martin, room N1 1-043, extension 39026 (August 1993, 29 pages). The Polish stabilization program implemented in stable. Large households, and children in particu- 1990 as part of the transition to capitalism lar, were especially affected. The poverty gap entailed unexpectedly high social costs. rose from an estimated 1.4 percent of GDP to 4.8 percent. The often unstated assumption had been that since central planning was intrinsically ineffi- Existing evldenct on income distribution cient, stabilization in Poland might be less costly shows that it did not change. There was a slight in terms of lost output than it would have been in compression of income among farmers, which a market economy. The idea was that recession has also occurred in the past when real incomes stemming from an overall decline in demand declined, and possibly some wage-stretching could be moderated by removing the administra- among workers. tive barriers that in a planned economy hindered the best deplo) ment of resources. What happened to the general welfare? Conclusive results are elusive. Personal con- The results were the reverse of expectations. sumption, ove-rall, decreased. Queuing also Unemployment reached 12 percent of the labor decreased, but utility gains from shorter lines force by the end of 1991, and real incomes were offset as real wages, and thus the opportu- plummeted (by about 40 percent). An estimated nity cost of waiting, declined. Real appreciation 17 percent of the population lived In poverty in of the exchange rate raised dollar wages substan- 1989. By 1991, that figure reached 34 percent. tially and led to an upsurge in consumer imports, The poverty rate more than doubled for all social thus increasing the utility derived from the groups except pensioners, for which it remained ownership of consumer durables. The Policy ResearcbWorking PaperSeries disseminate the findings of work underway in the BLnk. An objectiveof the series is to get these findings out quickly, even if presentations are less than fully polished. The findings, interpretations, and conclusions in these papers do not necessarily represent official Bank policy. Produced by the Policy Research Dissemination Center SOCIAL COSTS OF TRANSMTION TO CAPrrALISM: POLAND 1990-91 by Branko Milanovic* Tauie of Contents Section I. Polish Stabilization .togram of 1990: Effect on Incomes .............. 1 Section . Who Became Unemployed? ............................... 4 Section J. What Happened to Poverty? .............................. 5 Section IV. What H-appened to Income Distribution? ....................... 14 Section V. Did Welfare Really Decrease? ............................. 17 Conclusions ................................................ 28 References ................................................. 29 Table 1: Change in Non-Agricultural State-sector Employment (in percent): 1991 vs. 1989 ............................................... 5 Table 2: Poverty and Real Income: Change Between 1989 and 1991 ......... . 7 Table 3: The Determinants of Poverty. ............................. . 9 Table 4: Social Transfers in 1991 .......... 11 Table 5. Poverty Gap in Real Terms ..................... 12 Table 6: Estimated Percentage of Children Living in Poverty ....... .......... 13 Table 7: Income Distribution in 1989, 1990 and 1991: Gini Coefficients ... ....... 15 Table 8: Relative Wages by Occupation in State Sector ..................... 17 Table 9: Food Consumption and Ownership of Durables Among Workers' Households .......................................... 22 Table 10: Money Holdings of the Population ............ ............... 24 Figure 1. Poverty Rates for Different Social Groups, 1989-91 ................. 7 Figure 2. Poverty Rates for Households of Different Sizes, 1989-91 ........... .. 13 Figure 3. Distribution of Persons by Household Per Capita Income .............. 14 Figure 4a. Demand for Domestic Good .............................. 25 Figure 4b. Demand for Imported Good .............................. 25 * I would like to thank Alan Gelb, Emmanuel Jimenez and Bryan W. Roberts for very helpful comments. Section I. Polish Stabilizatioli Program of 1990: Effect on Incomes The implementation of the Polish stabilization program began on January 1, 1990. It was the boldest stabilization program implemented in any socialist country (defined in the sense that majority of assets outside agriculture is state-owned). The program involved substantial price liberalization with the cut in subsidies, incomes policy (wage caps), conveitibiliiy of the zloty, introduction of positive real interest rates, and the reduction of the budget &Aflcit. Some of these measures proved a success. Convertibility at the fixed exchange rate was maintained throughout ths year. Real interest rates, from being sharply negative in 1989, became positive. The budget deficit equal to 7 percent of GDP in 1989 was turned into a small suiplus in 1990 (0.3 percent of GDP). Real amount of subsidies was cut by more than half. The inflation rate shaxply decelerated from an average monthly rate of 19 percent in 1989 to 5 to 6 percent at the end of 1990. On the negative side, one can include a sharp decrease in real wages,' substantial increase in unemployment, and a collapse of industrial production. Real state-sector wages in 1990 were some 30 percent lower than the year before (the initial forecast was -10 percent). Unemployment rose from practically nil ia the beginning of 1991 to 1.2 million by the end of the year (6.1 percent of the labor force). The increase was three times greater than originally forecast (400,000). Finally, industrial production of the state sector plummeted, decreasing by about 25 percent compared to 1989. This was also a greater decline than anticipated (5 percent). Even accounting for the increased output of the private sector, gross industrial value added in 1990 was 22 percent less than in 1989 while GDP v as 11.5 percent smaller (GDP per capita 12 percent). The decline in industrial production generated a lively discussion and produced a number of theories as to why it was so dramatic, Wz. whether the slump was mostly due to external causes (breakdown of the CMEA market) or to internal ones (monopoly stmcture of the industry). Some questioned whether the decline was really of such a magnitude since output in the past was overstated and calculated at arbitrary (non-market) prices (Lipton and Sachs, 1990: 1 Unless explicitly stated otherwise, 'real always refers to 'non,jnjU deflaWt by the official consumer price index. 1 78-9). Winiecki (1991) proposed the following argument. According to Winiecld, two out of the tbree components of the registered output decline do not matter. One part of the decrease is purely a statistical artifact due to the earlier practice of padding the output figures to show fulfillment of plan targets. The second part of the decline is due to behavioral changes of enterprises and, to some extent, individuals. For example, better availability of goods, increased interest rates, and somewhat harder budget constraint, led entexprises to reduce their stocks of inventories which in socialist economies are inordiinately high.. The drawing-down of inventories produced a short-run output decline, but in reality it represented an adjustment to market conditions. Thus, the first component is fictitious, and the second, while entailing a real decline in output, has no effect on population welfare. Only the third component of output decline, caused by stabilization, is both real and does have an impact on welfare. Winiecki's argument shows that recorded declines are indeed overestimates of the actual ones, even if the break-down of the three components is impossible to make. It should be noted however that the so-called fictitious decline was probably very small in Poland because the scope of central planning in Poland was fairly limited even before the 1989 change of regime.2 The standard of living suffered seriously as a result of the stabilization. Several "macro" indicators show this decline: real average ivage in the state sector (where 87 pertcint of non-agricultural labor was employed) was reduced by 31 percent; unemployment emerged; real income of agricultural households was cut in half as terms of trade moved sharply against agriculture; and, finally, pensions decreased by 17 percent in real terms. lhe unly real increase was registered by private sector incomes: they went up by 33 percent in total while the number of private sector non-agricultural employees increased 22 percent. Private sector share in total non-agricultural employment was still small to make much difference overall. As a result, in 1990, real per capita consumption dropped by 13.5 percent. 2 Government contracts, the only form of central planing that existed, were introduced in 1982. By 1988, they covered only about 6 percent of GDP and in 1989 they became even further scaled down to include only some essential intermediate goods and energy (see Rzemzrpospolita, Economic Reform Supplement, September 6, 1988). State investments accounted for only 2 percent of GDP. 2 In 1991, the situation stabilized in the sense that the decline in industrial production, GDP, population incomes, and state-sector wages became less or bottomed out. GDP decreased by a further 7.6 percent. Industrial production of the state sector decreased by 13 percent. Real wages in the state sector stabilized at their 1990 level. Private sector incomes (outside agriculture) still continued to increase, by about 1 percent per capita. XJnemployment, however, continued to grow at almost the same pace as in 1990, reaching 2.3 million (11.8 percent of labor force) by the end of 1991. Several positive developments on the mdcro side were preserved: the inflation rate decelerated further to an average monthly rate in 1991 of 4.1 percent, real positive interest rate policy was continued, and the budget deficit, despite serious pressure, was maintained at less than 4 percent of GDP. However, recovery had not yet started in 1991. The first signs of the recovery in industrial production (the series that declined the most precipitously at thte inception of the stabilization program) were recorded only in the mid-1912. These macro indicators were known to the authorities and analysts as the stabilization program unfolded. What was not known was how these overall developments were reflected at the household level: who became unemploye6; how much did poverty increase and among what social groups; did wage distribution among state-sector workers become more unequal as entexprises restructured and tried to keep more productive workers from moving to the private sector; did overall income distribution become more skewed? The objective of the present paper is to tly to answer some of these questions on the basis of the 1990 and 1991 data provided by the regular Household budget surveys published by the Polish Central Statistical Office.3 The period of two years is sufficiently long as to allow all the essential effects of macro stabilization and system change to "trickle down" to the level of .souseholds. 3 The data for 1990 and 1991 are published in respectively Budzety Gospodartw Domowych w 1990 RoIu and Budzety Gospodarstw Domowych w 1991 Rolc, Warsaw: Central Statistical Office, 1991 and 1992. The surveys cover about 28,000 households (0.25 percent of all households in PolaA). The surveys are representative of about 88 percent of the population (see Gorecki, 1992, p. 4). They exclude non-agricultural private sector, Army and police personnel. Surveys include private sector workers only if the main earner is employed in the state sector. They do not include e.g. households whose both members are employed in the private sector, or whose one member is unemployed aud the other works in the private sector. Non-inclusion of private sector outside agriculture was n-t a major omission in the past when this sector was small. It is a serious defect now when the sector is growing. This shortcoming is being coirected in 1992 survey. 3 Section I. Who Became Unemployed? Polish stabilization program of January 1990 was followed by a swift increase in unemploycntt. As already mentioned, the percentage of the unemployed in non-agricultural labor force increased in two years between January 1990 and Janm'ry 1992 from less than 1 percent to 12 percent. In absolute numbers, the increase was from less than 200,000 to 2.3 million. Net job loss in the state sector was 20 percent or about 1.6 million workers.4 The question we address in this Section is how the decline in employment was reflected in the micro (household-level) data, and what types of households were primarily affected by it. The Survey provides information about the average number of state-sector non-agricultural employees by income group and by household size. The number of state-sector non-agricultural employees declined between 1989 and 1991, as shown in Table 1, by 5.3 percent. The decrease was more significant among mixed households (-8.2 percent) than among workers' households (-4.4 percent). This can be explained by two factors: (1) the tendency to fire first those with available outside work options (private agriculture in the case of mixed households) and/or (2) more prevalent voluntary quits among the mixed households who, due to their experience with private sector agriculture, may have greater confidence in starting own businesses. Some evidence of (2) is provided by the fact that state sector employment declined the most among the well-off households. Although this is true for both workers' and mixed households, it is more pronounced in the case of the latter: the participation in the state sector of the top income class among mixed households decreased by almost 10 percent. Interestingly, state-sector employment increased among the lowest income groups. Among pensioners participation rates were low (only about 6 percent of people living in pensioners' households were employed in state sector) and while they increased (to 6.- percent), their absolute contribution is minimal. Decrease in employment was the most severe among the "standard" household sizes (four- and five-person households). The state-sector participation rates for these households decreased by between 4.7 and 6.4 percent for workers, and 8.6 and 12.3 percent for mixed households. 4 Outside agriculure. See P. Suttcal Yearbook 1992, p. Xl. 4 Table 1: Change in Non-Agricultural State-sector Employmant (in percent): 1991 vs. 1989 Income Classes Workers Mixed All First group +4.7 |+2.9l Second +7.0 -6.9l ~~~~, 2, Third +4.4 -6.9 Fourth -3.0 -2.7 Fifth -2.9 -4.5 Sixth -7.9 -3.1 Seventh -6.4 -1.3 Eighth -|-1.9 -9.4 Household Size X 2-person -3.1 -0.9 l 3-person -4.9 -8.4 l 4-person -4.7 -8.6 5-person -6.4 -12.3 6-person _____ -3.4 -9.8 Total -4.4 -8.2 -S.3 Note: The value gives the percent change in people employed iD state sector. Income classes for each year are not &-d in real terms. They, however, contain similar percentages of recipients (the shape of income distribution curve is lognormal; see Figure 3). Section mI. What Happened to Poverty? Real state sector wages in 1990 dropped by 31 percent and in 1991 by a further 3 percent; real pensions declined by 17 percent in 1990 and rose by 15 percent in 1991, thus returning to their pre-stabilization level. Farmers suffered the most: their real income was cut by more than one-half in 1990, and by additional 36 percent in 1991. These are al "macro" data. How is the decline in real incomes reflected in the household data collected by Surveys? 5 Poverty rates went up in all social groups except pensioners (see Table 2 and Figutre 1).5 Among workers' households the share of the poor incr- sed, between 1989 and 1991, from 16 to 38 percent. The bulk of the increase occurred in 1990. The poverty headcount among workers' households remained broadly stable in 1991. Te percentage of the poor among farmers increased from 17 to almost 40 percent. For mixed ho'iseholds which have throughout the 1980 exhibited the lowest Doverty rates, the percentage of the poor also increased substantially: from 8 to 21 percent. The universal increas; in poverty rates had, as Figure 1 makes clear, the effect of equalizing poverty rates among different social groups. Thus, while in the 1980's, the proportion of the poor among pensioners ftuctuated between 30 and 40 percent, and was only around 10 percent for the mixed households,6 in 1991, the poverty rates of pensioners, workers, and farmers were very close. The poverty rate for the mixed households was also not far behind. 5 Poverty rates for different social grups are calculated as follows. 'he poverty line is assumed to be equal to the social minimum line calculated since 1980 by Ete Institute of Labor and Social Affairs in Warsaw. The line is calculated quarterly. It is based on a fixed basked of products. For the year as a whole, we take the simple average of the quarterly amounts. The social minima for one-person workers' and pensioners' households (calculated separatevy by the Institute) represent the poverty line for one male adult worker and pensioner. For mixed and farmers' households this amount is lowered by 20 percent, on account of lower prices in rural areas (the reduction suggested by the researchers in the Institute). The next step is to apply the poverty line to the income distribution statistics derived from the Surveys. The Surveys provide income distribution data (eight income clases arranged according to household per capia income) for six types of hc zseholds (from 1-person to 6 and more person-households) for workers, farmers and mixed household; and thrde types of households for pensioners. This gives a total of 21 incone distributions, or 21 x 8 income classes = 168 observations. For each observation, the Surveys also provide the data on average number of equivalent consumption units. All individuals belonging to a given income class are considered poor if the upper income limit of that class, adjusted for the number of equivalent consumption units, is less than the poverty line. To give an example. Let the poverty line for 1 adult be $10. Let now ten 3-member households have, on average. 2.2 adult consumptionunits and belong to the income class that ranges from $4 to $7 per capita. This yields lower and upper income limit per adu consmption unit of $5.45 and $9.54. We then classify all ten households or 30 individuals as poor because the upper adjusted income limit ($9.54) is less thin the poverty line ($10). Clearly, if the lower adjusted income limit is greater than the poverty line, none of the households is poor; if the poverty line is between the two limits, a proportion of individuals is considered poor. The poverty rates for social groups and the total poverty rate are then constructed from the individual 21 poverty rates. The poverty gap is calculated in a similar fashion. F For the exact data, see Milanovic (1992). 6 ,eoverty rates for different social groups, 1989-91 40 - 30- 20 10 W-vorkers mrxed Farmers penslonels able 2: Povertv and Real Income: Change Between 1989 and 1991 Decrease in real Poverty Rate in Poverty Rate in per capita income '91 '89 Workers 33.3 38.1 15.8 Mixed Households 41.0 21.2 7.9 Farmers 51.6 39.4 17.2 Pensioners 11.6 33.0 36.2 Total 41.5 34.4 17.3 Note: Poverty rae is the headcount index. Another interesting development is that farmers now have the highest proportion of the poor. This Gocurred as farmers' terms of trade collapsed by almost two-thirds in the two years 7 of the stabilization.7 The causes of the collapse of terms of trade were both economic and political. Significant real appreciation of the zloty at the inception of the program combined with trade liberalization shifted some of domestic food demand to inports. But rolitical factors were probably more important: farmers which in the previous regime had a very strong lobby were able to introduce the so-called "parity policy" whose objective was equalization of per capita incomes in rural and urban areas. The Communist iegime, which already faced strong workers' opposition, thereby sought to placate farmers in its bid to retain some social legitimacy. With thc new democratic goverment, farmers' political power waned. Their parties were divided, and farmers' unhappiness, manifested by strikes, failed to have much effect. Moreover, the leadirg parties in the shaping of the economic policy in the Suchocka government, namely Democratic Union and Liberal-Democratic Congress, are very much "anti-farmer". Holding that there is significant agricultural underemployment, they are in favor of forcing marginal farmers out of business.' The sharp decline in incomes of farmers' and mixed households had the following effect on the rankings of various social groups by their income levels. While throughout the 1980's, pensioners' per capita incomes were between 15 and 25 percent less than the average incomes of the other three active groups, in 1992, for the first tihzo since 1978 when the Surveys in their current shape started, the ava.rage income in pensioners households was higher than average income of mixed and farmers' households; it also came to within 5 percent of the average per capita income of workers' households. 7 As with other statistics, the collapse of terms of trade is somewhat exaggerated. Input shortages which existed prior to 1990, forced farmers to purchase part of inputs at much higher market prices. These higher prices were not fully captured by the statistics. 8 I am grateful for these points to Michnl Rutkowki. 8 The change in poverty rates can be expressed as a function of two viables: real per capita income, and income inequality. The first variable is expected to be negatively, and the second, positively, related to poverty. The results for the per.od 1978-91 are displayed in 'rable 3. All coeffilcients are statistically sigrificant and the 2 is around 0.9. The regression shows that Table 3: The Determinants of Poverty. Dependent Variable: Percentage of the Poor Constant Income Distrib. R2 DW Period Term Term Term (F) (SE) Urban Households . _ 1978-91 384.7** -50.15** 0.891* 0.88 2.07 (0.000) (0.000) (0.049) (56.0) (3.93) Rural Households _ 1 '-978-91 252.2** -35.25** 1.246** 0.90 1.86 (0.000) (0.000) (0.000) (74.1) (2.43) Notes: Equations are of the form: POOR = Bo + Bl*log(income) + B2*distribution. Autoregression coefficient is stat5stically sgnificant at less than I percent in the first equation. The number of observations is 28. Income is in 1978 constant zloty. Distribution term is the Gini coefficient for oach social group. All data are calculated from the Household Surveys according to the mothodology explainod in footnote 7. Data in brackets below regression coefficients show lovels of significance at which the null hypothesis is rejected. Two (one) asterisks show that the coefficient is significant at 1(5)porcent level. a ten percent uniform (across all income groups) reduction in real income of urban and rumal households is associated with respectively 5 and 3.5 percentage point increase in the poverty rate.9 Thus, the 30 to 40 percent reduction in incomes recorded since the transition began led to 15 to 20 percentage point increase in the poverty rate. Each point increase in the Gini 9 In order to increase the number of observatior, the data set for urban population is composed of 14 annual observations for workers and 14 annual observations for pensioners' households. The same applies to the rural population which is composed of farmers' and mixed households. 9 coefficient leads to approximately 1 point increase in the poverty rate. The implicit trade-off between growth and distribution is therefore the following. For the urban households 1.8 percent uniform real growth is, from the point of view of its impact on poverty, needed to offset 1 point increase in the Gini coefficient. For the rural households, 3.5 percent real growth is required for each point increase in the Gini coefficient. Higher growth-distribution trade-off (meaning that more growth is needed to offset a given increase in inequality) for the rural households is a reflection of greater dispersion of incomes among rural households. Thus, in order to "push" enough people over the poverty threshold, a greater increase in real incomes is needed. Total estimated number of the poor in 1990 and 1991 reached respectively 12 and 13 mnillion out of a population of about 38 million C(able 5).10 Approximately three-quarters of the poor (8.6 million) live in urban areas. The average number of the poor in the previous three years (1987-89) was somewhat less than 7 million. It divided between urban poor (about 5 million) and rural poor (2 million).11 Between 1989 and 1991, the total number of the poor increased consequently by about 6 million. The number of the poor in urban areas went up by 3.5 million in 1990 alone. The next year, their number remained constant. The number of the rural poor increased by over a million in 1990, and then by another million in 1991. The results clearly show that while in the flrst year of the transition the brunt of the social cost was borne by urban households, and in particular state-sector workers, the bulk of the cost in the second year shifted to farmers and mixed households. Poverty gap (income needed to bring all poor households to the poverty threshold level) almost tripled rising from an estimated 3.3 percent of total household income in 1989 to 7 percent in 1990 and to 9 percent in 1991. If Survey data are expanded to the whole population, the estimated poverty gap was approximately $1.8 billion or 3.0 percent of GDP in 1991, and $3.6 billion or 4.6 percent of GDP in 1991 (vs. $1.1 billion or 1.4 percent of GDP in 1989).12 10 Total number of the poor is calculated by extolating the poverty rates catculated from the Sunreys to tie whole population. '1 See Milanovic (1992, Table 4). 12 For 1992, this yields a per capita poverty gap of almost $100 per annum. 10 For comparison, in 1991 total expenditures for the unemployed amounted to $700 million or 0.9 percent of GDP (see Table 4). Table 4: Social Transfers in 1991 In $ Billion As Percent of GDP Pensions 9.6 12.3 Various family allowances 2.2 2.8 Other social transfers" 0.6 0.8 Unemployment benefits 0.7 0.9 Transfers in cash 13.1 16.8 Education 3.3 4.2 Health 3.7 4.7 Transfers in kind 7.0 8.9 Estimated poverty gap 3.6 4.8 a/ Includes scholarships, social aid given by enterprises, and social assistace at local level. Source: Poland StaZstdlo Yearbook 1991, Table 15 (242), p.152, and Table 1 (328), p. 237. In order to eliminate the effects of major exchange rate swings in the beginning of the transition, it is useful to express poverty gap in real zloty terms. Table 5 nhows that absolute poverty gap increased by about 75 percent while the estimated number of the poor more than doubled. The average real income shortfall of the people classified as poor, therefore decreased. The year 1990 is particularly interesting in that respect because the relatively low shortfall clearly indicates that many people just "slid" into poverty, falling slightly below the poverty line. Such a decline can socially be sustainable, if it is followed, within a relatively limited time, by an increase in income. In that case, the descent into poverty would have been only a transitory phenomenon. If this will indeed be the case is impossible to tell now. The necessary condition is that economic growth resumes. As Table 5 shows, the reverse was the case in 1990 and 1991. 11 Read GDP shnmk by almost 20 percent. Poverty gap as share of GDP therefore expanded because both the numerator and the denominator moved in the "wrong" direction. Had the economy during these two years remained stagnant, the same absolute poverty gap would have been equal to only 3.8 percent of GDP instead of the actual 4.8 percent. Table S. Poverty Gap in Real Terms Estimated number of Poverty gap Poverty gap the poor (1990 prices; per capita Real GDP (000) zl billion) of the poor (1990 prices) 1989 6,559 12,461 100 100 1990 11,947 17,462 76.9 88.2 1991 12,913 21,574 87.9 81.5 Now: Poverty gap deflatwd by the rotail pnce indox. Source: Calculated from Household Surveys. Real GDP from Poland Stadsfcal Yea'ook 1992, Table 2 (196), p.118. We have seen in Section I that participation rates in the state-sector declined the most among the "standard" (4- and 5-member) households. Among these households incidence of poverty increased most dramatically as well. For example, between 1989 and 1991, poverty rate for 4-member households went up from 11 to 31 percent; among 5-member households, poverty increased from 17 to 43 percent (Figure 2). The social implications of the increase in poverty for the typical household (husband, wife and two children) must not be overlooked because 37 percent of workers' population and 25 percent of the total population live in such households. In contrast, poverty rates among 1- and 2-person households remained almost unchanged. These rates are determined chiefly by pensioners households where the incidence of poverty after a slight increase in 1990 returned to its 1989 level. Because the increase in poverty was greater among large-size families, the change in the percentage of children who are poor is even greater than the change in the percentage of the poor households or the population. In 1989, about 17.5 percent of children under 6 years of age 12 lived in poor households; this increased to more than 50 percent in 1991 (Table 6). Poverty incidence among children of workers and mixed households tripled. More than one workers' child in two lives in a poor household. Poverty rates for households of different sizes, 1989-91 60. 40- 30 20. 1 2 3 4 5 6 Household size Figure 2 Table 6: Estimated Percentage of Children Living In Poverty 1989 1990 Workers households 19.1 57.9 Mixed households 10.3 30.4 Farmers households 20.3 50.9 Total 17.5 51.2 Note: Negligible percentage of children live in pensioners' households. 13 Section IV. What Happened to Income Distribution? The change in the headcount ratio of the poor is, in addition to the change in the overall income, influenced by the shape of the income distribution curve (how bunched are households near the poverty line?), and by the change in income distribution itself (did it become more unequal?). Between 1989 and 1991, income distribution, measured by the Gini coefficient, became slightly more equal in all social groups except workers (Table 7). Overall income distribution was practically unchanged. The increase in poverty was thus solely due to declining incomes. As Figure 3 makes it clear, the whole distribution practically shifted leftward between 1989 and 1991, pushing an additional 20 percent of the population below the poverty line. Dlstribullon of persons by household per capita Income (all householdsl i:S 20- 00 1000 90000 1000 4000 Income In 000 zloty per rmonth FIgur 3 The slight increase in the Gini coefficient among workers may be due to greater wage differentiation following stabilization, preparations for the privatization, and the onset of industWial StUcturing. The Gini coefficient of state-sector wages13 also increased, and there were some changes in returns to education and the stnucture of relative wages (see Table 8). These conclusions are necessarily very tentative since the increase in wage inequality is quite 13 Ihis is the distribution of all state-sector wages, and represents a disdinct (although Closely relatod) variable from income inequality among workers' households. Ihe latter includes all sourcs of income, not only wages, and is based on a sample. 14 moderate, and not out of line with what was recorded before (e.g. in 1986 and 1987 when wage inequality bounced back from extremely low levels reached during the "Solidarity" period and under the martial law). Table 7: Income Distribution in 1989, 1990 and 1991: Gini Coefficients 1989 1990 1991 Gii Point ________Change Workers 23.7 25.0 25.0 +1.3 Mixed households 24.7 23.9 22.2 -2.5 Farmers 35.7 32.5 30.1 -5.6 Pensioners 22.6 20.8 22.3 -0.3 All households 26.0 25.5 24.7 -1.3 State sector wages 19.7 23.6 22.8 +3.1 Note: Distribution of individuals ranked by their household per capita gross income. The Gini coefficient expressed as percentage (e.g. 30 instead of 0.3). Thi t3ini point change is the change between 1989 and 1991. Sources: Household Surveys, Poland Statiavcal Yearbooks (for the distribution of state sector wages). The ratio between wages of highly-skilled and medium-skilled workers also increased: in 1990, the ratio between these two categories was 1.35 to 1 while it was 1.16 to 1 in 1988.14 A similar process of increased returns to skills and education is noted by Gorecki and Peczowski (1992, Figure 2). They point out that the average income of households headed by university graduates was in 1990 about 25 percent higher that the average income of households headed by people with primary or vocational schonl. In 1988, for example, the average incomes were equal. This has produced a higher share of skilled individuals in top income groups. While in 1989, households headed by university and secondary-school graduates accounted for 43 percent 14 Polwnd Sta&sicdl Yearbook 1991, Table 7 (324), p. 234. The datum for 1988 is quoted in (Freeman. 1992:14). 15 of households in the highest income decile, two years later their share increased to almost 66 percent.1s More evidence on the changing wage distribution is obtained by looking at relative wages of different occupations. Such comparison is fraught with problems because the occupational structure presented for 1990 does not fully coincide with the occupational structure available for 1986 and 1987.16 What can, nevertheless, be concluded from Table 8 is a decline of relative wages for miners, routine accountants and economists, and some categories of manual workers (textile workers). In the previous system, miners occupied a very special position, for both ideological reasons (miner was the prototype of a socialist worker) and more pragmatic foreign exchange needs (Poland relied heavily on coal exports). That position was bound to end in the transition. Decline in relative wages of routine accountants and economists may be an indication of the deterioration of relative wages of the bulk of middle-level clerical personnel whose numbers were overblown in the socialist system. As demand for clerical staff decreases, their relative wages go down. Better or luckier among them may eventually move to positions of management specialists (such an occupation did not exist in the 1986-87 classification). Generally, however, the relative position of manual workers, with the exception of miners, did not change much. Resemblance between the 1986 and 1987 wage structures is much greater (correlation coefficient almost equal to unity) than among these two years and 1990 (r=0.85 and 0.8). Private sector wage structure (not reported here) is probably closer to that existing in market economies. Very high wages received by some occupations (business school graduates, auditors, some economists) provide a lot of anecdotal evidence that wage distribution must be widening. For example, government administration is outbid by local and foreign firms that are often able to hire its best people. 1s Quoted from Gorecki (1992, Table 7). " Data for 1991 ate not published. 16 Table 8: Relative Wages by Occupation in State Sector (average wage for all state sector workers = 1) 1986 1987 1990 Mining engineer n.a. n.a. 2.42 Management specialist n.a. n.a. 1.74 Miner 2.15 3.21 1.60 Geologist/engineer 0.98 1.02 1.35 4Accountant/economist 1.15 1.28 1.06 Electrician 0.88 1.02 1.06 Welder 1.00 1.15 1.00 Locksmith 0.94 1.04 1.00 Bricklayer 0.79 0.89 0.91 Weaver (textiles) 0.84 0.95 0.78 Cashier/shop assistant 0.60 0.68 0.66 Correlation coefficients 0.99 1 0.85 0.80 Soume: Poland 8rad sdcaI Yearbook 1991, Table 10(327), p.235, and Po6 andfStadical Yearbook 1988, Tables 7C224) and 8225):162-3. Section V. Did Welfare Really Decrease? An important question is whether the welfare, on average, has declined or not following the introduction of market-oriented reforms. Particularly relevant for such an analysis is the first year of the stabilization (1990) when, as indicated above, both the supporters and the opponents of the reform were suiprised by the extent of the decline of incomes, industrial production and GDP. The extremely severe decline was taken by some as an argument that the "Big bang" market- oriented reforms cause a lot, and it is argued, some probably unnecessary, hardship. On 17 the other hand, pardsans of the "Big bang" approach contend that theie is a cumber of data biases which lead to an exaggerated view of the hardships. Lipton and Sachs (1990) argument is emblematic of this second strand of thought. They argue that earlier statistics had an in-built bias toward presenting a rosier picture of the reality and that once this bias is eliminated, as it happens during the transition, the new situation by contrast appears worse. Measuring of output was geared to take account of physical quantities which made sense in a planned economy but not in a market economy where many of these goods are not wanted. Lipton and Sachs (1990:79) quote a Polish joumnalist who writes: "For the entire period of real socialism, investments were poured into a closed production cycle that offered no profit: coal was necessary to produce electricity; electricity was necessary to produce steel; and steel was necessary to mine coal. All that produced a statistical growth in national income, a growth which, as we now see, actually meant a decline in national wealth." State production alone was recorded in statistics; production of small private-owned enterprises was often unrecorded. This was not a problem so long as these entexprises' share in value added was negligible. But during the transition their importance rises rapidly. Hence, it is argued, underestimation of some output gains. Finally, one of the most visible aspects of stabilization in Poland and elsewhere has been a dramatic increase in trading. Often, on extremely small scale, used as a complement to jobs in state enterprises, alegal if not illegal, private trading was almost entirely bypassed by official statisticians. All of these are reasons wny the reported declines in output may look worse than they are in reality. Lipton and Sachs (1990), however, make a more ambitious attempt to show that utility of an average consumer might have gone up (even if his income appears to be less) 18 because of the disappearance of excess demand and consequent waiting that imposes real cost in terms of resources (time) spent. I shall first present their argument and then point to some problems with their claim. Lipton and Sachs (1990) define excess demand as a situation when aggregate demand (Y) excess aggregate supply (S) at the official -price level (P) which for simplicity we can set at unity. Percentage excess demand (ea) is equal to (Y-S)IY. Aggregate demand and supply are equilibrated through the mechanism of black market. At the black market price Pb (Pb> 1), aggregate demand and supply are equal: S Pb= Y. If we replace the last relation into the previous equation for ex we obtain: ex= S(Pb -1) = Pb-l (1) Pb s Pb- Representative consumer's utility is equal to U= U(S) + wL, where S=real consumption equal to aggregate supply, w=exogenous marginal utility of leisure, and L=the amount of leisure. Also, leisure L is reduced by the amount of time the consumer must spend in waiting lines. Let us denote time spent queuing per unit of output as q. Then the utility equation becomes U=S + w (N- qS) where N=maximum amount of leisure if q=O, and, for simplicity, we assume direct proportionality between consumption and utlity so that U(S) =S. Utlity from spending a unit of time queAing must, in equilibrium, be equal to the marginal utility of leisure:17 I? Note that the implicit assumption is that the marginal and average utility from queuing are the same, i.e. that the marginal utility does not decrease. 19 w - U(I,) - U Pb - 11i Pb -1 (2) q J q where T = profit per unit of queuing (Pb-I is the monetary gain, that is the difference between the market and official price of output), and, again, for simplicity we assume proportionality between utility and monetary return, w. In Lipton and Sachs (1990) equation (2) is divided throughout by Pb. This, however, is incorrect since the same deflation would either have to be done consistently (e.g. for w) or not done at all, thus implicitly assuming that the deflator is 1 as we are doing here. The mistake though does not affect the rest of the derivations. From (2) it follows that Pb-l =qw. Replacing this into (1) yields: Pb-1 qw (3) Pb Pb From (3), q=Pb(exlw) which, when if w=constant, shows that queuingper unit of output is proportional to excess demand and to the excess of black market price over the official price. Finally, we can substitute q=Pb(exlw) in the utility equation to obtain: U - S + w (N-qS) = S + w (N - Pb xST = S (1 - Pb =) + wN (4) w which is the final expression obtained by Lipton and Sachs (except for Pb which is canceled out in their derivation). They write: "when the system is characterized by excess demand, increases in real income lead to decreases in utility" (1990:92). That statement as well as their Figure 1 (1990:93), implicitly assume that S is constant. If it were the case, obviously, the greater the excess demand (ex) the less the utility because more time is spent in "directly 20 unproductive" waiting.1' However, what characterized Polish stabilization was a declne in S and utility in equation (4) must then decrease on account of lower S (BY/8S>O). According to official statistics the decline in S per capita was 13.5 percent. Even according to Berg and Sachs (1992-141), who strongly dispute the official statistics, real consumption decreased by 4.8 percent. Indeed, household surveys show that per capita consumption of most food items decreased (see Table 9).19 The decline in S was moderated, however, by a marked increase in ownership of, and thus utility derived from, some products that are almost entirely imported (e.g. video recorders, color TVs; see Table 9). If consumption is divided into two parts Sd = domestic consumer output and M = consumer imports, there was an increase in consumption of M both in terms of physical quantities of goods (mostly consumer durables) and non-factor services (foreign tourism). Increase in M was due to real appreciation of the exchange rate and increase in dollar wages. The fact that Polish dollar wages went up does not matter for that part of consumption that is used to buy domestic outputF° because the purchasing power of the dollar in terms of domestic commodities has gone down by even more. It is relevant, however, for import demand because an average Polish worker, whose wage increased from $39 in 1989 (estimated at the 18 Pb is also a fimction of ex and need not be discussed separately. '9 The same results are obtained if one uses the macro consumption data divided by the population. We prefer to use survey data to avoid the charge made at the macro data, namely that they miss some of the retail trade. 10 Sachs (1990:8) seems to imply this but is rightly criticized by Secretariat of the Economic Commission for Europe (1992: 54). 21 Table 9: Food Consumption and Ownership of Durables Among Workers' Households Per Capita Per Month 1989 1990 % Change Meats (kg) 2.81 2.71 -3.6 Butter (kg) 0.73 0.73 0 Milk (1) 7.26. 6.78 -6.6 Cheese (kg) 0.92 0.76 -17.4 Eggs (pieces) 15.82 14.45 -8.7 Sugar (kg) 2.06 1.92 -6.8 Total Calories Per Day 2542 2408 -S.3 Percentage of Households with: _ ____ _ Color TVs 50.7 67.1 +16.4 Video-recorders 4.7 20.1 + 15.4 Washing machines 59.1 63.5 +4.4 Cars 30.7 33.2 +2.5 Note: For consumer durables the data show percentage point increases in ownership. Sources: Stadsdcal Yearbook 1991 Tables 25(313), 27(315) and 28(316); SAZ*dcal Yearbook 1990 Tables 26(322) and 30(326). 1989 parallel rate) to $108 in 1990 did feel richer and indeed increased his or her consumption of importables. This is a feature common to many stabilization problems: instant prosperity can be created through real appreciation of the exchange rate. The obvious examples are Chile in the mid eighties, Peru under the Garcia government in 1985-87, and former Yugoslavia during the Markovic progrm in 1990. The problem is whether such a rate can be sustained. Most often, barring continuation of substantial inflows on the capital account, it cannot. The Polish situation, from the point of policy-makers, was better. Persistent excess demand and lack of alternative financial instruments created a lastng high demand for foreign exchange. Foieign exchange market was the n-th market that equilibrated aggregate demand and supply. Zloty was 22 thus chronically undervalued. Unification of the exchange rate in January 1990 permitted to poLicy-makers to provide "instant prosperity" at an exchange: ate that was basically correct. How the decline in real Incomes and consumption of the essentials can be reconciled with an increase in consumption of imported consumer durables is shown in Figures 4a and 4b. As before, suppose that total personal consumption is composed of Sd domestically produced goods and M imported consumer durables. Since the system is demand-driven, demand automatically determines the supply. For simplicity we also assume that all of Sd are the essentials (e.g. food) and all of M are consumer durables. Figure 4a shows the downward shift of domestic demand from DDo to DD, due to the decline in real income (Y, yo) went up shifting in the process the demand curve for imports from DM0 to DM1. Dollar prices of 0 Retail prices of food increased by 575 percent in 1990 vs. overall retail price increase of 585 perent. 24 Price (in zLoty) l\ I\(YOWO) I I\ 001 (YI,WI) I I I I ~~DDI (Yl,,Wl) I I . Ql1 Qo Quantity figure 4a - Oemand for domestic good Price Cin doLLars) I l> Ma(yl) I I MOCyQ) Q°3 q1 Quantity Figure 4b - Demand for imported good 25 imports remained unchanged and the consumption of importables went up from qo to ql. This is why we observe much greater ownership of consumer durables in the Surveys. This explains why the two components of S have moved in different directions. Since we ignore the shape of the utility function we cannot determine relative importance of partial derivatives of U with respect to Sd, M and ex. However, in addition to these variables, other variables in (4) also changed. w represents the real wage rate (=opportunity cost of leisure). Real wage declined in 1990 (even if one may dispute the exact amount of the decline).' This further reduces utility as the gain from the elimination of excess demand is lowered: if opportunity cost of leisure is less, people do not mind as much waiting in line. Finally, N probably increased (on the account of increased unemployment). This, of course, raises utility (people work less). However, if, for simplicity, we allow that the effects of w and N are too small and uncertain, we can take a reduced utility function of the form U= U(S,q) where the effect of reduction of S probably dominates the effect of lower q. The above analysis deals with the representative consumer. It thus fails to take into account the issues of distribution. If elimination of subsidies and excess demand (1) affects lower income households more than higher income households (note that their opportunity cost of queuing is less and that, according to all empirical studies, subsidies are pro-poor)'; and (2) 21 Real wage should be measured as nominul wage divided by Pb. 25 For Poland, it is calculated that inclusion of consumer subsidies reduces income inequality, measured by the Gini coefficient, from 21.8 to 20.0. Kupa and Fajth (1990: 37) similarly find for Hungary that the Gini coefficient is reduced from 23.1 (for disposable income) to 22.0 (for disposable income plus subsidies). Finally, for CSFR evidence points to the sae conclusion: negative tunover tax (a type of consumer subsidy) represents 7.1 percent of households' expenditures in the lowest and 4.4 percent in the highest income decile (World Bank, 1991:59). On the basis of household expenditure surveys, Ve6ernik (1991: 17) calculates that lowest quartile of households received per capita 7.5 pecent more food subsidies than the average while the top quartile received 6.1 percent less than the average. Similar results were obtaned for Algeria (Stanova .. :'991: 41). Subsidies are thus pro-poor in relative terms, meaning that they account for a greater proportion of poor households income or expenditures, but are not normally pro-poor in absolute terms, meaning that the poor receive more from subsidies in absolue value. (The implicit assumption is that households with different incomes pay the same average price for the subsidized good. In other words, the percentages of consumption at subsidized and free-market price are independent of the level of income). 26 marginal utility of income decreases with the level of income, then the overall utility must have been depressed even further. In conclusion, we can write the representative consumer utility function as follows: U = U (Sd, M, q, w, N (5) where all the symbols are as explained before. (Note that the change in wealth is left out because it affects utility only indirectly through its effec:: on real consumption: people consume less because they feel poorer.) Consumer utility in 1990 decreased on account of reduction in Sd. The decline, was moderated by the elimination of queuing and increased consumption of importables (A). What the balance of these effects is, is impossible to determine. What is sure, however, is that observing only S, implicitly assumes that the utility function is U= U(S) which is incorrect. Lipton and Sachs' (1990) contr5bution therefore consists in bringing in another important element, namely q. However, they fail to acknowledge the fact that S decreased (an important omission) and that at least two other elements changed as well: wage rate declined and thus the opportunity cost of waiting was reduced, and total leisure increased. 27 Conclusions The effects of the Polish stabilization program implemented in 1990 generated a widespread discussion. The discussion was motivated by the unexpectedly high social costs associated with transition to capitalism. The often unstated assumption at the inception of the stabilization program was that since centmal planning was intrinsically inefficient, stabilization in Poland may even be less costly in terms of lost output than in market economies. According to these views, recession stemming from the overall decline in demand could be moderated thanks to the removal of administrative barriers that hindered a better deployment of resources in a planned economy. In the event, the results were rather the reverse. Population welfare declined as unemployment reached 12 percent of labor force by the end of 1991 while real incomes plummeted (by about 40 percent per capita). Poverty increased from an estimated 17 percent of the population in 1989 to 34 percent in 1991. Poverty rate more than doubled in all social groups except among pensioners where it remained stable. Larger households and children in particular were affected. The estimated poverty gap rose from 1.4 to 4.8 percent of GDP. The existing evidence on income distribution shows that it did not change; there was some slight compression among farmers which occurred also in the past whenever real incomes declined, and some possible wage-stretchiwg among workers. What happened to the overall welfare? Conclusive results are difficult to obtain. While overall personal consumption decreased, utility derived from increased ownership of consumer durables (imperfectly covered by the statistics) increased; while queuing decreased, utility gains from shorter lines were lessened as real wages and thus the opportunity cost of waiting declined; real appreciation of the exchange rate that raised dollar wages substantially and led to an upsurge of consumer imports inflicted a capital loss on domestic holders of foreign exchange; total leisure increased on the account of higher unemployment but anxiety and uncertainty went up as well. 28 REFERENCES Berg, A. and J. Sachs (1992), "Structural Adjustment and Intemational Trade in Eastern Europe: the Case of Poland", Economic Policy, No. 14. Gorecki, B. (1992), "Evidence of a New Shape of Income Distribition in Poland", paper presented at the International Economic Association World Congress in Moscow, 26 August 1992. Gorecli, B. and M. Peczowski (1992), "Polish Household Panel: Preliminary Information", paper presented at the Workshop on Polish household panel in Warsaw, 20-21 November 1992. Lipton, D. and J. Sachs (1990), "Creating a Market Economy: The Case of Poland", Brooking Papers on Economic Activity, 1:75-148. Kordos, J. (1991), "Pomiar ubostwa w Polsce", Wiadomoffi Statystyczne, 11:1-6. Krzyzkiewicz, Z. (1989), "Uwarunkowania polityki piene±no-kreditowej", FYnanse, No. 10-12. Kupa, M. and 0. Fajth (1990), "Hungarian Social Policy Systems and Distribution of Incomes of Households", mimeo, Budapest. Milanovic, B. (1992), "Poverty in Poland, 1978-88", Review of Income and Wealth, September, pp. 329-340. Panek, T. and A. Szulc (1991), "Income Distribution and Poverty: Theory and a Case Study of Poland in the Eighties", Research Centre for Statistical and Economic Analysis of the Central Statistical Office of the Polish Academy of Sciences, Warsaw. Sachs, J. (1990), "A Tremor, Not Necessarily a Quake, for Poland", International Herald Tribune, 30 November. Secretariat of the Economic Commission for Europe (1992), Economic Survey of Europe in 1991-1992, New York: Economic Commission for Europe. Stanovnik, T. "Consumer Subsidization in Algeria", Ljubljana: Institut za ekonomska raziskovanja, 1991. Vecernik, J. "Income Distribution in Czechoslovakia", BIRG Bulletin, p.14. Winiecki, J. "The Inevitability of a Fall in Output in the Early Stages of Transition to the Market: Theoretical Underpinnings", Soviet Studies, 43(4):669-667. World Bank (1991), Czechoslovakia: Transition to a Market Economy, A World Bank Country Study, Washington, D.C.: World Bank. 29 Policy Research Working Paper Series Contact Title Author Date for paper WPS1 151 Is Growth Bad for the Environment? Charles van Marrewijk July 1993 J. Verbeek Pollution, Abatement, and Federick van der Ploeg 33935 Endogenous Growth Jos Verbeek WPS1 152 Population, Heafth, and Nutrition: Denise Vaillancourt July 1993 0. Nadora Annual Operational Review for Fiscal Stacye Brown 31091 1992 and Others WPS1 153 North American Free Trade Alberto Musalem July 1993 P. infante Agreement: Issues on Trade in Dimitri Vittas 37664 Rnancial Services for Mexico Ashl DemirgO9-Kunt WPS1 154 Options for Pension Reform in Tunisia Dimitri Vittas July 1993 P. Infante 37664 WPS1155 The Regulation and Structure of Martin F. Grace July 1993 P. Infante Nonlife Insurance in the United Michael M. Barth 37664 States WPS1156 Tropical Timber Trade Policies: What Panayotis N. Varangis July 1993 D. Gustafson Impact Will Eco-Labeling Have? Carlos A. Primo Braga 33714 Kenji Takeuchi WPS1157 Intertemporal and Interspatial Sultan Ahmad July 1993 E. O-Reilly- Comparisons of Income: The Meaning Canmpbell of Relative Prices 33707 WPS1158 Population Growth, Externalities, Nancy Birdsall July 1993 E. Homsby and Poverty Charles Griffin 35742 WPS1 159 Stock Market Development and Ashl Demirg0g-Kunt July 1993 P. Sintim- Financial Intermedia'y Growth: Ross Levine Aboagye A Research Agendia 38526 WPS1 160 Equity and Bond Plows to Asia and Punam Chuhan July 1993 R. Vo and Latin America: The Role of Global Stijn Claessens 31047 and Country Factors Nlandu Marningi WPS1 161 Increasing Women's Participation in Molly Maguire Teas July 1993 L. Maningas the Primary School Teaching Force and 80380 Teacher Training in Nepal WPS1 162 The Slovenian Labor Market in Milan Vodopivec July 1993 S. Moussa Transition: Issues and Lessons Samo Hribar-Milic 39019 Learned WPS1163 Domestic Distortions and James E. Anderson July 1993 D. Gustafson International Trade J. Peter Neary 33714 Policy Research Working Paper Series Contact Title Author Date for paper WPS1164 Power, Distortions, Revolt, and Hans P. Binswanger July 1993 H. Binswanger Reform in Agricultural Land Relations Klaus Deininger 31871 Gershon Feder WPS1 165 Social Costs of the Transition to Branko Milanovic August 1993 R. Martin Capitalism: Poland, 1990-91 39026 WPS1 166 The Behavior of Russian Firms in Simon Commander August 1993 0. del Cid 1992: Evidence from a Survey Leonid Liberman 35195 Cecilia Ugaz Ruslan Yemtsov WPS1 167 Unemployment and Labor Market Simon Commander August 1993 0. del Cid Dynamics in Russia Leonid Liberman 35195 Ruslan Yemtsov