76599 HOUSING AND INCOME DISTRIBUTION IN RUSSIA: ZHIVAGO'S LEGACY Robert M. Buckley Eugene N. Gurenko "I'm very glad you've given up those rooms. ... We should give up still more." Dr. Zhivago, p. 170 The inadequacy of housing and its effect on the quality of life go a long way toward explaining many aspects of Russian life, but little data has been avail- able to determine how housing affected the relative well-being of the citi- zenry. This paper presents comprehensive data examining for the first time the effect that seventy years of a socialist housing allocation scheme had on the distribution of income. It seems clear that housing provided by the gov- ernment or by employers has a value that can be measured to yield useful inferences about the distribution of income and wealth. This article shows that housing allocation had a progressive effect on the distribution of income in Russia. In addition, when the imputed value of housing is added to household income, the increase in income inequality that occurred in recent years is significantly reduced. The analysis concludes with a discussion of how housing policy could be used to address poverty concerns, an important aspect of the transition process. O n his return home from World War I, Boris Pasternak's fictional char- acter Dr. Zhivago finds that he has to share his family's Moscow man- sion with workers and their families. Since at least that time Russians have been sensitive to the effects of housing on living situations. The housing shortage and the difficulties of living in cramped quarters have been a funda- mental aspect of Russian life and writings about it.1 Unfortunately, aside from figures on the number of square meters of housing space produced and surveys based on interviews with emigres, little statistical material has been available to determine how housing affected the well-being and social position of Russian The World Bank Research Observer, vol. 12, no. 1 (February 1997), pp. 19-32 O 1997 The International Bank for Reconstruction and Development / THE WORLD BANK 19 citizens. Unlike Hungary, where analysts have produced a considerable body of statistical work describing housing conditions since the late 1960s (Szelenyi 1983; Hegedus 1987; and Daniel 1985), Russian researchers were not able to sift through the data to determine the role of housing in the economy.2 The only data available from the former Soviet Union were based on the Berkeley-Duke survey of Soviet emigre households, conducted in the late 1970s. As Alexeev (1988,1990), who examined these data, observes, this survey is not fully repre- sentative of the Soviet population. Atkinson and Micklewright (1992) suggest that data on income distribution in the former Soviet Union appeared next to alcoholism and drug addition on the censor's list of prohibited subjects. Not until 1989 did Russia begin to count as income the overall imputed value of housing in the national income accounts (Ivanov, Rjbushkin, and Homenko 1993). And until now comprehensive data on how housing affected income dis- tribution were not available. The article relies on data from the Russian Longitudinal Monitoring House- hold Survey conducted in 1992, the year before the passage of significant re- forms in housing policy. The survey, funded by the World Bank (World Bank- Goskomstat 1993), was the first nationally representative household survey conducted in Russia and covered 6,128 households.3 Our analysis of these data indicates that housing allocation had a progressive effect on the distribution of income. In addition, when the imputed value of housing is included as income, the increase in income inequality recorded in recent years is significantly re- duced. When housing is excluded from income, the Gini coefficient (a conven- tional indicator of social inequality that equals zero in the case of perfect equal- ity and one in the case of total inequality) for Russia more than doubled between 1987-88 and 1993-94, according to a study by Milanovic (1996). That was the largest increase in the eighteen countries in the study. When housing is included in income, the increase in inequality, while still large, is significantly reduced. Our analysis suggests that when income is adjusted for housing, the distribu- tion of income under the Soviet regime was much more compressed than is commonly thought (Bergson 1984; Gregory and Stuart 1989). Indeed, efforts to fulfill Khrushchev's dream of realizing the communist ideal by 1980, to provide workers with compensation according to need rather than performance, resulted in an extensive housing-based income transfer mechanism—a mechanism that was a fundamental part of socialist wage policy and one that significantly re- duced income inequality. To better understand housing's role in Russian welfare, we describe housing conditions and tenure patterns there and show how the inclusion of housing affects various measures of income distribution. We then consider how housing allocation affected the distribution of income and discuss the methodology used to measure imputed income. Finally, we discuss some of the policy implications of the socialist system's allocating such a large share of income through this in- kind payment, and we then speculate about how this aspect of income distribu- tion might be affecting the transition process. 20 The World Bank Research Observer, vol. 12, no. 1 (February 1997) The Russian Housing Stock The housing conditions of most of Zhivago's contemporaries in pre- revolutionary Russia were dismal. A 1912 census of St. Petersburg and Moscow shows eight occupants per apartment; the comparable figure for Paris was 2.7 (Alexeev, Baker, and Westfall 1991). The average living space per urban dweller was estimated at seven square meters. For workers housing was often a bunk in a rough dormitory attached to a factory. These substandard conditions are no longer the case for the majority of the Russian population. A massive, post-Stalin construction project—one of the larg- est government-sponsored investment programs ever undertaken—built nearly 20 million apartments in eight years (Morton 1980). Between the end of World War II and the late 1970s, the average living space per capita more than doubled, reaching 16.1 square meters in 1992 (World Bank-Goskomstat 1993). Despite this enormous investment, the housing shortage was by no means resolved. In 1992 almost 10 percent of households lived either in overcrowded communal flats or dormitories with less than eleven square meters of floor space for each household member (table 1). More than 20 percent of the residents of St. Petersburg and 17 percent of those in Moscow lived communally. For com- parison, in 1974 almost 30 percent of Moscow's residents lived communally, and in 1960, 40 percent of Soviet citizens did so (Morton 1980). Although the vigorous construction programs substantially reduced the short- age of self-contained housing units, the housing shortage—caused at least in part by the massive destruction of World War II—has never been overcome. In contrast to Western countries, where the ratio of households to the number of housing units—a simple measure of overcrowding—was about 0.9 by the early 1970s, the ratio in Russia stood at 1.17 and has increased since then (Alexeev, Baker, and Westfall 1991). Indeed, by 1992 according to this gross quantitative measure, housing conditions were worse in Russia than in either Poland or Al- bania, countries noted for housing shortages (Guarda 1993). A recent comparative study of housing conditions shows that qualitative in- dicators of housing in Russia are far below the average of those for reforming Table 1. Housing Provision in Russia, 1992 (percent) Urban Share of households living in Russia settlements Moscow St. Petersburg Rentals or sublets 2.1 1.7 0.3 0.6 Dormitories 3.0 3.9 0,9 0.0 Communal apartments 6.6 7.3 17.0 20.3 Self-contained apartments 58.9 69.1 81.8 78.8 Single-family houses 23.0 14.0 0.0 0.3 Shared houses 5.9 3.7 0.0 0.0 Not known 0.5 0.3 0.0 0.0 Source: World Bank-Goskomstat (1993). Robert M. Buckley and Eugene N. Gurenko 21 socialist economies in Europe (Hegedus, Tosics, and Mayo 1996). In 1992 only 44 percent of Russian households had access to running hot water; 66 percent, central heating; and 67 percent, indoor sewerage (World Bank-Goskomstat 1993). Even those who managed to escape from communal flats found their new dwellings poorly planned and constructed. As Ruble (1993, p. 234) noted: "Years of labor by residents are frequently required to correct a familiar litany of irksome deficiencies in the superficially modern Soviet high-rise apartment building of the 1990s; persistent elevator break-downs, plummeting water pres- sure, electrical surges, and upper-story windows that shatter in 'high winds' even though all is calm at ground level; the list goes on." These deficiencies were embedded in the tenure structure of Russia's housing sector. The state owned and maintained almost two-thirds of the housing stock, twice the publicly owned share in Poland, almost triple Hungary's 25 percent share, and considerably larger than the 17.1 percent share in France and the 2.3 percent in the United States (Buckley, Daniel, and Thalwitz 1996). In large cities such as Moscow and St. Petersburg, the state owned almost 90 percent of the housing, in part because private housing construction was prohibited in cities with more than 100,000 residents. The state allocated to households this publicly owned housing stock without regard to price. Indeed, for the most part, rents were set by a 1926 law, which remained nearly unchanged until 1992. Because the law held nominal rents and utility costs constant, the high inflation that occurred during the transition meant that by the end of 1992, gross housing services were essentially free. Our approach to measuring housing income allows us to use the available data to show how this housing subsidy was distributed according to income levels and to examine how this transfer affected the distribution of income. As shown in table 2, income had little or no observable effect on distribution of housing space in 1992. The relationship between income and housing allocation was apparently random: 27.9 percent of households in the lowest income cat- egories and 23.4 percent of households in the highest income category lived in overcrowded housing conditions—less than fourteen square meters per capita. A detailed comparison of the housing available to other income groups also fails to reveal any noticeable patterns. Table 2. Distribution of Urban Housing by Household Income,1992 Average monthly household income per capita (rubles) Housing space per Less 1,501 3,001 4,501 6,001 Mort household member than to to to to than (square meters) 1,500 3,000 4J00 .6,000 7,500 7J01 Fewer than 7.0 3.8 1.6 2.7 1.9 1.6 1.1 7.1 to 14 24.1 19.8 31.1 28.5 27.1 22.3 14.1 to 25 35.0 37.6 43.1 48.9 49.2 48.0 More than 25.1 37.1 41.0 23.1 20.7 22.1 28.6 Total (percent) 100 100 100 100 100 100 Source: World Bank-Goskomstat (1993). 22 The World Bank Research Observer, vol. 12, no. 1 (February 1997) The Measurement of Imputed Income from Housing Despite his caveats about potential biases in his data, the distribution of housing space that Alexeev (1990) gleaned from the 1976-79 survey of emigres was still valid approximately fifteen years later. Table 3 shows that the Gini coefficient for housing space did not change from that survey to the 1992 survey. Con- versely—and not surprisingly, given the changes in wage policy under Gorbachev and the economic shocks associated with the transition—the 1992 survey found that Alexeev's measures of the dispersion and variability of all nonhousing in- come had changed. No longer was there the same compression and constancy of wages. The Gini coefficients and measures of income dispersion both increased sharply, indicating that inequality had grown. The data show that a rapid surge in income inequality occurred after 1989.4 The 1985 Gini coefficient for per capita household income is 90 percent of that for 1989 and only 60 percent of the 1992 figure. Further disaggregation shows that in 1992, 31.7 percent of the country's total net income—excluding housing- related income—went to the richest 10 percent of the population, while only 2.5 Table 3. Distribution of Household Income and Housing per Capita, 1976-92 Gini coefficient Coefficient of variation 1976-79 1992 1976-79 1992 Variable survey survey survey survey Income from all sources 0.34 0.42 79.3 99.0 Socialist income 0.29 0.38 61.7 84.0 Living space 0.28 0.28 67.1 72.0 Number of rooms 0.24 0.20 46.8 57.8 Note: The coefficient of variation provides a standardized measure of income distribution. It is equal to the standard deviation divided by the mean. The definition of household income per capita is the same as that given by Atkinson and Micklewright (1992). The household is defined as a unit comprising one or more persons, including pensioners, chil- dren, and the sick and disabled. Household members may have income from sources other than primary employment, such as a second job or self-employment or income in cash or kind from farming private plots; it also includes capital income and state transfers such as pensions or child benefits. The income is considered shared within a household. Thus, the figure is the sum of household-based and individual- level income. First, the wages, net profits from entrepreneurial and individual economic activities, pen- sions, and unemployment benefits are aggregated over all individuals in a household. This intermediate total is added to household-level income, which consists of net profits from farm activities, subsidies from enterprises and local authorities, family allowances, income from property sold, scholarships, and pri- vate transfers (cash amount plus in-kind valuation by respondent). A more detailed breakdown of the household income components can be obtained from the questionnaire and the Statistical Package for the Social Sciences program written to generate this variable. It should be noted that the definition does not account for imputed rental subsidies on owner-occupied property. The total household floor space per capita is defined as the total housing space occupied by a family, that is, the total floor space of all bedrooms, living spaces, kitchen, bathrooms, lavatories, entry halls, closets, and storage rooms, including unheated areas. The definition also includes any living space at the disposal of any household member in addition to the residence they share. The per capita levels of hous- ing space are obtained by dividing the total housing space per household by the household size. Source: For 1976-79, Alexeev (1988); for 1992, World Bank-Goskomstat (1993). Robert M. Buckley and Eugene N. Gurenko 23 percent went to the poorest 10 percent. Cumulatively, these results indicate that by 1992 Russia had, as Atkinson and Micklewright (1992) suggest, come to look like a market economy. The Gini coefficient for 1992 was almost identical to the United States figure for 1991 (U.S. Department of Commerce 1995). But these figures do not include the value of subsidized housing, which was a significant component of the distribution of income. Rents were approximately zero for all but the poorest households and were only slightly higher for them. Citizens who lived in state or enterprise-owned housing paid low or no rent, in effect receiving substantial in-kind (noncash) transfers. To calculate how the benefits of unpriced housing affected the distribution of income, we derived a measure of the value of the rents and added it to each household's income to obtain a more accurate measure of total household income. We began with Hicks' (1946, p. 172) formula: "A person's income is the maximum value he can consume during a week and still be as well off at the end of the week as he was at the beginning." It is a short definition, one based on theory, and it nicely links current income from wages (or from a profit and loss statement) to the stock of wealth and the consumption associated with changes in it, such as our measure of imputed housing income. Analyses of household income must be based on official wage statistics. If we augment that information by including the amount that would otherwise be spent on housing, we can link the notion of income to household wealth. Hous- ing is the largest component of noncash income received by households, and it is one that the United Nations recommended be included in household income statistics. The approach used here to impute the value of housing follows the United Nations (1968, p. 6.22) definition: "The total of owner-occupied dwellings which is to be included in gross output should, in principle, be valued at rent on the market of the same facilities. It may be necessary to approximate the market rent by an estimate which should cover items such as operating, maintenance and repair outlays.... depreciation and mortgage interest in addition to interest on the owner's investment in the dwelling. . . . " Thus, we must first measure how much rent a household would have to pay to enjoy the same housing ser- vices provided by the state or enterprise. One way to make this notion concrete is to consider an individual who sells financial assets to buy a house with cash. Apart from such considerations as risk, liquidity, and administrative costs, if the person were as "well off" after buying the house as he was before, the imputed income (rent that the owner of the house no longer pays) would be equal to the annual interest that would have accrued on the financial assets. In terms of either the United Nations or the Hicks definition, this imputed value must be included as income if income is to be measured correctly. We used two approaches—the first based on market value and the second on opportunity cost—to estimate the income equivalent of subsidized rent. To ac- count for the possibility that shortages produced by the rationing system might induce households to pay more, we also used a third conceptual approach based 24 The World Bank Research Observer, vol. 12, no. 1 (February 1997) on the "minimum" cash compensation required to obtain adequate housing. If this approach were used, the value of imputed rents could well be higher than that estimated by the methodologies described above. The Market Value Approach The most commonly used approach to measuring imputed housing income is to assume it is equal to the market rental value of the housing. This estimate takes the value of imputed income to be equal to the market value of an analo- gous good. Although a rental housing market did not exist in Russia in 1992, it did exist in industrial nations and several developing countries. Horioka's (1994) careful analysis of household consumption and expenditure patterns in mem- bers of the Organization for Economic Cooperation and Development in 1989 found that the average household's expenditure on housing was 18.5 percent of household income, and the median was 19.2 percent. Those averages varied little from country to country. The World Bank Housing Indicators survey (Mayo and Angel 1993) for 1991 yielded similar results. Covering housing characteristics in fifty-two cities in industrial and developing countries, this survey provides a comprehensive set of data on the housing sector, including the cost and availability of key inputs such as land, infrastructure, building materials; the regulatory environment; demo- graphic variables; finance and subsidies; and the qualitative and quantitative features of the housing stock. These data were used to infer the median value of housing services received by a median-income household in countries whose per capita gross domestic product (GDP) was similar to Russia's. This figure was computed in the form of a rent-to-income ratio and was drawn from a regres- sion of rent-to-income ratios on per capita GDP in terms of purchasing power parity. The Opportunity Cost Approach The opportunity cost approach is the one used by Smeeding and others (1993) and Yates (1994) in measuring the imputed net income from owner-occupied housing in several Western countries. As noted above, the estimates of rental value for Russia take into account the provision not only of housing but also maintenance, depreciation, and utilities at essentially no cost. Hence, except for households in cooperatives, residents effectively received a 100 percent subsidy of gross rent. Based on Laidler's (1969) or Poterba's (1992) analysis, these gross rents can be converted into a rate of return that is approximately three times the 2 percent net rate of return used by Smeeding and others (1993). The gross rate of return on housing is much higher than the net real rate of return on other assets because operation and maintenance costs (about 2.2 percent) and depre- ciation (1 to 1.5 percent) reduce housing returns. Thus to realize a 2 percent net real rate of return requires a 5.7 percent gross rate of return (Alexeev 1990). Robert M. Buckley and Eugene N. Gurenko 25 Renters will now have to cover the costs of operation and maintenance, multi- plying their rental charges by a factor of three. Using the assumption of Smeeding and others (1993) that an average household spends about 7 percent of its in- come on housing, the rental of a dwelling that had cost a household 7 percent of its income will now cost it approximately 20 percent of its income. Both approaches, then, suggest that 20 percent of income approximates the amount that the median household would have had to spend to rent the median amount of housing space it currently occupies. Because this figure may under- state the amount people would actually be willing to pay, we also assumed that gross imputed housing income accounted for 25 percent of income, and consid- ered a lower rent-to-income ratio of 15 percent (as estimated by Alexeev 1990). The direction of change in the Gini coefficients after including housing income is very similar under all three assumptions. As would be expected, the Gini coef- ficient is lower—that is, income inequality is more equally distributed—when housing accounts for a larger share of income, and conversely. Because there are no recorded sales prices for housing in 1992, the technique we used ignored variations in the quality of housing. Kaganova and Malgin (1994) and Pusanov (1993) recently examined these variations for St. Peters- burg and Moscow, respectively. Their work shows that the assumption of in- variant quality overstates the equalizing effects of housing income on the distri- bution of income. With variation in quality considered, the Gini coefficient for the distribution of housing in St. Petersburg rises from 23.9 to 30.7. That is, higher-income households had significantly better quality units. To compute the value of imputed gross rent for other households at various income levels, we assumed that the amount of imputed income received was directly proportional to the square meters of housing provided. In other words, a family with twenty square meters of space per capita received 11 percent more in imputed housing income than residents with the eighteen square meters per capita—the median amount of housing space. Because the market rent for the median level of hous- ing consumption per capita is equal to 20 percent of median family income, the imputed rent for a twenty-square-meter space amounts to about 22 percent of median family income. We then added this measure of income to the total in- come of each household in our data set. Admittedly, this measurement values housing services solely on the basis of space. Previous estimates by the U.S. Central Intelligence Agency (1982) and Bergson (1961) rely on the same approach, however, although the latter used what he terms a "quite arbitrary" means of discounting the quality of private housing relative to public housing. In contrast Prell (1989) attempts to estimate how qualitative improvements in the housing stock in Russia, of the sort that occurred in the United States in the 1960s, may have affected the growth rate of this capital stock. We did not attempt to make such quality adjustments for two reasons. First, we focus on one point in time rather than on how the level of investment may have changed over time as a result of unmeasured changes in quality. Second, 26 The World Bank Research Observer, vol. 12, no. 1 (February 1997) without more data on housing prices, we are unable to sort out how housing was distributed across households. It is by no means obvious that simple mea- sures, such as the age or size of a building, had an effect on housing quality. Di Mao (1974), for instance, discusses the decline in quality in the larger build- ings—contradicting the observations suggested by Prell (1989). Reconsidering the Distribution of Income Table 4 shows the cushioning effect that the imputed amount of housing in- come had on the distribution of income in 1992. In contrast to Yates' (1994) findings for Australia, in which inclusion of imputed housing income changed the Gini coefficient by one percentage point (from 37 to 38), the change for Russia was almost 6 percentage points. One reason for this was that Russian households did not pay for maintenance, depreciation, and utilities. The Gini for combined housing and nonhousing income is about 80 percent of that for nonhousing-related income. In other words, when these in-kind benefits are in- cluded in measures of income, there is a substantial reduction in inequality. Finally, our measure of imputed housing income accounts for about 60 percent of total income for the poorest 20 percent of the population, suggesting that these households were relatively "house rich." These households almost cer- tainly would have preferred to receive their income in a less specific form. Housing allocation also had significant effects on horizontal equity, that is, among persons with similar income but different in other respects (Bergson 1984). For example, we found that gender inequalities in the distribution of individual earnings were significantly offset by imputed housing subsidies because female- headed households have more housing space per capita. The inclusion of im- puted housing subsidies raises the median female-to-male ratio by 7 percentage points—from 0.54 to 0.61. This effect was even more pronounced for the bot- tom 10 percent of the population, for which the female-to-male ratio went up by 11 percentage points (see Buckley and Gurenko 1995). Such changes in the distributional picture are likely to have affected elements of the transition process. One possibility is that the relatively high share of hous- ing income as a proportion of the total income of lower-income families shielded these households in their adjustment to the postcommunist depression. Milanovic Table 4. Inequalities in the Distribution of Combined Household Income and Income per Capita, 1985-92 Coefficient of Income category Gini coefficient variation Income per capita (including housing) 35.4 83 Income per capita (excluding housing) 41.7 99 Source: World Bank-Goskomstat (1993). Robert M. Buckley and Eugene N. Gurenko 2 7 (1996), who discusses the differences between labor market adjustments during the Great Depression in the United States and other market economies and that of the postcommunist depression, finds that in the former case the adjustment took place through job cuts, while in the latter it was through wage adjustments. In market economies, wages were broadly stable in real terms while unemploy- ment multiplied. In Russia the opposite occurred: real wages declined between 40 and 60 percent, and unemployment remained relatively low. How much of this difference in labor market adjustment is attributable to the fact that much of the income of lower-wage workers was unaffected by wage cuts? Did enterprises in Russia, which had traditionally provided most of the social safety net, respond in the way they did because employees' household income would not be affected by such cuts? Conversely, does the illiquidity of the imputed housing income affect the adjustment of labor markets? One recent study of Poland, for example, estimates that as much as 25 percent of its unem- ployment rate in 1992 occurred because workers could not find housing near the available jobs (Coricelli, Hagemejer, and Rybinski 1995). Although we do not know the answers to these questions, the adjustment is a part of the eco- nomic landscape that should not be ignored. As shown in table 4, the distribution of combined income (including housing) was substantially more compressed than the distribution of income derived from wage data or total earnings in the techniques used by Atkinson and Micklewright (1992). Moreover, one has the sense that if such information could have been spliced onto the 1989 comparisons, the Soviet Union would not have looked so nearly like the Western economies but instead more like the socialist regime that it was. Conclusion The distribution of housing in Soviet Russia reduced income inequality and provided a strong cushion against the consequences of the transition. The effects were similar to those described by Smeeding and others (1993) for the OECD economies, but that study found that housing subsidies repre- sented the smallest share of in-kind compensation. Moreover, in their find- ings the total share of all nonwage compensation, 21 percent, was only slightly more than our measure of imputed income from housing. In Russia, in con- trast, it can be inferred from Aim and Sjoquist (1995) that housing accounted for the largest share of in-kind compensation. They found that expenditures on housing-related maintenance alone, which are just a small share of in- kind housing compensation, accounted for nearly half the financial obliga- tions for social services that firms were now transferring to local govern- ments. Housing is certainly not an income source that can be omitted from consideration (as does Kakwani 1995) on the grounds that the rents charged by the state are low. 28 The World Bank Research Observer, vol. 12, no. 1 (February 1997) An important manifestation of the cushioning provided by housing can also be seen by considering housing's role in national wealth, particularly in light of the greater dispersion of the distribution of wealth relative to that of income. In France and the United States, for example, housing accounts for one-half and one-third, respectively, of household wealth. In both countries, wealth is far more dispersed than income (Kessler and Wolff 1991). In a society like Russia's, in which inflation has eliminated most savings of the household sector, imputed income from tangible wealth such as housing is likely to be an even larger com- ponent of wealth than it is in market economies. And because of high rates of inflation, nonresidential wealth is likely to have been more widely redistributed than in market economies. As a result, the almost "give-away" privatization of housing that began in Russia in 1993 gave households at least some savings to help cushion the costs of the transition. This program was an important way of permitting households to exploit the distributional benefits of the old system, a step that has not been pursued nearly as aggressively in many other reforming economies. Because the distribution of housing income in Russia is so much less dispersed than is wage income, while at the same time the distribution of non- residential wealth is almost certainly much more dispersed, housing's privatization had a strongly progressive effect on the distribution of wealth. The privatization of housing may also help address poverty concerns. For example, in 1992 the elderly in the lowest-income quartile had less than one- seventh of the income of young households in the highest-income quartile. But they had five square meters more housing space per person. Because up to 30 percent of the elderly were below the poverty line, being able to liquidate this additional housing wealth could go a long way toward addressing their poverty problem. Devising means by which the poor and the elderly can gain access to this illiquid wealth is by no means a simple task. Complicated questions of property rights, registration, and enforcement as well as such issues as defining condo- minium rights and responsibilities are fundamental management issues that be- devil simple resolution. Nevertheless, at the very least, this source of income should not be ignored, as it has been in recent analyses (see, for example, Barr 1993). We recognize that our data could be significantly improved, particularly as market transactions permit better calculations, but given the scale of the effects involved, the evidence supports a 1977 United Nations recommendation that housing income be included in analyses of income distribution. The housing policies that caught Dr. Zhivago's attention more than seventy years ago continue to affect the basic fabric of life in Russia. Decisions with respect to marriage, children, profession, and job opportunities are no doubt affected by housing considerations. The impact of housing policies on so many aspects of economic behavior makes social compensation systems more difficult to reform. Thus, at a minimum, the former socialist economies should place a greater emphasis on integrating housing decisions into broader measures of eco- nomic activity. Robert M. Buckley and Eugene N. Gurenko • - 29 Notes Robert M. Buckley is with the Operations Evaluation Department of the World Bank, and Eugene N. Gurenko is with Planning and Development Collaborative (PADCO), Wash- ington, D.C. This article was initiated as a component of a World Bank research program. The authors would like to thank Michael Alexeev, Patricia Annez, Lev Freinkman, Jozsef Hegedus, Stephen Mayo, and Blair Ruble for their valuable assistance. 1. Pasternak's interest in housing has been a constant theme of Russian literature. See, for example, Brodsky's "In a Room and a Half," in Less than One: Selected Essays (1985) and the chapter called "Housing the People," in Khrushchev Remembers: The Last Testa- ment. (1980). 2. Szelenyi was exiled from Hungary when his studies, conducted in the late 1960s and early 1970s, showed that the Hungarian mechanism for allocating housing resulted in a regressive distribution of benefits. (His work was published in English in 1983.) 3. The Gini coefficient for 1985 was 25.6; for 1989, 28.9; and for 1992, 41.7. See Atkinson and Micklewright (1992) for 1985 and 1989 figures; World Bank-Goskomstat (1993) for 1992 estimates based on the same definition. 4. A detailed description of the data set is available upon request. References The word "processed* describes informally reproduced works that may not be commonly available through library systems. Alexeev, Michael. 1988. "The Effect of Housing Allocation on Social Inequality: A Soviet Perspective." Journal of Comparative Economics 12:228-34. . 1990. "Distribution of Housing Subsidies in the USSR, with some Soviet-Hungarian Comparisons." Comparative Economic Studies 32(3):138-57. Alexeev Michael, Lee Baker, and Michael Westfall. 1991. "Overview of the Soviet Housing Sector." Paper prepared by the Planning and Development Collaborative, Washington, D.C. for USAID Office of Housing and Urban Program. Processed. Aim, James, and David Sjoquist 1995. "Social Services and the Fiscal Burden in Russia." Comparative Economic Systems 37(4): 19-30. Atkinson, Allan, and J. Micklewright. 1992. Economic Transformation in Eastern Europe and the Distribution of Income. Cambridge, U.K.: Cambridge University Press. Barr, Nicholas. 1993. Income Transfers and the Social Safety Net in Russia. Studies of Economies in Transformation 4. Washington, D.C: World Bank. Bergson, Abraham. 1961. The Real National Income of Soviet Russia Since 1928. Cam- bridge, Mass.: Harvard University Press. Brodsky, Joseph. 1986. Less than One: Selected Essays. New York: Farrar, Strauss, Giroux. Buckley, Robert M., and Eugene N. Gurenko. 1995. "Distribution of Income and Housing in the Period of Economic Transition in Russia: Discussion of Data and Further Analy- sis." World Bank, Operations Evaluation Department, Washington, D.C. Processed. Buckley, Robert M., Zsuzsa Daniel, and Margret Thalwitz. 1996. "The Cost and Trans- parency of Housing Subsidies in Socialist Economies." In Robert M. Buckley, ed., Housing Finance in Developing Countries. London: Macmillan Press. Coricclli, Fabrizio, Krzysztof Hagemejer, and Krzysztof Rybinski. 1995. "Country Stud- ies: Poland." In Simon Commander and Fabrizio CoriceUi, eds., Unemployment, Re- structuring and the Labor Market in Eastern Europe and Russia, EDI Development Stud- ies. Washington, D.C: World Bank. Daniel, Zsuzsa. 1985. "The Effect of Housing Allocation on Social Inequality in Hun- gary." Journal of Comparative Economics 9:391—409. 30 The World Bank Research Observer, vol. 12, no. 1 (February 1997) DiMao, Alfred. 1974. Soviet Urban Housing: Problems and Policies. New York: Praeger Publishers. Guarda, G. C. 1993. "Housing Finance in Albania: On the Path of Economic Reform." Housing Finance International (December):5-10. Gregory, Paul R., and Robert C. Stuart. 1989. Comparative Economic Systems. 3d ed. Boston: Houghton Mifflin. Hegedus, Jozsef. 1987. "Reconsidering the Roles of the State and the Market in Socialist Housing Systems." International Journal of Urban and Regional Research 11(1, March):79-97. Hegedus, Jozsef, Ivan Tosics, and Stephen Mayo. 1996. "Transition of the Housing Sector in the East-Central European Countries." Review of Urban and Regional Studies 8:101-36. Hicks, James R. 1946. Value and Capita. London: Clarendon Press. Horioka, Charles. 1994. "Japan's Consumption and Saving in International Perspective." Economic Development and Cultural Change 42(2, January):293-317. Ivanov, Yuri, Boris Rjbushkin, and Tarjana Homenko. 1993. "The Introduction of the SNA into the Official Statistics of the Commonwealth of Independent States." Review of Income and Wealth 39(3, September):279-95. Kaganova, Olga, and Alexander Malgin. 1994. "What Is Happening in St. Petersburg's Housing Market?" Voprosi Economiki 10:120-131. Kakwani, Nanak. 1995. "Income Inequality, Welfare, and Poverty: An Illustration Using Ukrainian Data." Policy Research Working Paper 1411. World Bank, Policy Research Department, Washington, D.C. Processed. Kessler, Denis, and Edward N. Wolff. 1991. "A Comparative Analysis of Household Wealth Patterns in France and the United States." Review of Income and Wealth 37(3, Septem- ber):246-64. Khrushchev, Nikita. 1974. Khrushchev Remembers: The Last Testament. Translated and edited by Strobe Talbott. Boston: Little, Brown and Co. Laidler, David. 1969. "Income Tax Incentives for Owner-Occupied Housing." In Arnold Harberger and Martin Bailey, eds. The Taxation of Income from Capital. Washington, D.C: Brookings Institution. Mayo, Stephen, and Shlomo Angel. 1993. Housing Indicators Program. A joint program of the U.N. Centre for Human Settlements and the World Bank, Washington, D.C. Processed. Milanovic, Boris. 1996. "Income, Inequality, and Poverty during the Transition." Research Paper Series 11. World Bank, Washington, D.C. Processed. Morton, Henry. 1980. "Who Gets What When and How? Housing in the Soviet Union." Soviet Studies 32(2):235-59. Pasternak, Boris. 1991. Dr. Zhivago. 14th ed. New York: Ballanrine Books. Poterba, James. 1992. "Taxation and Housing: Old Questions, New Answers." American Economic Review 82(2, May):237-42. Prell, M. A. 1989. "The Measurement of Housing Output: U.S. and Soviet Case Studies." Review of Income and Wealth 35(3, September):297-316. Pusanov, Alexei. 1993. "Housing Quality in Moscow and Its Relation with Households' Resources." Voprosii Ekonomiki 7:89-99. Ruble, Blair. 1993. "From Khrushcheby to Korobki." In William C. Brumfield and Blair Ruble, eds., Russian Housing in Modern Age. Cambridge, U.K.: Cambridge University Press. Smeeding, Timothy, Paul Saunders, John Ceder, Stephen Jenkins, Johan Fritzell, Aldi Hagenaars, Richard Hauser, and Michael Wolfson. 1993. "Poverty, Inequality, and Family Living Standards Impacts across Seven Nations: The Effects of Non-Cash Subsi- Robert M. Buckley and Eugene N. Gurenho 31 dies for Health, Education, and Housing." Review of Income and Wealth (Septem- ber):229-56. Szelenyi, Ivan. 1983. Urban Inequalities under State Socialism. Oxford: Oxford University Press. United Nations. 1968. A System of National Accounts. Studies in Methods, Series If, No. 2, Rev. 3. Department of Economic and Social Affairs. New York. . 1977. Provisional Guidelines on Statistics of the Distribution of Income, Con- sumption, and Accumulation of Households. Studies in Methods, Series M, N . 61. New York. U.S. Central Intelligence Agency. 1982. Soviet Statistics on Capital Formation. SOV 82- 1000093, August. U.S. Department of Commerce. 1995. Personal communication. Research Department, Washington, D.C. World Bank-Goskomstat. 1993. Russian Longitudinal Monitoring Household Survey. Washington, D.C. Yates, Judith. 1994. "Imputed Rent and Income Distribution." Review of Income and Wealth 40(1, March):43-66. 32 The World Bank Research Observer, vol. 12, no. 1 (February 1997) AN EXCHANGE ON PROJECT EVALUATION Because of the importance of investment to development, the economic ap- praisal of projects has long attracted the attention of development theorists and practitioners alike. The subject has recently been enjoying a revival of interest, as several studies revealed a decline in the quality of economic analysis used, both in the World Bank and elsewhere. The following three articles reflect some of the views expressed in the recent resurgence of interest in project analysis. We hope that these articles will stimulate further contributions, as the develop- ment community grapples with this important and evolving subject.