VI How Large Was the Output Collapse in Russia? Alternative Estimates and Welfare Implications
  • 1 0000000404811396 Monetary Fund
  • | 2 0000000404811396 Monetary Fund


Since the onset of economic transition in Russia, the perception has become increasingly widespread that output and living standards are highly unlikely to have dropped as much as the official numbers indicate.2 Many observers find it hard to believe that the size of the Russian economy really halved between the late 1980s and 1994. This judgment, however, remains very much an impressionistic one, relying on anecdotal or partial evidence rather than on a documented set of alternative estimates. The ambition of this paper is to show that the output decline was much less than what is recorded in the national accounts data published by the Russian State Committee for Statistics (Goskomstat). The paper also discusses some of the welfare implications of the fall and recomposition of output. It concludes that too many tears have been shed on measured output losses, and that the transition process itself should not be blamed for the dismal heritage with which it was endowed.

Since the onset of economic transition in Russia, the perception has become increasingly widespread that output and living standards are highly unlikely to have dropped as much as the official numbers indicate.2 Many observers find it hard to believe that the size of the Russian economy really halved between the late 1980s and 1994. This judgment, however, remains very much an impressionistic one, relying on anecdotal or partial evidence rather than on a documented set of alternative estimates. The ambition of this paper is to show that the output decline was much less than what is recorded in the national accounts data published by the Russian State Committee for Statistics (Goskomstat). The paper also discusses some of the welfare implications of the fall and recomposition of output. It concludes that too many tears have been shed on measured output losses, and that the transition process itself should not be blamed for the dismal heritage with which it was endowed.

Output, of course, really did collapse in many sectors, owing, interalia, to the breakdown of internal and external trading arrangements; the contraction in demand (for example, military procurement and investment programs); the compression of imports of some intermediate goods; price liberalization, insofar as changes in relative prices rendered some activities inviable; the archaic financial system, which limited enterprises’ ability to engage in intertemporal substitution and hampered settlements; and domestic and cross-border “domino” or contagion effects. It cannot be ruled out a priori that the cumulative fall in production was even larger than the one experienced in the United States during the Great Depression of 1929-33 (Table 1)3 and larger than any downturn registered in Russia during the previous seventy years.4

Table 1.

Selected Countries in Transition: Cumulative Officially Registered Real Output Declines

(Percent change compared to base year)

article image
Sources; Central Statistical Administration of the U.S.S.R.; Statistical Committee of the Commonwealth of Independent States; Goskomstat of the Russian Federation; Rajewski (1993); United States Bureau of the Census (1975); World Bank (1994); statistical offices of Albania, Poland. Romania, Bulgaria, the Czech Republic, and the Slovak Republic; and authors’ calculations.Note: In some cases, pretransition output peaked before 1989-90.

Gross sales used for Albania; for the other countries, gross output used.

The figures shown for Bulgaria do not reflect the large upward revision of the real GDP series for 1990-91 implemented by the statistical authorities in 1992.

There are a number of reasons, however, to suspect that the official real GDP statistics are overstating the output collapse.5 The traditional statistical apparatus, based on a census rather than a sampling approach, is obviously missing an increasingly large portion of economic activity, as publicly acknowledged by both the former head of Goskomstat (Guzhvin (1992)) and his successor (Yurkov (1994)).6 Furthermore, the official data show an increasing discrepancy between output and consumption in many sectors, with the latter falling much less (or rising more) than the former.7 Likewise, the official data imply a puzzling divergence between the huge decline in output and the much more resilient behavior of household incomes, which have tended to recover and by late 1994 had reverted to the level of the late 1980s.

Other indications supporting the presumption that output is higher than reported by Goskomstat include evidence from other transition economies. For example, Rajewski (1993) shows that the cumulative fall in real GDP in Poland from 1989 to 1991 was probably of the order of 5-10 percent, in stark contrast to the officially registered 18 percent drop, and that actual consumption also fell less, a result consistent with the reestimation of the decline in 1990 carried out earlier by Berg and Sachs (1992). According to the “Czech Economic Monitor” (1994) and the “Slovak Economic Monitor” (1994), similar results obtain for the Czech Republic and the Slovak Republic. Finally, it could be conjectured that, if half of the economy had genuinely evaporated, major social upheavals would already have been observed.

Some of these reasons may be less than fully compelling. If underground activities represent a broadly constant share of the officially measured economy, as they do in many countries, ignoring the shadow sector will not bias growth rates. However, underrecording is most blatant in those sectors that are expanding, particularly the new private entities, reflecting the highly plausible motivation of tax evasion, as well as the attempt to avoid all kinds of government regulations. Consumption may be declining less than output because less of the latter is wasted on the way to the final user, because the share of investment goods is falling, or because of destocking. Also, the monetization of transactions associated with the move to a market economy would in itself increase household money incomes and expenditures. Lastly, the Russian population has endured so much hardship in the past that it may be showing more patience than outsiders would expect.

This paper focuses on the size of the output collapse and on its welfare implications but will not dwell on its causes. Much of the discussion is relevant for the other countries of the former Soviet Union, where large declines in output have analogously been recorded (Table 1).8 The paper proceeds as follows. The first section discusses the extent to which pretransition output is overstated by the official statistics, and what the magnitude of the understatement of economic activity may have been in the course of the transition. In the process, alternative real GDP estimates are proposed. The next section discusses the benefits of moving away from a teratogenetic system. Some of the tremendous inefficiencies characterizing the centrally planned economy lost their prominence as the latter was dismantled, but the associated welfare gains are often overlooked—even though they should be part and parcel of any analysis of the output collapse. Finally, some concluding remarks on welfare are offered.


Growth rates were probably overstated in the old regime, as argued by Khanin and others.9 While, under central planning, incentives were mostly geared toward overreporting, they are now reversed, suggesting that the rates of output decline since the onset of transition are likely to be overstated.

Overstatement of the Base

Official output data under the Soviet regime suffered from manipulations by the statistical agency and from falsification by the reporting units themselves. Perversely, cheating was sometimes necessary for the reporting firms to overcome the inefficiencies of central planning and meet plan targets. A variety of tricks were used to document the achievement or over-achievement of the objectives laid out in the Plan, ranging from sheer tampering with the raw data to the inclusion in finished goods totals of brak (spoilage, rejects, and substandard goods) and incompletely assembled articles.10

Insofar as the resulting officially recorded pretransition output levels were artificially high, a portion of the decline recorded for 1990-94 could constitute a belated recognition of earlier overstatements. Khanin (1992), for example, argues that actual output in the late 1980s has been vastly overstated by Goskomstat. If, indeed, pan of the measured output around 1989 was essentially fictitious, part of the subsequent collapse may also have been an illusion. Importantly, this would imply that the centrally planned economy had exhausted its growth potential well before the start of radical reforms.

It should be recognized, however, that enterprise managers in the centrally planned economy also faced incentives to underreport, not least to conceal the illegal appropriation or diversion of production. Another incentive to underreport was related to the so-called ratchet effect: as firms’ targets were set on the basis of registered performance, it was prudent not to advertise success too boldly. Managers perhaps also underreported unintentionally, owing to prior theft or pilferage of the product. Nevertheless, on balance, the overall impression was that traditionally the underreporting bias was smaller than the overreporting one (Grossman (1960)).

It should also be acknowledged that the overstatement of activity in the official sector in the late 1980s was offset to some extent by the understatement of the growing activity in the “informal” sector, notwithstanding the attempts initiated by the U.S.S.R. Goskomstat at that time to include estimates of the second economy in the computation of GDP.11

On the whole, it seems fair to conclude that, although the pretransition peak of output in the registered economy was overstated by Goskomstat, the actual size of overall GDP in the late 1980s, including the shadow economy, may have been smaller or larger than indicated by the official statistics.

Understatement of Output Since the Onset of the Transition

Reform of the statistical apparatus used by Goskomstat, which had originally been set up as a central planning instrument, has lagged the breakdown of the command economy. In particular, exhaustive reporting of production outcomes, as opposed to business surveys, has so far continued to be a key data collection tool.12 As a result, whole chunks of economic activity have vanished from the official statistics even as many new entities escape recording. Moreover, reporting incentives are presumably inversely correlated with output expansion, as enterprises would readily report sharp output declines in order to justify claims for subsidies, tax exemptions, cheap credits, or other favors but would be reluctant to advertise relatively buoyant performances, which would imply substantial tax liabilities. Thus, the official rates of decline in production tend to depict what is happening in the shrinking state sector more than the movements in overall economic activity.

A number of inconsistencies between the official data collected from various sources point to a significant understatement of output. One of the most prominent inconsistencies is the sharp divergence between the output and consumption of various food items (Table 2). Changes in net imports, dishoarding, and declines in waste are unlikely to account for the full discrepancy between the very large drop in production and the small decrease—or, in some cases, the increase—in consumption. As the underreporting bias is less of a problem for consumption data, the odds are that output is substantially understated.

Table 2.

Russia: Measured Output and Consumption of Selected Items

(Percent change)

article image
Sources: Goskornstal of the Russian Federation (production); and Center for Economic Analysis (consumption).

In the case of bread, it is reasonable to believe that, as indicated by household budget surveys, household consumption increased because bread is a typical inferior good. Imports and exports are virtually nil, implying that changes in foreign trade flows cannot be invoked to reconcile changes in output and consumption. Similarly, the scope for intertemporal substitution in the form of hoarding and dishoarding is negligible. One rationale for the observed divergence could be that the quantity of bread used as fodder declined considerably, thus presumably offsetting to some extent the increase in household consumption. However, as the price of bread did not rise more than that of grain, and as privately owned livestock increased, this is unlikely to be a plausible explanation, given that feeding bread to animals was done only by households, not state or collective farms. Indeed, Goskomstat seemingly continues to rely exclusively on production figures collected from factories, even though small-scale bakeries have sprung up in many cities.

In the case of meat and milk, imports represented no more than 10-15 percent of consumption in 1994. Again, changes in net imports cannot account for the gap between production and absorption. Most probably, the missing output of milk and meat is produced but either not reported or underreported by privatized enterprises, new small farms, or even collective farms equipped with their own processing facilities.

More generally, published Goskomstat national accounts data (Goskomstat of the Russian Federation (1994), for example) suggest that the decline of real GDP based on end-use categories is smaller than the drop derived from the production side. These data indicate, interalia, that Goskomstat has started to make upward adjustments to better measure consumption but has not correspondingly amended its techniques to measure the production side. This discrepancy will be exploited below to produce an alternative GDP series.

A second type of inconsistency pertains to financial stocks and flows by main sectors, as compared to these sectors’ shares in value added (Table 3). It appears that a major portion of the credits (50 percent) go to the category “other sectors,” which contributes only marginally to GDP (about 12 percent). Similarly, current accounts with the banking system are predominantly held by these same “other sectors.” In all likelihood, these sectors comprise entities that either are not registered with or do not report to Goskomstat, or both.13

Table 3.

Russia: Sectoral Shares in Credits, Current Accounts with Banks, and GDP

(In percent)

article image
Sources: Central Bank of Russia; Center for Economic Analysis; and authors’ estimates.

As of January 1, 1994.

Provisional, in 1993.

Derived as a residual in the first two columns; to estimate the share of these same other sectors in the GDP decomposition, it is necessary to subtract the contributions to value added of the banking sector and of the budgetary, public, and government organizations (which are also excluded from the first two columns).

A third source of doubt stems from the comparison between the path of electricity consumption and that of output. A priori, one would expect a high degree of positive correlation between these two variables. However, electricity consumption both for industry and overall declined much less than officially measured output (Chart 1).14 Of course, the divergence between the two series partly reflects the existence of fixed costs, but its magnitude and persistence may also indicate that the actual collapse in output was not as large as the official data state, especially considering that the emerging private sector is likely to be less energy intensive than the old production apparatus. The year-on-year elasticity of electricity consumption with respect to actual output may be quite volatile, however; it depends on parameters that are hard to quantify, such as sectoral price elasticities of electricity use, ongoing changes in the output mix, and the technological scope in the short run for reducing wasteful usage and substituting one energy source for another. The buildup of large arrears on electricity bills should also be taken into account. It would therefore be difficult to go one step further and, as Dobozi and Pohl (1995) have done, estimate the bias in the output series based on electricity consumption alone.

Chart 1.
Chart 1.

Russia: Officially Measured Output Versus Electricity Consumption

(In percent)

Source: Goskomstat of the Russian Federation.

Besides the above inconsistencies, it is clear that numerous activities cannot be captured through the traditional reporting channels, particularly in the service sector.15 Some of the areas where the problem is most acute include street trading, private taxi services, real estate services, individual translation services, tutoring, home manufacturing (such as handicrafts), small-scale private manufacturing (for example, of furniture and clothing), construction and renovation (such as provided by crews working on private housing projects), repairs of various kinds, private-practice medicine, poaching, pilfering of material inputs, such as timber or gasoline, from state-owned enterprises for private resale, small-scale smuggling (by so-called business tourists), illegal exporting of raw materials (particularly petroleum and nonferrous metals), private “security” services, and production of “moonshine” vodka.16

The importance of these activities is reflected in labor market statistics: according to Goskomstat, secondary employment is rising and reached 8 million people by mid-1994, out of a total labor force officially estimated at 75 million persons. Half of these 8 million held secondary employment on a permanent basis, of which 2.5 million were involved in “middleman trading activities.” The survey data cited by Rose (1994) suggest that involvement in the second economy is probably even more widespread than Goskomstat’s published estimates indicate. Another sign of the importance of this type of hidden output is the rapidly growing share in household incomes of the category of “business activities, interest and dividends, and other sources”—which accounted for 38 percent of the total by 1994—and the correlative sharp decrease in the share of wages.

While many of these activities never entered the national accounts in the first place, others apparently disappeared from the Goskomstat totals. Inspection of the deseasonalized monthly industrial output series shows very substantial declines in January 1993 and January 1994.17 These falls are not observed for sectors such as the fuel and energy complex, where output can be monitored relatively closely, but rather for sectors such as light industry, where the extent of privatization and the nature of the production process make control more difficult. This suggests that, with the advent of a new year and the changing of tax and other rules, a number of enterprises decided to stop reporting and simply dropped out of the official statistics.18

As mentioned earlier, Goskomstat officials have long recognized that underrecording is a major problem. Indeed, several adjustments have started to be made by Goskomstat to the output data provided through the traditional reporting channels. Specifically, as of mid-1994, the raw numbers collected for trade, construction, and agriculture were supplemented by information from other sources (household budget surveys, customs surveys, building permits, and other survey data),19 These adjustments added up to about 10 percent of GDP at that time. Another innovation has been the inclusion from July 1994 onward in the official monthly data on gross industrial output of an estimate for the heretofore excluded production of small enterprises. This adjustment has raised the level of that series by some 10 percent. While such efforts are most welcome, they are long overdue and apparently cover only a few sectors.

Alternative Real GDP Estimates

The most significant adjustments carried out so far by Goskomstat pertain to retail sales, which form the bulk of household consumption. In 1991, before revisions, they represented 83 percent of household consumption. As shown in Table 4, a drastic revision was implemented, implying that, instead of a cumulative decline of over 40 percent between 1990 and 1994, the volume of retail sales fell by only 2 percent. The revision reportedly involved three markups to the “old” series, which, for 1993, amounted to 5 percent on account of undercoverage of outlets, 20 percent for underreporting, and some 36 percent to account for street trading and supply through “hectic trade” (that is, imports by individuals in their private capacity), respectively. Paid services were also adjusted by Goskomstat. Nevertheless, the resulting cumulative 70 percent decline still seems implausibly large. The size of the revisions shown in Table 4 is of the same order of magnitude as the gap between the sum of the old retail sales and paid services series and consumption as estimated via household budget surveys.

Table 4.

Russia: Sales of Goods and Services

(Annual percent change)

article image
Source: Goskomstat of the Russian Federation.

Goskomstat, however, neither carried out the corresponding adjustments on the supply side nor compiled real GDP anew from the demand side.20 As a result, the early estimates for real GDP continue to be used even though the evidence on expenditures calls for a re-estimation of production.21 The exercise could be attempted by adjusting the raw output data for those sectors in which underrecording seems to be most acute, using, for example, such data as employment (including secondary employment), combined with some assumptions about productivity or even information obtained from enterprise surveys. This route is left to those who have access to the unpublished data collected by Goskomstat.

The corrections proposed below rely solely on public information and deliberately err on the conservative side. In other words, the adjustments are of a partial nature, and, when in doubt, the smaller volume estimates were used. Moreover, no conjectures about the size and growth of the shadow economy are introduced, nor is there any speculation about the quantitative effect of improvements in the quality of output.22 Therefore, the resulting real GDP path should be interpreted as a lower bound rather than as the most plausible estimate of actual GDP.

Starting from 1990,23 and using that year’s prices, the expenditure side of GDP was adjusted in the following way (see Table 5 for details). Household consumption was augmented by taking into account the revisions appearing in Table 4. No allowance was made for the likely increase in households’ domestic production and consumption of food. Changes in public consumption were estimated on the basis of budget execution data. Changes in fixed investment were derived taking into account the rising share of capital repairs (as opposed to the installation of new capacity). Lastly, given that considerable uncertainty surrounds any estimates of stock building and net exports, alternative assumptions with respect to their real rates of change were tried out. The estimates actually used involved a downward correction of the official Goskomstat data for inventory accumulation and net exports. In any event, the contribution of these two components to the rate of change of GDP was small, as they represent a relatively modest fraction of total GDP.

Table 5.

Russia: Adjusted Real GDP at 1990 Prices

(In billions of rubles)

article image
Sources: Goskomstat of the Russian Federation; and authors’ estimates.

Based on revised Goskomstat retail sales series (seeTable 4).

Based on revised Goskomstat paid services series (see Table 4).

Including capital repairs.

See Koen (1994) for a behavioral rationalization of stockholding in 1991 and 1992.

Allowance is made for the underreporting of imports and exports (see International Monetary Fund (1992), (1993), and (1995) for details on the measurement of balance of payments flows).

Based on this approach, real GDP appears to have declined significantly less than indicated by the official Goskomstat statistics. By and large, the latter overstate the annual rate of decline over the years under consideration by at least 4-7 percent (Table 6 and Chart 2); the cumulative decline in real GDP from its pretransition peak to 1994 would thus be of the order of one third, rather than one half.

Table 6.

Russia: Alternative Real GDP Estimates

(Annual percent change)

article image
Sources: Goskomstat of the Russian Federation; and authors’ estimates.

Quantities are valued at the prices of the previous year. Decimals are not published.

Chart 2.
Chart 2.

Russia: Alternative Estimates of Real GDP1

(In percent)

Sources: Goskomstat of the Russian Federation; and authors’ estimates.1 No correction is made for 1990.

The re-estimation carried out above at 1990 prices was then repeated, using prices of the previous year. This second estimation produced a series that is more directly comparable to Goskomstat’s.24 The resulting series differs only marginally from the one compiled by using 1990 prices; cumulatively, they lead virtually to the same estimate.


Notwithstanding the substantial bias distorting official real GDP statistics, there is no doubt that the total volume of gross and net output has dropped very steeply in Russia since the late 1980s. It is often posited that the welfare impact must have been of the same order of magnitude as the actual decline in gross output. However, as emphasized by Winiecki (1991), for example, such an inference is misleading because it ignores the ongoing changes in the composition of output and the sharp reduction in deadweight losses associated with price liberalization and hardening budget constraints.

Machines to Make Machines

One of the characteristics of the centrally planned economy was a massive overaccumulation of capital and, therefore, a slowdown of growth despite very high investment rates. The glorification of gross industrial output, the reliance on an extensive and semi-autarkic growth strategy, the correlative use of antiquated technologies and emphasis on installation of new capacity at the expense of maintenance and distribution, and the constraints imposed by the Plan resulted in the use of more production goods per unit of consumer good than would have been necessary under more efficient arrangements. Not only were productivity levels low, but the decline in the efficiency of investment also accelerated during the second half of the 1980s, as evidenced by the rise in both the ratio of unfinished construction and uninstalled equipment to gross fixed investment and the ratio of investment to gross output.25

When the tyranny of plan objectives and constraints ceased, the incentives to continue with overinvestment receded, and the composition of domestic absorption changed. From 1989 to 1993, the share of fixed investment in GDP (at current prices) fell from 32 percent to 21 percent. At the same time, the output mix shifted from producer goods in favor of consumer goods (Chart 3). The steepest declines were recorded in industry, for investment goods such as locomotives, freight wagons, bulldozers, cranes, tractors, machine tools, electric motors, turbines, and steel pipes. However, a key reason for the dramatic drop observed in the production of many of these items is that they were of very poor quality or obsolete in their design. Table 7 suggests, for example, that about six times as many tractors were needed per unit of grain output in Russia as in the United States.26 That such equipment virtually ceased to be produced should not be lamented as an output “loss”; rather, it should be interpreted as a sign that market forces were beginning to operate, Correlatively, as can be inferred from Table 7 for the case of steel, output for some items declined from excessively high levels.

Chart 3.
Chart 3.

Russia: Industrial Output

(In percent)

Source: Goskomstat of the Russian Federation.
Table 7.

Production of Selected Goods in Russia and in the United States, 1989

(Per capita)

article image
Source: Goskomstat of the Russian Federation.

Military Output

Given the highly militarized structure of the Soviet production apparatus, a particularly important dimension of the shift in the composition of output is related to the conversion process in the military-industrial complex launched in the late 1980s. Traditionally, enterprises in the military-industrial complex produced, in addition to armaments, the bulk of civilian intermediates, such as presses for the shoe industry, spinning looms, and agro-industrial equipment, as well as the bulk of civilian final goods, such as refrigerators, ovens, vacuum cleaners, television sets, videocassette recorders, cameras, radios, tape recorders, and sewing machines. Civilian products represented some 44 percent of the output of defense enterprises in 1988.

The overall collapse in output did not spare defense enterprises (Table 8). However, the blow was cushioned by a rapid shift in production from military to civilian goods. Output declined only moderately or barely at all between 1989 and 1993 for a number of the consumer goods manufactured by defense enterprises (for example, refrigerators, vacuum cleaners, television sets, and sewing machines).27 This shift in production, combined with drastic cuts in defense procurement, caused the share of civilian products in total output to approach 80 percent by 1993. Notably, defense enterprises not only stepped up the production of existing goods, but also attempted to diversify and innovate, for example, in the areas of electric household appliances and medical equipment.28

Table 8.

Russia: Conversion of Enterprises in the Military-Industrial Complex

(Annual percent change)

article image
Source: Center for Economic Analysis.

Queues and Shoddy Consumer Goods

The liberalization of internal and external trade in 1992 resulted in a dramatic expansion of consumer choice and an equally spectacular curtailment of queues.29 Under central planning, enterprise managers had an incentive to choose the intracommodity assortment that would maximize plan fulfillment in terms of the specified physical unit of measurement rather than the one that would satisfy consumers. The death of the Plan, the liberalization of prices, and the opening up of the economy meant that a wide range of consumer goods and services that were previously unavailable or restricted to an elite were henceforth on sale in kiosks and private shops or offered by private companies. Moreover, as the real exchange rate of the ruble appreciated,30 Russian consumers gained access to a widening array of foreign goods, which contributed to the switch from a sellers’ to a buyers’ market.

The availability of goods continued to improve over time: white the overall retail availability coefficient (calculated as the average for selected items of the percentage of the main cities in which those items were available) was approximately 50 percent for food products and 70 percent for nonfood goods in 1992, it exceeded 90 percent by December 1994.31 This statistic is consistent with the evidence from opinion surveys, which show a sharp decline in the time spent in queues: according to one such poll,32 the proportion of respondents spending at least one hour a day in a queue fell from 67 percent in January 1992 to 23 percent two years later.33 At the same time, anecdotal evidence strongly suggests that the availability of services also improved considerably.

As stressed by Lipton and Sachs (1992), among others, traditional indicators of household welfare do not capture the benefits entailed by the broadening of the consumption set and by the reduction of the time spent searching and queuing. It remains a moot point whether, as suggested by Roberts (1994), these gains exceed the measured real income losses suffered in the wake of price liberalization, but they are clearly very substantial.

Efficiency in Consumption

The former economic system was characterized not only by chronic overinvestment, a high degree of militarization, and pervasive shortages of consumer goods, but also by waste on a large scale in the consumption of intermediates and final output. These characteristics are distinct but obviously interrelated, and examples of such behavior abound in the literature (see, for instance, Grossman (1960) and Liapis and Markish (1994)). They include the deliberate destruction of producer goods, which sometimes took place when performance was measured not by output but by the consumption of an input (scrapping of unused structural steel by construction enterprises or the wanton spilling of gasoline). Another example is the accumulation of unnecessary ton-miles (actual, not written-up) hauled by trucks. More generally, the rates of breakage and spoilage in the course of the distribution process were notoriously high, while the goods received by customers often deviated considerably from their alleged specifications, forcing end users to adapt them to their needs at considerable cost. These gross inefficiencies meant that some enterprises were actually engaged in value-subtracting activities. Declining raw output in such instances is value additive.

The rise in real interest rates and the hardening of budget constraints associated with the transition significantly increased the cost of the above-mentioned “losses.” To the extent that their occurrence lessened, there was scope for consumption to decline less than gross output. Admittedly, some inefficiencies were too deeply embedded in the organization of the economy to disappear overnight. Moreover, the disruption of the traditional linkages caused by the abandonment of the old rules entailed significant but temporary coordination failures and the transitory persistence of wastage. Thus, short-run rigidities meant that the gains allowed by the move to market-based arrangements would materialize only over a period of several years.


The earlier sections showed that output fell less in Russia than stated in the official statistics and that part of the decline was not detrimental to welfare. This concluding section tries to come closer to an overall judgment on aggregate welfare. A definitive verdict is, of course, bound to remain elusive, as some important dimensions cannot be quantified, but some of the oftencountered misconceptions about the welfare impact of the output collapse can be dismissed.

A very crude, extremely inadequate, but frequently cited welfare indicator is the officially measured volume of gross industrial output. Many commentators in the Russian press, in line with Soviet prejudices, continue to describe the evolution of living standards over the past few years as if they were perfectly correlated with this indicator and thus assert that, on average, these standards have tumbled by more than one half (Chart 4).34

Chart 4.
Chart 4.

Russia: Alternative Welfare Measures

(1989 = 100; in real terms)

Sources: Goskomstat of the Russian Federation; Center for Economic Analysis; and authors’ calculations.1 Defined as the product of average real per capita income and one minus the Gini coefficient.2 Revised yearly series, adjusted for undercoverage, underreporting, and “hectic trade” (see Table 4).3 Yearly series.4 Yearly series through 1991, quarterly series thereafter.

A more relevant production-based measure is GDP corrected for underreporting, as estimated above. On this account, aggregate welfare would have declined by no more than one third (Chart 4). However, if only because of the changes in the output mix, GDP may not be a satisfactory proxy for welfare. Another reason to prefer an absorption-based measure can be illustrated by the case of oil. A large portion of the oil produced in Russia was traditionally sent to the other states of the former Soviet Union at a price so low that these shipments in effect represented a huge subsidy. The steep decline in Russian oil production since the late 1980s was accompanied by a considerable compression of these subsidized deliveries. Hence, part of the output decline was the counterpart of a cut in subsidies to foreigners, which thus did not hurt domestic living standards.35

Average real income per capita may be considered a better proxy for welfare than GDR. This measure suggests that, by the second half of 1994, welfare had broadly reverted to its 1989 level. It could be argued, however, that the sharp increase in income inequality registered since 1991 mitigated the recovery in average real incomes. A synthetic measure taking into account both the level of total household income and its distribution can be defined as W = μ(1 - G), where μ, is an index of average real income per capita and G is the Gini coefficient.36 This measure, depicted as the Kakwani measure in Chart 4, increases with the aggregate income level and decreases as inequality rises. By the end of 1994, it stood at about 85 percent of its 1989 level.37

An alternative set of indicators could be based directly on household per capita consumption, the bulk of which is constituted by retail sales of goods. If Goskomstat’s adjustments for informal trade (discussed in the first section of the paper) are taken into account, this indicator declined only moderately in 1991 and 1992 and recovered thereafter, while remaining above its 1989 level throughout the period under consideration. Sales of paid services declined much more, but they represent only a small fraction—about one tenth—of household expenditures. Aggregate private household consumption thus did not collapse. This finding is consistent with the observed increase in household ownership of consumer durables, such as cars, washing machines, and television sets (Table 9).

Table 9.

Russia: Ownership of Selected Consumer Durables

(End-year; units per 100 households)

article image
Source: Goskomstat of the Russian Federation (1995).

It could be argued, however, that the current dynamism of aggregate consumption is not sustainable because it is being paid for by the dilapidation of the capital stock, as evidenced by asset-stripping behavior or the deterioration of important segments of the infrastructure (which was never that good to begin with). In this regard, using a measure such as GDP as a proxy for welfare has the merit of including investment and thus of reflecting this trade-off. In principle, of course, welfare would be better captured by the present value of current and future consumption flows than by current consumption flows alone.

Both income- and consumption-based measures project an overly bleak image of the evolution of welfare insofar as they ignore the gains associated with wider choice and shortened searching and queuing. In addition, as emphasized by Illarionov, Layard, and Orszag (1994), services such as housing remained vastly underpriced while their consumption stayed broadly unchanged, implying that total consumption, properly defined (that is, at market-based shadow prices), declined much less than indicated by the real money income or real consumption measures described above. In this light, it may not be unreasonable to claim that by 1994 welfare had actually improved, compared with the late 1980s.

Finally, there are a number of costs accompanying the transition, such as the deterioration in the sanitary situation, the increase in morbidity rates, the generalization of corruption, and greater uncertainty, that should be incorporated in more comprehensive welfare measures. Similarly, a number of benefits, such as the newly gained political freedoms or the reduction in pollution levels mirroring the decline in industrial production, ought also to be factored in. In some cases, it is not clear how intimately these costs or benefits are related to the output decline per se; in any event, an extended welfare analysis of this type lies beyond the scope of this paper.


  • Berg, Andrew, “Measurement and Mismeasurement of Economic Activity During Transition to the Market,” in Eastern Europe in Transition: From Recession to Growth? World Bank Discussion Paper No. 196, ed. by Mario I. Blejer, Guillermo A. Calvo, Fabrizio Coricelli, and Alan H. Gelb (Washington: World Bank, 1993).

    • Search Google Scholar
    • Export Citation
  • Berg, Andrew, and Jeffrey D. Sachs, “Structural Adjustment and International Trade in Eastern Europe: The Case of Poland,” Economic Policy, No. 14 (April 1992), pp. 11773.

    • Search Google Scholar
    • Export Citation
  • Center for Economic Analysis, Russia—1994 Economic Outlook, No. 3 (in Russian) (Moscow: Center for Economic Analysis, September 1994).

    • Search Google Scholar
    • Export Citation
  • Clark, Colin, A Critique of Russia Statistics (London: MacMillan, 1939).

  • “Czech Economic Monitor,” PlanEcon Report, Vol. 10 (September 28, 1994).

  • Dobozi, Istvan, and Gerhard Pohl, “Real Output Decline in Transition Economies—Forget GDP, Try Power Consumption Data” Transition, Vol.6 (January-February 1995), pp. 178.

    • Search Google Scholar
    • Export Citation
  • Ericson, Richard E., “The Soviet Statistical Debate: Khanin vs. TsSU,” in The Impoverished Superpower: Perestroika and the Soviet Military Burden, ed. by Henry S. Rowen and Charles Wolf, Jr. (San Francisco: Institute for Contemporary Studies Press, 1989).

    • Search Google Scholar
    • Export Citation
  • European Union, International Monetary Fund, Organization for Economic Cooperation and Development, United Nations, and World Bank, System of National Accounts 1993 (Brussels/Luxembourg, New York, Paris, and Washington, 1993).

    • Search Google Scholar
    • Export Citation
  • Gavrilcnkov, Evgeny (1994a), “Russia: Out of the Post-Soviet Macroeconomic Deadlock Through a Labyrinth of Reforms,” Review of Economies in Transition, Bank of Finland, No. 3 (1994), pp. 5980.

    • Search Google Scholar
    • Export Citation
  • Gavrilcnkov, Evgeny (1994b), “The Crisis of Expenditure Economy in Russia” (in Russian), Problemy Teorii i Praktiki Upravleniya, No. 5 (1994), pp. 2834.

    • Search Google Scholar
    • Export Citation
  • Gavrilcnkov, Evgeny, and Vincent Kocn, “How Large Was the Output Collapse in Russia? Alternative Estimates and Welfare Implications,” IMF Working Paper, WP/94/154 (Washington: International Monetary Fund, December 1994).

    • Search Google Scholar
    • Export Citation
  • Goskomstat of the Russian Federation, The Russian Federation in Figures, 1993 (in Russian) (Moscow: Goskomstat, 1994).

  • Goskomstat of the Russian Federation, Living Standards of Households in the Russian Federation (in Russian) (Moscow: Goskomstat, 1995).

  • Grossman, Gregory, Soviet Statistics of Physical Output of Industrial Commodities; Their Compilation and Quality (Princeton, New Jersey: Princeton University Press, 1960).

    • Search Google Scholar
    • Export Citation
  • Grossman, Gregory, “The Second Economy in the U.S.S.R. and Eastern Europe: A Bibliography,” Berkeley-Duke Occasional Papers on the Second Economy in the USSR, No. 2) (Bala Cynwyd, Pennsylvania: WEFA Group, July 1990).

    • Search Google Scholar
    • Export Citation
  • Guzhvin, P., “Trust in Statistics” (in Russian), Statistical Bulletin, Goskomstat of the Russian Federation, No. 9 (1992), pp. 321.

    • Search Google Scholar
    • Export Citation
  • Illarionov, Andrei, Richard Layard, and Peter Orszag, “The Conditions of Life,” in Economic Transformation in Russia, ed. by Anders Ɨslund (London: Pinter Publishers, 1994).

    • Search Google Scholar
    • Export Citation
  • International Monetary Fund, Russian Federation, IMF Economic Review (Washington: International Monetary Fund, 1992, 1993, and 1995).

  • International Monetary Fund, World Bank, Organization for Economic Cooperation and Development, and European Bank for Reconstruction and Development, A Study of the Soviet Economy, Vol. 1 (Washington: International Monetary Fund, 1991).

    • Search Google Scholar
    • Export Citation
  • Ivanov, Yu., and A. Ponomarenko, “Gross Domestic Product: Definition, Estimation and Projection” (in Russian), Ekonomist, No. 3 (1994), pp. 1120.

    • Search Google Scholar
    • Export Citation
  • Kakwani, Nanak, “Measurement of Welfare with Applications to Australia,” Journal of Development Economics, Vol. 18 (August 1985), pp. 42961.

    • Search Google Scholar
    • Export Citation
  • Khanin, Girsh Itsykovich, “Economic Results for Russia in the First Quarter of 1992” (in Russian), EKO, Nos. 6 and 7 (1992), pp. 2638 and pp. 5469.

    • Search Google Scholar
    • Export Citation
  • Koen, Vincent, “Measuring the Transition: A User’s View of National Accounts in Russia,” IMF Working Paper, WP/94/6 (Washington: International Monetary Fund, January 1994).

    • Search Google Scholar
    • Export Citation
  • Koen, Vincent, and Steven Phillips, Price Liberalization in Russia: Behavior of Prices, Household Incomes, and Consumption During the First Year, IMF Occasional Paper No. 104 (Washington: International Monetary Fund, June 1993).

    • Search Google Scholar
    • Export Citation
  • Koen, Vincent, and Eric Meyermans, “Exchange Rate Determinants in Russia: 1992-93,” IMF Working Paper, WP/94/66 (Washington: International Monetary Fund, June 1994).

    • Search Google Scholar
    • Export Citation
  • Kostinsky, Barry L., and Misha Belkindas, “Official Soviet Gross National Product Accounting,” in Measuring Soviet GNP: Problems and Solutions, A Conference Report (Washington: Central Intelligence Agency, 1990).

    • Search Google Scholar
    • Export Citation
  • Liapis, Peter S., and Yuri Markish, “Demand for Purchased Inputs Continues to Decline: Yields Yet to Be Significantly Affected,” in International Agriculture and Trade Reports, Former U.S.S.R., Situation and Outlook Series (Washington: U.S. Department of Agriculture, May 1994).

    • Search Google Scholar
    • Export Citation
  • Lipton, David, and Jeffrey D. Sachs, “Prospects for Russia’s Economic Reforms,” Brookings Papers on Economic Activity, No. 2 (1992), pp. 21383.

    • Search Google Scholar
    • Export Citation
  • “Polish Economic Monitor,” PlanEcon Report, Vol. 10 (November 4, 1994).

  • Rajewski, Zenon, “Gross Domestic Product,” in Polish Economy in 1990-1992. Experience and Conclusions, ed. by Leszek Zienkowski (Warsaw: Research Center for Economic and Statistical Studies of the Central Statistical Office and the Polish Academy of Sciences, 1993).

    • Search Google Scholar
    • Export Citation
  • Roberts, Bryan, “Welfare Change During Economic Transition: The Effects of Ending Shortages in Russia” (unpublished; Miami, Florida: University of Miami, October 1994).

    • Search Google Scholar
    • Export Citation
  • Rose, Richard, “Getting by Without Government: Everyday Life in Russia,” Daedalus, Vol. 123 (Summer 1994), pp. 4162.

  • “Slovak Economic Monitor,” PlanEcon Report, Vol. 10 (September 9, 1994).

  • Treml, Vladimir, “Problems with Soviet Statistics: Past and Present,” in Defense Conversion, Economic Reform, and the Outlook for the Russian and Ukrainian Economies, ed. by Henry S. Rowen, Charles Wolf, Jr., and Jeanne Zlotnick (New York: St. Martin’s Press, 1994).

    • Search Google Scholar
    • Export Citation
  • United States Bureau of the Census, Historical Statistics of the United States, Colonial Times to 1970 (Washington: U.S. Department of Commerce, Bureau of the Census, 1975).

    • Search Google Scholar
    • Export Citation
  • Winiecki, Jan, “The Inevitability of a Fall in Output in the Early Stages of Transition to the Market: Theoretical Underpinnings,” Soviet Studies, Vol. 43 (No. 4, 1991), pp. 66976.

    • Search Google Scholar
    • Export Citation
  • World Bank, Statistical Handbook 1994: States of the Former U.S.S.R. (Washington: World Bank, 1994).

  • Yurkov, Yurii, “The Statistical Mirror of the Market” (in Russian), Ekonomika iZhizn, No. 35 (August 1994).


Evgeny Gavrilenkov is Acting Director of the Center for Economic Analysis under the Government of the Russian Federation. This paper was written while he was a visiting scholar at the Research Department of the IMF, Vincent Koen is an Economist in the Research Department, Comments from Anders Aslund, Andrew Berg, David T. Coe, Francois Lequiller, Bryan Roberts, Doris Ross, Mark Schafter, and participants in an IMF seminar are gratefully acknowledged. However, the authors remain responsible for any judgments formulated in the paper, as well as for any remaining errors.


See, for example, “Russian Agency Tracks Soviet-Style Economy as Free Market Thrives,” Wall Street Journal, July 6, 1994, and “Services Take Lead in Russia,” Financial Times, July 14, 1994.


In Table 1, published yearly rates of change are chained to derive cumulative declines. As these rates of change are computed in different ways across countries, the resulting cumulative declines are not strictly comparable, but they are nevertheless relevant as a first approximation.


See Gavrilenkov (1994b). Consumption, however, fell much more in 1941-42 than output, and much more than in the course of the transition.


There are also reasons to doubt the accuracy of real GDP data, which for the past few years have been published two weeks after the end of the monthly period to which they relate, without any subsequent revisions. The only revision published so far was for 1991 as a whole, in which the decline originally estimated at 9 percent was adjusted to 13 percent.


See also Koen (1994) and the references therein.


The same divergence was noted for Poland by Berg (1993).


In those countries, unlike in Central Europe, officially measured GDP fell as much as or more than industrial output, thus contradicting the presumption that the share of industry shrinks during the transition.


See Ericson (1989), Kostinsky and Belkindas (1990), and, for an official recognition, Guzhvin (1992).


Grossman (1960) provides a wealth of examples.


See Treml (1994) and, for more general references, Grossman (1990). It may be no coincidence that such methodological innovations were undertaken as growth in the state sector could on some measures be seen to be coming to a hall. The effort to measure underground activity collapsed, however, with the dissolution of the Soviet Union in 1991 and resumed only slowly. It should be noted that the System of National Accounts 1993 (Sees. 6.30-6.36) recommends that underground and illegal production of goods and services be included in GDP.


Surveys are admittedly difficult to implement in a context of high rates of exit from, and entry into, any pool of enterprises.


Admittedly, this phenomenon could also reflect an inability or unwillingness on the part of banks to classify enterprises in well-defined categories.


Incidentally, the fact that the turnaround in the consumption of electricity by industrial enterprises preceded that of industrial output could be interpreted as a sign that output growth may have been overstated in the late 1980s.


More than half a century ago, Clark (1939, p. 5) already deplored that “the Soviet authorities have adopted a somewhat limited and materialistic definition of national income.”


Ivanov and Ponomarenko (1994) indicate that thus far illegal activities are not included in official Russian GDP estimates. However, many of the occupations listed above would not be considered illegal in a market economy, even though some of them are admittedly more value additive than others.


These series are published in the monthly Goskomstat bulletins and in the quarterly reports of the Center for Economic Analysis.


Another sign that some entities stopped reporting could be that nominal gross industrial output is smaller than what would be implied by real output and producer price data. As noted in Koen (1994), however, this discrepancy may also be caused by differences in sample coverage.


However, it should be remembered that, while using information on building permits is sensible, unreported renovation activity would still go uncaptured, as would construction carried out in the absence of permits.


No decomposition of real GDP changes by sectors of origin or by end uses was ever published.


Official nominal GDP estimates were revised for 1992 (from Rub 15 trillion to Rub 20 trillion, and subsequently back to Rub 18.1 trillion), but, strangely enough, the revisions were not accompanied by changes in the associated real GDP numbers.


The “Czech Economic Monitor” (1994), the “Slovak Economic Monitor” (1994), and the “Polish Economic Monitor” (1994), for example, based their alternative GDP estimates for the Czech Republic, the Slovak Republic, and Poland, respectively, on hypotheses regarding these two factors.


No adjustment was attempted for the rate of change from 1989 to 1990. As most of the factors underlying the undenrecording of output only became important around 1991, one would not expect the methodology followed here to produce a result very different from the official 3 percent decline.


Goskomstat never published a time series for real GDP based on prices of some fixed year. The cumulatives appearing in Goskomstat documents are derived by chaining year-on-year estimates, based on prices of the previous year.


See, for example, international Monetary Fund, World Bank. Organization for Economic Cooperation and Development, and European Bank for Reconstruction and Development (1991), and Gavrilenkov (1994a). Also striking and paradoxical is the disregard of planners for long horizons, which is obvious, for example, in the way that Russia’s natural resource base was exploited.


A more precise comparison would have to be based on stocks of tractors rather than on the domestically produced flow.


However, the sharp real appreciation of the ruble in the second half of 1993—analyzed by Koen and Meyermans (1994)—contributed to substantial declines in the production of some of these items in 1994 (bottom panel of Chart 3).


More detailed information and methodological comments are provided in Center for Economic Analysis (1994).


Koen and Phillips (1993) analyze price liberalization at length. The economic reviews of Russia by the IMF describe the evolution of the rules governing external trade (International Monetary Fund (1992), (1993), and (1995)).


Largely reflecting the massive overshooting at the time of the price jump associated with the freeing of most prices, the real exchange rate vis-à-vis the U.S. dollar on the interbank market appreciated by close to 1,000 percent between January 1992 and December 1993. In 1994, it remained within a 10 percent band around the end-1993 level.


The unavailability of items in some cities three years after price liberalization probably reflects insufficient demand or administrative restrictions on price setting, rather than persistent supply failures.


New Russia Barometer III. conducted nationwide in March-April 1994 under the aegis of the Paul Lazarsfeld Society of Vienna.


Rcsidual queuing can be associated with maintained local price controls or, in a newer way, with the (most often illusory) perception of golden investment opportunities, as witnessed by the long lines to purchase the shares offered by some investment funds.


In the process, these commentators overlook the fact that industrial output, valued at world prices, declined less than at domestic prices (see an earlier version of this paper, Gavrilenkov and Koen (1994, Section 11.4)).


A more complete analysis would have to encompass the changes in terms of trade and volumes exported and imported for all goods and services.


One among several rationales for this measure is the following: in a society with n individuals arranged in ascending order of their incomes xi (x1x2≤…≤xn), a welfare function can be defined as ∑xivi where vi is proportional to the number of individuals whose income is at least equal to xi; it can then be shown that aggregate welfare equals μ(1 - G) (see Kakwani (1985) and the references therein).


The Gini coefficients published by Goskomstat and by the Center for Economic Analysis differ at times, but the use of one or the other series only marginally affects the evolution of the Kakwani measure.