Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Meeting Sessions INTRODUCTION The National Research Council (NRC) hosted a meeting of specialists concerned with developing estimates of the size of the Soviet economy in order to clarify the fundamental methodological issues that lie at the base of the current debate over the reliability of the estimates produced and utilized by the U.S. government. The goal of this meeting was not to pro- duce a better estimate of the size of the Soviet economy, or to attempt to reach a concensus concerning the validity of various current estimates, but rather to investigate the problems inherent in efforts to produce such estimates. The participants in the meeting sought to clarify what is known and what is knowable about the Soviet economy, to analyze the assump- tions on which the major methodologies rest, to critique the methods used by practitioners in the field, and to suggest adjustments, alternative strategies, and methods--the adoption of which might lead to improved calculations. As this summary report makes clear, the meeting constituted only a preliminary step in the process, focusing on the identification of areas of controversy, the specification of issues meriting further study, and offering suggestions concerning possible alternatives. SESSION I: THEORETICAL FOUNDATIONS OF CURRENT METHODOLOGY The methodology currently used by the Central Intelligence Agengy (CIA) to measure both Soviet gross national product (GNP) and Soviet growth rates is largely based on the adjusted factor cost (AFC) method. Abram Bergson argued that although a case could still be made for using the AFC method to measure both the relative size and rate of growth of Soviet GNP, one of the main problems with this method is its quasi- conventional nature as an adaptation of the ideal standard. Theoretically, the AFC method corresponds to this ideal standard only if there is no misallocation of resources in production in the Soviet system. This raised the question as to whether it would be possible to obtain a better measure of marginal rates of transformation (MRT) in the Soviet economy if differ- ent techniques were applied. One reason advanced against the use of alternative techniques was the fact that the AFC standard incorporates a
2 THE SOVIET ECONOMY number of adjustments (such as removing taxes and subsidies from prices) that are similar to those used in calculating GNP and national income in the West. It was suggested that these similarities provide a basis for the comparison of Soviet GNP with the GNPs of other countries. In addition to the traditional adjustments performed in measuring GNP in Western countries, the AFC methodology as applied by the CIA removes profits from Soviet prices and factors in a uniform for all sectors average return on capital. Bergson noted that he made a series of alter- native calculations on the basis of various adjustments, but that he found the standard AFC calculations to be remarkably insensitive to such alterations. He suggested that this robustness might be due to the highly aggregated nature of the available data. Bergson also mentioned that the CIA decision to eliminate profits from ruble prices had been challenged by Judith Thornton, who argued that disparate profit rates in different sectors might reflect real MRTs in the Soviet economy. The meeting did not produce a consensus on the best way to adjust Soviet prices. One view was that Soviet prices were so divorced from reality that no adjustments would produce reasonable MRT estimates. However, if one were to operate on the basis of this assumption, one would have to resort to current dollar prices for similar goods or to Soviet pre-S-year plan prices as a base from which to measure Soviet economic growth. The participants agreed that one of the most important problems underlying the contemporary debate about the size of the Soviet economy was the popular confusion between GNP estimates and measures of social welfare. The debate has been skewed by a fundamental misunderstanding of what it is that GNP is designed~to measure. Estimates of the Soviet economy that placed its aggregate size at roughly half that of the United States were not intended to suggest that individuals living in that economy lived half as well, but that total Soviet economic production was roughly half that of the U.S. economy. Although GNP offers little or no infor- mation as to the distribution of income, access to public and private goods, or to the noneconomic priorities that vitally affect social welfare, it remains a useful tool when used with a proper understanding of what it does and what it does not seek to measure. It was also noted that some analysts have attempted to introduce adjustments into consumption cal- culations in order to derive better measures from a welfare point of view, but it was agreed that more work will be necessary in this area. One of the first theoretical issues explored at the meeting was the issue of Cost maximizations in the Soviet economic system. Vladimir Kontorovich argued that one difficulty that arises in dealing with inflation in the Soviet case is the assumption of cost-minimizing producers. Even the most cursory examination of the Soviet case, however, demonstrates
THE SOVIET ECONOMY 3 that this assumption is not validated by the data. Kontorovich cited a series of examples that he contended demonstrated that policies of Cost maximizations prevailed in construction, machine building, and a number of other Soviet industries. Other participants queried the use of this term, arguing that "cost maximization" was theoretically impossible in that it would lead to infinite costs. Even though the assumption of cost efficient behavior is central to all modern methodologies of measuring economic aggregates, Richard Kaufman pointed out that similar practices occur in certain sectors of Western economies as well (notably defense). He further noted that the attempt to measure and/or make adjustments for inefficiency is a general problem in GNP accounting procedures, that it is unrelated to the issue of inflation, and that the problem is not limited to the Soviet case. One major area of criticism of CIA estimates discussed at the meeting concerned the issue of inflation in the Soviet economy. The CIA attempts to deal with this issue by using the most extensive sample of goods and prices that it can obtain and by basing its calculations on physical units wherever possible. It was pointed out that this approach does not work well in sectors such as machine building where the rate of introduction of new products is particularly high. In addition, the peculiarities of new product pricing in the Soviet Union complicate measurements of inflation. Although there was substantial agreement that hidden inflation poses a significant problem in attempting to measure the Soviet economy, none of the participants were able to suggest mechanisms that would reliably exclude it from GNP calculations. It was noted that one of the main problems in dealing with Soviet data is that of attempting to monitor the quality of output, both over time and in comparison with other economies. The quality issue is particularly difficult to resolve because different sectors sometimes exhibit divergent trends with respect to the quality of their output. For example, re- searchers who focus on the service and defense sectors have claimed that the CIA estimates systematically understate improvements in quality, whereas those who investigate other sectors of the Soviet economy contend that the CIA has overstated Soviet GNP because it does not make allow- ances for the generally poor and apparently deteriorating quality of Soviet goods. Since the AFC method substitutes an assumed rate of return on capital for various kinds of nonlabor charges, variations in the quality of capital goods can be particularly important for the validity of AFC-based results. In this regard, Dmitri Steinberg noted the conclusion of a study conducted by a group of Soviet economists that stated that returns on capital invest- ment (the present value of output per ruble of cost) varied significantly across different sectors of the Soviet economy. Others pointed out that
4 THE SOVIET ECONOMY it is frequently difficult to distinguish between the low relative quality of capital goods and the misallocation of capital resources within the economy. The central issue raised at the session, however, was whether the uniform 12 percent rate of return on capital that is currently used by the CIA should be replaced with a number of variable rates of return for different sectors. Richard Ericson pointed out a more general problem with the AFC method. He argued that any attempt to automatically impute output from factor inputs assumes that a useful output will be produced and that the relationship between its value and the value of the inputs can be reliably established. In the Soviet case it is possible that neither of these holds true. The AFC method uses adjustments that are based on labor pay- ments, but these labor payments are determined in accordance with administrative plans and often remain faxed for long periods of time. Although some researchers have argued that competition in the Soviet labor market alleviates this problem, Ericson emphasized that no competition takes place on the demand side of the market. Ericson also pointed out that the attempt to determine the cost of capital on the basis of the AFC method presents even greater problems. Ideally, one would want to measure the expected present value of future services on the basis of the output they will produce, rather than on the cost of the inputs. Unfortunately, a measurement technique that would accomplish this task, and that would be applicable to a Soviet-type economy, does not yet exist. A number of other issues were touched on at the opening session, although they could not be explored in any depth due to time limitations. One of the most important of these was the extent to which the structural changes introduced in the Soviet system during the recent period of perestroika and increased glasnost have made the utilization of alternative measurement techniques and methodologies possible. Uncertainties as to the future course of change in the Soviet system prevented any resolution of this issue. SESSION II: AN ANALYSIS OF THE DATA USED IN ESTIMATING THE SIZE OF THE SOVIET ECONOMY Until quite recently the conventional wisdom in the field maintained that with few exceptions, the data published by the Soviets in physical units were generally reliable, but the data furnished in current prices were of only limited utility. The first assumption has been seriously undermined by Goskomstat's recent report (Pravitel'stvennyi Vestnik, No. 12, 1990) that
THE SOVIET ECONOMY s cites an aggregate employment figure for the Soviet economy that includes 4 million people who had not appeared in any previous report. Although the source of this discrepancy is not known, Dmitri Steinberg contended that the increase in this employment figure validated his long-standing contention that Soviet official statistics on material production have consistently excluded a large part of defense output. The issue remains unsettled, but it underlined the more general problem of how to reconcile the old data with much of the new information that is becoming available. With the onset of glasnost, large quantities of new data on the Soviet economic system have become available. These new data come from a variety of sources including Goskomstat, local, regional, and republican sources, as well as individual Soviet economists. One of the central issues before the meeting was the question of how these new data might affect Western estimates of the size and rates of growth of the Soviet economy. It was suggested that these new data might shed some light on the degree of statistical overstatement in the Soviet system and provide more infor- mation on data expressed in physical units. It was also noted that given the post-1985 reforms and the general decline in central control in the Soviet economy, inflation in the data expressed in physical units may have been increasing throughout the period. The traditional point of view has been that even if the physical data were inflated to a significant degree, it would not affect the estimates of growth rates, as long as the degree of inflation did not change over time. No one expressed confidence that this condition held true during the period in question. It was pointed out that even if the official Soviet data have become increasingly inflated, it does not necessarily follow that GNP estimates have to be adjusted downward. For an example, the padded distances and loads recorded for motor transport do not necessarily represent complete fabri- cations as these vehicles are frequently engaged in hauling freight for the second economy. Another phenomenon affecting the reliability of physical unit calculations is the double counting of output that may occur when producer cooperatives are attached to state enterprises. In these cases the output of the cooperative may be counted twice, once in the figures for the output of the state enterprise itself (for the purposes of plan fulfillment) and once in the figures for total cooperative output (for statistical purposes). Examples of cheating and the fabrication of physical data are well known, but the meeting produced no consensus on how to deal with these widespread phenomena or the degree to which they might effect GNP esti- mates. Campbell pointed out that there were incentives in the Soviet system that led to the underestimating of economic production and for an example cited sheep herds that included uncounted, "private, sheep. Those familiar with the CIA research effort stated that the agency had not
6 THE SOVIET ECONOMY been able to produce any reliable way to make the adjustments needed to incorporate these phenomena. The problems with cheating aside, it was generally accepted that the physical data could provide useful information about the structure of the Soviet economy and the size of particular sectors. For example, Steinberg suggested that although the value of inputs is a notoriously unreliable guide to the output of the construction industry, largely due to waste and the lack of incentives for cost minimization, one can reliably use the square footage of industrial floor space as a direct measure. However he also noted that this measure was less than fully satisfactory in that it did not take into account differences or changes in the quality of output. Steinberg suggested that, in general, indices based on comparable prices should be replaced by indices based on the available output data in physical units (e.g., square meters of fabric should be used to estimate increases in clothing production). Steinberg's suggestions were criticized in particular for the lack of uniformity in his approach; he uses inputs to measure growth in apparel production but employs physical measures of output to evaluate trends in construction output. Richard Ericson suggested that some of the problems with both value series and physical data series could be alleviated if researchers were able to obtain access to individual ministries' accounting data. Vladimir Treml pointed out that in some cases these data remain classified and that in others Goskomstat does not have the authority to publish them. However, he indicated that in those instances where ministries have been abolished their records may become available. Some participants expressed the hope that the increasing decentralization in the Soviet system and the con- comitant increase in the power of regional authorities might result in improved access to the highly disaggregated data that are collected at this level. It was noted, however, that not only the reliability of these data would be questionable, but that their value would be limited in that they would lack comparability with other data and old time series. Gur Ofer suggested that Western researchers could benefit not only from greater glasnost in the Soviet Union but also from more openness from the CIA. Although he noted that the so-called CIA estimates should more accurately be termed the estimates of the field of Sovietology--since many researchers had a hand in establishing the general framework--he argued that the agency had taken the generally accepted mechanism and proceeded to work on its own, making decisions about the details without either adequate scrutiny or feedback from other practitioners in the field. He further asserted that neither the papers contributed by CIA analysts to the Joint Economic Committee's publications on the Soviet economy nor the CIA's own publications made completely clear the rationale behind particular decisions as to how the agency handles specific data. Other
THE SOVIET ECONOMY 7 participants suggested that the agency should provide other researchers with access to much of the raw data that it collects on the Soviet economy, produce comparable estimates across sectors, and provide an integrated accounting of the economy as a whole. Steinberg suggested that one way to improve Western estimates of the size and growth of the Soviet economy would be to compare the sector- of-origin estimates with end-use estimates. According to his calculations, the two sets of CIA estimates produce significantly different results with the discrepancies ranging from 1-2 percent a year. Bergson asserted that it would be useful for the CIA to produce an alternative estimate on the basis of the deflation of current ruble data, and Steinberg added that from a strictly methodological point of view double deflation would be even better. Ofer identified a need to "predict the pasty He suggested that it would be extremely valuable to produce a comprehensive picture of what the Soviet economy looked like in the early 1980s, before the onset of perestroika. Asserting that it was time to "think anew," Ofer stated that 1985 should be the last possible date for the utilization of the old procedures, and he called for the application of a completely new structure in the subsequent period. Barry Kostinsky stated that the Department of Commerce was about to publish a very detailed series of accounts for 1985. He also noted that the Center for International Research (Bureau of the Census) had produced an estimate for Soviet GNP in 1985 on the basis of established prices that was very close to the official Soviet figure for GNP, although they had derived different subtotals for different sectors. SESSION III: COMPARISON AND CRITIQUE OF ALTERNATIVE METHODS OF ESTIMATING THE SIZE OF THE SOVIET ECONOMY The examination of various issues associated with the use of the AFC methodology was followed by a discussion of possible alternatives. Richard Ericson posed a series of questions at the outset of the session that served to focus the discussion. Are there alternatives to the AFC method? Are the alternative estimates that are at the center of the current debate based on alternative methodologies? How should we account for the second economy? How important a factor is hidden inflation in the Soviet eco- nomic system? Some participants maintained that at this point there were no real alternatives to the AFC-based methodology, and that what passed for alternative estimates had in fact been derived from various makeshifts and ad hoc extrapolations. Richard Judy stated that although the CIA esti-
8 THE SOVIET ECONOMY mates could be marginally improved, it was not at all clear that the results would change. Vladimir Treml stated that in view of the inherent limita- tions on the data available, the estimates produced by the AFC method provided the best approximations possible, and he suggested that efforts should be concentrated on attempting to improve rather than replace the standard methodology. Other participants asserted that better practical alternatives to the AFC method already existed, and that they had demonstrated their capacity to produce more accurate estimates. William Lee stated that he had been able to produce better estimates for the defense sector both now and in the past by basing his calculations on the official ruble numbers published by Soviet statistical agencies. He contended that the estimates produced in this way were not only far closer to the mark than those produced by the CIA, but that they also provided a far superior measure of the effec- tive resource costs of production. Igor Birman stated that he generated an alternative estimate on the basis of the same data and the same basic methodology by altering the imputed values assigned to investment and capital stock in the Soviet economy. Contending that much of what the CIA counted as investment should be considered as waste or written down as inflated, Birman suggested that if the CIA's investment figures had been accurate that the Soviet economy would have long since outstripped that of the United States. A number of participants suggested ways in which the production of estimates of the Soviet economy could be improved Whereas the CIA bases its estimates on an elaborately derived calculation for a single base year and then estimates the size of the economy in subsequent years on the basis of rates of growth from that base, it was argued that it would be better to calculate GNP annually. Ericson suggested that the economy be treated as a whole, that the defense sector be integrated into the general accounts, and that input-output tables should be compiled in order to produce a more accurate depiction of the Soviet economy. Steinberg contended that Soviet "comparable" prices should not be used at all, but he contended that there are compelling reasons for using "established" prices in calculating the weights of various branches within the economy. Although he was willing to accept the argument that established prices contain certain distortions, he argued that they still provide important indicators of financial flows and resource allocation within the economy. He also noted that value added should not be assumed to grow at the same rate as the gross value of output. Steinberg asserted that the utilization of his method of calculating the size of the Soviet economy produced estimates of Soviet growth rates that were considerably smaller than those produced by the CIN He argued that there had been no real growth either in productive investment or in
THE SOVIET ECONOMY 9 consumption in the U.S.S.R. since 1976, and that most of the growth that had occurred was limited to the military sector. He stated that his results did not take into account the reported deterioration in the quality of Soviet consumer goods that may have taken place during the period under study, and suggested that if such a tendency was indeed present, that Soviet consumption levels had actually declined since 1976. Steinberg suggested that two separate sets of estimates should be produced and their results compared. One set should be based on the AFC methodology, but it should also introduce a series of improvements identified at the meeting: adjustments in the capital data to account for variations in the productivity of capital in different branches of the economy, adjustments in wage data that would take into account the difference between nominal and real wages (including differential access to goods in short supply), and a mechanism to account for unfinished con- struction projects and inventories. The second set of estimates should be produced on the basis of black market and world market prices. A number of the alternatives proposed at the session raised problems of their own. It was noted that black market prices may be quite different from free market prices, that world market prices may be difficult to use over long periods of time due to fluctuations that have nothing to do with conditions on internal markets and that the valuation of services in terms of comparable world prices may obscure important differentials in quality and productivity between different countries. Moreover, there are goods and services within the Soviet Union that are not traded on the world market and for which there are no adequate world price equivalents. Adjustments in nominal wages are also difficult to make in view of the differential access to goods of different population groups and the inherent problems associated with assigning numerical values to privileges or perquisites. Most fundamentally, neither the methodology that would be appropriate for making adjustments for differential rates of productivity of capital nor the data that might be required for such adjustments are presently available. Robert Campbell pointed out that the conventional treatment of in- vestments and many types of services as final goods does not take into account their differential productivity across countries. He also noted that investment goods are final goods only if one adopts a relatively short time horizon. Considered over a longer period of time, investment goods are intermediate goods that are used up in production. Given that the goal is to measure the output of genuinely final goods, one should adjust the comparative GNP calculations to take into account the fact that Soviet investment goods represent a smaller stream of future services than would a similar stock of U.S. investment goods. This approach would essentially require measuring per capita consumption, rather than the more conven-
10 THE SOVIET ECONOMY tional approach of measuring GNP. Others pointed out that consumption measures welfare, not total economic performance, and is therefore not commensurate with GNP. This approach also encounters conceptual difficulties. It is not always clear what should be counted as consumption. For example, such conven- tional consumption goods as bread and sugar constitute intermediate goods when they are used to feed livestock or to produce moonshine for the second economy. More importantly, in order to make productivity adjust- ments for investment goods, one must first be able to calculate produc- tivity. If productivity is calculated on the basis of conventional GNP accounting techniques, the process becomes circular and fails to advance the analyst's work. Others asserted that this was not an irremediable problem, and that the comparative consumption approach would be a useful tool for mea- suring and describing the size of the Soviet economy. For example, if you have two economies in a steady-state equilibrium, each producing 100 units, but with one producing 80 percent consumer goods and 20 percent producer goods, while the other produces only 70 percent consumer goods and 30 percent producer goods, the first economy is clearly operating far more efficiently than the second. The most frequently used ways of determining the relative size of Soviet and U.S. GNPs are product comparisons and dollar/ruble ratios. Abram Bergson suggested that official U.S. agencies would benefit from an extended analysis of alternative dollar/ruble ratios, but he also noted that the level of sophistication and detail in the alternative approaches he had examined was considerably inferior to that in the AFC method. Several participants countered that such comparisons were necessarily skewed by the fact that no individual researcher has had the opportunity or the funding to produce estimates of a comparable range or complexity. Others raised questions as to the extent to which the alternative methods under examination actually provided estimates that were significantly different from those produced by the conventional methodology. The CIA generated estimate of Soviet GNP as roughly 50 percent of that of the U.S. has sometimes been contrasted with Igor Birman's estimate that Soviet per capita consumption is approximately 22 percent of the U.S. level. However, the two figures are not strictly commensurate with each other. In order to make a valid comparison one would have to make adjustments for the following factors: (1) the CIA's estimate represents a geometric mean of vastly different ruble and dollar estimates; (2) the share of investment in Soviet GNP is considerably greater than that in U.S. GNP; and (3) the Soviet population is almost 20 percent greater than that of the U.S. Making these adjustments would substantially diminish, but by no means eliminate, the gap between the two
THE SOVIET ECONOMY 11 estimates. The remaining gap appears to result from different estimates of the appropriate dollar/ruble ratio, from Birman's ad hoc adjustment for largely nonquantifiable Other factors, and from the different treatment of trade services. Regarding the treatment of trade services, Bergson pointed out that his own calculations of their effect on total consumption implied the need for much smaller adjustments than those used by Birman. To a large extent the recent intense debate about the validity of official U.S. estimates of the size and growth of the Soviet economy and the methodologies used to generate them is a result of what Gregory Grossman has termed Cognitive dissonances This dissonance resulted from the apparent contradiction between the prima facie evidence available to both experts and casual observers that demonstrate that the Soviet economy is in a crisis and the CIA's analyses throughout the Cold War period that maintained that the Soviet economy was performing reasonably well and exhibiting respectable growth rates. A variety of factors were suggested to account for this: . First, the units of comparison are clearly incommensurate. Second, the Soviet economic system is currently performing much worse than before, due to the disruption of the command-administrative system, changing national objectives, and the chaos introduced by the perestroika reforms. ~ The generally accepted accounting conventions used to derive GNP, that use a single year as the standard for calculations, necessarily lead to counting investments as final goods. Utilizing a longer time frame could substantially alleviate this problem. Repressed inflation in the Soviet economy was not effectively Inflation was engendered by unusually large budget deficits and unrestricted monetary growth and has been manifested in longer queues, increased rationing through the work- place, and a greater propensity on the part of consumers to hoard goods. All of these factors tend to mislead casual observers and to create the impression that the economy is performing less well than may be the case. ~ The casual observations that influence specialist and nonspecialist perceptions in the West are usually made in Moscow and Leningrad, and this may seriously misrepresent trends in the country at large. accounted for in the official estimates. It was also noted that the CIA's estimates of the rates of growth of Soviet consumption may actually be consistent with the current state of affairs since growth rates have been calculated relative to the abysmally low level of Soviet consumption in 1953. No consensus as to the general applicability of particular alternative methodologies was achieved during this session. Several participants
12 THE SOVIET ECONOMY stressed the importance of keeping in mind the particular goal of the exercise in evaluating the utility of alternative methodologies; it would not be reasonable, for example, to expect a method designed to measure the welfare of the population to adequately encompass the total production of goods and services in the Soviet economy. It was suggested that alter- native methodologies can coexist and should be used to answer different questions.
