#14 - JRL 7069 - RAS 16
HISTORY OF THE COLD WAR: THE CIA ESTIMATES OF SOVIET ECONOMIC STRENGTH

SOURCE. John Howard Wilhelm, "The Failure of the American Sovietological Economics Profession," Europe-Asia Studies, Vol. 55 No. 1, January 2003, pp. 59-74

In 1991 a committee of academic specialists submitted to the Permanent Select Committee on Intelligence of the House of Representatives of the US Congress a report evaluating the works on Soviet economic performance that the CIA had published annually during the 1970s and 1980s. Why did the CIA experts fail to predict the collapse of the Soviet economy? The committee argued that the collapse had been a low probability outcome: the CIA had failed to predict it, but so had American academic specialists. Nobody had anticipated the political developments that triggered the collapse. And it was too much to expect that anyone should have predicted what happened. So nobody was to blame.

John Wilhelm regards this position as dishonest because it obscures the fact that there WERE some specialists who defied the complacent consensus and argued that collapse was a real possibility. Above all, this was the view taken by émigré economists with direct personal experience of the object of their study. Yet far from trying to learn as much as they could from the experience of these émigrés (as the author himself did), established US academics resented their criticism and tried to exclude them from the debate. As a result, they were unable to publish much of their work in the US. Instead it was published in Britain, (1) where the climate in academia was more open and tolerant.

Wilhelm examines two cases in detail. One is that of Naum Jasny, a refugee first from the Russian Revolution and later from Nazi Germany who specialized in Soviet agriculture and Soviet consumption levels. Jasny's critical approach to the Soviet statistics led him to conclude that the Soviet economy in the 1930s was in a much worse state than most Western economists believed. His most intense conflict was with Abram Bergson, at that time a consultant for the Rand Corporation, and the group of people around him. It was this group who pioneered the techniques of Soviet national income analysis and comparisons that became standard usage both in American academia and in the CIA.

Thus in one of his early studies of Soviet national income Bergson argued that the level of Soviet consumption in 1937 had surpassed the level of 1928 -- that is, the level at which it peaked just before the sharp decline associated with forced collectivization and industrialization. Jasny presented calculations indicating that at best consumption had reached 60 percent of the 1928 level. (Everyone agreed that consumption had risen again in the mid-1930s: the question at issue was by how much.)

The second case is that of Igor Birman, a Soviet economist who continued research on the Soviet economy after emigrating to the US in 1974. Birman also came into conflict with the Rand group. Like Jasny, he thought that American specialists were overstating Soviet consumption levels. The CIA thought that Soviet per capita consumption was about one third of the US level; Birman concluded that it was at most one fifth. He also believed that Soviet GNP was nowhere near 55-60 percent of US GNP, as claimed by the CIA; he did not venture an alternative figure of his own, but did suggest that Japan had overtaken the USSR in economic strength.

How did the Rand group, despite their sophisticated computerized mathematical models, manage to get it so wrong? The author points to two major factors:

[1] The models were designed for a market economy, and were invalid when applied to the Soviet-type administrative ("planned") economic system.

[2] The modelers relied for their data on grossly unreliable Soviet statistics. "Garbage in, garbage out!"

It was the commitment to modeling that was at the root of the problem. Predictive models adequate to the Soviet-type economy were never developed; arguably they were impossible in principle. (2) So if you had decided in advance that you had to have models, they had to be market models. And the models required data that at least looked precise, even if you really knew that they weren't. The critics could demonstrate how awful the Soviet statistics were, but they were rarely able to come up with any alternative figures -- and when they did the figures looked so approximate as to seem arbitrary, even though they were actually LESS inaccurate than the misleadingly precise Soviet statistics. Then once you had invested a huge amount of money and effort into the models, you naturally didn't want to jeopardize the investment by admitting how shaky the models were.

I spent some years in the early 1980s (3) studying the open Soviet literature on statistical methods -- that is, on how data were collected and processed to derive statistics. A critical study of such literature was quite sufficient to make one appreciate just how bad Soviet statistics were, for all sorts of reasons. It was not essential to be an émigré with direct personal experience, though I'm sure that helped.

My Ph.D. thesis was on the methodology of the Soviet Family Budget Survey, which was supposed to determine the size and structure of household incomes and expenditures. I also studied the Soviet survey of the quantities and prices of food products sold in "kolkhoz trade" -- that is, on the semi-free peasant markets. I came to the conclusion that the figures published by the Central Statistical Administration (TsSU) underestimated the scale of this trade by a factor significantly greater than two, in large part due to an archaic approach to sampling. (4) I recall arguing the point at a conference with a colleague connected with the Rand group who refused to accept that published Soviet statistics could be wrong by such a large margin.

NOTES

(1) Much of their work appeared in "Soviet Studies," the precursor of "Europe-Asia Studies." US establishment academics were later to argue that dissident analyses had not been published in US peer-reviewed scholarly journals and therefore need not be taken into account. But it was they who controlled those journals and had blocked the publication of dissident analyses!

(2) Models to illuminate certain aspects of the functioning of the Soviet-type economy, such as the ratchet, were possible and did exist in the specialist Western literature, but they were not the kind of models that could be used to predict future economic performance.

(3) At the Centre for Russian and East European Studies (CREES) of the University of Birmingham (England). My supervisor was Philip Hanson. I had previously trained as a mathematical statistician and worked in the British Government Statistical Service.

(4) See my articles "A Note on Data Quality in the Soviet Family Budget Survey," Soviet Studies, vol. XXXV, no. 4, October 1983, pp. 561-8; and "How Reliable Are Soviet Statistics on the Kolkhoz Markets?" Journal of Official Statistics (published by Statistics Sweden), Vol. 2, No. 2, 1986, pp. 181-91. When I submitted the latter article, I received a reply from the journal's editor from which it was clear that he was upset by my poor opinion of Soviet statistics, but to his credit he still published the article. I still have spare off-prints of both these articles and shall provide them on request to anyone who sends me a stamped addressed envelope.