Over several decades, Western governments and scholars puzzled over the Soviet defense budget. The puzzle was a rather obvious one. The official figures were impossibly small, and they failed to increase in step with the visible expansion of Soviet military capabilities.
The baseline for the alternative Western estimates was set by the US Central Intelligence Agency. The CIA maintained an Office of Soviet Analysis charged with evaluating the military activities and overall economic potential of the Soviet Union. Working out the likely Soviet military budget in both rubles and dollars was an essential part of this work.
In addition to the deceptive Soviet published statistics, the CIA had access to its own intelligence data. At infrequent intervals, its few human sources provided what purported to be inside information about the Soviet Union’s factual defense costs. The “benchmark data,” as William T. Lee, a former CIA analyst, called them, are listed in Table 7.4. They painted a picture of Soviet defense outlays rising rapidly from 50 billion rubles in 1970 (12.5 percent of GNP) to 100 billion in 1980 (16 percent) and to 150 billion in 1982.[385]
To provide cross-validation without undue reliance on Soviet official statistics, the CIA developed a direct costing approach or “building block” methodology. The building blocks were estimates of quantities of Soviet defense resources used or stockpiled in each year. These were based
| table 7.4. The CIA’s “benchmark data” for Soviet defense outlays, 1970-1982, selected years | |||
|---|---|---|---|
| Billion rubles | |||
| 1970 | 50 | ||
| 1972 | approx. 58 | ||
| 1980 | approx. 100 | ||
| 1982 | more than 150 | ||
| Note: The “benchmark data” were those supplied to the CIA by intelligence sources and defectors. | Source: Lee, CIA Estimates, 145. To these figures might be added a figure for 1990 of “more than 200 billion rubles” provided by Soviet economic journalists Gams and Makarenko, “Razmyshleniia о rak-hodakh obshchestva na oboronu.” | ||
on another kind of intelligence data: estimates of the scale of the Soviet armed forces’ acquisitions, deployments, and operations. The building blocks were valued by working out how much it would have cost to provide them in the United States in some base year, such as 1960. For each block of a given resource, the quantity multiplied by the base-year cost gave its dollar value, and the sum of values of the blocks gave the CIA an estimate of overall Soviet defense outlays in constant dollars.[386] Dollars were converted to rubles by applying either ruble prices of the base year (where known) directly to the same blocks of quantities, or by converting dollar values based on estimated purchasing-power parities. Finally, the current ruble value of Soviet defense activity for a given year could be derived by applying estimates of ruble price changes since the base year.
On that basis, the CIA estimated that the Soviet defense outlays at around 47 billion rubles in 1970, and 94 billion in 1980. These figures, included in Table 7.5, were close to the “benchmark data” if slightly below them, and they showed a similar trend over time.
At first the CIA had a monopoly in the field. Before long, outsiders’ doubts about CIA analysis developed into keen rivalry. Other players included the Pentagon’s Defense Intelligence Agency (the DIA, established in 1961) and the Stockholm Independent Peace Research Institute (SIPRI, established in 1966). Outside the institutions worked a handful
| table 7.5. A bewildering range: Competing estimates of Soviet defense outlays, 1980 (billion rubles and current prices) | |||||
|---|---|---|---|---|---|
| 1970 | 1980 | ||||
| DIA (high) | 107 | ||||
| Lee (high) | 49 | 106 | |||
| DIA (low) | 96 | ||||
| CIA | 46.9 | 94 | |||
| Lee (low) | 44 | 93 | |||
| Steinberg | 48.2 | 81.2 | |||
| Rosefielde | 49 | ||||
| SIPRI | 42 | 48.7 | |||
| Soviet official | 17.9 | 17.1 | |||
| Wiles | 16.5 | ||||
| Note: GIA figures (in bold) are used as a baseline estimate in the text. | Sources: DIA from Michaud, “Paradox of Current Soviet Military Spending,” 120. Rosefielde, False Science (second edition), 186. Wiles from Rosefielde, “Soviet Defence Spending,” 70 (interpolated by Rosefielde on his own estimate for 1970). Other figures from Noren, “Controversy over Western Measures,” 269. | ||||
of independent scholars and other mavericks. Among them were Lee (a former CIA analyst who was hired by the DIA), Steven Rosefielde and Peter Wiles (university-based economists), and Dmitri Steinberg (an independent economist). A bewildering range of competing estimates emerged, as shown in Table 7.5.
While admitting that its own figures were of uneven reliability, the CIA maintained that they were corroborated by other sources and accepted by other intelligence services.[387] But this self-defense was challenged, most fiercely by Lee and Rosefielde. They charged that the CIA’s direct costing procedures were incomplete and insufficiently transparent; they were insufficiently robust to the incorporation of new information and understated the growing quality and cost of Soviet weapons.[388]
In the CIA’s exchanges with Lee and Rosefielde, there was bad feeling on each side. Lee and Rosefielde both charged that the CIA had colluded in a Soviet strategy of disinformation, resulting in understatement
of the Soviet military-economic effort. The implications that they drew diverged, however. According to Rosefielde, the outcome was insufficient United States preparedness for war.[389] According to Lee, the outcome was excessive United States engagement in arms control. If (as he assumed) the Soviet economy was already at the limits of its military-industrial potential, restraining the arms race would one-sidedly advantage the Soviet Union since only the United States was capable of further economic mobilization.[390]
In his own working, Rosefielde hewed closest to the lines of the CIA’s direct costing methodology. He started from the same quantity building blocks as the CIA, supplemented or corrected by those of the CIA, and from base-year prices or costs of i960 that the CIA first adopted, then later abandoned. In addition, Rosefielde claimed a superior methodology of price adjustment for quality change in later years. The result was a measure of defense spending in rubles that approximately matched that of the CIA, but Rosefielde claimed the match was no accident: the CIA estimates were being forced to adapt to realities, but the adjustment was being done dishonestly under a pretense of scholarship. Moreover, the implications of the CIA’s and Rosefielde’s estimates for military stock building in real terms were quite different: in Rosefielde’s view the Soviet Union had substantially overtaken the United States in its military capabilities by the 1980s.
385
Here and below, defense shares of Soviet GNP are based on the retrospective estimates used in Table 7.7.
386
The CIA’s dollar estimates raised specific issues that I do not discuss here. A persistent critic of the dollar estimates was Franklyn D. Holzman, who argued that, by neglecting substitution (or index-number) biases, the CIA overestimated the real value of Soviet military outlays. Holzman, “Are the Soviets Really Out- spending the U.S.?”
388
Central Intelligence Agency of the United States, Estimated Soviet Defense Spending; Lee, Estimation of Soviet Military Expenditures; Rosefielde, False Science (first edition). It was suggested that the building-block methodology could only give a lower bound on Soviet spending because its coverage of the building blocks was inevitably incomplete. For the categories of equipment that were covered, the CIA was alleged to understate the rates of growth of true quantities and costs. That is, it attributed too much of observed price changes to hidden inflation and so failed to capture the full improvement in the quality of Soviet weapons through time; this led to understatement of the value of Soviet military stocks relative to the United States. At the same time, it failed to observe the full extent of price inflation, and this led to understatement of the cost of Soviet defense activity to the economy supplying it. Both Lee and Rosefielde argued that the CIA methodology lacked transparency, withholding the evidence base of prices and quantities from which the building blocks were valued, and the procedure for consistent revision of serial data when new information was factored in. Finally, they diagnosed lack of robustness from the character of CIA responses to new information. In the most celebrated case, when the “benchmark” figure of 50 billion rubles in 1970 transpired, it was more than twice the previous CIA estimate. The latter was then revised upward. The CIA claimed that this revision would have happened anyway in the course of repricing and was independent of the “new information”. Burton, “Estimating Soviet Defense Spending,” replied for the CIA, but neither Rosefielde, False Science (second edition) nor Lee, CIA Estimates of Soviet Military Expenditures, was reconciled. Although the CIA’s direct costing exercise was wound up after the fall of communism, those formerly engaged in it continued to defend their position. Noren, “Controversy over Western Measures”; Firth and Noren, Soviet Defense Spending.