PREDICTING THE SOVIET STRATEGIC THREAT

• Collection: FOIA Collection
• Collection: Declassified Articles from Studies in Intelligence: The IC’s Journal for the Intelligence Professional
• Document Number (FOIA) /ESDN (CREST): 0006183129
• Document Page Count: 9
• Document Release Date: July 30, 2014
• Case Number: F-2013-02322

Approved for Release: 2014/07/29 006183129Improving forecasts-SteREfi/PREDICTING THE SOVIET STRATEGIC THREAT*(b)(3)(c)_(b)(3)(n)-The most common measures of the Soviet intercontinental attack force involve the size ofthe force. Since the early 1970s, however, the aggregate size of the force has been governed bythe SALT agreements, so predictions of it usually have been accurate. In this arms controlenvironment, issues of force modernization have become as important as issues of force size. Topredict the Soviet threat, US analysts today have to predict how quickly the Soviets willimprove their force. These improvements take the form of new technologies and the newweapon systems that use them, and US analysts thus have to predict the rate of forcemodernization. Consequently, new measures of the Soviet strategic threat have been developedthat capture the issues of force modernization, and these new measures have been used toevaluate the forecasting record on modernization. The new measures focus on the number ofweapon systems that enter the Soviet force during some period rather than the total numberexisting in the force at the end of the period.All of the National Intelligence Estimates (NIEs) on Soviet strategic forces have offeredalternative projections based on various assumptions about the general pattern of Sovietbehavior, but some consistent criteria were needed to construct a unified forecasting record.Most NIEs since 1970 have contained two alternative sets of forces to incorporate bothpossibilities with regard to arms control: -SALT- force projections within the numerical limitsand -No-SALT- force projections exceeding those limits. Because the Soviets have basicallyadhered to the key numerical SALT limits, we used the SALT force projections for ourevaluations whenever they were available. We used the unconstrained force projections onlywhen an NIE did not contain any SALT forces.Similarly, the NIEs expressed uncertainty in the future Soviet level of effort by presentingmore possible force projections. The NIE authors first constructed a -Moderate- or -Best- forcethat was consistent with past Soviet efforts; next, they constructed "High- and "Low- forces asexcursions on either side of the first. In general, the high and low forces reflect high and lowlevels of effort and success with development and deployment. We used two forces from eachNIE, a high force and a low force. Some NIEs did not contain a force labeled -Low,- and, inthose cases, the force labeled -Moderate- or -Best- was used.The Forecasting RecordFor this evaluation, we introduce the parameter new Strategic Nuclear Delivery Vehicles(SNDVs) as an aggregate measure of the Soviet intercontinental attack force. It is the numberof new ICBMs, new SLBMs, and new heavy bombers entering the force each year. Thismeasure has been implicitly predicted in each NIE, but it has not been emphasized. It also hasbeen implicitly measured by National Technical Means each year. The rate at which newSNDVs enter a force is a measure of how quickly the threat is growing, as well as a measureof the level of effort being applied to upgrade the force.? This article is based on a Research Paper published by CIA's Office of Soviet Analysis in April 1989.(b)(3)(n)1 1Approved for Release: 2014/07/29 006183129Approved for Release: 2014/07/29 006183129(b)(3)(n)Figure 1 compares the number of new SNDVs projected in each NIE from 1970 through1988 to those that were actually deployed. The horizontal scale shows NIE dates and thenumber of years into the future each comparison extends. The number of years compared iseither the NIE's total projection period or the number of years to 1989, whichever comes first.Comparisons begin in 1970 because the Soviets had deployed most of their initial interconti-nental attack force by then; from that point on, they have been modernizing that force. Theheights of the bars represent the numbers of new weapons entering the force during each NIE'sspecified projection period. The blue and green bars are the low and high projections,respectively, and the red bars in between are the actual historical deployments. For example,the chart shows that the 1970 NIE projected that during the next eight years the Soviets woulddeploy between 1,521 and 2,726 new SNDVs and that they actually deployed 1,880 newSNDVs.Figure 1Forecasting Record on, Modernization: New Soviet SNDVsNew SNDVs entering forceduring projection periodiiNIE low ormoderateprojectionActualn NIEhighU projection5.0004,0003 0002,0001,0000Date of projectiom_ _ 1970_Yeag_in_jarojection penal_ 8(b)(3)(n)Iihbiiii717173_74_75___7677__ 787980_8112_ 8385889_ 1110101010_10_98_36_ _4__ 86 87_ 883_2323377 9-89The High force projections always depicted a modernization rate in excess of whatactually occurred, often by a factor or two or more, even though for, most of them the totalSNDV count was assumed to be limited by the SALT agreements. In all of the NIEs from 1974through 1986, the modernization rates derived from the Low/Moderate force projections alsoexceeded actual Soviet deployments. In some cases, even the Low/Moderate projections offorce modernization were high by a factor of two.'The Low/Moderate forces in Figure 1 are the lowest forces projected in each NIE.However, as the name implies, not all of the Estimates contain forces based on an assumptionof a "low level of effort." The 1971 lowest force projection is described as a "likely" force. TheThe heights of the bars in Figure 1 generally increase from 1970 to 1974, then decrease to 1988. These trendsshould not be interpreted as changes ,in projections or history from year to year. The increase is due to thelengthening of the NIE projection period from 8 to 11 years, while the decrease is due to the gradual reduction inhistorical data available as one approaches the current year.OCCRCI(b)(3)(n)Approved for Release: 2014/07/29 006183129Approved for Release: 2014/07/29 006183129SECRET (b)(3)(n)1977-1979 lowest projections were labeled "Moderate," and they were based on theassumption of a moderate level of effort and technological success. The lowest force projectionsin the 1981 and 1982 NIa were based on the assumption that the SALT era was nearly over,so they cannot properly be interpreted as low-effort or SALT forces. The 1983-1986 lowestforce descriptions are similar to the earlier descriptions of "moderate" forces because theycontain statements such as "a pace reasonably consistent with that observed over the past 10years." The series of high force projections does not suffer from such ambiguity; every NIE hasat least one force based on a high level of effort.The fact that about half of the Low/Moderate forces were intended to be "Moderate"rather than "Low" does not explain why every, one of these force projections turned out to behigh in terms of modernization. Theoretically, the record on these "Moderate" forces shouldhave been equally split between high and low. Nor does it explain why several of them werehigh by more than ,a factor of two. Regardless of whether the frequent omission of a true "Low"force was proper or improper, the Intelligence Community did not project modernization ratesfor Soviet intercontinental attack forces accurately from the mid-1970s to the mid-1980s.The conclusion reached from the comparison in Figure 1 requires further justification, todemonstrate that it is not merely ,a consequence of our procedures or assumptions. To that end,our inquiry is broadened to include, in addition to new systems, the major modifications tothose systems that the Soviets routinely deploy later. These results are shown in Figure 2. Themodernization rates derived from the High force projections again exceeded the actualmodernization rates every time. In all of the NIEs from 1973 through 1986, the Low/Moderateforce modernization rates also exceeded the actual modernization rates. Thus, the problem isnot simply one of interpretation of the degree to which the Soviets would modernize eachweapon system. If some of the major modifications had been labeled follow-ons or vice versa,the trend would still be evident.Figure 2Forecasting Record on Modernization: New Soviet SNDVs and Major ModificationsaIIME low ormoderateprojectionActualNIE highprojectionNew SNDVs and major modifications entering force during projection period5,0004,000liDate- of projeetign? 1970_ 71 72 73 74 75 __76 77 78 29 80 _Al_ 82 81___85 87 = J8Pfsars_in_projecrion period: 8 8 ____8_ 9 II__ _10_ 10 _ _10_ 10 10 __9_ 8 7____ 6 _4 __ 3 2, 1s=7(b)(3)(n)SCCRETi(b)(3)(n)323378 9.883Approved for Release: 2014/07/29 006183129Approved for Release: 2014/07/29 0061831293ECRETi (b)(3)(n)Each aggregate projection is the summation of many individual weapon system projec-tions, each of which is composed of an Initial Operational Capability (IOC) date projection anda deployment rate projection. These two component projections were compared to historicaldata for all the intercontinental attack weapon systems deployed since 1970. The record on thefirst component, IOC dates, is summarized in Table 1. Ten of the 17 weapon systems weregenerally predicted to reach IOC earlier than they actually did. In six cases, the projectionswere fairly accurate, while in only one case were the projected IOC dates later than the actualSoviet performance. Thus, since 1970 the NIEs have generally overestimated IOC dates of newSoviet weapon systems.Table 1. The Record on IOC Dates Since 1970Projected IOC Projected IOC Projected IOCGenerally Generally GenerallyEarly Accurate LateSS-16 SS-17 or 19SS-17 follow-on* SS-18SS-18 follow-on SS-25SS-19 follow-on* SS-N-8SS-24 SS-N-18SS-N-6 follow-on* SS-N-23SS-N-8 follow-on'SS-N-17SS-N-20Blackjack? IOC never actually achievedBear HThere were three specific reasons for the trend of early IOC date projections. First, someweapon development programs were overestimated because US analysts misperceived Sovietmilitary requirements. Second, sometimes the Soviets adjusted their weapon developmentplans for arms control reasons. Third, the Soviets sometimes delayed or canceled a developmentprogram in progress because it had serious and expensive technical problems. Often, two ofthese reasons were operative on the same weapon system at the same time.In addition to predicting when a new weapon system will reach IOC, analysts have toproject that system's deployment rate. Only six Soviet intercontinental attack weapon systemscan be used to evaluate the record on new system deployment rates since 1970 because theyhave been in the field long enough?at least five years?to provide adequate historical data. Inthree cases the projections were generally high, in two cases the projections were generallysuccessful, and in one case the projections were generally low. These results are summarized inTable 2. Of course, all the projected systems from Table 1 that did not actually achieve IOChad deployment rates equal to zero, so the projections of those systems' deployment ratesturned out to be high.Table 2. The Record on Deployment Rates Since 1970Generally Generally GenerallyHigh Accurate LowSS-17SS-194SS-18SS-N-8SS-N-18Approved for Release: 2014/07/29 006183129(b)(3)(n)Approved for Release: 2014/07/29 006183129SECRET(b)(3)(n)The principal cause of the trend of high deployment rate projections was the use of thethird-generation ICBM program, the SS-9 and the SS-11, as a model for the fourth-generationICBM program, the SS-17, SS-18, and SS-19. Current projections of the fifth-generation ICBMprogram do not use the fourth-generation ICBM program as a model to such an extent, butthere is not enough historical data to date to evaluate these projections.Popular NotionsWhen one first recognizes the consistency of the Quantitative overestimates, it isintellectually tempting to rush to an obvious explanation, such as analysts -playing it safe" byshading their projections toward the high side or analysts bending under bureaucratic pressure.While it might be "safe" to shade the high force toward the high side every year, to makelow force projections -safe" analysts would shade them in the opposite direction, toward thelow side. The record does not show both high forces that are very high and low forces that arevery low, which is what one would expect if analysts had been -playing it safe." In addition,during the late 1970s, President Carter was negotiating and then seeking ratification of theSALT II treaty. There is certainly nothing -safe" about shading the SALT-limited projectionstoward the high side when the Administration is committed to a treaty. Finally, even if wecould identify some specific instances of -playing it safe" by yesterday's analysts, thisknowledge would not necessarily inspire tomorrow's analysts to perform better.Regarding the second popular notion, the military intelligence services, civilian policy-makers, legislators, and Administration officials occasionally bring strong views to bear on theformat and sometimes the content of the NIEs. This practice is well known in the IntelligenceCommunity, and it is generally understood to be part of the working environment. It reflectsthe legitimate desire to make NIEs truly -national" by providing a forum for alternative viewsand addressing a wide variety of policy concerns.For example, the Team "B" Experiment in Competitive Analysis, conducted from Junethrough December of 1976, addressed the concern that the NIE 11-3/8 series might have beensignificantly underestimating the Soviet threat. This exercise was designed from its inception toapply pressure on the authors of NIE 11-3/8, as evidenced by the following quote from one ofthe Team "B" reports:The mandate of Team -B" was to. . . determine whether a good case could be made thatSoviet strategic objectives are, in fact, more ambitious and therefore implicitly morethreatening to US security than they appear to the authors of the NIEs.However, we find no evidence that this pressure produced any substantial effects. Theprojections of the Soviet force modernization rate did not increase dramatically in the wake ofthe Team "B" experiment, and US perception of Soviet strategic doctrine had already begunto change before the experiment. The projected threat to Minuteman silos did increase at thistime, but this change was due to new evidence, not Team "13" analysis.'The numerous footnotes and alternative views taken by the various intelligence agenciesin the texts of the NIEs year after year also indicate that they often did not achieve the degree2 In the spring of 1978, DCI Stansfield Turner charged NIO/SP Howard Stoertz, Jr. with -running scared ofTeam 'B'," when Stoertz proposed changing the estimates of future Soviet ICBM accuracy, thus increasing theexpected threat to Minuteman. Stoertz denied bending under any such pressure, and he showed Turner the evidencefor the revised assessment. This incident is significant because it illustrates how important it was to those involvedto remain objective and to base the NIEs strictly on evidence.(b)(3)(n)Approved for Release: 2014/07/29 0061831295ccnET(b)(3)(n)?Approved for Release: 2014/07/29 006183129of influence they desired. Finally, as is the case with playing it safe, even if we could identifysome specific instances of unwarranted influence of bureaucratic pressures on yesterday's NIEauthors, this would not preclude tomorrow's authors from repeating the error.Corrective ActionsThere is no procedure that analysts can use to make correct projections every time.Inevitably, some intelligence projections will be wrong. Nevertheless, our study of the NIEs hasprovided some insights into the projections process, and several steps recently have been takentoward projecting aggregate modernization more accurately.First, a low force projection has been restored to the NIE series. In the 1987 NIE, we wereable to use some of what had been learned in this study about the trend of overestimates in theaggregate number of new SNDVs. For this NIE, the decision was made to include a low forceprojection, which had happened only once in the previous 10 years.'Second, increasing credence is being given to aggregate economic factors. Evidence on theSoviet economy usually has provided a restraining influence, but the Intelligence Communitybelieved that the Soviets placed a higher priority on achieving their strategic objectives and thatthose objectives were ambitious. Beginning in the mid-1970s, the NIEs contained statementssuch as, "We believe that economic difficulties will have little or no impact on Soviet strategicprograms during the period of this Estimate" and that these programs were intended to achievea "capability to fight and survive a nuclear war."In the early 1980s, evidence became available that the Soviet economy had deteriorateddramatically in the second half of the 1970s. But even though analysts realized the depth ofSoviet economic difficulty, they still saw little reason for it to influence their projections. Theycontinued to emphasize the Soviets' desire for "superior war-fighting capabilities" and did not"believe that domestic economic difficulties (would) bear significantly on the size andcomposition of future Soviet strategic forces because of the high priority the Soviets place onsuch forces."In 1985, the NIE authors began to reevaluate the impact of economic factors andconsequently the priority of strategic programs. It took another two years, however, before thereevaluation was explicitly translated into a force projection. The recognition that in somecircumstances Soviet economic problems could have a major impact on future forces was animportant consideration in the decision to restore a low force projection to the NIE series in1987.As a result of these two corrective actions, the 1987 NIE's low force projection is truly low.Figure 3 shows that it contains an aggregate modernization rate measured in terms of newSNDVs that is lower than any other NIE published during the SALT era. A figure using ouralternative measure?new SNDVs plus new mods?would be similar. Figure 3 also shows thatthe aggregate modernization rate derived from the low force continued to be low in the 1988NIE. (Because of time constraints, the 1989 NIE was not evaluated for this article.)3 The low-level-of-effort force projections were discontinued in the 1977 NIE. From that time until 1987, the onlyNIE to contain a force labeled -low- had been the 1980 NIE.6?sfeRE-7 (b)(3)(n)Approved for Release: 2014/07/29 006183129Approved for Release: 2014/07/29 006183129SECRET(b)(3)(n)Figure 3Recent Reductions in Projections of Modernization: New Soviet SNDVsI NIE low or1111 moderateprojectionNew SNEWs entering force during projection period(b)(3)(n)  323379 9.89Third, more attention is being focused on aggregate quantitative force modernization. USanalysts are now using the measures we introduced?new /SNDVs and new SNDVs plus newmods?as additional analytical tools when constructing Sovietiorce projections. Before a forceis completed, analysts compute the number of new SNDVs'and the number of new SNDVs plusnew mods contained in that force for each year and for. the full 10-year projection period; thenthey compare those values to past Soviet accomplishments. Analysts may then choose similarmodernization rates or different ones, depending on their judgments about 'how similar theSoviets' current political, military, and economic conditions are to those in the past.Interviews with the analysts involved indicate that aggregate modernization was notreceiving explicit attention in the 1970s and early 1980s. Thus, it was difficult for theIntelligence Community to recognize the trend as it developed. If the focus of interest had beenon aggregate modernization earlier, then the tendency to overestimate aggregate Soviet forcemodernization could have been detected earlier.Figure 4 illustrates how this tendency Might have been recognized. It displays anaggregate modernization bar chart of the type displayed, earlier, but it was drawn using dataavailable in 1980. As of 1980, the Estimates through 1974 usually contained the correct numberof new SNDVs?the high and low projections bounded history. Beginning in 1975, however,the modernization rates derived from both the high and low/moderate force projections.consistently exceeded history. Another chart based on the alternative modernization measure,new SNDVs plus new mods, again using only data available in 1980, would display similarcharacteristics?accurate projections from 1970 to 1.974 and high projections thereafter. Thus,US analysts could have detected the trend as early as 1980.(b)(3)(n)7Approved for Release: 2014/07/29 006183129 Approved for Release: 2014/07/29 006183129  4EGRET(b)(3)(n)Figure 4Forecasting Record on Modernization as of 1980: New Soviet SNDVsNew SND Vs entering force during projection period(b)(3)(n)323380 9.89,As a corollary to the above, it appears that the Intelligence Community has neither aninstitutional mechanism nor a strong built-in incentive to seek out overestimates vigorously. Onthe other hand, because the Community ,and its principal consumers of Soviet strategic?estimates are charged with protecting the nation against external threats, underestimates arelikely to be viewed as serious mistakes. Although underestimates are sought out and corrected,this may not be the case with overestimatesFourth; the projections are being integrated more completely. The analysts responsible forpredicting Soviet weapon system IOC dates and deployment rates are coordinating theirprojections with each other in greater detail. These discussions encourage each analyst tounderstand how. his projection influences other projections. This also encourages analysts todevelop a better understanding of the directions of total Soviet force modernization, ratherthan focusing on their own components of modernization.Until recently, US analysts had predicted system IOC dates and deployment ratesseparately by projecting the IOC dates first and then by projecting the deployment rates. Thisprocedure has worked well for individual weapon systems, but, if we consider a system and itsimmediate follow-on, we can find instances when this procedure may have caused problems.When the IOC' date for a specific replacement system follows close on the heels of the currentsystem, the existing system has to be deployed quickly or be overtaken. The early projectedIOC dates for the follow-ons probably have contributed to making .some projected deploymentrates high.RecommendationTo prevent the development of, persistent error trends in the future, studies like this oneshould be performed periodically. Such studies also should be conducted on other forecasts,including theater-range weapons ,and strategic defense forces If a study detects a 'growing8 3CCRET(b)(3)(n)Approved for Release: 2014/07/29 006183129Approved for Release: 2014/07/29 006183129sEGRET (b)(3)(n)trend of misestimates, then analysts may be able to halt it early. But those who perform theevaluations of projections should not restrict their inquiry to only those parameters that wereemphasized in the NIEs. If this had been done in this study, we would not have noticed thepattern of aggregate modernization overestimates. In addition to describing the accuracy ofpast projections, the studies should serve as checks on the Intelligence Community's focus ofinterest.This article is classified SECRET-ftelitEr(lb)(3)(1)(b)(3)(n)Approved for Release: 2014/07/29 0061831299