The Last Warrior

by Andrew F. Krepinevich

  Marshall, never inclined to accept the conventional wisdom at face value, decided to think the problem through and draw his own conclusions. Starting in 1984 he entered into a teaming arrangement with Guy Barasch from the Los Alamos Scientific Laboratory to explore the potential effects of missile defenses on the US-Soviet strategic nuclear balance. Barasch was working with the newly established Strategic Defense Initiative Office (SDIO). The thrust of their effort was to use seminar war-gaming to assess various missile-defense architectures.52 The games were played using Booz Allen Hamilton’s Strategic Analysis Simulation (SAS), which Mark Herman, a former colleague of Phill Karber’s at BDM, and others had automated on a personal computer.53 Charlie Pease served as Marshall’s point man for the games.

  Going into the SDI games the players, like most observers, estimated that US missile defenses would need effectiveness levels of 90 percent or better in order to have a significant effect on the US-Soviet nuclear relationship.54 The working assumption was that as little as ten percent leakage would still find hundreds of Soviet warheads getting through to US targets. To explore this assumption a total of fourteen of SDI games were conducted with Soviet specialists from the intelligence community playing the “Red” or Soviet side.

  When the “Blue,” or US, side had layered defenses capable of reducing leakage to 10 percent (that is to say, when 90 percent of the incoming warheads were successfully engaged), the Red players were “deterred at every level of war much more” than they were in games in which the Blue side lacked missile defenses.55 The real surprise, however, came when Barasch and Pease decided to run a game to investigate the effectiveness threshold for a “Small Strategic Defense.” What they discovered was that SDI effectiveness levels as low as 15 percent—that is, leakage as high as 85 percent—exerted a “profound impact” on the US-Soviet nuclear balance.56 The reason, as John Battilega’s efforts have shown, was the uncertainty that even minimally effective defenses imposed on the attacking side. Soviet military theory and practice demanded very high probabilities of mission success in nuclear as well as in conventional conflicts. The Red players used these planning norms in the games.57 The upshot was that Red players had to allocate so many warheads to ensure a sufficiently high success rate against the US Minuteman ICBM force that they risked exhausting their own ICBM forces. Even modest missile defenses combined with the Soviet military’s overly ambitious planning norms created a “missile sink” for the Soviet ICBMs. This suggested that even very “leaky” US missile defenses could have a strong deterrent effect on the Soviets.58

  As with a number of other Marshall initiatives, this insight about the potential influence of missile defenses on attacker uncertainty never became a major part of the public debate over SDI. Moreover, as in the cases of competitive strategies and the US-Soviet investment balance, Marshall could provide senior policy makers with insights derived from ONA’s assessments, yet he could not compel these officials to act on them. Put another way, he could lead the policy-making horses to water, but could not make them drink—or, perhaps, think.

  It remains something of a mystery as to why the findings of ONA’s SDI games were not used to counter the criticism that missile defense systems needed to be near perfect, if not impenetrable, to affect the balance between US and USSR nuclear forces, or to influence Soviet thinking about the risks associated with attempting a disarming first-strike against US land-based missile forces. Interestingly, while Marshall’s assessment did not enter the mainstream of discussions among senior US policy makers, there is at least circumstantial evidence that he had a good fix on how the USSR viewed the prospect of American missile defenses. Like their US counterparts, the Soviets knew the limitations of prospective missile defense systems, which were the subject of intensive debate. But despite these limitations, from 1983 until the USSR’s collapse constraining SDI became “the single most important object of Soviet diplomacy and covert action.”59 And even after the Soviet Union’s collapse in 1991, Russian concern over even limited US missile defenses against a small attack from, say, North Korea or Iran has not abated. As recently as February 2011, when the New Strategic Arms Reduction Treaty (New START ) officially entered into force, Soviet deputy prime minister Sergey Ivanov warned that any attempt to build a shield against ballistic missiles would inevitably provoke the creation of a “better sword” in the form of increasingly accurate Russian ICBMs.60

  Particularly for those outside ONA, the varied and complex relations between its external research program and its formal balances were often opaque. But as Marshall and Schlesinger recognized at the outset, the research program was a necessary, vital component of Marshall’s development and maturation of net assessment. Without it, ONA’s assessments would have been confined to official government data. But as Marshall found time and again, the data available from official sources could be nonexistent, incomplete, or simply wrong.

  The debate between the CIA and ONA over the burden that Soviet military programs imposed on the USSR’s economy is perhaps the most consequential example of official intelligence estimates simply getting it wrong and, hence, the need for independent research outside the government. During the 1970s most of ONA’s efforts to estimate the ratio of Soviet military spending to the USSR’s GNP focused on getting the numerator—how much the Soviets were spending on their military forces—right. Again, Marshall’s view was that the CIA’s estimates of Soviet defense spending concentrated on the visible elements of the USSR’s military forces: ICBMs, naval combatants, shipyards and naval bases, fighter and bomber bases and the aircraft on them, surface-to-air missile batteries, tanks, armored fighting vehicles, military design bureaus, manufacturing plants, and so forth. With the advent of satellite photographic reconnaissance in 1960, the US intelligence community was able to build up relatively accurate estimates of the Soviet military’s equipment holdings. Rightly suspicious of official Soviet figures for defense expenditures, the CIA adopted a building-block approach to estimating the USSR’s military spending, based on the concept that prices × quantities = spending.61

  Identifying prices presented the greatest challenge because the USSR spent rubles, not dollars, on its military programs. Establishing a valid ruble–dollar exchange ratio proved difficult, as ruble prices in the Soviet Union were not set by market forces as were dollar prices in the United States.62 In contrast, getting the quantities of military goods to be priced was simpler because over time the CIA was increasingly able to estimate the numbers of units in the USSR’s order of battle, tables of organization and equipment for Soviet units, manpower levels, and production rates for weapons and equipment.63 In 1967 the CIA’s Office of Strategic Research (OSR) was responsible for estimating Soviet defense expenditures. A different Agency staff element, the Office of Economic Research (OER), was tasked with estimating Soviet GNP.64 Because of this division of labor there was ambiguity about which of the two offices was responsible for estimates of the USSR’s military burden. This ambiguity seems to have persisted even after OSR and OER were absorbed into the CIA’s Office of Soviet Affairs (SOVA) in 1981.

  In 1983 ONA analyst David Epstein completed a two-volume assessment of the US-Soviet investment balance. Epstein’s assessment built on the efforts of Major Lance Lord and retired Navy captain William Manthorpe who had inherited the military investment portfolio from Major Robert Gough in the late 1970s. The assessment broadly accepted CIA estimates of the direct costs of Soviet weapon systems that had been generated with the Agency’s building-block approach. But Soviet production rates were hard to square with estimates of the USSR’s military burden. In 1986 the CIA and the DIA estimated that over the years 1974–1985, the USSR had procured three times as many ICBMs and SLBMs as had the United States, nine times as many surface-to-air missiles, three times as many tanks, and ten times as many artillery pieces.65 Even though US systems were generally more costly than their Soviet counterparts, the USSR’s dramatically higher production rates suggested that the burden imposed on
the Soviet economy should be higher than the CIA’s estimate of less than 14 percent, which had been revised upward from its earlier estimates of roughly 6–7 percent.66

  There were also at least two sources of “indirect” military costs. One consisted of costs incurred outside the country’s defense budget where the Soviet military acquired resources or influenced their allocation to satisfy wartime requirements. Here Marshall cited as examples the fact that Aeroflot passenger aircraft and the USSR’s merchant marine fleet incorporated additional features—at additional cost—to enable them to transport men and equipment in time of war. Marshall also pointed to the costs incurred by the Soviets in providing economic and military aid to its satellites and clients—in effect, the costs of sustaining the Soviet Union’s external empire.67

  There was yet another tranche of expenditures beyond the direct costs of Soviet military programs. In the spring of 2001 Marshall met in Paris with Colonel Vitaly V. Shlykov, who had been in one of the major planning sections of the Soviet General Staff. Shlykov revealed that the Soviet military leaders had accumulated and incurred the cost of maintaining “gigantic war reserve stocks” intended to help match their estimates of US production if the country mobilized fully as it had done in World War II.68

  Epstein contended that the CIA’s estimates of the numerator for the Soviet defense burden clearly left out the indirect costs of the USSR’s military efforts. He estimated that these indirect costs added another 5 to 8 percent of GNP to the USSR’s defense burden.69 After the Cold War former CIA analysts Noel Firth and James Noren recalculated the CIA’s historical burden estimates. They estimated the average burden for 1981–1990 to have been 14.8 percent, using constant 1982 ruble prices (and 15.8 percent for 1981–1989, using current prices). Their revised burden for the 1970s was 16.8 percent, using constant 1982 ruble prices. Firth and Noren conceded, therefore, that the 6 to 7 percent in the early 1970s had been a substantial underestimate.70

  There was, however, another issue—estimates of the denominator (the actual size of the Soviet economy)—where the disagreement between the CIA and ONA was even more pronounced. Soviet GNP eventually turned out to be not 55 to 60 percent of US GNP, as the CIA believed during most of the Cold War, but closer to 25 or 30 percent. Igor Birman was one of the first to raise this possibility with Marshall. Before immigrating to the United States in 1974, Birman had been a member of the Soviet Communist Party and had worked as an economist in several Soviet industrial institutes; he had also published a number of books in Russian on economic issues.71 Sometime in the early 1970s Birman got into trouble with the Party over provocative statements he made about the Soviet economy and centralized planning during a lecture on Gosplan.* As a result he decided to leave the Soviet Union.

  In the early 1980s Birman began arguing that Soviet GNP was as low as 20 to 25 percent of US GNP, thereby directly challenging the CIA’s much higher estimates and the prevailing conventional wisdom in the West concerning the Soviet economy.72 Consequently he found himself engaged in numerous professional disputes with Western Sovietologists and economists, not least those at the CIA. Marshall, long accustomed to challenging the conventional wisdom and, given his abiding interest in how the world really works, was one of the few who took Birman seriously enough to support his research.

  Time proved Marshall—and Birman—right. After the Cold War ended, records released from Russian state archives showed Birman’s analysis to have been largely correct. Yet even he actually underestimated the proportion of national wealth being swallowed by the Soviet military.73 As late as 1990 Birman put the USSR’s defense burden as a least 25 percent of Soviet GNP.74 Marshall had reached roughly the same estimate in 1987, prompted in large part by articles such as Vasiliy Selyunin and Grigoriy Khanin’s “Cunning Figures,” which appeared in the February 1987 issue of Novyy Mir (New World). Their contention was that Western estimates of recent Soviet economic growth were overstated; in fact, the USSR’s economy had actually stagnated.75

  Birman was by no means the only researcher who developed contrarian views on the size of the USSR’s economy during the 1980s. Other researchers who did so included Henry Rowen, Charles Wolf, Anders Åslund, and Stephen Meyer.76 All received encouragement and in some cases funding support from Marshall’s office. Inevitably these outside research efforts reinforced Marshall’s long-standing conviction that the CIA’s estimate that Soviet GNP was 55 percent of US GNP was “total hokum.”77

  In his efforts to alert senior US policy makers to his conclusions, Marshall also found some allies at the CIA whose judgments were closer to his than those of the SOVA economists. Among them were Herbert Meyer and, ultimately, William Casey himself. Casey had hired Meyer as a special assistant and deputy director of the NIC. Prior to joining the CIA Meyer had been an associate editor at Fortune magazine, where he specialized in the Soviet economy. He had also traveled extensively in the USSR. In late 1983, Meyer wrote a memorandum to Casey arguing that the Soviet economy was “heading toward calamity” and, with an average annual growth rate of less than 2 percent, could not sustain an estimated 4 percent growth rate in military spending.78 Another notable skeptic about SOVA’s burden estimates was Robert Gates, who served Casey as deputy director for Intelligence and, later, as the DCI and eventually as secretary of defense.79

  In the end, intelligence community estimates of the USSR’s military burden by the 1980s were understated by at least a factor of two. In 1975, Marshall had estimated the burden as “10 to 20 percent of Soviet GNP” versus the 6 to 7 percent then estimated by the CIA.80 By 1987 he thought the burden had been “somewhere between 20 and 30 percent” during the preceding ten years.81 In 2001, after Shlykov’s revelations, he concluded that the burden had probably been even higher: “somewhere in the range of 35 to 50 percent of Soviet GNP.”82 Without an office as independent of the intelligence community as ONA and a director with Marshall’s emphasis on rigorous analysis and facts, US senior policy makers would have had to rely solely on the CIA’s flawed estimates of the USSR’s military burden.83

  The economic burden imposed by Soviet military programs was not the only area where ONA’s study funds for independent research paid dividends. Starting in 1970, the Soviet Union temporarily loosened emigration restrictions for Jewish emigrants, allowing nearly 250,000 people to leave the USSR. By the mid-1970s the flow of former Soviet citizens into the United States had reached significant numbers.

  A young RAND analyst, Enders Wimbush, approached Marshall and suggested that these émigrés might prove a source of useful information regarding the role of ethnic factors in the Soviet Army and, by extension, its effectiveness as a fighting force.84 Marshall asked Wimbush to put together a proposal for an émigré interview project. Marshall sent it to both the CIA and the DIA. Neither intelligence agency expressed any interest in it. Their view was that any information gleaned from Soviet émigrés could not be trusted.

  Marshall decided to fund the project himself. Eventually some 200 to 250 émigrés were interviewed about their military experiences. The effort, Wimbush concluded, “taught us not just about the ethnic factor in the Soviet armed forces, it taught us about their staffing requirements; it taught us about violence in the ranks; it taught us about the alcohol problem; it taught us about the political oversight problem. And when we finished with this, we had a profoundly different view of what the inside of the Soviet military looked like. . . . ”85

  The conscript personnel that formed the backbone of the Soviet armed forces clearly were not comparable to the volunteer soldiers, sailors, marines, and airmen that composed the US military. Moreover, in the aftermath of Vietnam the US military began investing in realistic combat training with such innovations as the Navy’s Topgun training program for its fighter crews, the Air Force’s Red Flag exercises at Nellis Air Force Base in Nevada, and the Army’s National Training Center at Fort Irwin in California. The Soviets had nothing to match these initiatives.

  By the mid-1980s President Rea
gan was halfway through his second administration. Having voted for dramatic increases in the defense budget throughout the early part of the decade, Congress was showing a growing reluctance to continue the rate of growth advocated by Weinberger. As a result, Marshall’s influence increased, especially during the Commission on Integrated Long-Term Strategy (CILTS). The Defense Department formed CILTS in the fall of 1986, stimulated in part by Weinberger’s efforts to silence critics who accused him of having no strategy.

  Chaired by Iklé and Albert Wohlstetter, both long-standing Marshall colleagues, CILTS was intended to chart those broad trends shaping the future security environment that were likely to affect US defense strategy over the next fifteen or twenty years, particularly with an eye to preventing catastrophic technological or geopolitical surprises.86 Among the commission’s eleven members were Zbigniew Brzezinski, Andrew Goodpaster, Samuel Huntington, Henry Kissinger, and John Vessey, former chairman of the Joint Chiefs of Staff. To support the Commission’s deliberations Iklé and Wohlstetter established four working groups. Retired Army general Paul Gorman examined low-intensity conflict; Fred Hoffman and Henry Rowen looked at offensive and defensive power projection on the periphery of the USSR; Charles Hertzfeld headed the technology working group; and Marshall and Charles Wolf explored the overall security environment into the first decade of the twenty-first century.

  Marshall had worked with Iklé, Wohlstetter, Hoffman, and Rowen at RAND and later in government. In 1987 Wolf was still at RAND. Herzfeld had inspired Marshall’s first net assessment in 1970 for Kissinger’s Special Defense Panel, and Marshall had collaborated with Gorman on a net assessment of US and Soviet tank crew training in the early 1970s. So the CILTS chairmen and the heads of its working groups knew one another well and had extensive experience in strategy.