Hostname: page-component-cd9895bd7-8ctnn Total loading time: 0 Render date: 2024-12-28T03:49:24.310Z Has data issue: false hasContentIssue false
  • English
  • Français

Do Bureaucratic Politics Matter? Some Disconfirming Findings from the Case of the U.S. Navy

Published online by Cambridge University Press:  13 June 2011

Edward Rhodes
Edward Rhodes
Affiliation:
Associate Professor of Political Science at Rutgers University
Get access

Abstract

Allison's Model III (governmental or bureaucratic politics) suggests that state policies reflect the parochial concerns of intragovernmental “players in positions” and the relative power of these players. This article offers a critical test. It examines a policy area—U.S. decisions on the composition of naval forces—in which we would, a priori, expect bureaucratic politics to have a maximum effect and which participants, observers, and scholars have routinely described as critically influenced by bureaucratic politics. This article employs statistical methods to assess whether outcomes have indeed been affected by the parochial priorities and perceptions of individuals who, because of their relative power and the rules of the game, have dominated the relevant bureaucratic action-channels. Contrary to the expectations of the bureaucratic politics literature—indeed, contrary to the reports of firsthand observers and the actors involved—bureaucratic politics do not seem to have mattered: knowledge of bureaucratic interests and power does not permit us to predict outcomes. The article then proceeds to suggest an alternative model of state behavior that does provide significant explanatory power: the article demonstrates that shifts in force posture can be modeled as a function of ideas and images rather than of interests. This gives rise to speculation that in explaining American foreign and security policy the name of the game is not, as Allison suggests, politics, but the competition of ideas for intellectual hegemony.

Type
Research Article
Information
World Politics , Volume 47 , Issue 1 , October 1994 , pp. 1 - 41
Copyright
Copyright © Trustees of Princeton University 1994

Access options

Get access to the full version of this content by using one of the access options below. (Log in options will check for institutional or personal access. Content may require purchase if you do not have access.)

References

1 Allison, Graham, Essence of Decision (Boston: Little, Brown, 1971), 144.Google Scholar

2 Ibid., 164, 166, 168, 169, 170, 171. “Action-channels” are institutionalized procedures for facilitating or implementing governmental decisions. Allison distinguishes Model III (bureaucratic politics) from Model I (unitary rational actor) and Model II (organizational process). The latter, which Allison bases on the organizational behavior and bounded rationality literature of James March, Herbert Simon, and Richard Cyert, describes state action as a product of rigid organizational routines rather than rational choice: policies were to be understood “less as deliberate choices and more as outputs of large organizations functioning according to standard patterns of behavior” (p. 67).

3 The basic lines of criticism were sketched out early. Two of the first and most important critiques are Art, Robert J., “Bureaucratic Politics and American Foreign Policy: A Critique,” Policy Sciences 4 (December 1973)CrossRefGoogle Scholar; and Krasner, Stephen D., “Are Bureaucracies Important? (Or Allison Wonderland),” Foreign Policy 7 (Summer 1972).Google Scholar

4 Welch, David A., “The Organizational Process and Bureaucratic Politics Paradigms: Retrospect and Prospect,” International Security 17 (Fall 1992).CrossRefGoogle Scholar

5 As Bendor and Hammond complain, it is not entirely certain that Allison requires his Model III players to be rational actors; this lack of absolute clarity makes Allison's model a somewhat problem atic basis for deriving testable propositions. See Bendor, Jonathan and Hammond, Thomas H., “Re-thinking Allison's Models,” American Political Science Review 86 (June 1992), 304.CrossRefGoogle Scholar Nonetheless a balanced reading of Essence of Decision strongly suggests that, as Welch puts it, “Model III does not suppose that the individual players behave irrationally in the games in which they participate, merely that the net effect of those games is to deflect state behavior from the course that would have been chosen by a unitary rational actor.” Welch (fn. 4) 118.

6 On the notion of critical cases, see Eckstein, Harry, “Case Studies and Theory in Political Science,” in Greenstein, Fred and Polsby, Nelson W., eds., Handbook of Political Science 7 (Reading, Mass.: Addison-Wesley, 1975).Google ScholarWorld Politics 47 (October 1994), 1–41.

7 The alternative model is developed more fully in Rhodes, Edward, The Pursuit of Hegemony (New York: Columbia University PressGoogle Scholar, forthcoming).

8 Zumwalt, Elmo R. Jr, On Watch (New York: Quadrangle, 1976), 63–64.Google Scholar This particular passage from Zumwalt's memoirs has been widely cited by U.S. naval scholars without comment or dispute. See, for example, Hartmann, Frederick H., Naval Renaissance: The U.S. Navy in the 1980s (Annapolis, Md.: Naval Institute Press, 1990), 14.Google Scholar

9 A review of documents in the Lyndon Baines Johnson Library, for example, fails to provide any evidence that during his White House years Johnson ever considered the question of what overall U.S. naval posture should look like, despite the fact that the navy was facing the looming bloc obsolescence of its existing forces.

10 See, for example, Halperin, Morton H., Bureaucratic Politics and Foreign Policy (Washington, D.C.: Brookings Institution, 1974), 26–28Google Scholar, 39–40, 49–60.

11 Allison (fn. 1), 176; emphasis added. Critics of bureaucratic politics have shared this general appraisal—in fact, suggesting that it may only be on issues like budget and procurement, which strike to the heart of institutional essence and survival, that we will see an impact of bureaucratic politics. Robert Art, for example, concludes that on “institutional bread and butter issues” in which “neither the President nor the relevant Congressional figures are particularly interested in what the bureaucrats want … bureaucrats will likely get their way.” He adds that “what is clearly predictable is that lower-level bureaucrats will assume policy stances that stem from their estimate of what will best serve the long-term interests of the institutions to which they belong. Whether they succeed depends primarily on three factors: how strong are their allies in Congress, how willing the senior players attached to their institutions are to fight for these interests, and how committed a President is to getting his way, that is, how the President perceives bread and butter issues affecting his domestic program and foreign policies.” Art (fn. 3), 484. As we will see below, naval force posture fits perfectly Art's characterization of a case in which bureaucratic politics would be most likely to matter.

12 See Rhodes, Edward, “Domestic and External Determinants of Force Posture: Preliminary Findings from the Case of the U.S. Navy” (Paper delivered at the CUNY Political Science Conference, New York, March 1990).Google Scholar

13 Halperin (fn. 10), 32–33. See also, for example, Bendor, Jonathan and Hammond, Thomas H., “Rethinking Allison's Models” (Paper delivered at the American Political Science Association, Atlanta, Ga., August 31-September 3, 1989), 38.Google Scholar

14 Indeed, Allison himself confessed this problem, noting the massive informational requirements needed to make predictions about something as simple as a game of poker and observing that “the extraordinary complexity of cases of bureaucratic politics accounts in part for the paucity of general propositions.” Allison (fn. 1), 173–74. See also Bendor and Hammond (fn. 5), 318; and Welch (fn. 4), 120.

15 Allison (fn. 1), 173.

16 Ibid., 164.

17 Ibid.

18 Ibid.

19 Ibid., 165.

20 Ibid.

21 Ibid., 166.

22 Ibid., 166–67. Allison argues that this endemic parochialism may not even be conscious: “Members of an organization, particularly career officials, come to believe that the health of their organization is vital to the national interest. The health of the organization, in turn, is seen to depend on maintaining influence, fulfilling the mission of the organization, and securing the necessary capabilities” (p. 167).

23 Ibid., 167.

24 Ibid., 167–68.

25 Ibid., 168–69.

26 Ibid, 169.

27 Ibid, 170.

28 See the comment of CNO Watkins, James D., in Hartmann (fn. 8), 55.Google Scholar See also Zumwalt (fn. 8), 66.

29 See, for example, Hanks, Robert J., American Sea Power and Global Strategy (Washington, D.C.: Pergamon—Brassey's, 1985), 81Google Scholar; or Lehman, John F. Jr, Command of the Seas (New York: Charles Scribner's Sons, 1988), 99.Google Scholar

30 Hart, Gary with Lind, William S., America Can Win (Bethesda, Md.: Adler and Adler, 1986), 108.Google Scholar

31 Enthoven, Alain C. and Wayne Smith, K., How Much Is Enough? (New York: Harper and Row, 1971), 229.Google Scholar

32 Ibid. The denouement of this story is worth noting. Given the resulting analytical gridlock, the CNO proceeded with his own preferred mix of forces, blithely asserting that “ASW forces and war planning have been well balanced and are essentially correct” (p. 230).

33 See, for example, Hadley, Arthur T., The Straw Giant (New York: Random House, 1986), 69Google Scholar; and Builder, Carl H., The Masks of War (Baltimore: Johns Hopkins University Press, 1989), 25.Google Scholar These authors both suggest that the unusual severity of the navy's rivalry is a direct consequence of the nature of naval duty and the self-contained quality of naval vessels.

34 Breemer, Jan S., U.S. Naval Developments (Annapolis, Md.: Nautical and Aviation Publishing Company of America, 1983), 9.Google Scholar

35 Jordan, John, An Illustrated Guide to the Modern US Navy (New York: Prentice Hall, 1986), 8.Google Scholar

36 Hadley (fn. 33), 239–40.

37 See, for example, Jordan (fn. 35), 11; and Breemer (fn. 34), 9. For illustrations, see also Zumwalt's discussion of union-based assignment of key commands (i.e., commands that would be crucial to further promotion) and Zumwalt's discussion of his own propensity to select members of his union for key positions; Zumwalt (fn. 8), 44–45, 66. Or see Secretary of the Navy John F. Lehman's tale of the successful efforts of CNO Carlisle Trost, a submariner, to ensure that 80 percent of the submariners eligible for promotion to captain would be promoted, while only about 50 percent of the surface sailors and aviators would make the cut; Lehman (fn. 29), 36–38.

38 Among the reported reasons CNO Thomas H. Moorer opposed the appointment of Zumwalt to succeed him was the fact that Zumwalt was not an aviator. See Kenneth McDonald, J., “Thomas Hinman Moorer,” in Love, Robert William Jr, ed., The Chiefs of Naval Operations (Annapolis, Md.: Naval Institute Press, 1980), 362.Google Scholar A perhaps more interesting case is that of Louis Denfeld in 1947. In 1947 the navy split along union lines; Denfeld emerged as something of a compromise candidate. It has been widely reported that Denfeld, a destroyerman and personnel specialist, struck a bargain with key members of the naval aviation community to gain their support. See Clark, J. J. and Reynolds, Clark G., Carrier Admiral (New York: David McKay, 1967), 253Google Scholar; Palmer, Michael, Origins of the Maritime Strategy: American Naval Strategy in the First Postwar Decade (Washington, D.C.: Naval Historical Center, 1988), 39Google Scholar; Coletta, Paolo E., “John Lawrence Sullivan,” in Coletta, , ed., American Secretaries of the Navy (Annapolis, Md.: Naval Institute Press, 1980), 752Google Scholar; and idem, “Louis Emil Denfeld,” in Love, 194. See also Reynolds, Clark G., The Fast Carriers (New York: McGraw-Hill, 1968), 393.Google Scholar

39 This is most clearly illustrated in Zumwalt's own case. As one close observer of the navy reported, the secretaries of the navy and defense “picked Zumwalt, whom they believed would bring new ideas to the billet and a different perspective, given his background in the surface line” (emphasis added). See Friedman, Norman, “Elmo Russell Zumwalt, Jr.,” in Love (fn. 38), 367.Google Scholar Zumwalt himself reports that he was explicitly informed that one of the criteria leading to his selection was that he was not an aviator. See Zumwalt (fn. 8), 46.

40 Apparent examples of this are Secretary of the Navy James Forrestal's frustrated attempts to appoint an aviator to succeed first Ernest King and then Chester Nimitz as CNO. See Reynolds (fn. 38), 385–86; Taylor, Theodore, The Magnificent Mitscher (New York: W. W. Norton, 1954), 5Google Scholar; Clark with Reynolds (fn. 38), 253. For a different account, see King, Ernest J. and Whitehill, Walter Muir, Fleet Admiral King: A Naval Record (New York: W. W. Norton, 1952), 636.Google Scholar Even when they choose CNOs who have been moderates rather than extremists in the navy's intraservice political battles, the navy's civilian masters frequently appear to be quite aware of the intraservice implications of their choices. The desire to appoint union moderates seems to have been a factor, for example, in the appointment of Louis Denfeld (rather than William Blandy or Dewitt Ramsey), Forrest Sherman (rather than Arthur Radford), and possibly William Fechteler (again rather than Radford). Regarding Sherman's appointment, see Palmer (fn. 38), 33–53; regarding Fechteler's appointment, see Kennedy, Gerald, “William Morrow Fechteler,” in Love (fn. 38), 236–37.Google Scholar

41 Jan Breemer has suggested that the fact that no submariner occupied the CNO's office between Chester Nimitz's retirement in December 1947 and James Watkins's appointment in 1982, despite the navy's steadily increasing interest in submarines, was no accident. Rather it may have been a reaction to the power wielded by Admiral Rickover, who as the tsar of naval nuclear reactors was a major proponent of nuclear submarines (as well as nuclear-powered cruisers and aircraft carriers) and who, through a complicated system of oversight, exerted substantial influence over the promotion and careers of submariners. The venomous relationship between Rickover—who, because of his congressional support was untouchable—and his nominal masters, uniformed and civilian, in the Pentagon meant that only after Secretary of the Navy Lehman successfully forced Rickover's long-delayed retirement in 1981 was it likely that those civilian masters would appoint a submariner as CNO. See Breemer (fn. 34), 11.

42 For example, Coletta observes that the “poor status of naval aviation” in the fierce fight for funds during the 1949 belt-tightening can be attributed to several factors, specifically including “‘black-shoe’ or ‘battleship’ admirals who preferred to expand surface forces rather than aviation; and the fact that the chief of naval operations was not an aviator.” Coletta, Paolo E., “Francis P. Matthews,” in Coletta (fn. 38), 786.Google Scholar On the rise and fall of the naval aviators in intranavy bureaucratic struggles in 1948–49, see also Palmer (fn. 38), 46–48.

43 Carrison, Daniel J., The United States Navy (New York: Praeger, 1968), 78–80.Google Scholar Regarding the responsibilities of the CNO, see also U.S. Navy Department, Office of the Management Engineer, “The United States Navy: A Description of Its Functional Organization,” March 1947 (NAVEXOS P-435, Rev. 3–47), 11; U.S. Navy Department, “Report of the Committee on Organization of the Department of the Navy” (“Gates Committee Report”), April 16, 1954, 10–11; U.S. Navy Department, “Report of the Committee on Organization of the Department of the Navy” (“Franke Committee Report”), January 31, 1959 (NAVEXOS P-1996), 53–86, 147–48; U.S. Navy Department, Navy Management Office, “The Department of the Navy: A Description of Its Functional Organization,” May 1962 (NAVEXOS P-435, Rev 5–62), 7–9, 15; and Department of the Navy, “Review of Management of the Department of the Navy” (“Dillon Committee Report”), December 15, 1962 (NAVEXOS P-2426A), 135–38. For comparison with the prewar power of the CNO, see U.S. Congress, Senate, The United States Navy: Information Relative to Organization, Personnel, Fleet, and Shore Establishments of the United States Navy, 75th Cong., 1st sess., 1937Google Scholar, Document no. 35, pp. 6–9. On the early evolution of the CNO position, see Hartmann (fn. 8), 45–51; and Love (fn. 38). On the evolution of the CNO's position during the war, see Davis, Vincent, Postwar Defense Policy and the U.S. Navy, 1943–46 (Chapel Hill: University of North Carolina Press, 1966), 41, 65Google Scholar; Davis, Vincent, The Admirals Lobby (Chapel Hill: University of North Carolina Press, 1967), 158–62Google Scholar; and Connery, Robert H., The Navy and the Industrial Mobilization in World War II (Princeton: Princeton University Press, 1951), 434.Google Scholar

44 Lehman (fn. 29), 241.

45 Regarding Rickover, see, for example, ibid., 1–38; Zumwalt (fn. 8), 63–64, 72–74, 85–122, 154–60; and Polmar, Norman and Allen, Thomas B., Rickover (New York: Simon and Schuster, 1982).Google Scholar

46 Kerr, Andy, A Journey amongst the Good and the Great (Annapolis, Md.: Naval Institute Press, 1987), 166–67.Google Scholar Kerr is describing the views of Captain, later Admiral, Isaac Kidd, one of the major “players” in the navy in the late 1960s and early 1970s. Kidd was a principal contender for the CNO's job in 1974. Hartmann also stresses the CNO's authority on a wide range of issues. In Hartmann's account, “The solution to these [force posture and manpower] and similar problems, as far as the uniformed navy is concerned, is in the hands of the chief of naval operations.” Hartmann (fn. 8), 4.

47 Hartmann (fn. 8), 50.

48 Rosenberg, David Alan, “Arleigh Albert Burke,” in Love (fn. 38), 279.Google Scholar

49 Potter, E. B., Admiral Arleigh Burke (New York: Random House, 1990), 399.Google Scholar See also Rosenberg (fn. 48), 277–79, on the opposition of the relevant deputy chiefs of naval operations.

50 Kerr (fn. 46), 167 and 150, where he recounts the reaction of Secretary Korth to the predominant role of the CNO: “The CNO commands almost all of the resources of the navy. The secretary quickly finds himself isolated. Korth described the sensation like the driver of a huge truck that is thundering down the highway. The secretary sits high in the cab, thrilled with the majesty and power of it all. Then suddenly he realizes that his steering wheel is not connected to anything. Others, down below, are steering the vehicle and deciding where it will go.” For an excellent discussion of the service secretariat's lack of independent analytic capacity and its “rubber-stamp” quality, see also Hartmann (fn. 8), 42–45.

51 Lehman (fn. 29), 116.

52 Ibid., 244. On Lehman's bureaucratic reforms, including his successful effort to retire Admiral Rickover, abolish the Navy Material Command, and gain control over sources of information and analysis, see ibid., 1–38, 228–65; Kennedy, Floyd D. Jr, “From SLOC Protection to a National Maritime Strategy: The U.S. Navy under Carter and Reagan, 1977–1984,” in Hagan, Kenneth J., ed., In Peace and War: Interpretations of American Naval History, 1775–1984, 2d ed. (Westport, Conn.: Greenwood, 1984), 355–56Google Scholar, 359; Polmar, Norman, Ships and Aircraft of the U.S. Fleet, 14th ed. (Annapolis, Md.: Naval Institute Press, 1989), 1.Google Scholar

53 Thomas Etzold, as quoted in Richard A. Stubbing with Mendel, Richard A., The Defense Game: An Insider Explores the Astonishing Realities of America's Defense Establishment (New York: Harper and Row, 1986), 122.Google Scholar

54 Kaufmann, William W., A Thoroughly Efficient Navy (Washington, D.C.: Brookings Institution 1987), 2.Google Scholar

55 Stubbing with Mendel (fn. 53), 73.

56 Ibid., 122. The study in question was “Sea Plan 2000: Naval Force Planning Study.” For an account of “Sea Plan 2000” from a very different perspective, see Lehman (fn. 29), 129.

57 Stubbing with Mendel (fn. 53), 326–27.

58 For a summary of the evolution of action-channels, see Hartmann (fn. 8). The most important changes in the last forty years have involved the relationship between the CNO, the secretary, and procurement bureaus and systems commands.

59 Indeed, the problem becomes even trickier, because the presence of submariners as CNOs correlates closely with the absence of Admiral Rickover. Rickover, as noted above, was the only rival to the CNO in force-posture decisions; he was also an advocate of submarine power. Thus our case is not so sweet for making predictions about submarine forces: plausible Model III post hoc rationalizations can be made for whatever we find.

60 Data are from budgets from fiscal year 1956 through fiscal year 1988, inclusive. The navy procurement budget is taken as the sum of monies appropriated for navy procurement accounts, as reported after the completion of the fiscal year, exclusive of Marine Corps procurement. Aircraft procurement is calculated as the “aircraft” procurement account or as that portion of the “aircraft and related” or “aircraft and missiles” procurement account devoted to aircraft procurement. In the latter case indirect expenses have been assigned on a proportional basis. Since 1951, new CNOs have been sworn into office in July or August; in this analysis, new CNOs are associated with the budget presented to Congress the following January-March and defended before congressional committees over the spring and summer.

61 In this model, our right-hand side variables are a dummy variable taking unitary value in years in which an aviator was CNO; a dummy variable taking unitary value in years in which a submariner was CNO; a dummy variable taking unitary value in years in which the United States was at war; and an arithmetically increasing time variable to control for a long-run secular trend. Obviously, even though procurement decisions are theoretically made independently each year, there is always a concern that budget shares in any given year are, in fact, a function of the previous year's shares, raising concerns about autocorrelation. To reassure ourselves that this is not in fact a problem here, we can include on the right-hand side the left-hand side variable lagged one time period. This revised model has a slightly higher adjusted R-square, but the parameter estimate for the lagged left-hand side variable is not statistically significant, nor do the parameter estimates or standard errors of any of our other right-hand side variables change significantly. In other words, our findings are robust to this sort of manipulation.

62 Our only statistically significant finding in this model is that there has been a secular trend to ward a smaller share of the procurement budget going to aircraft. This is interesting but provides neither support nor challenge to Model III propositions. As for Zumwalt's more specific claim that he redressed an overemphasis on aviation that existed under his three predecessors, even for this confession of parochialism we can find no support. The average aircraft budget share under Zumwalt was .345; while the average share under his three predecessors was a marginally higher .355, the average share under Zumwalt's immediate predecessor, at .316, was actually slightly lower than under Zumwalt. Given the standard deviations associated with these averages (.021, .051, and .031 respectively), none of the differences between these averages is significantly different from zero. Using OLS regression techniques, modeling budget shares as a function of Zumwalt's tenure compared to that of his predecessors, and controlling for whether the nation was involved in war, we again find that there is no statistically significant impact from Zumwalt, compared to his three predecessors, on the proportion of the navy's procurement budget devoted to aircraft.

63 Data are from fiscal years 1953–88. Calculations of the tonnage of annual construction programs were based on data in Silverstone, Paul H., U.S. Warships since 1945 (Annapolis, Md.: Naval Institute Press, 1987)Google Scholar and were updated by me.

64 There are, of course, some interesting findings: they just do not support Model III's propositions. Not surprisingly, war has an impact on what is constructed: during wartime (that is, Korea and Vietnam), the United States constructed more mine and amphibious warfare vessels. Similarly, there are some long-run trends that seem to be consistent with technological changes. And, anomalously, there seems to be a negative impact of submariner CNOs on the acquisition of surface warships. As noted above, interpreting this finding is difficult, given that submariners have been CNO only in recent years and only since Rickover's removal. Our skepticism that bureaucratic politics are at work here is enhanced by the fact that the presence of submariner CNOs has no significant effect on the share of new construction devoted to submarines; indeed, to the extent there appears to be an impact, submariners would appear to be biased against submarine construction. Tests of alternative models suggest that the explanatory power of our model of share of new construction going to surface vessels appears to come from two factors, neither of which has anything to do with bureaucratic politics: first, during wartime, construction shifts from surface warships to amphibious and mine warfare vessels; second, there is a strong secular trend away from surface warships. We also find no statistically significant support for the proposition that, compared to his three predecessors, Zumwalt reduced carrier construction or increased surface warship construction. Indeed, though the results are far from statistically significant, an OLS model like the one above suggests that Zumwalt may have increased the share of new construction tonnage going to aircraft carriers.

65 The general-purpose fleet is the fleet excluding ballistic missile submarines, which are designed and operated solely for strategic nuclear deterrence purposes. Force requirements for these ballistic missile submarines are, by all accounts, determined by the National Command Authorities in consultation with Congress. Therefore, to make the strongest possible “easy” test case for a bureaucratic politics model, we exclude these from our calculations.

66 For data on retirement patterns, including the very different history of sister ships, see Silver-stone (fn. 63).

67 Data are from calendar years 1950–86. Fleet composition is based on commissioning status data reported in Silverstone (fn. 63). Our model uses the difference between the percentage of the commissioned general-purpose fleet (excluding auxiliary vessels and minor patrol vessels) “belonging” to a particular union at the beginning and end of the calendar year as the dependent variable. This is modeled as a function of three arguments: a dummy variable measuring whether the nation was at war, a dummy taking unitary value when the CNO was a submariner, and a dummy taking unitary value when the CNO was an aviator. This permits us to measure the impact of aviators as compared to surface sailors, and vice versa. As noted above, CNOs have typically assumed office in July or August. Since commissioning and decommissioning decisions take a matter of months to implement, responsibility for the changes implemented during the calendar year is assigned to the outgoing CNO.

68 The findings of the “self-perception theory” literature are intriguing in this regard. For an excellent discussion, see Larson, Deborah Welch, Origins of Containment (Princeton: Princeton University Press, 1985), 45–50.Google Scholar

69 Art (fn. 3), 476. Art, it should be emphasized, uses this critique as a basis for arguing the utility of a rational unitary-actor paradigm. But the observation, and the challenge it poses to any “rational actor” model of behavior, is a profound one—as Art, by grounding his argument in the work of the “first-wave” foreign policy theorists such as Warner Schilling, makes clear. Schilling argued, Art observes, that while institutional interests would indeed be a factor, the principal influence on state behavior would be not interests but ideas: “The kind of defenses a budget provides will be primarily a reflection of the kinds of ideas people have about the political-military world in which they are living.” See Schilling, Warner R., “The Politics of National Defense: Fiscal 1950,” in Schilling, Warner R., Hammond, Paul T., and Snyder, Glenn H., Strategy, Politics and Defense Budgets (New York: Columbia. University Press, 1962), 15.Google Scholar

70 The phrase is Michael Shafer's. See Shafer, , Deadly Paradigms (Princeton: Princeton University Press, 1988), 32.Google Scholar

71 In recent years, the impact of ideas on behavior has become a major topic in the international relations literature. A complete list of relevant work is hardly possible. Certainly, however, among the important contributions to our understanding are Goldstein, Judith, “The Impact of Ideas on Trade Policy,” International Organization 43 (Winter 1989)CrossRefGoogle Scholar; Haas, Ernst B., When Knowledge Is Power (Berkeley: University of California Press, 1990)Google Scholar; Hall, Peter, ed., The Political Power of Economic Ideas (Princeton: Princeton University Press, 1989)Google Scholar; Odell, John, U.S. International Monetary Policy (Princeton: Princeton University Press, 1982)CrossRefGoogle Scholar; and Shafer (fn. 70).

72 For an early exposition of the implications of cognitive psychology's notions of simplicity and stability for understanding state behavior, see Steinbruner, John D., The Cybernetic Theory of Decision (Princeton: Princeton University Press, 1974), esp. 101–3.Google Scholar The most influential work on the topic remains Jervis, Robert, Perception and Misperception in International Politics (Princeton: Princeton University Press, 1976).Google Scholar

73 Considerable literature exists on the phenomenon of learning. For an excellent overview, see Levy, Jack S., “Learning and Foreign Policy: Sweeping a Conceptual Minefield,” International Organization 48 (Spring 1994).CrossRefGoogle Scholar Perhaps the most intriguing approach for modeling the spread of ideas within a group, however, involves the application of analogies from evolutionary biology. See Farkas, Andrew, “State Learning and International Change” (Ph.D. diss., Princeton University, 1994).Google Scholar

74 Jervis (fn. 72), 217–87.

75 On the possibly pernicious effect of Mahan's teachings, the divergence between Mahan's image of naval warfare and the reality of America's historical experience, and the plausibility and historic reality of alternative images of naval warfare, see, for example, Hagan, Kenneth J., This People's Navy: The Making of American Sea Power (New York: Free Press, 1991).Google Scholar

76 For an account of Mahan's views on naval warfare and a discussion of the extent and limits of his influence on the American navy, see, for example, Crowl, Philip A., “Alfred Thayer Mahan: The Naval Historian,” in Paret, Peter, ed., Makers of Modern Strategy from Machiavelli to the Nuclear Age (Princeton: Princeton University Press, 1986), 444–77Google Scholar; Margaret Tuttle Sprout, “Mahan: Evangelist of Sea Power,” in Earle, Edward Meade, ed., Makers of Modern Strategy: Military Thought from Machiavelli to Hitler (Princeton: Princeton University Press, 1943), 415–45Google Scholar; Livezey, William E., Mahan on Seapower (Norman: University of Oklahoma Press, 1981), 40–98Google Scholar; or Seager, Robert II, Alfred Thayer Mahan (Annapolis, Md.: Naval Institute Press, 1977), 160–253.Google Scholar Again, it should be underscored that, for reasons of space, this analysis has been deliberately reduced to its most simplistic version: a slightly more complex model would take into account intellectual divisions within the government that made the Mahanian image relatively weaker in the period prior to 1968 than in the period following. The more complex model offers substantially more predictive power; the simple one will do well enough here, however.

77 On the legacy of Vietnam for policy-making, see, for example, Herring, George C., America's Longest War, 2d ed. (New York: Knopf, 1986), 272–81Google Scholar; or Holsti, Ole R. and Rosenau, James N., American Leadership in World Affairs: Vietnam and the Breakdown of Consensus (Winchester, Mass.: Allen and Unwin, 1984).Google Scholar

78 Builder (fn. 33), 21. Throughout the 1930s the United States operated fifteen battleships, which were the capital vessel of the day. See Friedman, Norman, U.S. Battleships (London: Arms and Armour Press, 1985), 418–20.Google Scholar

79 This calculation is based on standard navy accounting practice during the period. The United States has extended carrier service life through periodic substantial rebuilding of existing vessels. Nondeployable carriers undergoing service life extension or major reconstruction are excluded from this count, regardless of whether they are technically decommissioned.

80 For our calculations, we have taken the battle fleet as composed of attack aircraft carriers, battleships, warships designed for fleet escort duties, and warships designed for fleet radar picket duties. These are the forces principally designed to operate as a unit to destroy or contain the adversary's concentrated surface fleet and its bases ashore.

81 This is calculated here as the sum of ASW carriers, surface ocean-escort types, and modernized diesel-powered attack submarines suitable for ASW barrier duty.