I would like to thank Joe Grieco, Stephan Haggard, John Ikenberry, Robert Keohane, Cliff Kupchan, Ed Rhodes, Stephen Van Evera, and the editors and reviewers of IO for their comments on earlier drafts of this essay. I wrote the article while a fellow at Harvard University's Center for International Affairs, with financial support from the Ford Foundation and NATO.

1. I recognize that the theoretical scope of this essay is limited by focus on one case study. My aim is not to present a definitive examination of theories of alliance behavior, but to synthesize and put forth a set of clear hypotheses for studying alliances, to show how they can be applied to empirical materials, and to establish causal statements about the sources of cooperation and noncooperation within the confines of the case study. While the causes of aliance cohesion and discord in this specific episode may not apply to other cases, the approach to studying alliance behavior I set forth should be generally applicable. Within Harry Eckstein's classification of case studies, this is a “heuristic” one. In Alexander George's words, a heuristic case study “is used as a means of stimulating the imagination in order to discern important new general problems, identify possible theoretical solutions, and formulate potentially generalizable relations that were not previously apparent. In other words, the case study is regarded as an opportunity to learn more about the complexity of the problem studied, to develop further the existing framework, and to refine and elaborate the initially available theory.…” See George, , “Case Studies and Theory Development: The Method of Structured Focused Comparison” in Lauren, Paul Gordon, ed., Diplomacy: New Approaches in History, Theory and Policy (New York: Free Press, 1979), pp. 51–52Google Scholar.

2. I use “political” to refer to the realm of issue-areas concerned with general alliance policy and military strategy; “cooperation” is measured primarily in terms of mutual concessions and policy adjustments. I use “economic” to refer to the realm of issue-areas concerned with the distribution and allocation of goods and services; cooperation is measured primarily in terms of a reasonable sharing of the collective defense burden.

3. There is a small but growing body of literature on the out-of-area problem. For a general background on intra-alliance controversies over peripheral conflicts, see Grosser, Alfred, The Western Alliance (London: Macmillan, 1980)Google Scholar, especially chaps. 5, 8, and 10. For works focusing specifically on the Middle East, see Kupchan, Charles, The Persian Gulf and the West—The Dilemmas of Security (Boston: Allen & Unwin, 1987)Google Scholar and Spiegel, Steven, ed., The Middle East and the Western Alliance (London: Allen & Unwin, 1982)Google Scholar. Articles dealing exclusively with the out-of-area problem include: Douglas Stuart, “NATO Out of Area Disputes: From the Washington Talks to the RDF,” Atlantic Quarterly 2 (Spring 1984)Google Scholar; Tow, William, “NATO's Out-of-Region Challenges and Extended Containment,” Orbis 28 (Winter 1985)Google Scholar; Treverton, Gregory, “Defense Beyond Europe,” Survival 25 (09–10 1983)CrossRefGoogle Scholar.

4. European and American views diverged widely on the Palestinian question as well as on the value of undertaking collective efforts to deal with oil shortages. The Europeans (with the exception of France) did join the International Energy Agency to establish guidelines for responding to emergency shortages, but they also insisted on maintaining an independent bilateral “Euro-Arab” dialogue with producer states.

5. The purpose of Article Six—the portion defining the formal treaty area—was not to preclude the possibility of allied actions outside Western Europe or to the south of the Tropic of Cancer; it was intended to ensure that an attack on colonial territories not be automatically considered as an attack on the alliance as such. See Theodore Achilles, “U.S. Role in Negotiations that Led to the Atlantic Alliance,” NATO Review 5 (10 1983), p. 17Google Scholar.

6. For the historical background to the out-of-area debate, see Kupchan, The Persian Gulf and the West, chaps. 7 and 8.

7. U.S. Congressional Budget Office, Rapid Deployment Forces—Policy and Budgetary Implications (Washington, D.C.: GPO, 1983), pp. xvi–xxxGoogle Scholar.

8. “U.S. Asking Allies to Assume More of Military Burden,” New York Times, 14 April 1980.

9. Between 12 and 15 March, 1984, 1 gathered details of these negotiations through personal discussions and interviews with NATO staff and delegates. European delegates, without exception, complained that the United States showed little tact in the timing or presentation of its request for a formal out-of-area policy in NATO.

10. The proposal for a NATO strike force was publicly dropped by the United States in October 1980. See U.S. International Communication Agency, 8 October 1980, “An Interview with U.S. Secretary of Defense Harold Brown.”

11. U.S. Congress, House Appropriations Committee, Military Construction Appropriations for 1983, Part 5 (Washington, D.C.: GPO, 1982), p. 257Google Scholar.

12. Luns quoted in “NATO Plans for Europe in Case US Switches its Troops to the Gulf, ” London Times, 7 May 1980.

13. NATO DPC, Final Communiqué (May 1980), paragraphs 5 and 6.

14. Ibid., paragraph 6.

15. North Atlantic Assembly, Political Committee, “Interim Report of the Sub-Committee on Out-of-Area Security Challenges to the Alliance” (Brussels, 11 1984), p. 5Google Scholar.

16. See, for example, NATO NAC, Final Communiqué (December 1981), paragraph 9.

17. The report was entitled “Study on the Implications for NATO of the U.S. Strategic Concept for South-West Asia.” See North Atlantic Assembly, “Interim Report” (11 1984), p. 7Google Scholar.

18. Under the current configuration of the RDF, a full deployment to Southwest Asia would involve over 400, 000 troops. Given that it could well use virtually all of the U.S. strategic lift capability, another six U.S. divisions might have no way to get to Europe. Among major European countries, the British, Italians, and Germans have substantially raised force levels (reserve and active) between 1980 and 1985, but only by about 20, 000, 17, 000, and 15, 000, respectively. Moreover, these increases reflect long-term programs and do not indicate specific efforts to respond to the depletion problem. Dutch force levels have actually declined by about 5, 000 troops. As far as lift capability is concerned, the British and French have both purchased a small number of new lift aircraft and are stretching existing planes. The overall impact, however, is minimal: the French and British each have one squadron of strategic lift aircraft while the United States has seventeen. International Institute for Strategic Studies, The Military Balance (London: IISS, 1980–1986)Google Scholar.

19. See, for example, U.K. House of Commons, “Second Report from the Defense Committee —Statement on the Defense Estimates 1980” (London: HMSO, 1980), p. 62Google Scholar; and U.S. Congress, House Armed Services Committee, “NATO—An Alliance of Snared Values, ” pp. 70–71Google Scholar.

20. Treaties and other formal arrangements among nations often lapse or fail to be implemented. This does not mean, however, that the original agreements were not the result of cooperation or that the states involved had no intentions of ever honouring their commitments.

21. These four hypotheses have been culled from the literature on alliances and they encompass most of the discrete propositions that have been put forward to explain intra-alliance behavior. The best review of the literature can be found in Holsti, Ole et al. , Unity and Disintegration in International Alliances (New York: Wiley, 1973), especially pp. 16–24Google Scholar. The only variable they raise which lies outside the scope of these hypotheses is bureaucratic processes. I consider the overall impact of bureaucratic structures on outcomes to be minor when compared with the roles played by the relevant state actors. See also Ward, Michael, Research Gaps in Alliance Dynamics, Monograph Series in World Affairs, vol. 19, no. 1 (Denver: Colorado Seminary, University of Denver, 1982)Google Scholar. For an insightful study of the formation of alliances, see Walt, Stephen, The Origins of Alliances (Ithaca, N.Y.: Cornell University Press, 1987)Google Scholar.

22. See, for example, Wolfers, Arnold, Discord and Collaboration (Baltimore: Johns Hopkins University Press, 1962), p. 207Google Scholar; Liska, George, Nations in Alliance: The Limits of Interdependence (Baltimore: Johns Hopkins University Press, 1962), pp. 97–100Google Scholar.

23. See Snyder, Glenn, “The Security Dilemma in Alliance Politics, ” World Politics 36 (07 1984)CrossRefGoogle Scholar. The notion of “entrapment” usually refers to involvement in unwanted conflict or the assumption of what are perceived as unnecessary and excessive defense responsibilities. “Abandonment, ” in its extreme form, refers to realignment and the breaking of defense commitments, but it may also take more moderate forms, such as the alliance leader moving closer to the adversary, imposing sanctions on its weaker allies, or ignoring the interests of small powers in the designation of alliance policy and strategy.

24. See Snyder, , “Alliance Security Dilemma, ” pp. 483–84Google Scholar; and Waltz, Kenneth, Theory of International Politics (Reading, Mass.: Addison-Wesley, 1979), chap. 8Google Scholar.

25. Alternatively, the alliance leader may dispense with bargaining leverage and resort to outright physical coercion and invasion, as the Soviet Union has done to preserve “cohesion” in the Warsaw Pact.

26. The basic argument was first put forth by Kindleberger, Charles, The World in Depression, 1929–1939(Berkeley: University of California Press, 1973)Google Scholar. Robert Keohane was the first to call the general proposition “hegemonic stability theory” and evaluated its premises in “The Theory of Hegemonic Stability and Changes in International Economic Regimes, 1967–1977, ” in Holsti, Ole et al. , eds., Change in the International System (Boulder: Westview, 1980)Google Scholar.

27. The seminal article applying the notion of collective action to alliances is Olson, Mancur and Zeckhauser, Richard, “An Economic Theory of Alliances, ” in Russett, Bruce, ed., Economic Theories of International Politics (Chicago: Markham, 1968)Google Scholar. For empirical assessments and modifications of the original model, see Sandier, Todd et al. , “In Defense of a Collective Goods Theory of Alliances, ” Journal of Conflict Resolution 24 (09 1980)Google Scholar. They elucidate a “joint product model” that considers impure public and private as well as public outputs of defense expenditure. See also Kennedy, Gavin, Burden Sharing in NATO (New York: Holmes & Meier, 1979)Google Scholar; and Brown, Stephen et al. , “Public-Good Theory and Bargaining between Large and Small Countries, ” International Studies Quarterly 20 (09 1976)CrossRefGoogle Scholar.

28. Systematic study of the domestic determinants of alliance behavior has been neglected in scholarly works until quite recently. See Ward, , Research Gaps, pp. 60, 76Google Scholar. For a review of approaches to the topic and presentation of recent public opinion surveys on NATO, see Flynn, Gregory and Rattinger, Hans, eds., The Public and Atlantic Defense (Totawa, N.J.: Rowman & Allanheld, 1985)Google Scholar; and Domke, William, Eichenberg, Richard, and Kelleher, Catherine, “Consensus Lost? Domestic Politics and the ‘Crisis’ in NATO, ” World Politics 39 (Spring 1987)CrossRefGoogle Scholar.

29. These explanations are by no means mutually exclusive. As the following analysis reveals, however, the empirical evidence can be used to rule out all but one of the hypotheses in each case.

30. To avoid circularity, it is important to establish an independent measure of threat—one that is not related to alliance cohesion or increasing defense preparation—when testing the external threat hypothesis. We can achieve this by consulting official statements and documents pertaining to threat assessment and by examining events and their implications for the changing strategic environment.

31. North Atlantic Assembly, Political Committee, “Interim Report of the Sub-Committee on Out-of-Area Security Challenges to the Alliance, ” (Brussels, 10 1983), p. 3Google Scholar.

32. “Threat to Gulf Diminishes, ” Guardian, 17 February 1982.

33. For a general discussion of these differences, see Congressional Research Service, NATO After Afghanistan, Report to the House Foreign Affairs Committee (Washington, D.C.: GPO, 1980)Google Scholar.

34. U.K. House of Commons, Afghanistan: The Soviet Invasion and its Consequences for British Policy, Fifth Report from the Foreign Affairs Committee (London: HMSO, 1980), p. xxxGoogle Scholar.

35. Rummel, Reinhardt, “Public Perspectives in NATO on the ‘Out-of-Area’ Question, ” paper prepared for the Workshop on the Atlantic Alliance, University of Pittsburgh, 13–15 03 1986, p. 15Google Scholar.

36. “Menaces sur les Puits du Golfe, ” Defense Nationale, July 1981, p. 42.

37. See Labrousse, Henri, “Une Stratégie de Dissuasion pour le Golfe?” Défense Nationale, 04 1982, pp. 75–79Google Scholar.

38. See for example, “Menaces sur les Puits du Golfe, ” Défense Nationale, p. 43.

39. The British, for example, favored an independent EEC position because of the anti-American sentiment in the Middle East following the Camp David Accords. See U.K. House of Commons, Afghanistan, p. xxviiGoogle Scholar.

40. Though most European countries maintain both amphibious and air transportable forces, these units are limited in number and could add little to America's force projection capability. See Foot, Peter, “Beyond the North Atlantic: The European Contribution, ” Aberdeen Studies in Defense Economics, no. 21 (Aberdeen, 1982)Google Scholar. In my interviews, the European delegations repeatedly stressed that they resisted the adoption of a formal out-of-area strategy because they doubted America was willing to consult with its allies before acting.

41. A game theoretic representation will help illustrate the dynamics of the interaction. In Figure 1, defection (D), as usual, means to stand firm and cooperation (C) to accommodate. For the United States, a “C” strategy involves moderating its initial position on out-of-area policy. For the Europeans, a cooperative strategy entails a willingness to incorporate—at least to some degree—out-of-area considerations into NATO's planning cycle. The preference ordering for the United States is DC > CC > DD > CD. The United States most prefers to stand behind its original position; the second preference is for mutual accommodation; the third preference is for mutual defection; the least preferred outcome is CD because the United States would feel that it had been taken advantage of. The preference ordering for the Europeans is DC > DD > CC > CD. They do not fear abandonment and therefore associate few costs with Quadrants I and II; their third and fourth preferences are CC (moderate entrapment) and CD (entrapment), respectively. With the Europeans playing “deadlock” and the United States playing “Prisoner's Dilemma, ” there is little hope for a cooperative outcome.

42. Less disincentive to cooperate does not as effectively induce cohesion as more incentive to cooperate. To use an economic analogy, a consumer will not be induced to buy an item that he does not want because it costs less, but only because there is a cost associated with not buying it. Reducing the cost will, however, make the item more attractive once the incentive to buy is evident.

43. U.S. Congress, House Armed Services Committee, “NATO—An Alliance of Shared Values, ” Report of the Delegation to NATO Countries—1982 (Washington, D.C.: GPO, 1982), p. 69Google Scholar.

44. Griffiths, Eldon, cited in U.K. House of Commons, Afghanistan, pp. 189–90Google Scholar.

45. Gary Hart cited in U.S. Congress, Senate Armed Services Committee, Department of Defense Authorization for Appropriations for Fiscal Year 1983, Part 6 (Washington, D.C.: GPO, 1982), p. 3748Google Scholar.

46. See, for example, U.K. House of Commons, Afghanistan, p. 212Google Scholar.

47. This point runs counter to Snyder's assertion that smaller powers enhance their bargaining power by adopting a strategy of defection. In his words, “bargaining over the ally is enhanced to the extent he doubts one's commitment because one can then make credible threats of nonsupport.” (Snyder, , “Alliance Security Dilemma, ” p. 467.)Google Scholar The problem is that nonsupport may also reduce influence simply by alienating the hegemon. The solution is not to pursue consistently a strategy of cooperation; this would indeed reduce influence. A compromise or mixed strategy seems to offer the most potential for long-term leverage.

48. U.K. House of Commons, Afghanistan, p. xxxGoogle Scholar.

49. Howard, Michael cited in Afghanistan, pp. 212–13, 217Google Scholar.

50. Interviews revealed that many delegates believed in early 1980 that the United States would soon drop the issue. When it became clear that this was not the case, the prospect of iterated negotiations on the Persian Gulf illuminated both the future benefits of cooperation and the potential long-term costs of defection. This point corresponds nicely with the assertion that a longer “shadow of the future” facilitates the emergence of cooperative outcomes in certain games. See Axelrod, Robert, The Evolution of Cooperation (New York: Basic Books, 1984), pp. 13–14, 206–7Google Scholar; and Oye, Kenneth, “Explaining Cooperation Under Anarchy: Hypotheses and Strategies, ” World Politics 37 (10 1985)Google Scholar.

51. The “filling out” of the dilemma changed European preference orderings. They were left with the choice of moving from Quadrant I to either II or IV. For the sake of simplicity, I have ranked CC and DD equally, suggesting that they both provide options for avoiding the poles of the dilemma. Yet since the United States had adopted a cooperative strategy, the Europeans were forced either to reciprocate and compromise (CC) or to risk abandonment (DC).

52. Snyder, , “Alliance Security Dilemma, ” p. 485Google Scholar.

53. Edward Heath cited in U.K. House of Commons, Afghanistan, p. 181Google Scholar.

54. This analysis was strongly confirmed by interviews at NATO headquarters. Without exception, European delegations, the U.S. delegation, and officials in the international secretariat stressed that one overriding factor led to European acquiescence and compromise on the out-of-area issue: U.S. pressure and the consequent European fear of the costs of noncompliance.

55. Certain military assets—such as lift capabilities—qualify as “swing” forces; that is, they can be used for operations either in the periphery or in Europe and can be considered as impure public outputs of defense expenditure. Other assets—such as light divisions or pre-positioned materials—would be used only in the Persian Gulf or other non-NATO areas, thus assuming the attributes of quasi-private outputs.

56. The hypothesis does not suggest that the alliance should necessarily experience smooth political relations as long as the United States maintains or increases the level of its contribution to European security. It would, however, predict that a decline in the willingness and ability of the United States to maintain its contribution should lead to increasingpolitical tension and that the Europeans would not sign an agreement that effectively diminishes conventional force levels in Europe.

57. This divergence may point to an important difference in the conditions that facilitate cooperation in economic and security realms. A declining contribution by the dominant power may strengthen security regimes while it weakens economic regimes. For other differences see Lipson, Charles, “International Cooperation in Economic and Security Affairs, ” World Politics 37 (10 1984)CrossRefGoogle Scholar.

58. See Rummel, “Public Perspectives.”

59. See Williams, Phil, “Revitalizing the Western Alliance: Proposals for Change, ” Atlantic Quarterly 2 (Summer 1984)Google Scholar.

60. Apel cited in Congressional Research Service, NATO After Afghanistan, p. 10.

61. Giscard D'Estaing cited in ibid., p. 10.

62. U.K. House of Commons, Afghanistan, p. xxiiiGoogle Scholar.

63. Egon Bahr cited in Joint Publications Research Service (hereafter JPRS) 7632, no. 1616 (28 August 1980), Federal Republic of Germany, p. 29.

64. It is problematic to conceive of defense as a pure public good. Despite the existence of alliance commitments, member states can face aggression and yet not benefit from the alliance's collective defense capability. This qualification does not, however, impinge upon my general argument. For more detailed discussion of the concept of public goods and international cooperation problems, see Snidal, Duncan, “Public Goods, Property Rights and International Organizations, ” International Studies Quarterly 23 (12 1979)CrossRefGoogle Scholar; and Conybeare, John, “Public Goods, Prisoners' Dilemmas and the International Political Economy, ” International Studies Quarterly 28 (03 1984)CrossRefGoogle Scholar.

65. NATO members agreed that the need for military intervention by one or more NATO powers would increase were the war to spread or were either side to attain victory. See North Atlantic Assembly, “Interim Report” (11 1984), p. 12Google Scholar.

66. When current lift procurement programs are completed in 1988, the United States will be able to meet its requirements for Southwest Asia. This assumes, however, that all lift assets are devoted to the Southwest Asia mission. See McNaugher, Thomas, Arms and Oil (Washington, D.C.: Brookings Institution, 1985), p. 68Google Scholar.

67. In the worst case scenario, a full deployment of the RDF could lead to a 25 percent decrease in the total number of active divisions available to fight in Europe. This assumes that there would be roughly four U.S. divisions stationed in Europe fighting, with thirty-two European divisions.

68. “NATO Ministers Present a United Front on Iran and Afghanistan, ” London Times, 15 05 1980Google Scholar. See also NATO DPC Comminiqué (Brussels, May 1980), paragraph 6.

69. See note 19.

70. Olson and Zeckhauser note that alliances may solve burden-sharing disputes in this manner. They argue that “a nation can bear some of its alliance burden by making policy concessions rather than by providing additional forces.” “An Economic Theory of Alliances, ” p. 35, note 12.

71. Egon Bahr quoted in JPRS 7632, no. 1616 (28 August 1980).

72. U.K. House of Commons, “Second Report from the Defense Committee, ” p. 64Google Scholar.

73. U.K. House of Commons, Afghanistan, p. 181Google Scholar. One member of NATO's international secretariat who was responsible for dealing specifically with force planning and out-of-area problems stressed that the Europeans recognized the importance of compensatory steps but simply lacked the budgetary resources. Interview with George Olson, Brussels, 13 March 1984.

74. Heisbourg, François, “Conventional Defense: Europe's Constraints and Opportunities, ” in Pierre, Andrew, ed., The Conventional Defense of Europe: New Technologies and New Strategies (New York: Council on Foreign Relations, 1986), p. 72Google Scholar.

75. Domke, et al. , “Consensus Lost?” p. 394Google Scholar.

76. Heisbourg, , “Conventional Defense, ” p. 74Google Scholar.