Earlier versions of this article were presented at the University of Chicago, PIPES, February 1998, and at the Rational Design Conference, May 1998. I thank David Lake, Peter Gourevitch, Barbara Walter, Barbara Koremenos, Duncan Snidal, Kenneth Abbott, Beth Yarbrough, Charles Lipson, and Alexander Thompson for helpful comments. I thank Peter Cowhey for many useful conversations on related topics.

1. I examine scheduled aviation markets, not charter operations. Charter carriers largely did military contract work in the immediate aftermath of World War II and were therefore not considered in postwar institution building. Only later, in the 1960s, did charter operations become a significant part of international aviation markets.

2. Although the IATA was founded in 1919 as a trade association for European airlines, this early version of the organization is distinct from the post–World War II organization.

3. The bilateral agreements also retained authority over some aspects of service in international markets, such as the use of national ground-handling and catering services; the IATA was given authority over the most prominent features of in-flight services, including entertainment and food service.

4. For the politics behind U.S. efforts to dismantle the Bermuda institutions, see Kasper 1988; and Richards 1999.

5. Koremenos, Lipson, and Snidal this volume, 762.

6. One way to view the different institutional instruments available to states—scope, membership, and voting rules—is to consider how governments use fiscal and monetary policy as interchangeable tools to grow their economies (and win reelection); that is, different instruments are substitutes for each other depending on actors' goals and their ability to secure different institutional outcomes in pursuit of these goals. I return to this issue in the conclusion.

7. See Baxter 1983; Carlton and Klamer 1983; and Salop 1991.

8. For an overview of electronic data networks, see Guerin-Calvert and Wildman 1991. On Internet interconnection arrangements and the problems posed by closed networks, see Bailey 1997; and Srinagesh 1997.

9. International aviation markets were seen as an important piece of postwar reconstruction. The head of the U.S. delegation at Chicago, A. A. Berle, asserted that “aviation will have a greater influence on American foreign interests and American foreign policy than any other non-political consideration.” Berle and Jacobs 1973, 481. For a similar conclusion, see van Zandt 1944.

10. For detailed discussions of the Chicago Convention and U.S.-U.K. bargaining at Bermuda, see Dierikx 1992; and Haanappel 1980.

11. See Jönsson 1987, 98; and Sochor 1991, 3–16.

12. Berle and Jacobs 1973, 498–509.

13. Dempsey 1987, 10. See also Milner 1997, 158–78.

14. See Milner 1997, 158–78; and Sampson 1984, 77–82.

15. See Dierikx 1992; and Kasper 1988, 47–50.

16. Although all bilateral agreements were roughly similar, there were important differences among them in regard to the amount of competition and potential for government intervention in the marketplace. The ability of carriers to pool revenues was one area of difference; others included whether or not capacity was predetermined, the minimum/maximum traffic one bilateral partner could carry, the circumstances under which governments could intervene to set fares, and the potential for new entry into any given city-pair. Because of these differences, they were not simply a set of bilateral agreements that could be easily multilateralized.

17. The IATA did not “capture” fare-setting authority. Rather, the Bermuda bilateral agreements contained explicit language granting the IATA authority to set fares. In the United States, IATA authority depended on the Justice Department's granting anti-trust immunity, which was renewed annually until 1955, when it was granted indefinitely. For a discussion of airline motivations regarding the IATA, see Sochor 1991, 13.

18. The United States was the exception here since anti-trust laws precluded U.S. carriers from concluding side-agreements with their foreign counterparts. The Civil Aeronautics Board (CAB) did, however, occasionally grant permission for U.S. carriers to collude with foreign airlines.

19. On IATA traffic conferences, see Lowenfeld 1981, 455–76. I draw heavily on this source for the discussion of IATA rules, except where otherwise noted.

20. On the secrecy surrounding the IATA and the fare conferences, see Haanappel 1978, 35–37, 57–63.

21. All major international airlines were members of the IATA until the late 1960s, when many Asian carriers, including Thai Airways International, Malaysian Airline System, Cathay Pacific Airways, and Singapore Airlines, chose not to join the IATA. The role of these non-IATA carriers in the breakdown of the Bermuda regime suggests that universal membership was important for the success of the IATA in curbing incentives to break the cartel. See Dresner and Tretheway 1988, 8–9.

22. For the original distinction between fire alarms and police patrols as methods for political principles to monitor their agents, see McCubbins and Schwartz 1984. See also Lupia and McCubbins 1994.

23. Remember that the IATA fares conferences had to be approved by national governments (and that most airlines were state owned), so in effect states maintained veto power over fares as well as over entry and capacity.

24. There were clear rules for ending bilateral agreements, including time horizons and default rules once the agreements expired, which meant that states had a de facto veto over air services arrangements.

25. In the early 1990s, U.S. pressure for more competitive bilateral agreements led some states to renounce their Bermuda I agreements as a way to stifle U.S. initiatives. The U.S.-France market was perhaps the largest and most well known of those renounced. Of course, service still continued in the bilateral marketplace, but it was governed only by a set of regulations defined in the terms of the initial bilateral agreement (rather than through ad hoc bilateral bargaining).

26. See Spruyt 1994; North and Weingast 1989; Greif, Milgrom, and Weingast 1994; and Milgrom, North, and Weingast 1990.

27. For a discussion of the emergence of the state, see Spruyt 1994.

28. On the ability to regulate the Internet and efforts of both public and private actors to establish property rights, see Lessig 1999.

29. Akerlof 1970.

30. The principle of state sovereignty over airspace was part of the 1919 Paris Peace Conference, where states saw the military potential rather than the commercial potential of international aviation. Post–World War II institution builders drew on this prior set of agreements regarding state sovereignty over airspace.

31. See Oatley, this volume.

32. This was particularly true in the postwar years, when most travel was for business or government purposes, rather than for leisure, which meant demand for final destinations was relatively inelastic. In addition, capacity was restricted, so there was usually excess demand in any given bilateral market.

33. Keohane 1984.

34. Air France: $3.3bn State Aid Approved; and Europe: BA Angered at Air France Outcome, both in The Financial Times, 23 July 1998.

35. This is not to suggest that monitoring and sanctioning were irrelevant but that they were not as effective in aviation as in many other markets. Indeed, as discussed later the IATA compliance department monitored and sanctioned airlines. But the total failure of the IATA compliance machinery to slow rampant cheating once airlines had large incentives to cheat (in the early 1970s) suggests that making cheating less profitable was more important than any monitoring and sanctioning IATA did.

36. Although contractual rules could have been written for all these contingencies, this would have proven too costly or outright impossible in practice.

37. National carriers traditionally controlled travel agents through incentive schemes that provided large bonuses if travel agents sold a large number of tickets on a particular airline. Of course, this discouraged travel agents from booking tickets on airlines that had a smaller share of the market, which by definition meant foreign carriers.

38. For a discussion of incentive problems facing cartels, see McMillan 1991.

39. This is one reason most telecommunications alliances fail: monitoring traffic routing is difficult, and there is often no simple rule for allocating costs and profits. Aviation alliances are easier to monitor, so they have been more successful.

40. Revenue pooling was easiest with a limited number of city-pairs and large traffic flows (that is, passengers) to justify the time and expense of accounting and allocation. Since scheduled airline traffic was largely composed of large flows between European capitals, it therefore made sense that revenue pooling dominated European markets. Note that charter traffic, not discussed here, was governed by a different set of regulatory rules.

41. The telecommunications regime also used bilateral rent sharing within a multilateral arrangement. See Cowhey 1990. Recent efforts by the U.S. Federal Communications Commission to erode the traditional cartel in telecommunications have focused on ending bilateral revenue sharing, one of the benefits to major carriers of maintaining the cartel. In short, bilateral revenue and service pooling created incentives for large carriers not to cheat on existing arrangements. See Cowhey and Richards 2000.

42. Dempsey 1978, 393–449.

43. International carriers not members of IATA include Thai Airways International, Malaysian Airline System, Cathay Pacific Airways, and Singapore Airlines. Dresner and Tretheway 1988, 8–9.

44. Because selling tickets to consolidators at discount prices was illegal, these revenues were not included in revenue-sharing arrangements.

45. See note 25.

46. See Rosendorff and Milner, this volume.

47. The CAB could not directly set fares in international markets until 1972, when it was given greater authority over the fares of carriers on international routes.

48. See IATA Faces Increasing Pressures, Aviation Week and Space Technology, 7 November 1977, 27–28Google Scholar.

49. For a good discussion of these coordination functions, see Zacher 1996, chap. 4.

50. For a somewhat sensationalized version of how IATA inspectors worked, see Sampson 1984, 17–18.

51. Fines were due within sixty days, and the sizes of the fines were related to the breach and the disruptive effects on the traffic of interested parties.

52. Lowenfeld 1981, 471–72.

53. Taneja 1980, 91.

54. The potential for differences in entry and capacity in a single bilateral agreement to upset other bilateral agreements was limited for two key reasons. First, capacity was restricted, so demand almost always outstripped capacity, which meant in practice that there was excess demand for services in any given agreement. Second, most traffic (business and government travelers) in the early postwar years was very sensitive to time and distance, so alternative routes, regardless of the price, were unacceptable. Therefore, an increase in the number of seats in U.S.-U.K. markets, for example, had little impact on demand for seats in U.S.-French markets (as long as price was held constant across the markets, as it was by the IATA).

55. As quoted in Dresner and Tretheway 1988, 6–7.

56. For an excellent discussion of the standardization functions performed by the Bermuda institutions, see Zacher 1996, 94–109.

57. A multilateral apparatus governing fares had large advantages because inconsistent bilateral agreements on fares could have led to perverse route structures and inefficient growth of international markets.

58. Moreover, it could be argued flexibility substituted for centralization in response to uncertainty about the state of the world.

59. Cowhey and Richards 2000.