I thank the participants in the First Annual Conference on EU–U.S. Relations, European Union Center, Georgia Tech, March 1999, where I presented an earlier version of this article. I also thank the participants in the Rational Design project, the editors of IO, and two anonymous reviewers for their feedback. I especially thank Stephan De Spiegeleire, Frank Schimmelfennig, Charles Glaser, Barbara Koremenos, Dan Lindley, Charles Lipson, James Morrow, Duncan Snidal, Robert Pahre, David Pervin, and Peter Rosendorff for their comments and suggestions.
2. For the origin of the security community concept, see Deutsch et al. 1957.
4. For the origin of this literature, see Olson and Zeckhauser 1966; and for a survey, see Sandler 1993.
5. See Morrow 1994a; and Smith 1995.
6. See below for the relationship between trust and uncertainty about preferences.
7. For detailed accounts of the process leading up to enlargement, see Eyal 1997; and Goldgeier 1998.
10. See Mearsheimer 1990, 5; and Walt 1997, 171.
11. On the issue of monetary costs, for the optimistic side, see Asmus, Kugler, and Larrabee 1996; and for the pessimists, see Perlmutter and Carpenter 1998; and Rubinstein 1998.
13. Schimmelfennig 1998/99. See also his analysis of EU expansion in Schimmelfennig 2001.
14. Schimmelfennig 1998/99, 211.
17. For the beneficial effects of NATO enlargement on Hungarian democracy and Hungarian-Romanian relations, see Kramer 1999, 429–30.
18. See Brown 1996; and Chan 1997.
20. Schimmelfennig acknowledges this point. Schimmelfennig 1998/99, 230.
21. For an interesting argument that NATO enlargement has not actually accomplished these goals, in particular, has not fostered democracy, see Reiter 2001. Reiter argues that the countries admitted were solid democracies with civilian control of the military before NATO enlargement became a possibility, and hence that NATO enlargement was irrelevant in promoting cooperation in Eastern Europe. Even if one agrees with this point, which I do not fully, my analysis still can explain both the enlargement criteria and the enlargement dilemma, which are a function of policymakers' perception that NATO enlargement would promote democratization and trust building while harming NATO–Russian relations.
22. Deutsch et al. 1957, 5. For a constructivist take on security communities, see Adler and Barnett 1998.
26. See Pierre and Trenin 1997; Asmus and Larrabee 1996; Brown 1995; and Mandelbaum 1995.
29. See Jervis 1976, 62; and Kydd 1997.
30. For a contrary argument that offensive alliances are smaller than defensive ones, see Schweller 1998, 61.
31. See Coleman 1990, 91; Güth and Kliemt 1994; Watson 1999; and Kydd 2000a.
32. See Jervis 1976 and 1978; and Glaser 1994/95 and 1997.
33. Schweller 1998, 15–38.
34. On preventive war, see Copeland 2000, 11–34; on preemptive war, see Van Evera 1999, 35–72.
36. This raises a commitment problem. Given that NATO pays a cost (discussed later) to extend a security guarantee, it might be best for them to promise a security guarantee, and then renege on the promise after the allies have moved. I will assume that NATO faces reputational costs sufficient to render such a deceitful strategy unappealing.
37. In reality, of course, there is a much larger set of possible offers. Some states could be given guarantees even if they do not cooperate; others could be denied a guarantee even if they do. The three-part choice is the simplest framework in which we can examine how expansion could be threatening or reassuring, depending on whether it is conditional or unconditional.
38. For quasi-game-theoretic analyses of trust along these lines, see Bennet and Dando 1982; and Plous 1988. Glaser also suggests this strategy for modeling the security dilemma. Glaser 1997.
39. I use the name spiral because this equilibrium is sometimes provocative. Perhaps “conditional spiral” would be a more accurate, if more cumbersome, name.
40. The first and second are not possible for the parameter values illustrated in Figure 2. See the appendix for details.
41. For the debate on NATO and the Baltic states, see Asmus and Nurick 1996; Kamp 1998; and Blank 1998.