DANIEL W.DREZNER 711 pressure.Ceteris paribus,senders will be eager to coerce adversaries,and reluctant to coerce allies. Eagerness does not translate into greater concessions.The second effect of conflict expectations is paradoxical and surprising.While a robust anticipation of future disputes might make the sender prefer a coercive strategy,it also reduces its ability to obtain concessions.The target's conflict expectations determine the magnitude of concessions.Facing an adversarial sender,the target will be worried about the long-run implications ofacquiescing.Because it expects frequent conflicts, the target will be concerned about concessions in the present undercutting its bargaining position in future interactions.The sender might exploit the material or reputational effects from conceding in later conflicts.When relative gains concern is prominent,a concession represents a gain for the coercer and a loss for the coerced.When reputation is important,acquiescence bolsters the sender's credibil- ity as a tough negotiator while weakening the target's reputation.With allies,this concern is not prominent,because the target anticipates fewer zero-sum conflicts. Ceteris paribus,targets will concede more to allies than adversaries.Ironically,a sender will obtain the most favorable distribution of payoffs when it cares the least about relative gains. A conflict expectations model can explain the dynamics of economic coercion with more accuracy and parsimony than other theories.A major problem with the existing literature is that theories are developed from just one or two well-publicized cases of economic sanctions (Galtung,1967;Hoffman,1967;Schreiber,1973; Losman,1979;Renwick,1981;Doxey,1987).The very celebrity of these cases suggests they are atypical,that they stand out in some unusual way.In "classic"cases, the sender and target are usually adversaries.Thus,existing explanations overlook less contentious but more successful coercion attempts between allies.By choosing cases that take on extreme values of both the dependent variable and several independent variables,the literature commits two errors.First,there is a tendency to underestimate the main causal effects on the universe of events.Second,these studies will overestimate effects that are unique to the extreme set of cases (Collier and Mahoney,1996).In focusing on a limited subset of coercion cases,these writings have painted a distorted picture of economic sanctions.These arguments are not necessarily wrong,but the empirical evidence marshaled for their arguments is insufficient.In contrast to a conflict expectations model,they explain fewer cases, and less of the variation in outcomes. The following section further develops the notion of conflict expectations.The next section uses a simple coercion game to formalize the argument;the model generates testable propositions about the relationship between opportunity costs, alignment,and coercion events.The following three sections (a)consider what happens when conflict expectations change during the sanctions episode,(b) describe the data and testing procedures,and(c)examine the statistical results on sanctions events data and show they provide moderately strong support for the model.The final section summarizes and concludes. The Effect of Conflict Expectations For the purposes of this model,the actors are the foreign policy leaders of nation-states.It is assumed that the foreign policy regimes within these states act as a unitary,rational actor.This assumption has been made repeatedly in international relations theory (Krasner,1978;Bueno de Mesquita,1981),and the justification for it will not be delved into here. Conflict expectations influence state preferences about conceding to coercive pressure in a conflict in two ways.First,states are concerned that concessions made in the present can be used later to threaten their security.This possibility leads topressure. Ceteris paribus, senders will be eager to coerce adversaries, and reluctant to coerce allies. Eagerness does not translate into greater concessions. The second effect of conflict expectations is paradoxical and surprising. While a robust anticipation of future disputes might make the sender prefer a coercive strategy, it also reduces its ability to obtain concessions. The target's conflict expectations determine the magnitude of concessions. Facing an adversarial sender, the target will be worried about the long-run implications of acquiescing. Because it expects frequent conflicts, the target will be concerned about concessions in the present undercutting its bargaining position in future interactions. The sender might exploit the material or reputational effects from conceding in later conflicts. When relative gains concern is prominent, a concession represents a gain for the coercer and a loss for the coerced. When re~utation is im~ortant. acauiescence bolsters the sender's credibil- I 1 ' I ity as a tough negotiator while weakening the target's reputation. With allies, this concern is not prominent, because the target anticipates fewer zero-sum conflicts. Ceteris paribus, targets will concede more to allies than adversaries. Ironically, a sender will obtain the most favorable distribution of payoffs when it cares the least about relative gains. A conflict ex~ectations model can ex~lain the dvnamics of economic coercion I 1 i with more accuracy and parsimony than other theories. A major problem with the existing literature is that theories are developed from just one or two well-publicized cases of economic sanctions (Galtung, 1967; Hoffman, 1967; Schreiber, 1973; Losman, 1979; Renwick, 1981; Doxey, 1987). The very celebrity of these cases suggests they are atypical, that they stand out in some unusual way. In "classic" cases, the sender and target are usually adversaries. Thus, existing explanations overlook less contentious but more successful coercion attempts between allies. By choosing cases that take on extreme values of both the dependent variable and several indeeendent variables. the literature commits two errors. First, there is a tendencv to underestimate the main causal effects on the universe of events. Second, these studies will overestimate effects that are unique to the extreme set of cases (Collier and Mahoney, 1996). In focusing on a limited subset of coercion cases, these writings have painted a distorted picture of economic sanctions. These arguments are not necessarily wrong, but the empirical evidence marshaled for their zrguments is insufficient. In contrast to a conflict exeectations model, thev explain fewer cases, ' , A and less of the variation in outcomes. The following section further develops the notion of conflict expectations. The next section uses a simple coercion game to formalize the argument; the model generates testable propositions about the relationship between opportunity costs, alignment, and coercion events. The following three sections (a) consider what happens when conflict expectations change during the sanctions episode, (b) describe the data and testing procedures, and (c) examine the statistical results on sanctions events data and show they provide moderately strong support for the model. The final section summarizes and concludes. The Effect of Conflict Expectations For the purposes of this model, the actors are the foreign policy leaders of nation-states. It is assumed that the foreign policy regimes within these states act as a unitary, rational actor. This assumption has been made repeatedly in international relations theory (Krasner, 1978; Bueno de Mesquita, 1981), and the justification for it will not be delved into here. Conflict expectations influence state preferences about conceding to coercive eressure in a conflict in two wavs. First. states are concerned that concessions made in the present can be used later to threaten their security. This possibility leads to