752 Power or Plenty? following year from.108 to.109.For international crises,the probability of inter- vention would increase somewhat more,from .073 to.100.These substantively unimportant effects do not tell the full story,however.Trading relationships per- sist over time,and have cumulative effects on the probability of alliance forma- tion over many years.The difference between the .0001 and .0049 probabilities of alliance formation associated with relatively small and relatively large trading relationships is unimportant in the short run.However,if these annual probabili- ties persisted over the course of a 50 year trading relationship,the cumulative probability of alliance formation in the smaller trading relationship would be only about .005,while that associated with the larger would be about.22. The large effects of trade on the cumulative probability of alliance formation over time produce substantively meaningful indirect effects on the probability of intervention,but only in the long run.Strictly speaking,the cumulative change in the probability of intervention depends not just on alliance formation,but also on the probability that an alliance will be formed and then dissolve.Because the probability of such an event is so small,the cumulative probability of inter- vention depends almost entirely on the large but declining probability that no alliance has been formed.Thus,an approximation of the cumulative probability of intervention over time can be computed as follows: pr(intervention)=pr(intervention alliance)*[1-pr(~alliance) pr(intervention~alliance)*pr(~alliance) Figure 2 shows the probability of civil war intervention after 10,25,and 50 years of trade at the indicated level.As the graph indicates,the size of the long-run,indirect effect of trade on civil war intervention depends on how long the trading relationship has been sustained.If the relationship has been sus- tained for 10 years,exports one standard deviation above the mean would increase the probability of intervention,via alliances,only from.11 to.14.If the relationship were sustained for 25 years,the probability would increase from.11 to .18.After 50 years,it would increase from .11 to .29.The total effect of trade on crisis intervention is larger but more complicated,since it depends on the number of states involved in the crisis,and the probability of alliance formation with each one.The bottom line is that in some cases,most of the direct effect of alliances-the largest direct effect estimated earlier-is an indirect result of trade. How have the long-run and short-run influence of economic and security interests on intervention decisions played out in practice?In order to answer this question and summarize the evidence on the role of economic and security interests,it helps to return to the comparison between the civil wars in El Salva- dor and Morocco used earlier to illustrate the effect of alliances on the probabil- ity of intervention.At the end of that example,I noted that the large effect of alliances made sense if it was considered an indicator of a durable judgment about the national interest.The quantitative results concerning the influence of trade on alliance formation suggest that economic interests play an important role in shaping these assessments of national interest.In evaluating the quantita- tive results concerning trade and alliances in this example,the crucial question is whether trade played a role in the development of the U.S.-dominated alliance system in the Americas. 5 Part of this effect is due to the small and marginally insignificant direct effect of trade.The probability assumes that the international crisis involves only one state,and the trade values involved are those for individual states rather than for interational crises,which usually involve more than one.In cases where there was more than one "crisis actor,"the probability that one or more of these states would have an alliance with the United States would be correspondingly higher.The probability reported here thus represents a lower bound on the indirect effect of trade on the probability of American crisis intervention.following year from .108 to .109. For international crises, the probability of inter￾vention would increase somewhat more, from .073 to .100.5 These substantively unimportant effects do not tell the full story, however. Trading relationships per￾sist over time, and have cumulative effects on the probability of alliance forma￾tion over many years. The difference between the .0001 and .0049 probabilities of alliance formation associated with relatively small and relatively large trading relationships is unimportant in the short run. However, if these annual probabili￾ties persisted over the course of a 50 year trading relationship, the cumulative probability of alliance formation in the smaller trading relationship would be only about .005, while that associated with the larger would be about .22. The large effects of trade on the cumulative probability of alliance formation over time produce substantively meaningful indirect effects on the probability of intervention, but only in the long run. Strictly speaking, the cumulative change in the probability of intervention depends not just on alliance formation, but also on the probability that an alliance will be formed and then dissolve. Because the probability of such an event is so small, the cumulative probability of inter￾vention depends almost entirely on the large but declining probability that no alliance has been formed. Thus, an approximation of the cumulative probability of intervention over time can be computed as follows: pr(intervention) ¼ pr(interventionjalliance) ½1  prð allianceÞ t  þ pr(interventionj allianceÞ prð allianceÞ t Figure 2 shows the probability of civil war intervention after 10, 25, and 50 years of trade at the indicated level. As the graph indicates, the size of the long-run, indirect effect of trade on civil war intervention depends on how long the trading relationship has been sustained. If the relationship has been sus￾tained for 10 years, exports one standard deviation above the mean would increase the probability of intervention, via alliances, only from .11 to .14. If the relationship were sustained for 25 years, the probability would increase from .11 to .18. After 50 years, it would increase from .11 to .29. The total effect of trade on crisis intervention is larger but more complicated, since it depends on the number of states involved in the crisis, and the probability of alliance formation with each one. The bottom line is that in some cases, most of the direct effect of alliances—the largest direct effect estimated earlier—is an indirect result of trade. How have the long-run and short-run influence of economic and security interests on intervention decisions played out in practice? In order to answer this question and summarize the evidence on the role of economic and security interests, it helps to return to the comparison between the civil wars in El Salva￾dor and Morocco used earlier to illustrate the effect of alliances on the probabil￾ity of intervention. At the end of that example, I noted that the large effect of alliances made sense if it was considered an indicator of a durable judgment about the national interest. The quantitative results concerning the influence of trade on alliance formation suggest that economic interests play an important role in shaping these assessments of national interest. In evaluating the quantita￾tive results concerning trade and alliances in this example, the crucial question is whether trade played a role in the development of the U.S.-dominated alliance system in the Americas. 5 Part of this effect is due to the small and marginally insignificant direct effect of trade. The probability assumes that the international crisis involves only one state, and the trade values involved are those for individual states rather than for international crises, which usually involve more than one. In cases where there was more than one ‘‘crisis actor,’’ the probability that one or more of these states would have an alliance with the United States would be correspondingly higher. The probability reported here thus represents a lower bound on the indirect effect of trade on the probability of American crisis intervention. 752 Power or Plenty?