BENJAMIN O.FORDHAM 745 ally is an actor should be more likely to precipitate American action.In the ICB data,crisis actors are states that perceive a heightened threat of war.Interveners who do not experience such a threat are not coded as actors. Second,like ensuring the security of allies,countering the actions of rivals is an important security interest.Crises or civil wars in which American rivals have intervened should be more likely to draw American intervention as well.In order to ensure that rival intervention is influencing American actions rather than the other way around,only rival interventions that occur prior to any U.S.interven- tion are included.The strategic rivals identified by Thompson (2001)are used here.These include states that are"sufficiently threatening competitors to qual- ify as enemies''(Thompson 2001,557).In the ICB data set,intervention data were available only on the Soviet Union,China,Germany,and Japan.Cuba, which Thompson identifies as a U.S.rival from 1959 through the present,was therefore not included in the analysis of crisis intervention. Operationalizing economic interests required several additional judgments. Limitations on available data led to the selection of exports to represent the value of economic interests threatened by crises and civil wars.Economic interests in a civil war or international crisis stem from expectations about the loss of future income from a disrupted commercial relationship.Foreign direct investment is arguably more important in this respect than trade, because it reflects assets actually owned in the affected area rather than sim- ply a flow that could potentially be diverted to safer markets.Another prob- lem suggested by Copeland (1996)is that the current level of trade may not always reflect expectations about future trade.Given the difficulty of finding data on trade expectations and foreign direct investment for the time period considered here,exports are the best available proxy.Economists have found that exports and foreign direct investments are complementary.Lipsey and Weiss (1981,1984),examined a sample of nearly 1,300 U.S.firms in 1970, finding a positive relationship between exports and the output of local subsidiaries of those firms in each of 14 industries.Subsequent research sug- gests that a combination of exports and local production through foreign affiliates makes sense for multinational corporations (e.g.,Helpman 1985;Rob and Vettas 2003).Because exporters at one stage of the product cycle will often become foreign investors in the same markets (Vernon 1966),exports also reveal something about expectations for future investment income. The trade data used here are from the Department of Commerce statistics published in annual volumes of the Statistical Abstract of the United States from 1878 through 1949.Those from the post-1949 period are from the International Monetary Fund.In the case of some newly independent states,the International Monetary Fund coded trade as zero for a few years after independence when the Department of Commerce reported only regional aggregates.Because exports were not zero in these cases,these figures were estimated based on each coun- try's share of regional trade in later years.Because conflict tends to depress trade,figures from the year before the onset of the civil war or crisis are used here.Because additional trade should arguably have a declining marginal effect on policy makers'perceptions of the economic stakes,export values,expressed in millions of constant 2,000 dollars,were logged. Empirical Results This section first considers direct effects on the probability of intervention in crises and civil wars,then the relationship between trade and alliances.Finally,the analy- sis considers the total effect of the major independent variables on intervention.ally is an actor should be more likely to precipitate American action. In the ICB data, crisis actors are states that perceive a heightened threat of war. Interveners who do not experience such a threat are not coded as actors. Second, like ensuring the security of allies, countering the actions of rivals is an important security interest. Crises or civil wars in which American rivals have intervened should be more likely to draw American intervention as well. In order to ensure that rival intervention is influencing American actions rather than the other way around, only rival interventions that occur prior to any U.S. interven￾tion are included. The strategic rivals identified by Thompson (2001) are used here. These include states that are ‘‘sufficiently threatening competitors to qual￾ify as enemies’’ (Thompson 2001, 557). In the ICB data set, intervention data were available only on the Soviet Union, China, Germany, and Japan. Cuba, which Thompson identifies as a U.S. rival from 1959 through the present, was therefore not included in the analysis of crisis intervention. Operationalizing economic interests required several additional judgments. Limitations on available data led to the selection of exports to represent the value of economic interests threatened by crises and civil wars. Economic interests in a civil war or international crisis stem from expectations about the loss of future income from a disrupted commercial relationship. Foreign direct investment is arguably more important in this respect than trade, because it reflects assets actually owned in the affected area rather than sim￾ply a flow that could potentially be diverted to safer markets. Another prob￾lem suggested by Copeland (1996) is that the current level of trade may not always reflect expectations about future trade. Given the difficulty of finding data on trade expectations and foreign direct investment for the time period considered here, exports are the best available proxy. Economists have found that exports and foreign direct investments are complementary. Lipsey and Weiss (1981, 1984), examined a sample of nearly 1,300 U.S. firms in 1970, finding a positive relationship between exports and the output of local subsidiaries of those firms in each of 14 industries. Subsequent research sug￾gests that a combination of exports and local production through foreign affiliates makes sense for multinational corporations (e.g., Helpman 1985; Rob and Vettas 2003). Because exporters at one stage of the product cycle will often become foreign investors in the same markets (Vernon 1966), exports also reveal something about expectations for future investment income. The trade data used here are from the Department of Commerce statistics published in annual volumes of the Statistical Abstract of the United States from 1878 through 1949. Those from the post-1949 period are from the International Monetary Fund. In the case of some newly independent states, the International Monetary Fund coded trade as zero for a few years after independence when the Department of Commerce reported only regional aggregates. Because exports were not zero in these cases, these figures were estimated based on each coun￾try’s share of regional trade in later years. Because conflict tends to depress trade, figures from the year before the onset of the civil war or crisis are used here. Because additional trade should arguably have a declining marginal effect on policy makers’ perceptions of the economic stakes, export values, expressed in millions of constant 2,000 dollars, were logged. Empirical Results This section first considers direct effects on the probability of intervention in crises and civil wars, then the relationship between trade and alliances. Finally, the analy￾sis considers the total effect of the major independent variables on intervention. Benjamin O. Fordham 745