Power, Trade, and War

Rating:
Summary: Power, Trade, and War by Edward Mansfield
Review: Edward Mansfield argues in this book that there is a curvilinear relationship between the concentration of power and war at the level of the international system, based on data for 5-year periods between 1825 and 1964. That is, war is most likely at intermediate levels of power concentration, least likely at either high or low levels. He also argues that war is more (less) likely when there are relatively low (high) levels of international trade flows. The two variables -- power concentration and trade -- explain a high percentage of variance in the war data. Thus, Mansfield concludes that a combination of neo-realist and political economy perspectives is needed to explain war at the level of the international system.

The path to these conclusions is lengthy and difficult, given the disparate approaches to measuring war, but Mansfield does an excellent job of explaining his steps along this tortuous route. An entire chapter (Chapter 2) is devoted to describing and comparing war data collections by previous scholars. Five different definitions of war and nine data sets derived from them are discussed in this chapter. It is interesting and somewhat disheartening that the correlations between system war measures in the different data sets is low. According to the author: "Given the low correlation between these data sets, analysts should be hesitant to use them interchangeably. This is not to imply that any of [them] is "wrong" or misleading. Each is useful contingent on the objectives of the particular analysis." [p. 43]

One of the major subsidiary arguments of the book is that scholars have paid too much attention to "polarity" -- the number of poles (sometimes equated with the number of great powers) in the system. Mansfield agrees with these authors that the distribution of power is a key potential causal variable but disagrees with their contention that polarity is the best way to measure that distribution. He argues instead for using a measure of concentration which takes into account both the number of great powers and the relative distribution of power across them. This argument is made quite persuasively.

Mansfield also suggests that other scholars have erred in testing only monotonic relationships between the distribution of power and war, demonstrating that a curvilinear model explains more variance. Again, I found this demonstration convincing.

Finally, Mansfield shows that multivariate models which combine economic variables (trade levels in particular) with political ones (the concentration of power) explain a higher percentage of variance in systemic war levels than models that do not. This suggests to him that "interdisciplinary research between political scientists and economists needs to be conducted, and is likely to foster a fuller understanding of the relationships among power, trade, and war." [p. 253] Again, the argument was quite convincing.

There is only a short discussion in the book of its implications for current policy. Mansfield implies that the breakup of the Soviet empire "bodes poorly for the avoidance of war in Europe," but that "continued expansion of international trade offers an avenue for improving political relations while, at the same time, increasing global welfare." [pp. 252-3]

What is missing here, unfortunately, is a careful discussion of how far one can generalize or extrapolate from the type of systemic data used in the various data analyses. For example, Mansfield mentions briefly that there are reasons to believe that the introduction of nuclear weapons may have changed the relationship between the distribution of power and war, but does not go on to explain why he fails to take the argument seriously.

Still, Mansfield should be praised for the care and skill he demonstrates in dealing with a wide variety of theories, data sets, and statistical methods. The prose is a bit tortuous, and therefore not suited to use in introductory courses, but as an example for graduate students about to undertake their own empirical quests, it would be hard to find a better exemplar.