The Insurance Factor? An Assessment of War Costs

Abstract

The arms racing literature includes studies of definitional questions, identification of racers, mathematical modelling, and assessments of the risks or benefits of ongoing races. The nature of the relationship between arms racing and the outbreak of interstate war has also received some attention. Relatively little theoretical or empirical work has been directed to the questions of whether wars following arms racing differ systematically from other conflicts. Contrary to some recent assessments, arms race-related wars are argued here to be on the average of greater duration, magnitude, and severity than are other wars. In the Correlates of War data, arms race-connected wars as a separate class of events exact significantly higher casualties and tend to be somewhat longer in both calendar-and and nation-months than other wars. Significant differences exist in the entire sample, and in the sample of wars remaining after the outliers have been removed, which argues that the effects uncovered are not statistical accidents and are not artifacts of unusually lengthy or extensive wars.