A Test for Reverse Causality in the Democratic Peace Relationship*

Several studies have suggested the possibility of reverse causation in the ‘democratic peace’ relationship: that the well-known extreme rarity of wars between democratic nations may be partially or wholly explained by a negative impact of war on democracy. Three kinds of war-on-regime effects are discussed. Anterior effects are regime changes that occur in preparation for wars; concurrent effects are those that occur during the course of a war; and posterior effects are regime changes that occur after a war concludes. Because studies have shown that democratic nations are rarely, if ever, on opposite sides in wars at their start, it is argued that reversed causation may affect the presence of causation from democracy to peace only if nations tend to become more autocratic as they prepare for impending wars. This proposition is examined with the observation of war events involving geographic neighbors or major powers, worldwide, from 1816 to 1992. With interrupted time-series analysis, it is found that nations are about as likely to become more institutionally autocratic as they are to become more democratic in the periods before the onset of wars. Moreover, this pattern holds even for the smaller subset of nations estimated to be democratic in the periods before major wars. These results indicate that studies of regime type and war participation have not been underspecified due to possible reverse causation before the onset of wars, and thus support the notion that the direction of causation in the democracy and war relationship is unidirectional from democracy to peace.