Conflict as an Identification Strategy

The consequences of conflict (international and domestic) for political and economic outcomes is an important concern for scholars and policymakers. But establishing causal relations with the data typically available is an effort that is plagued by endogeneity concerns, thus restricting the scope of empirical conflict studies. A prominent feature of most theories of conflict is uncertainty, which implies an inherent randomness that is not predictable ex ante to combatants or researchers. Consequently, there may be random variation which can provide a source of exogenous variation for some political and economic outcomes related to conflict. We identify two distinct sources of risk that produce qualitatively different kinds of natural experiments. First, incidence risk is the uncertainty regarding whether a crisis escalates into violence, and arises from private information between potential combatants. Second, result risk arises from imperfect information about the ultimate outcome of conflict, which implies that the initiation of conflict essentially triggers a lottery. We discuss the kinds of natural experiments that emerge from these sources of uncertainty and ways they can (and cannot) be used.