The Investigation of Substitutability in Foreign Policy

More than a decade ago, Benjamin Most and Harvey Starr argued that more attention needed to be paid to policy and to specific (Most and Starr 1984, 1989). Substitutability in foreign policy is idea that factors could plausibly be expected to trigger different foreign policy acts, while domain specific laws are the logical opposite ... and suggests that different processes would plausibly be expected to lead to similar results (Most and Starr 1989, 98). Together, these ideas imply a fundamental challenge to much of work in international relations. Our field's research frequently analyzes specific policies (e.g., initiation of conflict or formation of alliances) as a function of specific factors, such as power transitions, or an increase in capabilities of an adversary. We then see whether conflict is more likely when a power transition takes place or whether alliances are more likely to be formed when level of perceived threat is high. In other words, we often see research that theoretically concludes that A leads to B and seeks to find whether that conclusion is empirically correct. Most and Starr cautioned us that there are many As that could lead to a given B and there are many Bs that may be chosen by leaders in response to a specific A. A power transition may lead to a higher probability of war but it also may lead to a higher probability of accommodation; increased international threat may similarly lead to accommodation as well as new alliance membership, or even war. By focusing on single As and single Bs, we run danger of missing much of how international relations actually works. Nonetheless, work in field has been slow to accept challenges of substitutability, even though failure to account fully (or partly) for substitutability creates a number of problems. Most notable is problem caused by underidentification. That is, substitutability and domain-specificity require that researchers completely specify and identify set of possible substitutable policies as well as possible causes for a specific policy output and to do so for any model of international behavior. If alterna-