In recent years scholars have paid considerable attention to the ways in which resource dependence predicts civil violence. But by only focusing on the link between resource dependency and the outbreak of conflict, studies tend to ignore ways in which the same goods influence how conflicts end. Exploiting variation in world commodities prices as an instrument for changes per capita income, this paper presents a multi-method approach for testing the extent to which commodities markets influence the outcomes of civil wars. After testing predictions derived from a formal model, I find that the likelihood of rebel victory is significantly increased by negative shocks to per capita income.

We develop and test a new explanation for case outcomes on diverse three-judge panels. After confirming the results of Boyd, Epstein and Martin (2010), we clarify the process by which minority judges (racial, gender, or even ideological) affect their peers’ votes. We argue that a true “informational” story fits better than one based on “expert cues.” We also expand the informational theory beyond ascriptive characteristics such as race and gender, and demonstrate that it works for other judges who might be considered “career experts” or “information-providers” in a given issue area. We test this effect in federal appellate cases on campaign finance and capital punishment, and find that judges with career experience do have a treatment effect on their peers. Moreover, this effect appears to be independent of the expert judges’ ideological leanings, thus supporting our description of a pure “informational” effect.