• This research project extends a major theme of my book on the causes of war termination: the role of leaders in international and comparative politics. In International Relations and -- to a lesser extent -- Comparative Politics we are witnessing a shift in both the theoretical and empirical unit of analysis to leaders. In a sense, IR has come full circle since the pathbreaking work by Snyder, Bruck and Sapin of the early 1960s with its clear focus on decision-making by individuals and leaders. In the 1980s, largely due to Waltz, the focus of the field shifted to the system; in the 1990s much of the field took the state or the dyad as the primary unit of analysis. With the impact of the rational choice revolution and its emphasis on methodological individualism, IR scholars but also Comparative Politics scholars have begun to put leaders and their incentives to stay in office front and center in their analyses.
    • The project seeks to promote this ongoing shift by collecting and disseminating data on all leaders between 1875 and 2015, updated regularly, for a new data base: Archigos. (This data set won the 2014 Lijphart/Przeworski/Verba Dataset Award, sponsored by the APSA section on Comparative Politics.) This data makes it possible to match theoretical and empirical units of analysis. Archigos contains data on the date as well as manner of entry and exit, as well as the leader's age, previous times in office and post-exit fate. Archigos is a collaborative effort with Kristian Skrede Gleditsch (UCSD & Essex), Giacomo Chiozza (UC Berkeley) and Jinhee Choung (UCSD).
    • In the run-up to the book, several papers opened space for new theoretical research on the role of leaders by showing that
      • War is not necessarily costly for leaders. As a result theoretical room is created for leader-based explanations other than the unitary rational actor explanations for war of Fearon (1995).
      • It is essential to properly model the reciprocal relationship between the probability of losing office and the probability of conflict initiation. While theories of the diversionary use of force require a reciprocal relationship between the loss of office and international conflict, no such reciprocal relations had been explicitly modeled. Once modeled, we do indeed find evidence for a reciprocal relationship, but opposite from the one posited by theories of the diversionary use of force: Leaders initiate (and become targets) when they are secure in office, and as the risk of conflict initiation increases, leaders become more likely to lose office.
      • The fate of leaders after they lose office is significantly worse after they lose a war then if they had not fought a war. In other words, defeat in war makes punishment above the mere loss of office, e.g., exile, jail or death, significantly more likely. In other words, and rejecting a crucial assumption in the literature on the diversionary use of force, punishment is not truncated.
    • In a co-authored effort with Giacomo Chiozza, the project culminated in new theory that links leaders' tenure and anticipated post-tenure fate with their conflict behavior in Leaders and International Conflict (Cambridge University Press 2011). We show that diversionary incentives crucially depend on the anticipated consequences of losing office. Leaders who anticipate no post-tenure punishment have little to gain, but a lot to lose from international conflict. In contrast, leaders who anticipate post-tenure punishment have little to lose and much to gain from international conflict. Thus, leaders who anticipate a regular replacement will seek to avoid conflict, whilst leaders who fear a coup have incentives to seek conflict, for example, to send potential coup plotters to the front
    • In a separate paper with Nikolay Marinov Coups and Democracy, we should that the Post Cold War dynamics of international aid fundamentally affected the consequences of Coups. Coup plotters who do not hold elections within a short period after their coup will see their international aid restricted. As a result, the time-to-election after a coup was significantly shorter after the Cold War than during the Cold War. Note that this sets up a selection mechanism whereby only coup plotters closer to the median voter than the current leader will want to stage a coup.