Timing Is Everything: A New Way to Estimate Strategic Behavior

Cycles of violence are driven by actions and reactions, whether between or among governments, insurgents, or terrorists. In this working paper, authors Danny Klinenberg, Eli Berman, and Esteban Klor describe an approach for estimating strategic behavior rooted in game theory that can accommodate the messy reality of conflict.
DownloadVector autoregression (VARs) has been a workhorse in political science and economic research. But what does it actually estimate? In this working paper, authors Danny Klinenberg (IGCC), Eli Berman (UC San Diego), and Esteban Klor (Hebrew University of Jerusalem) investigate potential pitfalls when using VAR to study conflicts. They show why estimates of strategic behavior may be biased high, low, or flip signs if combatants vary when they attack each other (which they often do). The authors advocate for using action-level data, rather than traditional time periods, like days or weeks.
Thumbnail credit: Mark Knobil (Flickr)