Essay Splitting Logs: an Empirical Perspective on Employment Discrimination Settlements

Most cases settle, in employment discrimination litigation as elsewhere. Unfortunately, empirical knowledge of settlements remains limited. Data scarcity fuels untested perceptions and, all too frequently, misperceptions about how employment disputes are resolved. This Essay exploits a unique data set of successful settlements in the U.S. District Court for the Northern District of Illinois from 1999–2004 that includes for each case the plaintiff’s initial monetary demand, the defendant’s offer, and the resulting settlement. We find that in raw constant dollars, final settlements are typically far closer to defendant offers than plaintiff demands. After converting plaintiff demands, defendant offers, and final settlements into natural logs, however, the typical settlement splits the difference between plaintiff demand and defendant offer. We also find that settlement amounts rise if a trial date is set for a case. Finally, results from three-stage least squares models—that plaintiff demands influence defendant offers that, in turn, influence final settlement amounts—provide a glimpse into the structure of employment discrimination settlements.