Is there a strategic mechanism that explains role-contingent differences in conflict behavior? I sketch a theory in which differences in optimal behavior for attackers and defenders arise under initially symmetric conditions through the dynamic accumulation of differences in the distributions of traits in the subpopulations of potential opponents.

We present a model and a laboratory experiment on the informativeness ofdebate, varying both informational and institutional variables. Theinformational variable we focus on is a novel factor affecting the extent towhich audience members can learn from exposure to unpersuasive arguments.The more easily a listener can learn from an argument she findsunpersuasive, the greater the risk that the speaker will alienate thislistener when she fails to persuade her. We find a strong interactionbetween speakers’ responsiveness to that risk and the institutions ofdebate. When listeners can learn from unpersuasive arguments, many speakersare discouraged from attempting persuasion, irrespective of the debate ruleswe consider. In contrast, when listeners cannot learn from unpersuasivearguments, debate rules affect speakers’ willingness to engage inpersuasion.

Regulatory agencies impose costs and benefits tailored to individual firms through their discretionary enforcement activities. We propose that corporations use political expenditures in part to “flex their muscles” to regulators and convey their willingness to fight an agency's specific determinations in the political arena. Because the signaling function of political expenditures is strategically complex, we derive a formal model wherein we demonstrate the existence of an equilibrium in which (1) large political donors are less compliant than smaller ones, but the bureaucracy monitors them less, and (2) firms with publicly observable problems reduce their political expenditures. We test the empirical implications of the model using plant-level data from the Nuclear Regulatory Commission on the inspection of 63 privately operated nuclear power plants and the political expenditures of their parent companies. We find strong evidence for the first prediction and qualified support for the second.