Experimentation and Persuasion in Political Organizations

Abstract

Formal models in political science typically consider actors with differing interests or ideologies, but political actors often have “good faith disagreements” about how to achieve commonly shared goals. We develop a model of a hierarchical political organization in which a principal and an agent who disagree about the best policy for achieving their shared goals can experiment. Disagreement creates a motivational problem, and we demonstrate how both observing policy outcomes and experimenting with policies can ameliorate it. We also show that the principal will often defer to the agent in order to motivate him, thereby generating more informative policy outcomes and building future consensus. Moreover, she sometimes allows the agent to implement his desired policy solely to persuade him through failure that he is mistaken. We apply the model to shed light on the effects of performance measurement, adaptive management, and Presidential appointments in U.S. federal agencies.

Keywords: strategic experimentation, disagreement, beliefs, persuasion, responsiveness, learning by doing

JEL Classification: C73, D73, D74, D83