Is agency theory self-activating?

This article examines the conditions under which the principal-agent model is selfactivating/socially causal. We do so by exploring a principal-agent framework that allows for the possibility that rational agents may hold intrinsic preferences for autonomy in decision making and experience disutility from being monitored. Using a dynamic model of preference formation, we identify conditions under which the principal-agent model is self-activating in that, over time, the introduction of the model in an otherwise efficient monitor-worker relationship leads to the inefficient adoption of the agency model. We also examine the extent to which the agency model is robust when autonomy-preferring agents are introduced into the population.