Learning, Accounting Information, and Contracting Dynamics

Abstract

I develop a dynamic principal-agent model where short/long-term incentives and performance-based turnover are jointly used to motivate productive activity in a setting with learning. The agent controls underlying fundamentals and can manipulate earnings at the expense of future reversals. Long-term incentives are optimal due to the interaction between learning and the reversal of earnings manipulations. Positive manipulation becomes optimal as long-term incentives have vested and the agency approaches termination in order to reduce the probability of inefficient turnover. This paper shows how incentive dynamics and turnover decisions are shaped by learning and performance measurement. In doing so, it generates novel empirical predictions on the impact of learning and measurement on the duration of incentive pay, the relationship between contractual convexity and productive effort, and the use of implicit incentives over termination.

Keywords: Moral hazard, learning, performance measurement, long-term incentives, performance-vesting, turnover

JEL Classification: D83, D86, G34, M12, M41, M52