A Flexible Framework for the Examination of Production, Measurement, and Contracts in the Face of Moral Hazard
Bonham, Jonathan David
Doctor of Philosophy
I develop a framework to examine the effect of measurement on productive activity in the face of moral hazard. I allow an agent intricate control over the stochastic value of a firm's assets, and he is compensated based on a report produced by an accounting system that admits a large class of bias- and timing-oriented accounting measurement rules. When measurement error is unavoidable but is treated to address the moral hazard problem, (i) the fundamental earnings distribution develops asymmetric tails and discontinuities at predictable thresholds, (ii) measurement rules develop all-or-nothing recognition properties and are rarely unconditionally biased, and (iii) the contract develops caps, floors, and hurdle bonuses at predictable thresholds. In contrast, when measurement error can be reduced by delaying measurement until uncertainty has been resolved, historical cost accounting is unambiguously optimal in curtailing moral hazard. However, I show that an accounting regulator with alternative objectives can influence economic activity by mandating timely measurement. Specifically, I show that timely loss recognition induces firms that are more (less) averse to downside risk to contract for riskier (less risky) actions. Finally, I show that first best actions are implementable in my setting via a two-wage penalty contract only if the measurement rule is extremely noisy and unconditionally conservative. Furthermore, the agent charges a negligible risk premium if he is sufficiently optimistic about the odds of avoiding a penalty-triggering earnings report. In other words, unconditionally conservative measurement can disable moral hazard when the agent is optimistic.