Expert Incentives under Partially Contractible States

TL;DR

This paper analyzes how a principal can incentivize an agent to acquire costly information under partially contractible states, using noisy observations and contract design to optimize incentives and compare information quality.

Contribution

It characterizes the optimal incentive schemes under partial contractibility and introduces a new order on information quality based on likelihood ratios.

Findings

Optimal contracts depend on noisy observations and limited liability.

A new information order is established based on likelihood ratio differences.

Certain types of information are consistently better for incentivizing expert predictions.

Abstract

I study whether and which expert incentives can be provided at what cost when the states of the world become non-contractible, but there is some noisy observation about the states that can be contracted upon. A principal hires an agent to acquire costly information about the states, but it is not possible to pay the agent based on the realized states. Instead, the principal has access to a noisy (Blackwell) experiment about the states, and can pay bonuses based on its realization. The agent is risk neutral and protected by limited liability. I completely characterize what the principal can incentivize the agent to learn, and how to design contracts to minimize the costs to provide such incentives. I then study which contractible information is always better at incentive provision. This gives rise to a novel order on information. In the binary-binary case, this order is characterized by larger differences in the likelihood ratios of the two realizations. My results provide insights into what information is better for evaluating expert predictions.