Instrument-Based vs. Target-Based Rules

We develop a simple model to study rules based on instruments vs. targets. A principal faces a better informed but biased agent and relies on joint punishments as incentives. Instrument-based rules condition incentives on the agent's observable action; target-based rules condition incentives on outcomes that depend on the agent's action and private information. In each class, an optimal rule takes a threshold form and imposes the worst punishment upon violation. Target-based rules dominate instrument-based rules if and only if the agent's information is sufficiently precise. An optimal unconstrained rule relaxes the instrument threshold whenever the target threshold is satisfied.