We characterize how the process of publicly-gathering information via discovery affects strategic interactions between litigants. It allows privately-informed defendants to signal through the timing of settlement offers, with weaker ones attempting to settle pre-discovery. Discovery reduces the probability of trial. Properly designed limited discovery reduces expected litigation costs. Stronger defendants gain more (lose less) from a given amount of discovery. We find that the court should grant more discovery when defendants are believed to be stronger and should grant discovery on more efficient sources of information, leaving less efficient ones to trial.

To punish an agent, the principal often incurs costs. I study a principal’s least costly reward and punishment scheme for an agent whose effort the principal cannot observe. I find the principal’s cost is sometimes minimized by using both costly rewards and costly punishments because (1) the agent has an outside option, or (2) a principal without commitment ability repeatedly interacts with the agent. I also find that when an agent’s effort is better at increasing the probability of a good outcome for the principal, the agent’s payoff may decrease, because the principal replaces rewards with punishments.