The Design of Optimal Dependency and Rewards

We analyze a two-period principal-agent model in which the principal faces a budget constraint, and the agent’s private costs of performing tasks across the two periods may be correlated. We examine the optimal design of the reward scheme and the cost correlation structure. Our findings reveal that when the budget is low, the optimal reward scheme employs $\mathit{\text{sufficient}}\mathit{\text{performance}}\mathit{\text{targeting}}$, rewarding the agent’s first performance. Conversely, when the principal’s budget is high, the focus shifts to $\mathit{\text{sustained}}\mathit{\text{performance}}\mathit{\text{targeting}}$, compensating the agent’s second performance. Introducing a negative cost correlation proves particularly beneficial in both scenarios: it increases the likelihood of the agent performing at least once under low budgets and balances the agent’s total costs to facilitate consistent performance under high budgets. However, the optimal cost correlation structure can be more elaborate, especially for intermediate budget levels. Our results offer valuable insights for real-world applications, such as research funding allocation.