I develop a dynamic agency model to investigate optimal managerial authority and its interaction with managerial compensation. The model shows that when hiring a manager, the principal delegates authority that is unresponsive to either the manager's outside options or the firm's recruitment costs, in contrast to promised compensation, which increases in both. Over time, both the manager's authority and his compensation rise after good performances and decline after bad realizations. Authority-performance sensitivity decreases as the manager's authority grows, resembling entrenchment. In contrast, pay-performance sensitivity increases with the manager's authority. If managerial authority can be adjusted only infrequently, the optimal contract may allow for self-dealing. Moreover, the model reveals that early-career luck plays a disproportionate role in determining the manager's authority and lifetime utility.

Corporate bonds include action-limiting covenants that may prevent the firm from undertaking valuable growth opportunities ex-post but are virtually impossible to negotiate. We study the covenant defeasance option, which effectively mitigates this inefficiency by allowing the firm to remove the covenants. Our model predicts and our empirical analysis confirms that (1) financially constrained firms with high uncertainty are more likely to include this option; (2) with the defeasance option, issuers are willing to accept more action-limiting covenants ex-ante; and (3) investors require lower yield on non-callable bonds and a higher yield on standard callable bonds that are defeasible.

Will asset managers with large amounts of capital and high risk bearing capacity hold large blocks and monitor aggressively? Both block size and monitoring intensity are governed by the contractual incentives of institutional investors, which themselves are endogenous. We show that when high risk bearing capacity arises via optimal delegation, funds hold smaller blocks and monitor significantly less than proprietary investors with identical risk bearing capacity. This is because the optimal contract enables the separation of risk sharing and monitoring incentives. Our findings rationalize characteristics of real world asset managers and imply that block sizes will be a poor predictor of monitoring intensity.