We present a moral-hazard-based hierarchical contracting model, where investors con- tract the top manager and the top manager contracts all middle managers. We compare e®ects of hierarchical contracting on managerial contract sensitivities with those of a di- rect contracting benchmark where investors directly contract all managers. We argue that under hierarchical contracting, the top manager shifts his compensation risk to middle man- agers by providing middle managers with higher-powered incentive contracts than would be desired by investors under direct contracting. It is striking that this top managerial risk- shifting behavior motivates investors to design the top managerial contract in such a way that the top-managerial hierarchical contract sensitivity approaches either the ?rst best or zero, as the ?rm size grows. However, under some reasonable conditions such as correlated managerial e®ort outcomes, the top managerial sensitivity quickly approaches zero as the ?rm size increases, and consequently, the sensitivity for large ?rms can be far lower than predicted by the standard agency theory. This result can serve as an explanation of widely observed ?rm-size e®ects on CEO compensations, namely, lower pay sensitivities for large ?rms than those for small ?rms. We also argue that even when investors are risk-neutral and then individual performance outcomes of nonexecutive employees may be very weakly correlated to the total outcome of the ?rm, company-wide bonus plans for nonexecutive employees can still be justi?ed under hierarchical contracting.