Some bank regulators require that performance bonuses for risk managers (RMs) and for employees in front offices (FOs) be linked to distinct performance metrics, as correlated pay incentives could lead RMs to rubber-stamp risky investments. We theoretically show that a positive correlation between FOs and RMs is optimal for banks, but can be socially excessive in leveraged institutions. Using data from German bank employees, we show empirically that incentive pay is indeed positively correlated between RMs and FOs. Consistent with our predictions, bonus correlations are higher in banks with higher leverage and weaker performance during the Great Financial Crisis.

We investigate the claim that supervisors do not differentiate enough between high- and low-performing employees when evaluating performance. In a first step, this claim is illustrated in a formal model showing that rating compression reduces performance and subsequent bonus payments. The effect depends on the precision of performance information and may be reversed when cooperation is important. We then investigate panel data spanning different banks and find that stronger differentiation indeed increases subsequent bonus payments. The effect tends to be larger for larger spans of control and at higher hierarchical levels but is reversed at the lowest levels.

We argue that risk sharing motivates the bankwide structure of bonus pay. In the presence of financial frictions that make external financing costly, the optimal contract between shareholders and employees involves some degree of risk sharing whereby bonus pay partially absorbs negative earnings shocks. Using payroll data for 1.26 million employee-years in all functional divisions of Austrian, German, and Swiss banks, we uncover several empirical patterns in bonus pay that are difficult to rationalize exclusively with incentive theories of bonus pay but that support an important risk sharing motive.