Multitasking and subjective performance evaluations : theory and evidence from a field experiment in a bank

We study the incentive effects of grating supervisors access to objective performance information when agents work on multiple tasks. We first analyze a formal model showing that incentives are lower powered when supervisors have no access to objective measures but assess performance subjectively by gathering information. This incentive loss is more pronounced when the span of control is larger and incentives are distorted towards more profitable tasks. We then investigate a field experiment conducted in a bank. In the treatment group managers obtained access to objective performance measures which raised efforts and profits. We find that the effects are driven by larger branches and lower margin products.