Costs of Control in Groups

This paper explores the role of social groups in explaining the reaction to control.We propose a simple model with a principal using control devices and a controlledagent, which incorporates the existence of social groups. Testing experimentally theconjectures derived from the model and related literature, we find that agents in socialgroups (i) perform more than other (no-group) agents; (ii) expect less control thanno-group agents; (iii) decrease their performance substantially when actual controlexceeds their expectation, while no-group agents do not react; (iv) do not reciprocatewhen facing less control than expected, while no-group agents do.