“…and they are really lying”: Clean Evidence on the Pervasiveness of Cheating in Professional Contexts from a Field Experiment.

We investigate the pervasiveness of lying in professional contexts such as insurance fraud, tax evasion and untrue job applications. We argue that lying in professional contexts share three characterizing features: (1) the gain from the dishonest behavior is uncertain, (2) the harm that lying may cause to the other party is only indirect and (3) lies are more indirect lies by action or written statements. Conducted as a field experiment with a heterogenous group of participants during a University ‘‘Open House Day’’, our ‘‘gumball-machineexperiment’’ provides field evidence on how preferences for lying are shaped in situations typically found in professional contexts which we consider to be particularly prone to lying behavior compared to other contexts. As a key innovation, our experimental design allows measuring exact levels of cheating behavior under anonymous conditions. We find clean evidence that cheating is prevalent across all sub groups and that more than 32% of the population cheats for their own gain. However, an analysis of the cheating rates with respect to highest educational degree and professional status reveals that students cheat more than non-students. This finding warrants a careful interpretation of generalizing laboratory findings with student subjects about the prevalence of cheating in the population.