Testing Leniency Programs Experimentally: The Impact of “Natural” Framing

We study the effects of loaded instructions in a bribery experiment. We find a strong gender effect: men and women react differently to real-world framing. The treatment effect becomes significant once we allow for genderspecific coefficients. Our paper contributes to the (small) literature on experimental tests of (anti-)corruption measures and adds evidence to the (mixed) results on gender effects and the on-going discussion on the need for sociodemographic controls.