OB Research Seminar - Jan Schmitz, Radboud University

We report results from a large scale RCT testing strategies to improve payroll tax compliance of small firms in a middle-income country. We randomize announced audit probabilities (1%, 10%, 40% or 60%) on the firm level and implement several novel types of moral appeals (varying information on the benefits of tax-financed public goods). Our monthly tax return data shows that both deterrence and moral measures significantly improve payroll tax compliance. The treatment effect of a high audit probability is approximately 50\% larger than the effect of moral appeals. An additional treatment with ambiguous audit probability shows that behavioral factors play a minor role in the deterrence treatments.