Tax Evasion on the Micro: Significant Simulations or Expedient Experiments?

Abstract

In recent years various aspects of tax evasion (the effects of tax rates, audit probability and equity) have been investigated using simulation techniques. This paper reports a series of experiments which have looked at the usefulness of this approach. The simulations have all been implemented on a microcomputer. Study 1 attempted to replicate previous findings. The results revealed the higher the audit probability the less tax would be avoided but severity of fine had no effect. Study 2 involved interviewing subjects after they had taken part in a simulation study. Most subjects reported that the simulation was just a game and that their behaviour was unrelated to their ‘real-life’ tax behaviour. What seemed to be crucial was how the situation was defined. Study 3 investigated this directly. When the simulation was described as an “economic problem” subjects behaved more like rational optimisers than when it was described as an “economic game”. It is concluded that for simulations to make a significant contribution to our understanding of tax evasion they must be considerably more sophisticated.