Ambiguity in criminal punishment

There has been substantial prior field research on how the incidence of criminal behavior responds to changes in the probability of punishment. The results of this research are at best mixed in regards to its conformance to the predictions of standard expected utility theory. One possible cause for these mixed results is that punishment probabilities in the field are typically unknown making the choice environment one of ambiguity rather than uncertainty. The presence of this ambiguity could in part explain some of these conflicting results. As a step towards investigating this link, we conduct an experiment framed as being on tax compliance intended to try to understand how individuals respond to ambiguous punishment probabilities and in particular to how they respond to shifts in ambiguous versus known probabilities. We find that when probabilities are known and shift, the standard model works well to explain the response. When the probabilities are ambiguous and shift, the behavioral response is minimal. We also use these experiments as a means of testing whether ambiguity aversion might be present in sufficient degree to be exploitable in how enforcement procedures are advertised to increase their effectiveness at minimal cost. We find at best weak evidence in favor of ambiguity aversion and thus little support for the notion that enforcement regimes could take advantage of ambiguity aversion.