۱- Introduction

۲- Literature review

۳- A simple analytical model

۴- Background and data

۵- Empirical strategy

۶- Results

۷- Conclusion

References

Abstract

When the tax authority increases the enforcement for one tax, what happens to the level of compliance in other taxes (spillover effect)? In this paper, we present a simple analytical model that shows that the sign of the spillover depends on how taxpayers update their beliefs about penalties and detection probabilities for one tax after observing the deterrence actions the tax agency takes for another tax. As a result, when spillovers are present, penalties and detection may not necessarily be interchangeable policy tools. We evaluate the sign of the spillover in the context of a randomized field experiment in a municipality in Argentina in a sample of about 700 taxpayers who are liable for both the property and gross-sales taxes. The evidence from the intervention indicates that the spillover from a message that increases the salience of penalties and enforcement for the property tax on the declaration in the gross-sales tax is positive. Those in the treatment group increase their reported tax by two percentage points more than the control group. This result has ample implications for researchers bringing interventions to the field and for governments’ enforcement strategies.

Introduction

Empirical studies evaluating the direct effect of enforcement on tax compliance have blossomed in the last few years (Hallsworth, 2014; Mascagni, 2018; Slemrod, 2016). However, there is little evidence of the effect of enforcing one tax on the behavior of taxpayers in other taxes (spillover effect) even though it can determine the overall success of an enforcement strategy. Should we expect positive, neutral, or negative spillovers? We explore the answer to this question by using a very simple and easily generalized analytical model à la Allingham–Sandmo that approximates the setting in which we work. In the model, taxpayers face sequential decisions about whether to pay a tax that has neither reporting nor informational asymmetries (property tax), and then how much sales to declare in a self-reporting tax where there are informational asymmetries (gross-sales tax). In that simple setting, the comparative statics are straightforward. If there is an increase in penalties (or perceived penalties), which tend to be uniform across taxes, the spillover will be positive. If there is an increase in the perceived probability of detection in one tax, the effect on the declaration of other taxes depends on how taxpayers update their beliefs about overall detection probabilities. That is, if taxpayers extrapolate the higher detection in one tax to the other taxes they owe, spillovers will be positive. However, taxpayers could also assume that, given the limited resources of the tax administration, higher detection in one tax might imply lower enforcement in other taxes, which will generate negative spillovers (Advani et al., 2017; DeBacker et al., 2015; Maciejovsky et al., 2007). Adding cash constraints or an overall budget constraint for each taxpayer could reinforce these negative effects. Consequently, while interventions affecting either the penalties or the probability of detection will have positive direct effects, the spillover effects are independent neither of whether the tax authority signals higher penalties or detection nor the assumptions taxpayers make about the enforcement capacity of the government.