AdobeStock_73076909.jpeg
VoxEU Column Taxation

Shaming and tax compliance

Many tax authorities use public shaming as a penalty for tax non-compliance. Yet, we lack empirical evidence on how the introduction of naming and shaming affects behaviour. This column uses a new policy in Slovenia to argue that shaming is an effective tool for tax enforcement, reducing tax debt by about 8.5%.

‘Naming and shaming’ is widely used in many areas of policy. For instance, water wasters are shamed in South Africa, sex offenders in the US, and speeding drivers in Australia. Also, tax authorities make ample use of public shaming as a penalty for non-compliance. According to the OECD (2017), shaming is the fourth most used instrument of tax debt enforcement. 

Yet, beyond anecdotal evidence, we know surprisingly little about how the introduction of naming and shaming impacts overall compliance. Conceptually, the effect of public shaming on tax revenue is ambiguous. On the one hand, the literature on social pressure (surveyed in Bursztyn and Jensen 2017) highlights that many people care about what others think of them. Increasing the observability of non-compliance may thus prevent taxpayers from engaging in it. On the other hand, shaming can backfire if it informs taxpayers that others are non-compliant (Gino and Ariely 2009, Blaufus et al.2017) or if it crowds out a taxpayer’s intrinsic motivation (Bénabou and Tirole 2003, Boyer et al.2016). Whether shaming of tax delinquents increases or reduces tax revenue is ultimately an empirical question.

Evidence from Slovenia

To make progress on this question, we study a new policy that shames taxpayers with outstanding tax debt on the internet (Dwenger and Treber 2018). Our analyses show that taxpayers strongly respond to the shaming law. In particular, they pay tax debt in order to avoidshaming. Overall, we find that the shaming law reduces tax debt by about 8.5%. This is remarkable given that the bulk of tax debt was considered uncollectable with standard enforcement instruments. Shaming can thus be a very effective tool for improving tax compliance.

To reach this conclusion we exploit unique administrative data for the full population of taxpayers in Slovenia and study how naming and shaming affects taxpayers’ tax debt levels. Our data set contains tax and payment information on both individuals and firms.

In November 2012, the Slovenian parliament passed a new shaming law. According to this law, taxpayers with tax debt over €5,000 that is more than 90 days overdue are named-and-shamed on the Internet. The information published on shamed taxpayers unambiguously identifies them. It includes name, address, and tax identification number, as well as the name and address of beneficial owners (for legal entities only). 

The new shaming policy in Slovenia was adopted four months before its implementation in March 2013. The delayed implementation allows us to study the development of tax debt in three phases – before the adoption of the shaming law (‘baseline’), after the adoption of the shaming law but before the publication of the first shaming list (‘threat’) and after the publication of the first shaming list (‘actual shaming’). 

In what follows we focus on the shaming threat. Although the shaming law was fully applicable to all taxpayers, taxpayers were heterogeneously affected. We exploit the fact that taxpayers had different debt histories before the shaming law to identify causaleffects. Think of taxpayers who seldom incurred tax debt as they predominantly paid new tax items on time. They hardly have to change behaviour to avoid shaming. Now contrast these taxpayers to taxpayers that had tax debt on many occasions. They can only avoid shaming if they significantly adjust their payment habits. As a result, we expect those taxpayers to most strongly respond to the shaming threat that most frequently had tax debt before – that is, taxpayers that have a high probability of being shamed. In short, the shaming law introduces an incentive to pay tax debt before the law is implemented to avoid shaming, particularly for taxpayers with a high ex ante shaming probability.

Figure 1 Tax debt by shaming probability, corporations

Notes: This figure displays the average tax debt for different groups of corporations based on their shaming probability. “very low”: shaming probability in the interval (0%, 20%]; “low”: (20%, 40%]; “medium”: (40%, 60%] “high”: (60%, 80%]; “very high”: (80%, 100%). Tax debt refers to tax debt more than 90 days overdue (the relevant quantity for the shaming law).

Figure 1 presents evidence on the effect of the shaming threat. The figure shows the evolution of the average tax debt of corporations over time, differentiating between taxpayers with very high, high, medium, low, and very low shaming probability. In the months up to the introduction of the shaming law, tax debt of all taxpayers was pretty stable (at different levels of tax debt). However, after the adoption of the shaming law, we observe a sharp decline in average shaming debt. This decline continues until the first shaming list is compiled. Moreover, the effect is most pronounced among corporations with a very high shaming probability. Within four months, these corporations dramatically reduce average debt from above €80,000 before the adoption of the law to about €5,000 at the end of the period of threat. Also, corporations with a lower shaming probability reduce their tax debt but to a lesser extent. Regression analyses reveal that the average corporation in our data set reduces its tax debt by 8.5% during the period of threat. We obtain qualitatively similar results when we study the self-employed. The threat of shaming thus significantly reduces tax debt and improves overall tax payment discipline. 

The large response to the threat of shaming suggests that taxpayers have been aware of the social norm of paying taxes on time but failed to comply. This is consistent with the interpretation that social pressure (and not peer comparisons) is the mechanism through which shaming policies affect behaviour (Perez-Truglia and Troiano 2015). Digging deeper, we find no difference in the response between sexes but large differences between taxpayers with different reputational concerns. For instance, reputational concerns should be most important for taxpayers that mainly sell to end customers. And indeed, we find them to most strongly reduce their tax debt in response to the shaming threat. 

In further analyses, we assess the effect of actual shaming on tax debt. Taxpayers that have been named-and-shamed are more likely to reduce their tax debt. However, the overall impact of actual shaming on tax revenue is negligible. There are three reasons. First, only few taxpayers are named-and-shamed. Second, behavioural responses of these shamed taxpayers are small (some of them lack the funds to pay their tax debt). Finally, the effect of actual shaming is short-lived as the public lost interest in the shaming list (turning the shaming list ineffective).

Policy implications

Overall, introducing a naming and shaming policy is effective and relatively cheap to implement (Luttmer and Singhal 2014). Does this imply that its use is recommendable in terms of social welfare? Not necessarily. Paying tax debt shifts funds away from other uses. For instance, to pay tax debt, taxpayers may postpone investment decisions, which in turn negatively affects growth prospects. Moreover, some tax delinquents may be unable to pay as opposed to simply unwilling. Indeed, increasing the costs of tax debt by public shaming aggravates the situation for taxpayers that are financially constrained. Shaming leads to a decline of their social reputation among customers and business partners. It further impedes access to external funding because of deteriorated business prospects and the perceived signal of payment difficulties. The presence of taxpayers that are unwilling and unable to pay reveals a subtle trade-off. The net benefit of the shaming policy depends not only on the magnitude of the treatment response among unwilling taxpayers but also on the relative importance of unwilling and unable taxpayers in the population.

While we therefore refrain from advising for or against the use of shaming, we discuss four important issues on the optimal design of a shaming policy. First, the threat of shaming is effective and comes with fewer negative side effects than actual shaming. Thus, it is advisable to allow taxpayers to respond during a period of threat. This requires high visibility of the threat. For instance, a letter could warn all subjects with tax debt that they would be part of a shaming list. Second, the threshold amounts for the shaming policy need to be carefully designed. They determine both the effectiveness of the policy and its distributional effects. On the one hand, a low threshold amount may signal a strong social norm (as small tax debts are also socially unacceptable). On the other hand, a low threshold amount may imply that taxpayers with small tax debts are disproportionally affected by the policy (as the shaming penalty does not scale by the tax debt amount). Third, shaming needs an audience. To keep the public interested and thus shaming effective, it is advisable to shame subjects intermittently. Fourth, public stigmatisation is associated with significant pecuniary and non-pecuniary costs. If subjects change their behaviour, in response to the shaming, they should quickly be reintegrated into society. Therefore, it is recommended to immediately remove taxpayers with paid tax debt from the shaming list.

References

Bénabou, R, and J Tirole (2003), “Intrinsic and Extrinsic Motivation,” Review of Economic Studies 70: 489–520.

Blaufus, K, J Bob, P E Otto, and N Wolf (2017), “The Effect of Tax Privacy on Tax Compliance An Experimental Investigation,” European Accounting Review 26(3): 561–580.

Boyer, P, N Dwenger, and J Rincke (2016), “Do Taxes Crowd Out Intrinsic Motivation? Field-Experimental Evidence from Germany,” Journal of Public Economics 144: 140–153.

Bursztyn, L, and R Jensen (2017), “Social Image and Economic Behavior in the Field: Identifying, Understanding, and Shaping Social Pressure,” Annual Review of Economics 9(1): 131–153.

Dwenger, N, and L Treber (2018), “Shaming for Tax Enforcement Evidence from a New Policy,” CEPR Discussion Paper 13194.

Gino, F, S Ayal, and D Ariely (2009), “Contagion and Differentiation in Unethical Behavior: The Effect of One Bad Apple on the Barrel,” Psychological Science 20: 393–398.

2,438 Reads