Tax evasion is a pervasive phenomenon all over the word. Recent estimates for 38 OECD countries by Buehn and Schneider (2016) show that tax evasion ranges from a low of 0.5% of GDP in the US to almost 7% of GDP in countries like Mexico or Turkey. Tax evasion not only depresses the public budget, but it may also have negative distributional and efficiency consequences (Alstadsaeter et al. 2019).
To investigate the determinants of tax evasion, the traditional approach is to assume a direct relationship between the taxpayer and the tax authority – the tax authority decides the audit strategy and punishment policy, and taxpayers infer their probability of being audited. Conditional on the audit strategy, taxpayers choose tax evasion by weighing the gain from hiding income and the expected penalty if caught cheating when filing taxes (Allingham and Sandmo 1972). In reality, however, the compliance decision is mediated by the advice of an expert who helps the taxpayer to navigate the complexity of the tax rules (Andreoni et al. 1998, Slemrod 1989). But what is the role of the tax professional? Is it merely to offer a valuable cost-saving service that helps taxpayers economise information gathering to understand the intricacies of the tax-code (as most expert intermediaries do)? Or does their role go beyond this, possibly as a by-product of their role as tax experts?
Early lab experiments highlight that tax compliance reacts not only to monetary incentives but is also influenced by social norms and the behaviour of peers (Alm et al. 2017, Fortin et al. 2007). Recent field experiments show that compliance depends on the salience of the communication of enforcement (Meiselman 2018), on the publicity of the behaviour (Perez-Truglia and Troiano 2015), and on the exchange of information with individuals that are geographically close (Rincke and Traxler 2011).
To what extent do tax accountants contribute to shaping these social norms? Anthropological and social studies hint at the possibility that tax professionals influence the formation of expectations on enforcement probabilities (Braithwaite et al. 2005) and may even contribute to shaping the ethical standards of their clients (Raskolnikov 2007). While intriguing, this evidence is only suggestive. Yet, a better understanding of whether tax professionals play such a role is key not only to contain tax evasion, but also, and above all, to design a more equitable and efficient tax system.
In our recent paper (Battaglini et al.2019), we document that there is a specific role of the tax preparer in determining the compliance choice of the taxpayer.
We combine tax records of 2.5 million taxpayers in Italy with the corresponding audit files from the tax revenue agency. These data cover the entire population of sole proprietorship taxpayers in seven regions, followed over seven fiscal years. Crucially, we are able to track the identity of the taxpayer’s tax professional.
We find evidence that the levels of evasion of taxpayers served by the same accountant are strongly correlated. A one standard deviation increase in the share of evasion of the clients of the accountant is associated with an increase in the evasion rate of a single taxpayer of 8.7 percentage points (27.2% of the sample mean). This result holds after controlling for the characteristics, sector of activity and geographical location of taxpayers. Importantly, it holds only among the clients of the same tax accountant. No correlation is found between the evasion of a given taxpayer and that of clients of other accountants located in the same city, even if their firm belongs to the same sector and is of similar size. Put differently, the correlation arises because of a specific role of the taxpayer’s own accountant.
To understand this role, we set up a model where both taxpayers and tax preparers have heterogeneous morality and the tax agency audits strategically. Additionally, as a by-product of his/her role, the tax accountant observes all of the audits to his/her clients and can pool them when inferring the tax authority policy. The model highlights two main possible channels that can generate the documented correlation – tax accountants can serve as tax evasion facilitators, or as information hubs.
The first channel is generated by the heterogeneity of tax accountants in their ethical characteristics. Taxpayers that are more prone to evade taxes tend to hire tax professionals that are more accommodating in the interpretation of the tax rules. This sorting of unethical taxpayers with unethical tax professionals implies that it is likely that a client of a certain professional is an evader if there are many tax evaders among the other clients. To test for the presence of this channel, we exploit the panel structure of our data, its size and the fact that some taxpayers are forced to switch accountants upon closure of their existing tax preparer. In Figure 1, we compare the evasion rate of the clients of the old and the new accountant before the switch. It illustrates evidence of sorting.
Figure 1 Sorting of taxpayers into tax accountants
Notes: This figure plots a bin-scatter of the average evasion rates of clients of the old and new accountants for the sample of taxpayers switching accountant after the closure of the old accountant. Average evasion rates are computed before the switch. We partial out year, sector and municipality fixed effects and switchers’ characteristics to account for the possibility that evasion rates have a local and sectoral component.
The other channel (the one arising from tax preparers acting as information hubs) involves an active role of the tax professional in the formation of the enforcement probabilities. In the traditional representation of the taxpayer/tax authority relation, the taxpayer can only predict his/her audit probability based only on his/her past experience, and thus with high uncertainty. On the contrary, a tax professional observes the audits of all his/her clients. By pooling all these signals together, he/she can estimate the probability of an audit with much greater precision and can thus share this valuable information with other clients. The key implication of this channel is that a taxpayer’s filed income should respond not only to own audits but also to the audits of the other customers of his/her accountant.
We find strong evidence consistent with this effect. We document that, after an audit, a taxpayer files a higher income, with an effect on compliance that decays over time (see Figure 2, solid green line). The total cumulative effect of the taxpayer’s own audit is to increase the taxpayer’s reported income by 16% (dotted green line). A similar effect has also been detected by using randomised audits (Kleven et al. 2011). In addition, our results also show that taxpayers respond to the audits of the other fellow customers of their accountants. Differently from the effect on the taxpayer’s own audit, this spillover is small in the first year (2%) but increases and persists over time, with a cumulative effect over three years as large as 9% (Figure 2, dotted red line). The information hub channel, therefore, is as large as 60% of the disciplining effect of a direct audit.
We also show in placebo regressions that there is no significant change in declared income if another individual, potentially connected with the taxpayer because they belong to the same social network but with a different accountant, is audited. In particular, we find that there is no effect of audits on peers living in the same province and with the same age or firm dimension, or operating in the same sector. This evidence confirms that the reaction to the audits is truly motivated by the information conveyed by one’s own accountant.
Figure 2 Information sharing across clients of the same accountant
Finally, we document that the probability of switching accountants falls by 6% after an audit on other clients and the presence of this effect is only possible thanks to the information dissemination role of the tax accountant who reveals the audit event.
The existence of an intermediary between the tax authority and the taxpayer changes the process of tax collection in a fundamental way. On the one hand, the spreading of information through the tax accountant boosts the compliance effects of tax audits and the effects are too large to be ignored. On the other, the tax evasion facilitator role of the accountants – as suggested by the evidence on sorting – allows the tax authority to act strategically by targeting compliant accountants. This is indeed the optimal policy that our model predicts. The data also support the prediction that the audit policy is responsive to features of the tax accountants that predict evasion. More generally, the revenue agency can and probably should exploit the tax accountants’ role as an information hub to strategically disseminate adherence to the tax code regulation and achieve an equilibrium with higher tax compliance.
Allingham, M G and A Sandmo (1972), “Income tax evasion: A theoretical analysis”, Journal of Public Economics 1(3-4): 323-338.
Alm, J, K M Bloomquist and M McKee (2017), “When you know your neighbour pays taxes: Information, peer effects and tax compliance”, Fiscal Studies 38(4): 587-613.
Alstadsæter, A, N Johannesen and G Zucman (2019), “Tax evasion and inequality”, American Economic Review 109(6): 2073-2103.
Andreoni, J, B Erard and J Feinstein (1998), “Tax compliance”, Journal of Economic Literature 36(2): 818-860.
Battaglini, M, L Guiso, C Lacava and E Patacchini (2019), “Tax professionals: Tax-evasion facilitators or information hubs?”, CEPR discussion paper 13656.
Braithwaite, J (2005), “Globalisation, redistribution and tax avoidance”, Public Policy Research 12(2): 85-92.
Buehn, A and F Schneider (2016), “Size and development of tax evasion in 38 OECD countries: What do we (not) know?”, Journal of Economics and Political Economy 3(1): 1-11.
Fortin, B, G Lacroix and M-C Villeval (2007), “Tax evasion and social interactions”, Journal of Public Economics 91: 2089-2112.
Kleven, H J, M B Knudsen, C T Kreiner, S Pedersen and E Saez (2011), “Unwilling or unable to cheat? Evidence from a tax audit experiment in Denmark”, Econometrica 79: 651-692.
Meiselman, B S (2018), “Ghostbusting in Detroit: Evidence on nonfilers from a controlled field experiment”, Journal of Public Economics 158: 180-193.
Perez-Truglia, R and U Troiano (2018), “Shaming tax delinquents”, Journal of Public Economics 167: 120-137.
Raskolnikov, A (2007), “The cost of norms: Tax effects of tacit understandings”, University of Chicago Law Review 74(2): 601-685.
Rincke, J and C Traxler (2011), “Enforcement spillovers”, Review of Economics and Statistics 93(4): 1224-1234.
Slemrod, J (1989), “The return to tax simplification: An econometric analysis”, Public Finance Quarterly 17(1): 3-27.