William Watson: Tax audits make people do the weirdest things — or do they?
I’ve never had my taxes audited. I hope that doesn’t change after this column. Maybe the CRA will consider an interest in audits suspicious. In fact, it’s purely a question of something catching my eye in the latest email stream of research from the International Monetary Fund.
For reasons that aren’t entirely clear, four economists from the IMF’s Fiscal Affairs Department took it upon themselves to study the effects of audits on taxpayers’ subsequent declarations of income. The title of their paper is “Do audits deter or provoke future tax noncompliance? Evidence on self-employed taxpayers.”
That audits would deter tax noncompliance is clear. That’s why tax authorities do them, presumably: to catch noncompliance when it happens but also “pour encourager les autres” (as Voltaire explained in Candide regarding the execution of British Admiral John Byng in 1757, for insufficient determination in relieving the French siege of the British garrison in Minorca in the first battle of the war that led to this country’s switch from French colony to British).
Taxable income reported 64% higher in the first year after the audit with additional assessment
Using tax data on 7,500 self-employed U.S. taxpayers, the IMF economists find there’s a pretty big effect from an audit, especially if it dings you: “Among those taxpayers who receive an additional tax assessment, reported taxable income is estimated to be 64 per cent higher in the first year after the audit than it would have been in the absence of the audit.” Sixty-four per cent is not beanbag (as bean-counters will tell you). The effect does fade. Three years out the increase in reported taxable income is less: 44 per cent. But the effects are big and statistically significant (that is, not due to chance).
The economists can’t actually say what the effects are on “les autres” — on other taxpayers. They’ve only got data on their chosen 7,500. But word presumably gets around. The whole country learned about then-candidate Trump’s tax audit in 2015 and 2016, when his story was that of course he’d release his tax returns — but only after the audit was complete. There’s been no word since on how it’s going and more recently he’s been fighting all the way up to the Supreme Court to resist Congressional and other attempts to pry open his tax returns.
Taxable income reported 15% lower in the first year after the audit with no assessment
But there is another effect the paper does report on: people the IRS audited but in the end didn’t assess reported income “15 per cent lower the year after the audit than … had the audit not taken place.” That isn’t necessarily noncompliance: they may have been over-complying before. But there does seem to be a change in people’s behaviour once they’ve come through the undoubtedly unpleasant experience of an audit without additional tax damage.
The problem with analysis like this is dealing with “the counterfactual.” How do you know people who were audited and got through without an extra assessment wouldn’t have reported 15 per cent lower taxable income even if they hadn’t been audited? Or the reverse regarding people who did get assessed? The trick is to try to compare the auditees with other people who are similar to them in socio-demographic terms but who either didn’t get audited or didn’t get dinged in their audit. As with lots of social science research these days, the IMF economists spend much of their paper showing how they’ve tried to eliminate selection biases of this sort.
Assuming they’ve been successful, what do we make of their results? That people who got caught later declared substantially greater income is a no-brainer: they’re worried about getting caught again and they assume the IRS’s audit algorithm will be watching them even more carefully. But why do people who have sailed through an audit later declare 15 per cent less income, all else equal, than they did before?
Maybe they figure having cleared the audit gives them a free pass for the next few years, and they use that pass to be more aggressive in what they claim as expenses or declare as income. Or, as suggested, maybe they and their accountants decide they have been too conservative — overcompliant, as it were — and opt to be 15 per cent less scrupulous. Or, given that in this study final audit results often weren’t known before the following year’s taxes had to be submitted, it may simply be, as some other studies have suggested, that “experiencing coercive enforcement activity reduces tax morale, erodes trust, and crowds out the intrinsic motivation to comply among honest taxpayers.”
Let me make clear to any readers who may work for CRA that those are not my words. They are a quote. But, speaking of CRA, audits do take resources. In figuring how much to spend on them tax authorities will want to consider both their deterrent but now also their provoking effects.