“In order to prevent tax fraud, income tax withholding should be increased so that governments over-withhold and most taxpayers receive a refund.” This is the policy prescription in a new research about to be published. We argue that this conclusion is wrong. The authors have not proven the link between tax withholding and tax fraud, and even if, increased tax withholding creates more serious problems that would wipe out any anti-fraud benefits. Tax policy should not rely on fooling people.
What is tax withholding?
Income tax can only be settled ex-post, for the preceding tax year. This is because taxable income is not known until the complete year has passed.
However, during WWII, the US Treasury was so desperate for more money from the income tax that instead of waiting until the following March for the year’s income tax money, it instituted monthly pay-as-you-go deductions as pre-payment for the year’s tax bill. The practice of “withholding” started on July 1st, 1943. At the end of the year, the taxpayer either receives a refund (if more was withheld than he now owes) or merely pays the difference.
Study: withholding encourages expense claiming
The idea spread and now virtually all countries practice some kind of withholding, including Sweden. In a paper about to be published in AEJ Economic Policy journal, Scandinavian economists Engström, Nordblom, Ohlsson and Persson claim to have found evidence that the current Swedish withholding encourages cheating on tax returns, at least in those individuals who display “loss aversion” . In order to minimize this kind of cheating , withholding should increase. Their story is as follows:
Currently, the Swedish government sends in March to each taxpayer their preliminary income tax balance for the previous year, indicating a preliminary deficit or refund. The taxpayer can then deduct some “other expenses” from the bill and proceed to either pay or receive the adjusted balance.
In theory, how much “other expenses” a taxpayer is claiming should be independent of whether they face a preliminary refund or deficit. Those expenses either did or did not happen, and the taxpayer is claiming this expense back; irrespective of whether the government originally withheld too much or too little (producing a refund or a deficit).
The paper authors found that in the dataset of all 2007 Swedish taxpayers, about 4% of those entitled to a refund (of any size) made an expense claim, increasing their refund. . However, the proportion claiming a deduction was higher among those having to pay off a deficit, as illustrated in the graph below. Furthermore, the higher the deficit, the greater the proportion of those making a claim.
To account for this behaviour, authors invoke Prospect theory according to which individuals are really averse to the prospect of something they interpret as a “loss”. Here, “being in preliminary deficit” is interpreted as such “loss prospect”, and the extra expense claims are a way of getting more taxpayers away from the deficit territory: at least to a 0 balance and ideally to a complete conversion of preliminary deficit into actual refund.
That may well be so. However, the authors then make the extra step of suggesting that such extra expense claims are “almost certainly” cheating.
IREF: claiming does not imply cheating
To prove that the extra claims are bogus, they cite evidence that in a random sample of deductions in 2006, “there were mistakes in 93% of the cases” and these expense claims were dismissed by the tax authority.
Therefore, conclude the authors, claiming expenses implies cheating.
This conclusion is wrong; there is insufficient evidence in the paper that the expense claimers are cheating. One would have to study the details of the rejected claims to show that those facing a preliminary deficit make less credible expense claims than those facing a refund. Only then could we surmise that the “additional” claims (above the 4% rate) are bogus.
Furthermore, we don’t necessarily know that the “missing” claims from refund recipients are not genuine claims which go unreported for fear of invoking a tax audit.
IREF: no grounds for increased withholding
Regrettably, the labeling of deficit-facing claimers as “cheaters” leads the authors to a very dangerous policy proposition:
If facing a refund instead of a deficit makes people cheat less, let’s change the system so that many more people face a refund. In other words, let’s over-withhold and then give “everyone” a refund. Such system will “nudge” people out of the fearful “loss” territory, they won’t have to cheat to get out of it and the taxpayer will save monitoring and auditing costs.
This policy prescription is, again, quite unwarranted. Even if link between cheating and facing a deficit were found, other problems associated with withholding will overshadow any compliance gains and end up burdening the taxpayer more.
Economic costs of tax withholding
• It makes tax less visible (salient) since people never see the withheld money. This then contributes to higher taxes since voters do not object as much as they would if they had to physically pay the whole sum or at least the preliminary deficit. Switzerland seems a good case in point: Withholding is applied only to dividend income and to foreigners’ wages. Most Swiss still pay in bulk for the preceding year – making them less likely to approve any tax increase in a compulsory referendum.
• There is evidence that withholding increases the size of government.
• Government collects extra income by earning interest on the withheld money before returning some of it (sans interest) back to the taxpayer as a refund.
• It is anti-social as poorer people may have to get an expensive consumption loan to tie them over until their refund.
• When taxpayers irrationally consider a refund as a “win” or a popular tax “break”, it may, according to Damon Jones of the University of Chicago, make voters to encourage governments to over-withhold. (Over ¾ of all U.S. taxpayers receive a refund.)
• Introducing withholding transfers part of the burden of tax collection from government onto businesses since they have to deduct a proportional estimate from each paycheck and pay it to the government. (Admittedly, after withholding is introduced, mere increases will add smaller burden due to modern accounting software.) One famous Connecticut businesswoman’s struggle after refusing to be a government’s collector is described in her bestseller “Toil, Taxes and Trouble”.
Increasing withholding would just exploit people’s irrational joy of receiving a refund without realizing it’s their own money. Tax policy based on fooling people is very bad policy.