By Shankar Vedantam
Monday, April 16, 2007
Economists have long known there are two reasons that people cheat on their taxes. One is that they are poor and need the extra cash so badly they are willing to risk getting caught. The other is that they are rich and have lots of "non-matchable" income -- mostly investment income not directly reported to the government -- which makes it less likely they will be caught.
Taxpayers in the middle class are the least likely to cheat: They are not struggling to make ends meet, and their income is mostly wages, which are directly reported to the Internal Revenue Service. If you measured the likelihood of tax evasion by income level, in other words, the graph would look like a giant U.
What economists missed seeing until recently, however, is that both these explanations are really part of a single, larger theory. Essentially, said Kim Bloomquist, a senior economist at the IRS in Washington, the more people you have at the upper and lower ends of the income spectrum -- at the ends of the U -- the more tax evasion you are likely to see. A central cause of cheating, in other words, might be inequality.
Bloomquist tested his hypothesis. In 1980, Americans under-reported 3.6 percent of their income. In 2000, Bloomquist estimated that they underreported about 5.6 percent of income -- based on increases in the kinds of income not reported to the government on documents such as a W-2 form.
In the same period, the non-matchable portion of the income of the wealthiest 5 percent of Americans rose by 9.2 percent a year. In 1980, about 19 percent of the income of the wealthiest 5 percent Americans was non-matchable. By 2000, it was 37.9 percent.
Bloomquist also compared underreported income with other inequality analyses. University of California economists Thomas Piketty and Emmanuel Saez found that in 1970, the wealthiest 10 percent of Americans had 31.5 percent of all reported income. By 2000, they had 43.9 percent.
"Empirical results show a statistically significant relationship between a measure of pre-tax income inequality . . . and U.S. wage and salary reporting noncompliance for the period 1947 to 2000," Bloomquist concluded.
Bloomquist said he has found only a flag, not proof, that inequality causes tax evasion. What would settle the question is better data on tax evaders. He said the IRS plans to conduct random audits in a more systematic fashion in the future, which should provide better evidence.
But another line of evidence suggests the economist is on to something. International data suggest that the size of a country's shadow economy -- income not reported to authorities -- appears to be linked to income inequality.
In a study of 23 countries, in which nations got inequality ratings from 0 to 100, the Netherlands had an inequality score of 29.6 and Mexico had a score of 54.98 -- meaning Mexico is more unequal than the Netherlands. The shadow economy in the Netherlands was 13.4 percent of the legitimate economy, but it was 49 percent in Mexico. Thailand had an inequality score of 48.8 and a shadow economy that was 71 percent of the legitimate economy.
The United States had an inequality score of 37.8 and a shadow economy that was 10.5 percent of the size of the legitimate economy, which suggests that better monitoring of income may be compensating to some extent for the effects of inequality on tax evasion.
Bloomquist believes tax evasion can be reduced by further increasing the visibility of investment and small-business income. The IRS has found that wage income reported to the government produces a reporting accuracy of 99 percent, whereas the accuracy of reporting "non-matchable" income is only around 44 percent.
Other experts are considering more creative ways to improve tax compliance. One idea is to take advantage of people's desire to get a refund at the end of the year.
"What some people do when they are doing their taxes is they do a first draft and see how much they are getting back," said Richard Thaler, a University of Chicago economist who studies how people think about money. "If they owe money, then they do a second draft. They keep finding deductions until the refund is positive."
Thaler said mandatorily increasing withholding levels so more people get refunds could increase compliance because taxpayers would no longer have to go to great lengths to get a refund.
Economist John Carroll at the Massachusetts Institute of Technology once explored the same phenomenon: He audiotaped taxpayers as they were filing taxes, getting them to speak their thoughts aloud. While many recognized that getting a refund meant they had given the government an interest-free loan, they said they loved to get a check from the IRS.
Would the IRS want to encourage people to withhold more money and claim refunds? Bloomquist said no. For one thing, most people who get refunds are taxpayers in the middle class who are the least likely to cheat anyway.
Besides, Bloomquist added, compliance is not the only issue: Increased withholding would mean people had less of their own money to use during the year -- last year, 108 million taxpayers overpaid nearly $250 billion to the government. Nearly 75 percent of taxpayers got refunds, which averaged $2,264.
Finally, Bloomquist said, unscrupulous tax-preparation firms already charge needy people high interest rates to get instant refunds, meaning that if you encouraged more refunds, more people would end up paying to get their own money returned to them.