All the rich countries are successful in raising sizable amounts of revenue from taxes with only a rather little tax evasion. Tax avoidance is the use of legal means to reduce taxes, whereas tax evasion uses illegal means. The federal government of the US raises almost 20 percent of American GDP through taxes on personal and business income, capital gains, estates, and the sale of gasoline and some other goods. The estimates from the 2001 IRS National Research Program indicate that the percent of income not reported is quite low for wages and salaries, but rises to over 50 percent for farm income, and about 40 percent for business income. Income tax payments overall are under reported by about 13 percent. What determines the degree of tax evasion?
If taxpayers responded only to the expected cost of evading taxes, evasion would be far more widespread. The reason is that only about 7 percent of all tax returns are audited (over a 7 year period), and typically the penalty on under reported income is only about 20 percent of the taxes owed. Virtually no one is sent to jail simply for evading taxes unless that evasion is on a very large scale, or involves massive fraud. If a person were to evade $1,000 in taxes, his expected gain would be 0.93x$1000 -0.07x$200 (=$1000/5) = $916. On these considerations alone, he should not hesitate to evade paying the $1,000, and presumably much more.
To be sure, the expected gain is not the right criterion since most taxpayers would be risk averse regarding audits and punishments, especially if there is some chance of much greater than the average punishment or likelihood of an audit. However, if the expected gain from evading $1,000 were $916, the degree of risk aversion would have to be huge, far higher than the risk aversion that is embodied in pricing of assets, for risk to explain why there is so little tax evasion.
This is not to say that possible punishments have no affect on the amount of tax evasion. Compliance rates are much higher when governments have independent evidence on a person's income since then the probability of audit when he under reports his income is much higher than when they do not have this information. For example, income from independent consulting to companies is better reported than tips on earnings, or than the incomes of farmers and other small business owners because employers report how much they paid to independent consultants, whereas no one reports how much they paid in tips, or how much they bought from a local store. A PhD study in progress at the University of Chicago by Oscar Vela also shows that persons in occupations where integrity is a more important determinant of success, such as law or medicine, are less likely to evade taxes. Presumably, any publicity that an individual in these occupations was convicted of tax evasion would damage his reputation and earnings.
Vela finds that considerations of reputation, along with more traditional variables in the tax evasion literature do help explain how much evasion occurs for different types of income. These variables include the likelihood of audits that varies for different classes of taxpayers, punishments for those audited, marital status (not surprisingly, married persons are less likely to evade taxes), the marginal tax rate, and the ease with which governments can match reported incomes with independent evidence on incomes, such as from 1040 and 1099 tax forms,
Note that tax avoidance as well as tax evasion tends to rise as the marginal tax rate increases. That is, with higher tax rates, individuals and businesses are both more likely not to report some of their income to the tax authorities, and also to search harder for ways to reduce how much of their income they are obligated to report. This implies, for example, that flattening the income tax structure would increase the amount of personal income reported to tax authorities because both the amount of evasion and the avoidance of the personal income tax would be reduced.
However, audits, punishments, and the other deterrence variables mentioned in the previous paragraphs do not fully explain why there is not much more tax evasion. I believe it is necessary to recognize that most people believe they have a duty, moral or otherwise, to report their taxable income more or less honestly. I intentionally say "more or less honestly" because a little cheating on taxes is usually considered to be ok, as long as it does not go too far. Individuals might not pay social security taxes on their payments to workers who clean their houses, and they might pay a mason in cash because he then gives them a lower price, but these same persons would be very reluctant to engage in large-scale tax evasion.
Similarly, most people do not believe it is moral to steal money even when there is little chance they will be found out, and they feel obligated to obey many other laws, even when that entails inconvenience and cost to themselves. There would be considerably more crime if individuals only obeyed laws when the expected cost of being caught, adjusted for risk, exceeded the benefits from disobeying these laws. To some extent, people obey many laws, including tax laws, because most other persons are doing the same. If so, their behavior might change radically if they lost confidence that others would pay their taxes and obey other laws.
Clearly, morality about obeying laws does not apply to all types of taxes, or all laws-people often cross a street when the light is red, do not stop at stop signs when riding their bikes, and do not report much of their tips. Moreover, in many countries of Latin America, Africa, and Russia and other parts of Eastern Europe, individuals do not even feel much obligation to pay ordinary income and other taxes. They evade except when they expect the chances of being caught are high, as with businesses paying value added taxes. These countries are unable to raise substantial amounts from taxes on personal incomes or businesses except when marginal tax rates are low. Instead they rely greatly on value added and other more difficult to evade taxes.