WHY TAXES?

Taxes have three potential functions: to raise revenues, to modify incentives, and to redistribute income. The first function is accepted by everyone. Even libertarians agree that the state must perform minimal functions, such as national defense, police protection, and maintaining a court system. To pay for them, the government needs tax revenue. The big question, of course, is how much revenue should be collected and from whom.

The second function of taxes, often ignored, is to shape incentives. Usually this is an unwanted side effect of the first function: for example, as just noted, taxing income discourages people from working. Sometimes, however, modifying incentives is the goal of the tax and collecting revenue is merely a side effect. Most libertarians object to this kind of tax because it represents a not-so-subtle form of state coercion. Historically, for instance, Muslim countries allowed non-Muslims to practice their religion so long as they paid a tribute.¹ Still, that was a kinder form of coercion than the one Christian countries sometimes practiced: executing infidels. My opinion is that incentive-shaping taxes are legitimate if the alternative is some worse form of coercion such as restrictions or prohibition.

The third function of taxation, redistribution, is very controversial. While most people accept a minimal amount of redistribution (there is little support for a tax that is the same for everybody), the extent of redistribution is the subject of debate. It is useful to distinguish between two forms of redistribution. The first is an unequal allocation of fiscal burden that occurs in connection with a plan whose purpose is not redistributive—for example, the wealthy happening to pay a disproportionate share of the costs of police protection or national defense. This milder form of redistribution was apparently endorsed even by Adam Smith, who wrote, “It is not very unreasonable that the rich should contribute to the public expense, not only in proportion to their revenue, but something more than in that proportion.”² The second form is what I call Robin Hood redistribution, whose actual purpose is to redistribute money from the rich to the poor.

As the role of government grows, taxes increase. Thus the first step toward reducing the tax burden is restricting the number of tasks that government performs. Eliminating subsidies, obviously, is one way to do this. Still, taxes are a necessary evil, since government, most people today agree, should provide a social safety net and fund primary and secondary education. How should these functions be financed?

THE MORAL BASIS OF TAXATION

During the 2008 presidential campaign, a colleague of mine was watching the Democratic convention with his nine-year-old son. At first, the youngster supported Barack Obama’s idea of taxing the rich to help the poor. Then his father asked him what category he belonged to. When he realized that he belonged to the “rich,” his support for the Democratic candidate vanished. As this example illustrates, our view of the optimal degree of redistribution tends to correspond to where we are in the income distribution. Many people are delighted to tax the rich—but only as long as the rich are those who make more than they do.

One important insight of economics is that taxes should fall most heavily on people whose behavior is least affected by them.³ Suppose that houses are taxed on the basis of the number of windows they have. Homeowners will start to brick up windows to avoid the tax, even though doing so decreases their satisfaction with their houses—their utility, an economist would say. While each homeowner has an incentive to brick up windows and lower his tax bill, collectively homeowners would be worse off doing that. If the tax doesn’t generate the necessary revenue, each taxpayer will be forced to pay roughly the same amount in some other form of taxation, except he will be living in a dark house. Now imagine a different kind of tax, one that people cannot avoid by modifying their behavior—a tax on height, for example. Such a tax is preferable, an economist would say, because it cannot encourage taxpayers to distort their behavior and lose utility.

As we saw at the beginning of this book, people’s view of taxation is heavily influenced by their perception of how income is obtained. When they perceive unequal income distribution as the result of luck, they are more likely to accept redistribution. In this sense, the fight against crony capitalism will help relieve pressure for redistribution.

BAD TAXES

Most taxes designed to raise revenue have a distorting effect on incentives, especially on people’s willingness to work. Just as a tax on cigarettes reduces the amount of smoking, an income tax reduces the amount of work people do in the formal sector. My next-door neighbor chose not to work, and instead to stay home and take care of his kids, because his wife’s income put him in a high income bracket. Had he worked and hired a baby-sitter, he would have had to pay taxes on his income and on the baby-sitter’s income. By staying at home, he enjoyed tax-free baby-sitting.

Harvard economist Martin Feldstein, using the 1986 tax reform, estimated that when marginal tax rates on the top income bracket dropped from 50 percent to 28 percent, reported taxable income increased by 44 percent.⁴ This didn’t fully make up for the reduction in revenues caused by the reduction in the tax rates, but it did pay for half of it and exposed the inefficiency of high taxation.

Many economists have challenged or qualified Feldstein’s results. The main qualification is that much of the sensitivity of reported income to taxes is due to tax elusion and evasion, not to people’s working harder.⁵ But the ability to dodge taxes is a function of the legal code, one that we can reduce and possibly eliminate by getting rid of preferential treatment and loopholes, as I recommended in the previous chapter.

If we exclude tax dodging, taxable income doesn’t respond to lower marginal tax rates quite as quickly as in Feldstein’s study, but the effect is still there for earners who make more than $100,000 a year.⁶ As for outright millionaires, there are so few of them that conducting a large study is difficult. Still, it is legitimate to ask how much superstars—major-league athletes, movie stars, top lawyers and financiers, and so forth—are affected by the marginal tax rates. Would Tiger Woods skip the Masters Tournament if his tax rate went up? Probably not.

However, if Tiger Woods’s tax rate went up significantly, he could easily move to another country. A study of European soccer stars shows that they move disproportionately to lower-tax countries.⁷ Sweden finds itself unable to attract many foreign players and has lost several superstars. This correlation could be spurious, but Spain offered a kind of natural experiment in 2004 by introducing preferential tax treatment for foreign players. Sure enough, the number of foreign stars in Spain increased. Interestingly, the higher a European country’s highest marginal tax rate, the worse its club teams in the European Championship League tend to perform. If entrepreneurs behaved like soccer players, the cost of a high marginal tax rate would be very high indeed.

If the government incurred no expenses or had other sources of revenue (like Kuwait’s oil revenue), we could get rid of taxes altogether. Unfortunately, in most countries, including the United States, that is not possible. Yet we can reduce the burden by raising revenues through good taxes.

GOOD TAXES

“Good tax” may sound like the ultimate oxymoron at first. But there are certain cases in which taxes may correct distorted economic incentives, doing free markets a favor. Let’s begin by recognizing, as even the libertarian tradition acknowledges, that individual freedom has limits: one person’s actions may not interfere with another person’s freedom. A musician’s freedom to play his saxophone at night interferes with his neighbors’ freedom to sleep. For this reason we impose limits on behavior, such as making excessive noise at night.

Absolute prohibitions, however, tend to be very inefficient. For instance, if the musician’s only neighbor is deaf, then an absolute prohibition on excessive noise at night reduces the musician’s utility without increasing the utility of his neighbor. The prohibition becomes much more efficient if it is not absolute. If the damage the musician causes with his playing is less than the benefits that he receives from playing, he can offer a side payment to his neighbor, maybe a promise to water his garden when he travels, in exchange for his tolerance for late rehearsals. Such a contract will leave both of them better off. In the absence of any prohibition at all—this would be a laissez-faire approach—the outcome would again be efficient, though a change would take place in who pays whom: the neighbor who wants to sleep would compensate the musician for not playing, rather than the other way around.

Most electric power plants release pollutants. If we were to prohibit pollution absolutely, most power plants would have to be shut down. But if we take the laissez-faire approach and permit a plant to pollute without reservation, how can its neighbors combat excessive emissions? They cannot easily pay it to curb its emissions (the equivalent of the neighbor’s paying the saxophonist to be quiet). You might answer that the Environmental Protection Agency is supposed to intercede in such a case, by allowing the power plant to operate only when the cost of pollution to the neighbors is lower than the benefit of energy production. Unfortunately, as noted throughout this book, government agencies tend to be captured and represent the interests of industry, so there would be no guarantee such regulation would work.

In this case, a tax is an effective alternative. A tax equal to the negative impact that elevated emissions have on the local community will force the producer to factor that cost into his decision-making. When market prices fail to incorporate part of the cost that an activity generates, as is often the case with pollution, such a tax will be beneficial, regardless of the revenue it generates.

But is the tax-setting process less likely to be captured than the regulatory process? It is reasonable to assume so, since the tax-setting process is less demanding of information than the regulatory one is. When the need for information is smaller, lobbies have less power. When the discussion is simpler, democratic control is greater. A vote in Congress on taxing pollution has a better chance of not being captured than an obscure regulatory decision made by a captured agency. Even if the agreed-upon tax is too low, at least we face only the cost of pollution, not the cost of pollution plus the waste of taxpayers’ money represented by the cost of the government agency plus the cost of all the distortions to entry that the agency will introduce in the industry in the interest of the incumbents.

The goal of this type of tax (named Pigouvian after Arthur Pigou, the British economist who invented it) is to correct distorted incentives, not to raise revenue. If it does generate revenue, however, we have a double benefit: we improve efficiency and we save the need to impose distortionary taxes to pay for government expenses. The tax on tobacco is an example of a Pigouvian tax, if we consider the cost of Medicare and Medicaid to be a cost imposed on the rest of society.

TWO USEFUL PIGOUVIAN TAXES

A Pigouvian tax that has been proposed but not implemented is one on short-term debt held by financial institutions. One of the lessons of the 2008 financial crisis is that an excessive amount of short-term debt financing by financial institutions can easily transform a relatively small loss into a major crisis. When subprime losses hit financial intermediaries in 2007–2008, short-term lenders, fearing bankruptcy, refused to renew lending. This withdrawal of funds forced the intermediaries to sell assets, depressing prices further. It was similar to a traditional bank run, except initiated by short-term lenders.

So why did the financial intermediaries, which could have anticipated this risk, choose to borrow so much in the short term? Because it allowed them to borrow more and borrow more cheaply, increasing profits. The short-term lenders, meanwhile, felt confident that they could get out of troubled companies in time. But while the exit option is available to each lender individually, it is not available to all lenders together. When all short-term lenders try to leave, they precipitate a crisis. In other words, the incentives to borrow short-term exceed what is optimal from a social point of view.

Some have argued that the best way to solve the problem is to eliminate the debt deduction. However, the real problem is not all debt but short-term debt. In an environment of close-to-zero interest rates, eliminating the debt deduction would have the perverse incentive of favoring short-term debt (on which the lack of deductibility would have almost no impact) over long-term debt.

A better solution is a tax on short-term debt, especially that held by financial institutions. By taxing the use of short-term debt (with maturity of less than a year, for example), we can discourage both excessive leverage and short-term leverage, preventing a crisis. Also, a 1 percent tax on outstanding short-term debt would raise $21.5 billion annually just among the top nine institutions. This is equal to the total amount of taxes raised from the 65 million households making less than $30,000 a year. So this tax could exempt 65 million households from paying taxes, while stabilizing the financial system and preventing a new crisis.

The ultimate Pigouvian tax is a progressive tax on corporate lobbying. Lobbying per se is not bad, but the imbalance in lobbying is. Larger companies lobby disproportionately more and consequently obtain disproportionately greater benefits. Heavily taxing lobbying, and redistributing the proceeds to support the arguments of the more diffused interests, would help level the playing field. It would also reduce the incentive to lobby—both because it would act as a disincentive on the activity being taxed and because each lobbying firm would know that part of its money was being used to support the opposing side. Finally, a heavy tax on lobbying would reduce shareholders’ incentives to pressure their managers to lobby.

THE POLITICS OF TAX AND SUBSIDIES

In principle, government intervention should aim to alter prices in order to achieve some social benefit.⁸ The goal of the intervention is to change incentives, and income redistribution is just a side effect, which, at least in principle, could be eliminated. Our current housing policy, for instance, makes owning relatively cheaper than renting, on the theory that homeownership enhances social stability. Federal student loans reduce the cost of college, on the theory that a highly educated workforce is desirable.

When it comes to nudging people to change their behavior, relative not absolute prices matter. A tax on sugar is equivalent to a subsidy to corn syrup: they both have the effect of increasing corn syrup consumption and reducing that of sugar. From a political point of view, however, this equivalence does not hold. Think about the above subsidies and how they would sell politically if we recast them as taxes. The popular home owner subsidy would become an unpopular tax on home rental; the progressive college subsidy would become a regressive tax on people who do not go to college.

Any careful reader would be quick to point out that there is a big difference between a subsidy to home ownership and a tax on home rental: in the first the government is giving money to people, in the other it is taking it away. Yet, the government is us: it does not create wealth, it simply redistributes it, destroying some of it in the process. Thus a proper comparison between these two policies should keep constant the overall government budget. If the policy is a subsidy, it should be considered in tandem with some other tax that raises the necessary revenues.

In the political debate this compensation never occurs. Ethanol subsidies enrich ethanol producers and gasoline taxes make gasoline consumers poorer. As a result, income redistribution becomes not just a side effect but the goal of the policy intervention. The change in incentives is just the fig leaf that gives some intellectual justification to an otherwise unappealing redistribution. When ethanol producers lobby for a subsidy to ethanol, they are trying to redistribute income in their favor: the benefit to the environment is just a good excuse. They use the social argument to make it appealing. Ideas are very powerful instruments of lobbying, and the most devastating effects of lobbying have occurred when the ideas are most appealing.

This design naturally leads to massive lobbying incentives. The benefit of subsidies is concentrated, while the cost is distributed among taxpayers. Thus lobbying in favor of subsidies will be much stronger than lobbying against it.

INVERTING THE PERVERSE POLITICS OF SUBSIDIES

The political dynamics of Pigouvian taxes is exactly the opposite. The burden of taxation falls on a concentrated and politically powerful group, while the benefit is enjoyed by a dispersed and politically nonin-fluential one: taxpayers. Politically, it is preferable to raise general tax revenues. Since the burden of general taxation falls on a large number of dispersed voters, raising income taxes slightly is easier than imposing a tax on a specific product, such as pollution—even though the former tax is distortionary, inducing people to work less, while the latter eliminates a distortion. This is the reason Pigouvian taxes are so rare.

But this is also precisely why I propose that we ban any form of subsidy and restrict government policy to Pigouvian taxes: these would be few and really needed.

CONCLUSIONS

Taxes can be used to improve the system and raise revenue. Pigouvian taxes in particular can provide a solution to the problem of regulatory capture. They can be used to penalize excessive lobbying, and they can also be used as a substitute for regulation. Because it reduces the information needed to regulate and is comparatively transparent, a system based on Pigouvian taxes is less captured by vested interests—though like any system of taxes, it is not perfect.