II. Taxes and Economic Growth

Taxes impact human behavior. Should I take an overtime job paying $15 per hour? If I am taxed 80 percent of the additional income, my after-tax compensation of $3 an hour may well be too low to make it worthwhile for me to sacrifice leisure time for working. I am happier watching football games on television. In contrast, if the marginal tax rate is 20 percent, and I can keep $12 per hour, I am more likely to prefer working—the opportunity cost of watching football is perhaps too high. Likewise, businesses facing taxes on capital investments might make similar decisions—if marginal tax rates on profits are high, the expected after-tax return may be too low to justify the risk of making the investment, thus increasing the probability that an alternative use of the funds would be more desirable. Thus, taxes can have an adverse impact on the magnitude of labor and capital resources used in making goods. Moreover, such taxes can lower investment and capital formation over the long run, having further long-run consequences on output and income. For example, if income from savings is highly taxed, the reduced supply of savings may lead to higher interest rates, leading to lower investment. Conceivably, technological progress could be impeded as well, as investments in innovation and research and development become excessively risky relative to the after-tax expected rate of return.

However, there are other theoretical reasons to believe that new or increased taxes could conceivably have positive effects on output. Individuals facing higher taxes may feel the need to work more hours to match previous levels of disposable income—the “income effect” of taxes induces more work. Also, tax revenues are spent, and if spent in a way that enhances output, it is possible that the resource-reducing disincentive effects of taxes may be offset by positive effects of government spending and the income effect. Therefore, it is not obvious a priori if taxes will have positive, negative, or neutral effects on the growth in income and wealth.

What does the empirical evidence show? Obviously, such factors as the type of taxes imposed, the uses of funds, and the time horizon studied (short-run versus long-run effects) will have an impact on the tax-growth relationship. Moreover, on a variety of grounds one might expect that the impact of new taxes will depend in part on the magnitude of taxes and the overall size of government. A society with a small government and modest taxes might benefit from higher taxes to fund needed expenditures, such as protection of property rights by improving the effectiveness of the police and justice system. A society with a large government (and already relatively high tax burdens) might find that the gains in economic growth from added taxes are far more than offset by the adverse disincentive effects of the new taxes. With these caveats in mind, a large majority of the studies seen by this author show that new taxes have adverse effects on economic indicators such as income, output, employment, migration, business investment, and plant location.

The overwhelming evidence that taxes impact behavior, often adversely, is demonstrated by a little research done for this essay. I downloaded five working papers dealing with taxes issued in the past year from the website of America's most prestigious economic research organization, the National Bureau of Economic Research. Below are excerpts from the summaries of each study, in reverse chronological order:

Two of the studies (the third and last one cited above) show that high taxes have significant negative effects on the supply of labor, thereby lowering output. Two of the studies (the second and fourth) show that lowering taxes on dividends powerfully increases the amount of dividends, while the remaining study suggests that the elderly rich migrate to low-tax states, lowering tax revenues in high-tax states in the process. Without exception, these studies are consistent with the view that taxes significantly impact human behavior, often in ways that lower income and labor and capital inputs (and thereby output).

Literally scores of other studies reach the same general conclusion. Some of the studies look at different geographic areas (nations or American states) at one moment in time; others look at the same area over different periods in time; still others use questionnaires or other means to assess behavioral impacts. Similarly, alternative economic indicators are used: the growth in income, output, employment, new business start-ups, etc. Some studies find modest-sized relationships between taxes and the economic indicator examined, while others find stronger relationships. Most of them find, however, that taxes on balance have at least some adverse impact on the growth in incomes, output, jobs, business investment, human migration, etc.

A full review of the literature is beyond the scope of this essay, but some examples will strengthen the points made above. In the third quarter of the twentieth century, say around 1960, the prevailing view was that “taxes do not matter” much in economic decision making. Speaking of the business location literature, the distinguished public finance professor John Due said in 1961 that studies “suggest very strongly that the tax effects cannot be of major importance.”⁶

Within two decades, however, the evidence was strongly supporting a different conclusion: taxes matter. In the interim, tax levels had risen appreciably, particularly at the state and local levels in the United States, and econometric techniques and computer advances had led to more and better research. Contrary to Due's 1961 pronouncement, a number of studies concluded that high taxes retarded business location in a given area.⁷ A series of papers using American data appearing at roughly the same time similarly showed that high taxes more generally retarded growth in incomes and output.⁸ Other studies of that era and beyond reached similar conclusions using international data.⁹

Nor have the studies been confined to analysis of income or output growth. Just as Edward Prescott and others have noted that taxes reduce labor effort of those working, other researchers have studied the impact of taxes on the number of people holding jobs or, conversely, the number who are unemployed, most often finding that increased taxation has adverse employment effects.¹⁰

A classic study observed that high property taxation had an adverse impact on property values.¹¹ Other studies have gone beyond location decisions of larger American businesses to argue that taxation has an adverse impact on small business start-ups or foreign investment in the United States.¹² Finally, a large number of studies have shown that taxes have an impact on migration decisions and, therefore, population growth; high-tax areas grow slower than lower-tax ones.¹³

This huge body of evidence, only a small part of which is cited here, is confirmed by less rigorous, more casual observation. Western Europe, which in the 1950s and 1960s grew far faster than the United States, has grown far slower in recent years, as table 1 indicates.¹⁴

While the tax burden showed little change in the U.S., the burden rose sharply in most European countries. In the 1973–2000 period, growth rates were far higher in the U.S. than in its European counterparts, in marked contrast to the earlier period. This relationship, I think, is not coincidental. Low-tax Ireland is the “Celtic Tiger” that leads Western Europe in growth, with real per-capita GDP rising 6.03 percent annually from 1990 to 2001, more than triple the average of 1.72 percent for twenty-nine European nations.¹⁵ At the same time, very-high-tax Sweden has such low growth rates that its per-capita income now is actually below that of most of its Western European industrial peer nations, whereas in 1972 it had easily the highest per-capita income of the European members of the Organization for Economic Cooperation and Development (OECD), exceeding the average of that group of industrialized countries by 83 percent.¹⁶ Russia's economic growth has spurted in the years since it inaugurated a low (13 percent) flat-rate income tax in mid-2000 (effective in 2001).¹⁷ Within the United States, those nine states without general state income taxes have had a massive net in-migration (4,169,268 from 1990 to 2004) of native-born Americans from the forty-one states with such taxes.¹⁸ Low-tax states like New Hampshire have generally had higher rates of growth of income and in-migration of people than higher-tax nearby states such as New York, Vermont, and Maine. In 1970, Ohio had the fourth lowest state and local tax burden in the United States according to the Tax Foundation; by 2004, it had the fourth highest burden, raising its overall tax burden by well over 25 percent when the typical state changed its burden very little. It is probably not a coincidence that in the same period, the Buckeye State went from having a per-capita income level that exceeded the national average to one that was more than 5 percent below that average.¹⁹

Before moving on, it is worthwhile to briefly ponder why the strong negative relationship between taxes and growth exists. The disincentive effects of taxation on savings, investment, innovation, and work effort have already been mentioned. But in the most fundamental sense, taxation takes resources from the relatively productive (in most modern societies with market economies and a strong rule of law) private sector, and diverts those resources to the public sector, which is less productive, since it is less competitive (more monopolistic) and does not face the accountability, the “bottom line,” that markets impose. Private businesses strive for increased profits by becoming more efficient (lowering costs) and offering better products. The profit incentive is absent in most government enterprises.

III. Taxes and Human Welfare—Is There a Relationship?

It is one thing to present evidence that high and increased levels of taxation tend to reduce the levels of income, wealth, and employment, but quite another thing to demonstrate that high taxes lead to reduced levels of happiness or well-being. One fundamental problem is that utility (economist jargon for satisfaction or happiness) cannot be scientifically measured with any great precision—a major reason why, for example, economic historians still debate whether the Industrial Revolution that came to Great Britain was, in some ultimate sense, a success or a failure. Material gains do not necessarily equate with gains in happiness.

Economists say that “cardinal utility”—a precise quantitative measure of the satisfaction derived from an activity or good purchased—is not measurable, whereas “ordinal utility” (as determined by preferences for one activity or good relative to others) is. I can say with reference to new cars, “I prefer a Lexus to a Honda,” but I cannot meaningfully say, “I believe I will be 42.17 percent more satisfied with a Lexus than a Honda.”

There are two major arguments that can be used to suggest that the welfare of the general populace may be improved by large government financed by taxes even if the crowding-out of productive private-sector activity reduces levels of income and wealth. The first argument is that much of what government does has what economists call “positive externalities”—spillover benefits that accrue not only to the user of government-provided services, but to society as a whole. For example, education spending, it is argued, leads to higher literacy, which in turn leads to better communications, lowering the costs of transactions and making the market economy more efficient in allocating resources. Also, a well-educated electorate is more likely to pick leaders who are wise and moral, as opposed to corrupt charlatans. These positive spillover effects are not present, we are told, in most goods and services sold in the private sector.

The second argument states that the material well-being of individuals is related not only to the absolute amount of income or wealth that people have, but to the distribution of that income or wealth. As Karl Marx once said, “A house may be large or small; as long as the surrounding houses are equally small, it satisfies all social demands for a dwelling. But if a palace rises beside the little house, the little house shrinks into a hut.”²⁰

Therefore, it is at least plausible that a society with average household income of $45,000 where the overwhelming majority (say, 85 percent) of households have between $25,000 and $75,000 in income will have a higher level of aggregate happiness, and an absolutely larger number of happier people, than a society with average household income of $50,000, but where there are a large proportion of households (say, 30 percent) with less than, say, $25,000 in income and another good-sized group (say, 15 percent) with more than $75,000. It is argued that income is subject to diminishing marginal utility, meaning that adding $1,000 in income to households who already have $200,000 in income will increase their happiness far less than adding the same $1,000 in income to households with $20,000 in income. Since both taxation and tax-financed government spending can be used to reduce inequalities in disposable (after-tax) income, taxation (particularly progressive taxation that is directed disproportionately to upper-income persons) can serve to increase the well-being of a large majority of the population, causing only modest losses in well-being to a relatively small minority. This is in keeping with the utilitarian ideal, attributable to Jeremy Bentham, of “the greatest good for the greatest number.”²¹

Looking at the first argument, it is extremely hard to measure “positive externalities,” a somewhat imprecisely defined concept that ultimately depends on evaluating the utility derived from governmental activity. One might argue equally that governmental activities sometimes have negative spillover effects or externalities. The government constructs a sewage plant that emits an odor. The residents in the area surrounding the plant see their utility decline—as well as the value of their houses. It is not unambiguously clear a priori whether governmental activities in the aggregate have positive or negative spillover effects. Moreover, private expenditures can have positive spillover effects as well—my neighbor puts up a gorgeous Christmas holiday display in his front yard, which I enormously enjoy. His spending provides utility to me, although I cannot precisely measure it given the inability to measure cardinal utility. In contrast, some private activity has negative externalities—an entrepreneur opens a bar near my house that brings a lot of noise and alcohol-induced reckless driving to my neighborhood. My utility declines. All in all, it is hard to say whether governmental or private activities have more positive or negative spillover effects impacting on human welfare.

Yet there is one form of human behavior that is measurable that does relate quite directly to the issue of positive and negative externalities—human migration. When a person moves from A to B, he or she is revealing a preference for B—in effect saying, “I think life will be better in coming years in B than if I were living in A.” Implicitly, assuming people in general wish to maximize their satisfaction or utility in life, the mover says, “I believe I will derive more utility living in B than living in A.” Migration is an exercise in revealing one's ordinal utility rankings.

If area B has more taxes and governmental services than area A, but similar income levels, job opportunities, climate, etc., it may well be that the move to the high-tax area B reflects the fact that the governmental services provided there have external benefits not reflected in the dollar value of the services provided. If positive externalities result from governmental activity, this should be reflected in in-migration, as people move to acquire the benefits associated with the governmental services. If governmental activity produces net positive externalities, one would expect a positive relationship between the size of government (measured by tax burden or governmental expenditures) and the in-migration of people.

This is a testable proposition. Indeed, as mentioned above, a number of studies have looked at the relationship between government size and in-migration, and have found a negative or nonexistent relationship, inconsistent with the notion that tax-financed governmental activity produces large positive external benefits. I confirmed that directly by using regression analysis to explain variations in net domestic migration among the fifty American states and the District of Columbia over the period from 1990 to 2004. I regressed the tax burden of each state in the middle of that period (as measured by total state and local taxes as a percentage of personal income) against the net number of in-migrants. Doing so produced a statistically significant (at the 5 percent level) negative relationship—persons tended to move away from the high-tax states. Adding some other possible factors into the analysis for control purposes (e.g., average income levels and the number of days the sun typically shines), the relationship remains negative. This finding certainly is inconsistent with the notion that large positive externalities of government spending add significantly to the quality of life.²²

For readers less acquainted with multivariate statistical techniques, I made a simple calculation. I took the ten states that in the middle of the period examined had the highest tax burdens. I found that those states had a net out-migration of 2,976,513 persons from 1990 to 2004. I then compared that with the ten states with the lowest tax burden. They had a net in-migration of 4,199,194.²³

There was a massive migration away from high-tax states—into low-tax ones. This is highly inconsistent with the idea that high-tax states have higher external benefits because of generous governmental services.

I turn now to the second argument discussed above. This argument suggests that economic growth fostered by low levels of taxation might not be welfare-enhancing because the unequal distribution of income or wealth affects well-being. Studies based on average data ignore income distribution, and widening income inequality might lower utility as much as (or more than) rising general income levels raise it. The validity of this scenario, however, is questionable for a variety of reasons.²⁴

To begin with, the thesis that increasing economic equality associated with bigger government has positive welfare effects cannot be evaluated if we cannot agree on the appropriate measures of economic inequality. It is my contention that the measure most often used is inappropriate, and that alternative measures often show a dramatically different picture as to the amount and trends in inequality.

At least three different economic indicators are commonly used in evaluating inequality: personal income, wealth, and consumption expenditures. Of these, income is the most used indicator, consumption the least. Yet it is my contention that consumption is the best of the three indicators for measuring the welfare implications of the inequality of economic conditions in any given time period. People do not derive utility from the income they earn as much as from the income they spend. Income taken from individuals involuntarily in the form of taxes or voluntarily in the form of savings does not provide direct material satisfaction to those individuals, at least in the period in which the individual is taxed or saves. To be sure, savers are, in effect, bribed to forgo consumption by the prospects of receiving future interest, dividend, and capital-gains income. Yet the ultimate goal of economic activity is consumption, not the act of earning the income itself, and that is why people have to be induced (through interest and dividend payments) to save rather than consume.

Data on consumption spending is less readily available than income data, but the Bureau of Labor Statistics of the U.S. Department of Labor has a much respected Consumer Expenditure Survey that provides relatively good consumption data. Table 2 shows consumption data for 2002 classified by income quintiles, along with Census Bureau data on household income similarly classified.²⁵

The data show that consumption spending is far more compressed (equal) than income. The richest 20 percent have four times the share of total consumption compared to the poorest 20 percent, while the richest 20 percent of households have more than fourteen times as much income. While the poorest Americans have only 3.5 percent the nation's income, they are responsible for 9 percent of its consumption, a share two and a half times as large as their share of income. The rich save or pay taxes equal to large portions of their income, while the poor actually dissave significantly, spending far more than they earn in income.

The major reason why the distribution of consumption spending is only roughly half as unequal as the distribution of income is that consumption spending is heavily influenced by expected lifetime income, not just income receipts in a particular twelve-month period. Americans have extraordinary income mobility, moving up and down the economic ladder frequently and sometimes abruptly. A “poor” person one year may be a financially average American two years later, and in the following decade might be rich. The “poor” that the government might try to subsidize this year may be the “rich” that they “soak” through high taxes next year.

The data demonstrating high income mobility are striking. Treasury Department data from 1992 showed that 86 percent of the poorest Americans in 1979 (those in the bottom 20 percent of the distribution) moved to a higher quintile in the income distribution in the next nine years.²⁶

An Urban Institute study suggests that more than one-quarter of Americans change income quintiles (up or down) from one year to the next, and a majority do so over intervals of a decade.²⁷ Other studies done at the Federal Reserve Bank of Dallas and the Economic Policy Institute reach similar conclusions.²⁸

To the extent that incomes tend to equalize over time, a case can be made for public tax and spending policies that are relatively neutral toward the distribution of income, since single-year income differences are of little long-term meaning in terms of life income equality. While no one claims that incomes perfectly equalize over time, lifetime income differentials are sharply smaller than the single-year differentials most often cited.

This brings us to another problem with the arguments that tax-financed government expenditures might be on balance utility-enhancing. Suppose that, despite the argument made above, it were concluded that promoting income equality is utility-enhancing on balance. Such conclusions could not be made with any confidence based on empirical evidence, particularly given the extreme difficulty of measuring utility. But even if one were to accept the notion that income inequality is an obstacle in maximizing the community's welfare (or utility), what assurances are there that tax-financed government spending would reduce that inequality? As the economist Gordon Tullock pointed out more than two decades ago, large amounts of government spending are essentially middle-class or even upper-class entitlements—subsidies for pro football stadiums, National Public Radio, and even farmers come to mind, not to mention corporate welfare such as the Export-Import Bank.²⁹

It is even arguable that high public assistance grants to lower-income Americans beginning in the 1960s and 1970s actually had such adverse effects on behavior (by lowering work effort and leading to the disintegration of two-parent families) that they actually increased poverty rather than reduced it, worsening the income distribution.³⁰ The cutback in welfare benefits after federal welfare reform in the mid-1990s led to dramatically lower welfare rolls and falling rates of poverty. The poverty rate, for example, fell from 13.8 percent in 1995 (the year before the welfare reform) to 11.3 percent in 2000.³¹ During the same period, federal current expenditures fell from 21.29 to 18.60 percent of gross domestic product.³² Falling poverty was associated with declining government (in a relative sense). From 2000 to 2003, there was a marked increase in relative government spending—and the poverty rate rose—to 12.5 percent. While the 2001 recession was a contributing factor, it would be hard to argue that increased government spending reduced poverty.

The notion that government spending might increase poverty (and increase income inequality) is hardly new: in 1848, Frédéric Bastiat opined that the state “produces more poverty than it cures.”³³

Lowell Gallaway, David Sollars, and I have presented strong empirical evidence that suggests that tax-financed income transfers from government do reduce inequality when administered in moderation, but actually increase it when payments reach the high levels common in the modern welfare state.³⁴ Thus, the assumption that expansive government leads to greater income equality through poverty reduction is, at the very least, highly questionable.

As a general proposition, does economic growth (whether stimulated by lower taxes or other means) generally lead to greater or lesser amounts of income inequality? The most famous answer to that question comes from the painstaking work of economist Simon Kuznets (who netted a Nobel Prize for his efforts). Kuznets suggested, based on examining empirical evidence, that economic growth among poor nations may generate greater inequality, but that, as nations get wealthier, growth is associated with lesser inequality.³⁵

While the existence of such a “Kuznets curve” has been debated, there is a substantial body of evidence that supports Kuznet's contention.³⁶ In the context of developed nations like the United States or the nations of Western Europe, this suggests that fiscal policies promoting economic growth are, on average, likely to promote greater income equality. The income distribution argument against lower taxes is further weakened.

In short, arguments used to justify tax-financed expansions of government on the grounds that they promote the general welfare and human happiness appear to me to be largely statements of faith not justified by the realities of the modern world. To the extent that empirical evidence speaks to this issue at all, it suggests that Americans prefer to live in relatively low tax areas where economic prosperity is enhanced by the relatively less fettered operation of market processes.

IV. Markets and the Elimination of Spatial Income Inequality

As the foregoing discussion suggests, it is an unproven hypothesis that reductions in income inequality are associated with advances in human satisfaction, and, even if it were proven, governments have a dubious track record in promoting equality. Yet, given the difficulties of measuring utility, it is still an unsettled question whether happiness is promoted by reducing the gap between the rich and the poor. The fact that people tend to migrate over time from poor to rich areas (for example, far more persons migrate from India, China, or Mexico to the United States than in the opposite direction) suggests that the acquisition of income is perceived to be utility-enhancing. If low-income countries have income gains relative to higher-income ones, they should witness some utility gains similar to what happens with migration from low- to high-income areas. Because of this, a case can be made, at least on a global level, for reducing income inequality (by having higher economic growth in lower-income countries relative to more affluent ones). The really huge disparities in income are not found within nations like the United States, but between rich and poor countries.

Table 3 documents these large international income disparities. Taking the average income levels in seventy-six countries for which data were available for 2002, I have treated each country as if it were an individual. I have then compared the quintile shares of the total income of these seventy-six individuals (actually, countries) with the shares for individuals living in the United States in the same year. The data show that the richest 20 percent of countries have thirty times the proportion of total income as the poorest ones do, more than double the differential between the rich and poor for individuals within the U.S. The rich nations produce a larger proportion of total world income (so measured) than rich individuals do of total U.S. income. International inequality is greater than interpersonal inequality within the United States. Moreover, for a variety of technical reasons, the statistics in table 3 probably understate international inequality. Purchasing-power-parity income data are used instead of data based on exchange rates, which would show even greater international inequality, for example. Many nations are excluded, a majority of which are poor (e.g., Haiti). Since many poor countries are populous (e.g., India), a weighted (by population) table would probably show even more inequality than is indicated here.

It is arguable that the world would be a better place if these wide disparities were reduced. In a world with a strong rule of law and property rights, no significant barriers to moving people or resources, and only minimal governmental interferences and involvement in the economy, economists would predict that international differences in average income levels would be drastically reduced. In such a world, capital would migrate from high- to low-wage countries to minimize labor costs, while labor would migrate from low- to high-wage areas to increase worker earnings. The two moves would serve to equalize differences in the amount of capital available per worker, and thereby also help to equalize incomes. This basic framework was most explicitly developed in a model by Robert Solow and was an important factor in his receiving the Nobel Prize in Economic Science.³⁷

The convergence hypothesis has been strongly supported in important empirical work using both U.S. state and international data by Robert Barro and Xavier Sala-i-Martin.³⁸ Dozens of empirical studies in general confirm the existence of convergence, although often with some qualifications (e.g., during periods of financial crisis or wars, convergence may be interrupted).³⁹ Yet other scholars, using different data sets (different countries or regions, alternative time periods) or methodologies, find no evidence of convergence, or, at best, mixed evidence.⁴⁰

The reasons for this somewhat mixed evidence are explained by tables 4 and 5. In table 4, personal income per capita for seven states scattered around the U.S. is expressed as a percentage of the level prevailing in New York, the most populous and one of the most prosperous states at the beginning of the period examined. In every single state, a majority of the percentage income differential with New York was erased from 1929 to 2003—strong evidence of income convergence. In a majority of the states, income converged on the New York figure at every date included in the chart. Southern and border states like Georgia and Tennessee had less than one-third the per-capita income level of New York in 1929, but more than three-fourths that level by 2003. In Georgia, for example, 72.3 percent of the differential with New York was eliminated. To be sure, convergence was not always continuous or complete, but it was still substantial and real. Note that in California and Illinois income levels stopped converging on New York (at almost New York's level) around 1950 and actually have diverged slightly since. Convergence, then, does not appear to be complete, but the sharp narrowing of regional income differentials is still one of the most important dimensions of twentieth-century American economic history.

The international data, however, are far more mixed. Only three of the seven nations shown in table 5 show any convergence with the U.S. over the very long run (1913 to 2001), with two of them (Japan and Italy) showing the classic convergence similar to that observed by states in the U.S., and one, China, showing very limited convergence. Three nations (Great Britain, India, and Mexico) actually diverged significantly from the U.S. over time, while the seventh nation, Germany, showed little change (modest divergence). In short, table 5 is a mixed bag of results not very strongly supporting the convergence hypothesis.

Yet an examination of tables 4 and 5 suggests a reason for these diverging patterns. In the U.S., there is a relatively uniform rule of law, protection of property rights, a regulatory regime that is moderate by world standards, and taxation that is low by the standards of modern welfare states. In that environment, the forces of markets work to greatly reduce geographically-related income inequality over time. In nations where those characteristics (e.g., the rule of law, low taxation, minimal regulation, etc.) have been absent (e.g., Mexico, India, or until the past generation, China) divergence occurs, as it does when wars (a form of massive government spending) absorb a large portion of national resources (see, e.g., the German, British, and Japanese divergence in the 1913–1950 period in table 5). Labor resources are not free to leave and capital resources have no desire to migrate into authoritarian societies where there are few if any private property rights. Some recent empirical work confirms that increased government taxes and spending have adverse effects on economic growth in general (in keeping with the discussion in Section II of this essay) and on convergence in particular.⁴¹

Using American state data, I looked at the role of state and local tax burdens in income convergence over the period from 1970 to 2003, years for which good tax data are available.⁴²

I identified fifteen states that could be viewed as convergence success stories. Each of these states had below-average per-capita incomes in 1970, and by 2003 had closed the gap with the average by at least 30 percent, or raised its income as a percent of the national average by at least 10 percentage points (e.g., from 63 to 73 percent).⁴³ I then identified ten states that diverged from the average, starting with below-average per-capita incomes but falling even further below that average from 1970 to 2003.⁴⁴ A majority of the successful convergence states lowered the tax burden from 1970 to 2003. For the group as a whole, the tax burden fell on average by $2.80 for each $1,000 in income earned. By contrast, in the less successful, diverging states, a majority (60 percent) raised their tax burden, increasing it on average for the group by $1.50 for each $1,000 in personal income. While I would certainly not want to claim that “taxes alone matter,” tax-reducing states were more likely to have a successful story than tax-raising ones.

Income equalization between areas is a by-product of convergence. High-tax welfare states and authoritarian societies that reject market solutions to problems have been slow to converge on the most affluent societies like the United States, where the discipline of the market plays a dominant resource-allocation role—witness the end to convergence in Britain and Germany after episodes of large growth in government (in the 1930s and 1940s in Britain, after 1973 in Germany). Not only are market solutions more conducive to the growth of incomes and prosperity, they are also more powerful and successful in equalizing economic opportunities across time and space. It may seem ironical to some readers that greater income equality can be achieved with less government, less taxation, and less redistribution—but that is what the evidence suggests. Well-intended government policies often have behavioral effects that lead to unintended consequences.