Taxes

Lane Kenworthy, The Good Society
May 2023

Hardly anyone loves taxes. Yet most of us recognize that, as Oliver Wendell Holmes once put it, taxes are the price we pay for a civilized society.1 We want a government that performs a variety of services, and in order to have one we need to fund it. Moreover, as societies get richer, citizens tend to be willing to allocate more income to ensure economic security and to enhance fairness and opportunity for the less advantaged. That means government spending, and hence taxes, will tend to rise over time.

How much do Americans pay in taxes? Through what types of taxes is the money raised? Do taxes reduce income inequality? Do they harm the economy? If the US were to increase taxes, what would be the best way to do it?

Skip to:

TAX REVENUES

Figure 1 shows tax revenues and overall government revenues in the United States since 1900. With the introduction of the federal income tax in 1913, government revenues began a steady rise, from 8% of GDP to a peak of 35% in 2000. Since then they have been flat. Tax revenues, meanwhile, have hovered around 25% of GDP for the past 50 years.

Figure 1. Tax revenues and total government revenues in the United States
Share of GDP. Includes all types of taxes (income, payroll, consumption, property, others) at all levels of government (federal, state, local). Data source for 1900-1964: Christopher Chantrill, usgovernmentrevenue.com, using data from the Census Bureau and the Economic Report of the President. Data source for 1965ff: OECD.

How does the US compare to other affluent nations? Figure 2 shows trends in tax revenues since the mid-1960s. Half a century ago America’s taxation level was in the middle of the pack. It’s now at the bottom.

Figure 2. Tax revenues
Share of GDP. Includes all types of taxes at all levels of government. Doesn’t include nontax sources of government revenue. Data source: OECD. Thick line: United States. “Asl” is Australia; “Aus” is Austria.

Despite the country’s comparatively low tax level, and despite the fact that that level hasn’t changed for the past 50 years, a majority of Americans say they’d like to pay less, as we see in figure 3. Virtually none feel they currently pay too little.

Figure 3. Many Americans would prefer to pay less in taxes
Estimated share of American adults. Question: “Do you consider the amount of federal income tax you have to pay as too high, about right, or too low?” Gallup also has a “no opinion” response, which is excluded here. Data sources: Gallup, “Taxes,” gallup.com/poll/1714/taxes.aspx; General Social Survey, sda.berkeley.edu, series tax.

Why did tax revenues rise in most rich democratic countries in the 1960s, 1970s, and 1980s?

Among the world’s affluent democracies, tax revenues rose from 26% of GDP in 1965, on average, to 35% in 1990. The percentage point increase in Belgium, Denmark, Finland, France, Norway, and Sweden was about twice as large as in Australia, Canada, Japan, New Zealand, Switzerland, and the United States. Why did tax revenues rise during this period, and why did they grow more rapidly in some countries than in others?

One obvious possibility is that the quantity of revenues raised is determined by the quantity of government spending a country wants. As the rich nations expanded public insurance programs during this period, they increased taxation in order to pay for it.

An alternative hypothesis holds that a country’s ability to increase revenue depends on its tax mix. The nations that expanded taxation most rapidly in the 1960s, 1970s, and 1980s, in this view, were able to do so by relying on increases in consumption and payroll taxes rather than income taxes.2 Taxes on consumption and payroll are less visible to citizens — they tend to be paid in small increments, and there is no end-of-the-year settling up where we see the grand total — so they generate less political opposition. Also, income taxes may be more vulnerable than consumption and payroll taxes to cross-national competition. That is particularly true for corporate income taxes, since firms are more mobile than people, but individuals too are thought to be more likely to decamp in search of a lower income tax rate than a lower tax rate on consumption or payroll.3

Figure 4 shows over-time trends in total tax revenues, income tax revenues, and consumption-plus-payroll tax revenues in the rich longstanding democracies. Contrary to the prediction, income tax revenues increased at about the same pace as revenues from consumption and payroll taxes.

Figure 4. Tax revenues
Share of GDP. Average for 21 countries. Includes all levels of government. Data source: OECD.

Figure 5 offers another way to assess the hypothesis that consumption and payroll taxes were the key to expanded tax revenues in the 1960s, 1970s, and 1980s. If the hypothesis is correct, countries that increased tax revenues as a share of GDP during that period should exhibit a rising share of total tax revenues that come from consumption and/or payroll taxes. That’s not, however, what we see in the data. Sweden conforms to the prediction: consumption and payroll taxes rose significantly as a share of its tax revenues (horizontal axis), and its increase in total tax revenues was quite large (vertical axis). But Sweden is the exception, not the rule.

Figure 5. Change in total tax revenues by change in the consumption-and-payroll share of tax revenues, 1965 to 1990
Percentage point change, calculated as 1990 value minus 1965 value. Data source: OECD. “Asl” is Australia; “Aus” is Austria.

Together, figures 4 and 5 tell us that rising income taxes were as important as consumption and payroll taxes in facilitating the increase in tax revenues in the world’s affluent nations in the 1960s, 1970s, and 1980s.4

Why have tax revenues been flat since 1990?

Since 1990, tax revenues (as a share of GDP) in virtually all rich democracies have barely budged. After rising steadily for much of the twentieth century, and particularly in the 1960s, 1970s, and 1980s, they suddenly came to a standstill. Why?

Unions and left political parties were a key reason why government spending, and therefore taxation, grew more rapidly in some nations than in others leading up to the 1990s.5 They are less likely to be a major determinant of why taxation stopped rising, because union strength and the political influence of left parties continued to vary sharply across countries whereas the flatlining of tax revenues happened in nearly all.

The 1980s commenced an ideological shift among parties on the right, led by the Thatcher government in the UK and the Reagan administration in the US, in favor of aggressive tax reduction. Before long, a number of center-left parties had altered their thinking about the size of government. In the United States, the Democratic president Bill Clinton declared in 1996 that “the era of big government is over.” Yet here too, we observe considerable cross-country variation. To note just one example, the Blair and Brown governments in the United Kingdom, often viewed as kindred “third way” or “neoliberal” spirits to the Clintonites, oversaw a significant expansion of government spending during their three terms in office between 1997 and 2010. And in the United States tax revenues had already stopped rising by the mid-1960s, long before the Democratic Party’s neoliberal turn.

Public opinion shifted against further increases in taxation in some countries, and antitax social movements cropped up. In the United States, a revolt against property taxes succeeded in passing state-level referendums or legislation limiting such taxes. But this, too, varied significantly across countries.6

Financial globalization — the growing ability of firms and individuals to move their money from their home country to one with lower taxes — is likely the most important cause of the nearly-universal flatlining of tax revenues beginning around 1990. This significantly increases the possibility of gradual or rapid capital flight. In response, nearly all rich countries have reduced income tax rates (see below), though most have offset this by also reducing tax deductions and loopholes (“broadening the tax base”).7

How big a problem is tax evasion?

Research on tax cheating suggests it is largely a function of incentives rather than culture or individual virtue.8 When a person’s income is reported to the tax authority by a third party, such as an employer or financial company, the incidence and magnitude of tax cheating tends to be very low. It increases when income is self-reported, as is the case for many people who are self-employed. In Denmark, the tax evasion rate among those with only or mainly self-reported income is about 50%. In the United States it is similar, at about 56%.9

TAXES’ EFFECT ON THE DISTRIBUTION OF INCOME

It’s helpful to think about the distribution of income in two stages. One is the distribution of wages and other sources of “market” income. The other is government redistribution. Taxes can play a role in both.

Taxes and redistribution

Taxes can contribute to income redistribution in two ways. The first is direct: if the tax system is progressive, the posttax distribution of income will be less unequal than the pretax distribution. The second is indirect: tax revenues provide the funds for government transfers, and those transfers may reduce income inequality.

Let’s begin with taxes’ direct redistributive effect. If households with high pretax incomes pay a larger share of their income in taxes than do those with low incomes, we call the tax system “progressive.” If the rich and poor pay a similar share of their incomes, the tax system is termed “proportional.” If the poor pay a larger share than the rich, the tax system is “regressive.”

Income taxes almost always are progressive; those with higher incomes pay at higher rates. Deductions and exemptions reduce the degree of progressivity, but they don’t eliminate it. Consumption taxes and payroll taxes usually are regressive.10

Typically they are levied at a flat rate, which in principle should make them proportional. But the poor (by necessity) tend to spend a larger share of their income than the rich, which means a larger share of their income is subject to consumption taxes. And payroll taxes often are capped, with earnings above the cap not subject to the tax. This means a larger portion of the earnings of low and middle earners is subject to payroll taxes.

Most studies of taxes’ direct effect on the distribution of income use data that don’t, unfortunately, include the top 1%. The data come from household surveys, which “topcode” the information they get about incomes.11 These data provide us with information about the effect of taxes on income distribution within the bottom 99%. What they tell us is that, for this group, taxes tend to have little direct impact on the distribution of income. Figure 6 shows the reduction in the Gini coefficient for income achieved by income and payroll taxes in eighteen countries. In most of these countries taxes achieve only a little direct income redistribution. And the amount they achieve is overstated in this figure, because the data don’t include consumption taxes, which are regressive.12

Transfers, by contrast, do achieve considerable reduction in income inequality, albeit less in the United States than in nations such as Sweden, Belgium, and Denmark.13

Figure 6. Income inequality reduction via taxes and via government transfers
The data are for the mid-2000s. The calculation for taxes doesn’t include consumption taxes. Neither measure includes data for the top 1% of households. Income inequality reduction via taxes: Gini coefficient for posttransfer-pretax income minus Gini coefficient for posttransfer-posttax income. Income inequality reduction via transfers: Gini coefficient for pretransfer-pretax income minus Gini coefficient for posttransfer-pretax income. Incomes are adjusted for household size. In calculating the Gini coefficients, households are ranked according to their posttransfer-posttax income. Data source: OECD, Growing Unequal, 2008, figure 4.6. “Asl” is Australia; “Aus” is Austria.

Countries achieving the most income redistribution do so mainly via transfers. And most of them do so with a large volume of transfers, rather than with transfers that are heavily targeted to the poor.14 In order to transfer large quantities of income, a government needs a large quantity of revenues. Figure 7 shows that there is a strong correlation between tax revenues as a share of GDP and the amount of redistribution a nation achieves via government transfers.

Figure 7. Income inequality reduction via government transfers by tax revenues
Mid-2000s. Tax revenues: Share of GDP. Includes all types of taxes at all levels of government. Data source: OECD. Income inequality reduction via transfers: Gini coefficient for pretransfer-pretax income minus Gini coefficient for posttransfer-pretax income. Incomes are adjusted for household size. In calculating the Gini coefficients, households are ranked according to their posttransfer-posttax income. Doesn’t include the top 1% of households. Data source: OECD, Growing Unequal, 2008, figure 4.6. “Asl” is Australia; “Aus” is Austria. The line is a linear regression line. The correlation is +.69.

Rich nations have discovered that in order to get a large quantity of tax revenues, it’s best to tax everyone — not just the rich — fairly heavily. That’s partly because a not-too-progressive tax system will generate less political opposition from the rich, and it’s partly due to the need to go where the money is. So the rich nations that redistribute the most tend to have relatively proportional tax systems and high tax rates. This generates a large volume of revenues, which is used to fund a large volume of transfers, and those transfers accomplish the bulk of the income redistribution.15

Do we see an effect on the level of income inequality? Or is inequality mainly a function of the “market” distribution, with redistribution having just a minor impact? As figure 8 shows, countries with higher taxes revenues do indeed tend to have lower levels of income inequality. Part of this owes to differences in pretransfer-pretax inequality, but the redistributive transfers highlighted in figures 6 and 7 play an important role.

Figure 8. Income inequality by tax revenues
Mid-2000s. Tax revenues: Share of GDP. Includes all types of taxes at all levels of government. Data source: OECD. Income inequality: Gini coefficient for posttransfer-posttax income. Incomes are adjusted for household size. Doesn’t include the top 1% of households. Data source: Standardized World Income Inequality Database. “Asl” is Australia; “Aus” is Austria. The line is a linear regression line. The correlation is -.73.

As noted earlier, the estimates of redistribution and income inequality in figures 6-8 leave out the top 1%. Would the picture change if the top 1% were included? Are top incomes taxed at a much higher rate, increasing the progressivity of the tax system?

Figure 9 suggests that the answer probably is “no.” It shows average effective tax rates in the United States at various points along the pretax income distribution, including the top 1%. (An “effective tax rate” is calculated as taxes paid divided by pretax income.) These tax rates include all types of taxes at all levels of government. The top 1% pay about 34% of their pretax income in taxes. The overall tax system is progressive, though only mildly so.

Figure 9. Effective tax rates, United States
Taxes paid as a share of pretax income. The tax rates are averages for the following groups: p0-20, p20-40, p40-60, p60-80, p80-95, p95-99, p100 (top 1%). Includes all types of taxes (personal and corporate income, payroll, property, sales, excise, estate, other) at all levels of government (federal, state, local). The data are for 2020. Data source: Steven Wamhoff and Matthew Gardner, “Who Pays Taxes in America in 2020?,” Institute on Taxation and Economic Policy, 2020, figure 2.

Figure 10 gives us another way to answer the question. It shows the difference between the top 1%’s pretax income share and its posttax income share in the United States. The difference has been relatively small throughout the past several decades, suggesting, consistent with the picture in figure 8, that taxes on the top 1% do relatively little redistributive work.

Figure 10. Top 1%’s income share before and after federal government taxes, United States
Includes capital gains. Federal taxes only; doesn’t include state and local tax payments. Data source: Congressional Budget Office, “The Distribution of Household Income and Federal Taxes, 2013,” Report 51361, 2016, supplemental data, worksheet 3.

Figure 11 shows that the same is true for the top 0.01%. Even among these households with very high pretax incomes — about $35 million on average in 2005, according to the Congressional Budget Office — taxes do little to redistribute.

Figure 11. Top 0.01%’s income share before and after federal government taxes, United States
Includes capital gains. Federal taxes only; does not include state and local tax payments. Data source: Congressional Budget Office, “Historical Effective Tax Rates, 1979 to 2005: Supplement with Additional Data on Sources of Income and High-Income Households.”

What about cross-country differences in tax progressivity with the top 1% included? Unfortunately, for most other rich nations we don’t have comparable data. But figure 12 shows a somewhat-comparable set of effective tax rates for Sweden. The main source of noncomparability is that these Swedish rates leave out consumption taxes, which are sizable in Sweden and thus an important omission.

To help with comparability, I include another set of effective tax rates for the United States that leaves out state and local taxes such as consumption (sales) taxes. For the US, the effective tax rates that include consumption taxes (black dots) are much less progressive than those that don’t include consumption taxes (gray dots). From this we can infer that Sweden’s tax system, including its consumption taxes, probably is a good bit less progressive than the rates shown here imply.

Figure 12. Effective tax rates, United States and Sweden
Taxes paid as a share of pretax income. The tax rates are averages for the following groups: p0-20 (US only), p0-40 (Sweden only), p40-60, p60-80, p80-90, p90-95, p95-99, p100 (top 1%). United States data: 2015. Includes all types of taxes (personal and corporate income, payroll, property, sales, excise, estate, other) at all levels of government (federal, state, local). Data source: Citizens for Tax Justice, “Who Pays Taxes in America in 2015?,” using data from the Institute on Taxation and Economic Policy. United States (w/o state and local) data: 2011. Includes federal personal and corporate incomes taxes, payroll taxes, and federal excise taxes. Data source: Congressional Budget Office, “The Distribution of Household Income and Federal Taxes, 2011,” data set, alternative income definition, worksheet 13. Sweden (w/o consumption) data: 2009. Excludes consumption taxes. Ages 20-64 only. Data source: Niklas Bengtsson, Bertil Holmlund, and Daniel Waldenström, “Lifetime versus Annual Tax and Transfer Progressivity: Sweden, 1968-2009,” 2014, table 1.

The data in figure 12 suggest, then, that effective tax rates in Sweden are higher across the board than in the United States and that Sweden’s tax system is less progressive. This is consistent with the picture painted by the household survey data that don’t include the top 1%.

If other rich nations aren’t too different from Sweden and the US, then the conclusions we can draw about taxes and redistribution from household survey data also hold for data that include the top 1%. Those conclusions, to reiterate, are that taxes have little direct redistributive effect, because they aren’t especially progressive, and that taxes help with income redistribution to the extent they provide a large volume of revenues, which can be used to fund a large quantity of government transfers.

Taxes and the pretax income distribution

Taxes can influence the “market” distribution of income too. One prominent hypothesis suggests that when top statutory income tax rates are lower, people and households at the top have greater incentive to try to maximize their income. They may do so by working harder or smarter, or perhaps by grabbing more “rent.”16

Figure 13 shows over-time trends in the United States for the top statutory federal income tax rate and the top 1%’s share of pretax income. Consistent with the hypothesis, the trends tend to move in opposite directions. In the 1920s the top tax rate decreased and the top 1%’s income share shot up. The top tax rate rose sharply between 1929 and 1945, and the top 1%’s income share fell sharply. After 1981 the top tax rate decreased a good bit and the top 1%’s income share jumped. The correlation over the full period is -.69.

But there are notable exceptions. The 1963 Kennedy tax reform reduced the top statutory tax rate from 90% to 70%, yet the top 1%’s pretax income share continued its slow, steady post-World War II decline. In the early 1990s the (first) Bush administration and the Clinton administration increased the top tax rate from 28% to 40%, yet the top 1%’s market income share continued its sharp post-1979 rise. The (second) Bush administration’s cuts in the early 2000s and the Obama administration’s increase beginning in 2013 also appear to have had little or no impact.

Figure 13. Top statutory income tax rate and the top 1%’s income share, United States
Data sources: Economic Report of the President; World Wealth and Income Database.

Carola Frydman and Raven Molloy have looked closely at whether compensation for top executives in large US firms changes in response to shifts in top statutory tax rates. Drawing on data going back to the 1940s, they find no noteworthy correlation between top tax rates and executive compensation.17

What does the experience of other countries suggest? The appendix has time plots for 17 other rich nations, with data beginning in the mid-1970s. In some of them — Australia, Canada, Ireland, New Zealand, Norway, Portugal, and the UK, along with the US — we observe the predicted increase in the top 1%’s income share when the top statutory tax rate decreases. But in others — Denmark, France, Italy, Japan, the Netherlands, Spain, and Sweden — we don’t.

What does the cross-country pattern tell us? Figure 14 shows the top 1%’s pretax income share as of 2007 by the average top statutory income tax rate over the years 1979 to 2007. There is a negative association: the share of income going to the top 1% is lower in nations with higher top statutory tax rates.

Figure 14. Top 1%’s pretax income share in 2007 by top statutory income tax rate in 1979-2007
Data sources: World Wealth and Income Database; OECD. “Asl” is Australia. The line is a linear regression line. The correlation is -.46.

Yet there are many other contributors to the cross-country variation in top-end income inequality. One way to get closer to identification of a true causal effect is to examine whether changes in top tax rates are associated with changes in top income shares. All of these countries reduced top income tax rates during this period, but they differed significantly in the degree of reduction. Did the nations with larger decreases in top tax rates experience larger increases in their top 1%’s income share? As figure 15 indicates, here too the association is negative, but it is weaker and it disappears entirely if Denmark and the US are excluded. Particularly noteworthy is that four English-speaking countries — the US, the UK, Canada, and Australia — are among those with largest increase in the top 1%’s income share even though only one of them, the United States, enacted very large tax rate reductions.

Figure 15. Change in top 1%’s pretax income share by change in top statutory income tax rate, 1979 to 2007
Data sources: World Wealth and Income Database; OECD. “Asl” is Australia. The lines are linear regression lines. The solid line includes all 18 countries; the dashed line excludes Denmark and the United States. The correlation is -.36 for all countries and -.05 with Denmark and the US excluded.

Why isn’t the association stronger? Part of the reason, surely, is that statutory tax rates aren’t necessarily a good proxy for effective tax rates. Hiding behind statutory rates are an assortment of loopholes, deductions, and “tax expenditures.” These reduce the effective tax rate on persons or households with high incomes by shielding some, potentially much, of their pretax income from taxation. Warren Buffett’s famous discovery that he pays a lower effective federal income tax rate than his office staff illustrates the point. Moreover, different parts of high incomes — salary, business income, capital gains — may be taxed at different rates.18

Figure 16 gives a sense of this for the US case. It shows the top statutory federal income tax rate along with the effective federal tax rate paid (on average) by the top 1% of taxpayers. The top statutory rate dropped sharply in the early 1960s and in the 1980s. The 1%’s effective rate fell too, but far less.

Figure 16. Top statutory federal income tax rate and effective federal tax rate on the top 1%, United States
Effective rate on top 1%: tax payments as a share of pretax income for the top 1% of taxpayers. Federal taxes include personal income, corporate income, payroll, and excise. Data source for the top (black) effective tax rate line: Thomas Piketty, Emmanuel Saez, and Gabriel Zucman, “Distributional National Accounts,” Quarterly Journal of Economics, 2018, figure 9. Data source for the lower (gray) effective tax rate line: Congressional Budget Office, “The Distribution of Household Income, 2019,” Data Underlying Exhibits, exhibit 11.

On the other hand, Kenneth Scheve and David Stasavage have assembled estimates of effective tax rates paid by the top 0.01% in five other countries — Canada, France, the Netherlands, Sweden, and the UK — and they conclude that these effective rates have tended to move in sync with the top statutory rate.19

Another reason why the association in figure 15 isn’t especially strong may be that other institutions affect the degree to which those at the top are able to grab more income in response to lower tax rates. This might help account for why the large 1963 Kennedy tax cut in the United States had little or no impact on the top 1%’s income share: perhaps unions or large-firm norms were strongly opposed to high compensation levels for executives, so CEOs who wished to sharply increase their compensation were unable to do so.

Figure 17 offers some reason to think this is correct. It replicates figure 15 for the preceding period: 1960 to 1979. Here we see essentially no association between changes in top statutory income rates and changes in the top 1%’s share of pretax income.

Figure 17. Change in top 1%’s pretax income share by change in top statutory income tax rate, 1960 to 1979
Data sources: World Wealth and Income Database; Thomas Piketty, Emmanuel Saez, and Stefanie Stantcheva, “Optimal Taxation of Top Incomes: A Tale of Three Elasticities,” American Economic Journal: Economic Policy, 2014, online dataset. “Asl” is Australia. The lines are linear regression lines. The solid line includes all 14 countries; the dashed line excludes the United States. The correlation is -.27 for all countries and -.13 with the US excluded.

If the top statutory tax rate does affect pretax income inequality between the top 1% and the bottom 99%, does that carry over to posttax inequality? The answer is likely to be yes over time in the United States, as the trend in posttax top-end inequality, shown in figure 10 (also figure 11) above, has been driven almost entirely by the trend in pretax inequality. For understanding cross-country differences, we have no way to judge, as comparable data for the top 1%’s posttax income share in other nations aren’t available.

TAXES’ EFFECT ON THE ECONOMY

A key worry about taxation is that it weakens the financial payoff to work effort, investment, innovation, and entrepreneurialism, and may thereby reduce economic growth. Yet the experience of the world’s rich countries over the past century suggests little if any negative impact of higher taxes on growth performance.20

Peter Lindert has suggested that this may be a function of the tax mix. Countries with higher tax revenues may rely disproportionately on consumption taxes, and these, he says, create less in the way of investment and work disincentives than do taxes on individual and corporate income.21 Can we see this effect in the cross-country data? Figure 18 plots (catchup-adjusted) economic growth from 1979 to 2007 by consumption taxes’ share of total tax revenues. Ireland appears to fit the hypothesis nicely, but Ireland’s growth experience during these three decades was exceptional in so many respects that attributing it to the country’s tax mix would be dubious. If Ireland is discounted we see no association. The same is true if the quantity of total tax revenues is controlled for.

Figure 18. Consumption tax share and economic growth, 1979 to 2007
Economic growth: average annual rate of change in GDP per capita, adjusted for initial level (catch-up). Data source: OECD. Consumption tax share: consumption tax revenues as a share of total tax revenues. Data source: OECD. “Asl” is Australia; “Aus” is Austria. The line is a linear regression line, calculated with Ireland and Norway excluded. The correlation is -.38 (with Ireland and Norway excluded).

What about employment? Here research has tended to suggest reason for worry about heavy use of payroll taxes. The concern has to do mainly with the way in which these taxes raise the price of labor in low-productivity services. Fritz Scharpf put the point as follows: “The negative impact on service employment is particularly acute in those countries which, like Germany and France, rely to a large extent on payroll taxes for the financing of the welfare state…. In Germany, these contributions presently amount to about 42% of the total wage paid by the employer…. If the net wage of the worker cannot fall below a guaranteed minimum [the level of unemployment benefits and social assistance], the consequence is that any social insurance contributions, payroll taxes, and wage taxes that are levied on jobs at the lower end of the pay scale cannot be absorbed by the employee but must be added to the total labor cost borne by the employer…. As a consequence, a wide range of perfectly decent jobs, which in the absence of payroll taxes would be commercially viable, are eliminated from the private labor market.”22

Several studies have found supportive evidence: employment, particularly in low-end services but also overall, has tended to grow more slowly in nations with heavier payroll taxes.23

Figure 19 plots (catchup-adjusted) change in the employment rate between the peak business cycle years of 1979 and 2007 by the payroll tax share during this period. The pattern suggests a fairly strong negative association. There are a number of other policies and institutions that may influence employment growth and with which a high payroll tax share is correlated across these countries, but controlling for them doesn’t make the association go away.24

Figure 19. Payroll tax share and employment growth, 1979 to 2007
Employment change: 2007 employment rate minus 1979 employment rate, adjusted for initial level (catch-up). Data source: OECD. Payroll tax share: payroll tax revenues (including social security contributions) as a share of total tax revenues. Data source: OECD. “Asl” is Australia. The line is a linear regression line, calculated with the Netherlands excluded. The correlation is -.75 (with the Netherlands excluded).

Is there a rationale for heavy payroll taxes? In the countries in which payroll taxes are most significant — France, Germany, the Netherlands, and Spain — the welfare state is “Bismarkian.”25 Pensions, unemployment benefits, and sickness benefits are administered jointly by unions, employers, and the state. These programs are financed by payroll taxes, paid by both employers and employees. In these systems payroll tax payments are referred to as “social security contributions” rather than taxes. Though the programs operate on a pay-as-you-go basis — money paid in goes directly to current beneficiaries — they are popularly viewed as akin to private insurance. People tend to think of their contributions as investments set aside to benefit them directly in the event of job loss, illness, disability, or retirement. This is an illusion, but it’s an illusion that has, arguably, underpinned the generosity of these programs. Were these programs to be financed largely by income and/or consumption taxes, they might end up being less generous because public support would be narrower and shallower.

Nevertheless, there now is relatively widespread sentiment that the employment cost of such heavy reliance on payroll taxes outweighs this advantage, and so most of these countries have moved to alter this financing structure and/or its impact on employment.26 Most, for instance, have introduced partial or full exemptions on social contributions for certain types of low-end jobs. The shifts have been relatively limited, though, because this is a “insider-outsider” dilemma.27 While heavy payroll taxes impose costs on the nonemployed, they underwrite generous social insurance programs, so they are happily paid and strongly supported by those with steady jobs and their families. It’s in the (short-term) interest of these “insiders” to maintain the status quo.

WHAT’S THE BEST WAY FOR THE UNITED STATES TO INCREASE TAXES?

America’s policy makers may want or need to increase tax revenues in coming decades. The technical details of doing so, even if the aim were to boost revenues by as much as 10% of GDP, are not difficult.28 Nor would such an increase be unprecedented; the rise that occurred between 1915 and 1965 was far larger.

As a presidential candidate in 2008, Barack Obama pledged to not increase taxes for households in the bottom 95% of incomes. The Democratic nominee in 2016, Hillary Clinton, made the same pledge. In the contemporary U.S. context, there is some sense in focusing on the top in the search for more revenue. The chief rationale for progressive taxation is that those with more income can afford to pay a larger share of that income than those with less.29 While the incomes of Americans in the middle and below have risen slowly over the past few decades, for those at the top incomes have soared, so it’s reasonable to ask them to contribute a larger share of those incomes.

However, there’s a limit to how much additional tax revenue we can get from Americans in the top 5%. Figure 20 shows the effective tax rate on the top 5% of households going back to 1960. We have three estimates of this tax rate (two of the three include only federal taxes, not state and local). The dot for the year 2016 indicates what the effective tax rate on this group would need to have been in that year in order to increase tax revenues by 10% of GDP.30 It’s a very high rate, and one far above the actual rate at any point in the past half century. This seems neither desirable nor likely to find favor among policy makers.

Figure 20. Effective tax rate on the top 5% of incomes
Effective tax rate: tax payments as a share of pretax income. The chart has three estimates of the actual rate. The gray lines are for federal taxes. The black line is for all taxes (federal, state, and local). Data source for the top gray line: Thomas Piketty and Emmanuel Saez, data set for “How Progressive Is the U.S. Federal Tax System?,” Journal of Economic Perspectives, 2007, elsa.berkeley.edu/~saez. Data source for the lower gray line: Congressional Budget Office, “The Distribution of Household Income and Federal Taxes, 2011,” data set, alternative income definition, worksheet 13. Data source for the black line: Institute on Taxation and Economic Policy (ITEP), “Who Pays Taxes in America,” various years. Calculation of the rate needed to increase tax revenues by 10% of GDP is as follows: Get the total pretax income of the top 5% of households by multiplying this group’s average pretax income (from ITEP) by its number of households (from the Census Bureau). Then divide 10% of GDP by the group’s total pretax income.

Suppose instead that we increased taxes for all Americans, keeping the distribution of tax payments exactly the same as it is now while increasing revenues by 10% of GDP. What would that change look like for households at various points along the income distribution?

Start by recalling that the United States, like all affluent democratic nations, has a relatively flat tax system. Figure 21 shows average effective tax rates in the US at various points along the pretax income distribution (hollow circles). The effective tax rates paid by Americans are fairly similar up and down the income ladder. However, the distribution of pretax income is quite unequal. Households at the top get a much larger portion of the income than those in the middle or bottom.31 As a result, the distribution of tax payments (dark circles) also is very unequal. Households in the top quintile pay about 65% of all tax dollars, the middle fifth pay about 10%, and the bottom fifth pay 2%. Each is paying a similar percentage of their income in taxes, but the affluent end up paying a lot of the tax dollars because they have so much of the income.

Figure 21. Effective tax rates and shares of tax payments by pretax income quintile
Includes all types of taxes (personal and corporate income, payroll, property, sales, excise, estate, other) at all levels of government (federal, state, local). 2016. Effective tax rates: taxes paid as a share of pretax income. Data source: Institute on Taxation and Economic Policy (ITEP), “Who Pays Taxes in America: 2016.”

So if we were to increase tax revenues by 10% of GDP while keeping the distribution of tax payments exactly as it is now, households in the lowest fifth of incomes would account for about 2% of the added revenues, households in the middle around 10%, and households in the top quintile 65%. In dollar terms, households in the bottom fifth of incomes would pay, on average, about $1,400 more per year, those in the lower-middle fifth $3,600, those in the middle fifth $7,000, those in the upper-middle fifth $13,100, and those in the top fifth $46,200 more.32 With the top fifth, we can go further and break it down into subgroups. Those between the 80th and 90th percentiles would pay $21,100 more per year, those between the 90th and 95th percentiles $30,600, those between the 95th and 99th percentiles (average income $320,000) $54,600, and those in the top 1% (average income $1.8 million) $340,000.

SUMMARY

Many Americans would prefer to pay less in taxes, but they already pay less than their counterparts in nearly all other rich democratic countries. Moreover, after rising for a good bit of the twentieth century, US tax revenues have been flat since the mid-1960s, at about 25% of GDP. (Total government revenues continued to increase up to the year 2000.)

Some argue that this taxation stagnation is a product of our lack of a national consumption tax, such as a VAT. Adding such a tax surely would help to generate more revenue. But that doesn’t mean it’s necessary. Nations that increased tax revenues in the 1960s, 1970s, and 1980s did so just as much via income taxes as via consumption (or payroll) taxes.

Around 1990, tax revenues flattened out in all affluent countries. While many developments probably contributed — shifts in the ideology of right and left parties, union decline, public opposition to further tax increases, among others — the most important may have been the growing threat of capital flight due to financial globalization.

Taxes help to reduce income inequality. Their main contribution is indirect: taxes fund the transfers that redistribute income. For this reason, the quantity of revenues a country raises turns out to matter more for inequality reduction than the progressivity of its tax system.

High top statutory tax rates reduce top-end inequality in the pretax distribution of income, though this effect is likely small.

A key concern about taxation is that higher levels will hurt the economy. Heavy reliance on payroll taxes seems to reduce employment. But the record over the past century offers no observable adverse effect of high overall levels of taxation on either economic growth or employment growth.

APPENDIX

The appendix has additional data.

  1. Oliver Wendell Holmes, Jr., dissenting opinion, Compañía General de Tabacos de Filipinas v. Collector of Internal Revenue, 1927. The actual quote was “Taxes are what we pay for civilized society.” ↩︎
  2. Harold Wilensky, The Welfare State and Equality, University of California Press, 1976; Wilensky, Rich Democracies, University of California Press, 2002, ch. 10; Gary S. Becker and Casey B. Mulligan, “Deadweight Costs and the Size of Government,” Journal of Law and Economics, 2003; Junko Kato, Regressive Taxation and the Welfare State, Cambridge University Press, 2003; Thomas R. Cusack and Pablo Beramendi, “Taxing Work,” European Journal of Political Research, 2006; Pablo Beramendi and David Rueda, “Social Democracy Constrained: Indirect Taxation in Industrialized Democracies,” British Journal of Political Science, 2007; Achim Kemmerling, Taxing the Working Poor, Edward Elgar, 2009. In addition to the two reasons discussed here, Becker and Mulligan note that consumption and payroll taxes are thought to be more economically efficient than income taxes, because they vary less by income, and they hypothesize that more efficient taxes might induce less opposition from citizens and interest groups to increases in taxation. ↩︎
  3. In contrast, Steffen Ganghof notes that Denmark has very high tax revenues (as a share of GDP) and yet relies heavily on income taxation. Ganghof suggests that pressure for low income tax rates applies mainly to a particular type of income: corporate profits and capital income. There is much less pressure on taxation of wage and salary income. Hence, if policy makers are willing to tax wage and salary income at a different rate than capital income and corporate profits (a so-called “dual income tax”), as Denmark does, they can choose to rely mainly on income taxes rather than consumption and payroll taxes to finance a large welfare state. Steffen Ganghof “Globalization, Tax Reform Ideals, and Social Policy Financing,” Global Social Policy, 2005; Ganghof, “Tax Mixes and the Size of the Welfare State: Causal Mechanisms and Policy Implications,” Journal of European Social Policy, 2006; Ganghof, “The Political Economy of High Income Taxation: Capital Taxation, Path Dependence, and Political Institutions in Denmark,” Comparative Political Studies, 2007. ↩︎
  4. For more analysis and discussion, see Lane Kenworthy, Progress for the Poor, Oxford University Press, 2011, ch. 8. ↩︎
  5. Alexander Hicks, Social Democracy and Welfare Capitalism, Cornell University Press, 1999; Evelyne Huber and John D. Stephens, Development and Crisis of the Welfare State, University of Chicago Press, 2001. ↩︎
  6. Isaac William Martin, The Permanent Tax Revolt, Stanford University Press, 2008. ↩︎
  7. Steffen Ganghof, “Adjusting National Tax Policy to Economic Internationalization: Strategies and Outcomes,” in Welfare and Work in the Open Economy. Volume II: Diverse Responses to Common Challenges, edited by Fritz W. Scharpf and Vivien A. Schmidt, Oxford University Press, 2000; Philipp Genschel, “Globalization, Tax Competition, and Welfare State,” Politics and Society, 2002; Philipp Genschel and Peter Schwartz, “Tax Competition: A Literature Review,” Socio-Economic Review, 2011; Peter Egger, Sergey Nigai, and Nora Strecker, “The Impact of Globalization on Tax Structures in OECD Countries,” Vox, Center for Economic Policy Research, 2016. ↩︎
  8. Henrik Jacobsen Kleven, Claus Thustrup Kreiner, and Emmanuel Saez, “Why Can Modern Governments Tax So Much? An Agency Model of Firms as Fiscal Intermediaries,” Working Paper 15218, National Bureau of Economic Research, 2009; Henrik Jacobsen Kleven, Martin B. Knudsen, Claus Thustrup Kreiner, Søren Pedersen, and Emmanuel Saez, “Unwilling or Unable to Cheat? Evidence from a Tax Audit Experiment in Denmark,” Econometrica, 2011; Dina Pomeranz, “No Taxation without Information: Deterrence and Self-Enforcement in the Value Added Tax,” American Economic Review, 2015. ↩︎
  9. Henrik Jacobsen Kleven, “How Can Scandinavians Tax So Much?,” Journal of Economic Perspectives, 2014. ↩︎
  10. Neil Warren, “A Review of Studies on the Distributional Impact of Consumption Taxes in OECD Countries,” OECD Social, Employment, and Migration Working Paper 64, 2008; Monica Prasad and Yingying Deng, “Taxation and the Worlds of Welfare,” Socio-Economic Review, 2009. ↩︎
  11. Very high incomes are recorded by the surveyor as an upper limit number. For instance, in the US Current Population Survey, the upper limit for recorded earnings for most of the 2000s was $1.1 million. ↩︎
  12. Information on consumption taxes paid is very difficult to capture accurately in surveys. ↩︎
  13. Stein Ringen, The Possibility of Politics: A Study in the Political Economy of the Welfare State, Clarendon Press, 1987, ch. 8; Deborah Mitchell, Income Transfers in Ten Welfare States, Avebury, 1991; Sven Steinmo, Taxation and Democracy: Swedish, British, and American Approaches to Financing the Modern State, Yale University Press, 1993; Vincent Mahler and David Jesuit, “Fiscal Redistribution in the Developed Countries: New Insights from the Luxembourg Income Study,” Socio-Economic Review, 2006; Lane Kenworthy, Jobs with Equality, Oxford University Press, 2008; Kenworthy, Progress for the Poor, ch. 8; OECD, Divided We Stand, 2011, ch. 7; Koen Caminada, Kees Goudswaard, and Chen Wang, “Disentangling Income Inequality and the Redistributive Effect of Taxes and Transfers and Taxes in 20 LIS Countries Over Time,” Working Paper 581, Luxembourg Income Study, 2012. ↩︎
  14. Lane Kenworthy, “Public Insurance and the Least Well-Off,” The Good Society. Australia is a notable exception. ↩︎
  15. See also Sven Steinmo, Taxation and Democracy: Swedish, British, and American Approaches to Financing the Modern State, Yale University Press, 1993; Pablo Beramendi and David Rueda, “Social Democracy Constrained: Indirect Taxation in Industrialized Democracies,” British Journal of Political Science, 2007; Edward Kleinbard, We Are Better Than This: How Government Should Spend Our Money, Oxford University Press, 2014, ch. 12; Elvire Guillaud, Matthew Olckers, and Michael Zemmour, “Four Levers of Redistribution: The Impact of Tax and Transfer Systems on Inequality Reduction,” Working Paper 695, Luxembourg Income Study, 2017; Vincent A. Mahler and David K. Jesuit, “Indirect Taxes and Government Inequality Reduction: A Cross-National Analysis of the Developed World,” 2017. ↩︎
  16. Jesper Roine, Jonas Vlachos, and Daniel Waldenström, “The Long-Run Determinants of Inequality: What Can We Learn from Top Income Data?,” Journal of Public Economics, 2009; Thomas Piketty, Emmanuel Saez, and Stefanie Stantcheva, “Optimal Taxation of Top Incomes: A Tale of Three Elasticities,” American Economic Journal: Economic Policy, 2014; Kenneth Scheve and David Stasavage, Taxing the Rich, Russell Sage Foundation and Princeton University Press, 2016, ch. 3. ↩︎
  17. Carola Frydman and Raven S. Molloy, “Does Tax Policy Affect Executive Compensation? Evidence from Postwar Tax Reforms,” Journal of Public Economics, 2011. ↩︎
  18. Warren E. Buffett, “Stop Coddling the Super-Rich,” New York Times, 2011; Michael Förster, Ana Llena-Nozal, and Vahé Nafilyan, “Trends in Top Incomes and their Taxation in OECD Countries,” OECD Social, Employment and Migration Working Paper 159, 2014; T.R. Reid, A Fine Mess: A Global Quest for a Simpler, Fairer, and More Efficient Tax System, Penguin, 2017. ↩︎
  19. Scheve and Stasavage, Taxing the Rich, figure 3.4. ↩︎
  20. Jon Bakija, Lane Kenworthy, Peter Lindert, and Jeff Madrick, How Big Should Our Government Be?, University of California Press, 2016; Lane Kenworthy, “Is Big Government Bad for the Economy?,” The Good Society. ↩︎
  21. Peter Lindert, Growing Public: Social Spending and Economic Growth since the Eighteenth Century, two volumes, Cambridge University Press, 2004, pp. 235-45. ↩︎
  22. Fritz W. Scharpf, “Employment and the Welfare State: A Continental Dilemma,” Working Paper 97/7, Max Planck Institute for the Study of Societies, 1997. ↩︎
  23. Fritz W. Scharpf, “The Viability of Advanced Welfare States in the International Economy: Vulnerabilities and Options,” Journal of European Public Policy, 2000; Achim Kemmerling, “Tax Mixes, Welfare States, and Employment: Tracking Diverging Vulnerabilities,” Journal of European Public Policy, 2005; Kemmerling, Taxing the Working Poor; OECD, “Financing Social Protection: The Employment Effect,” in OECD Employment Outlook, 2009; Kenworthy, Jobs with Equality. ↩︎
  24. Kenworthy, Jobs with Equality, ch. 8. ↩︎
  25. Gøsta Esping-Andersen, The Three Worlds of Welfare Capitalism, Princeton University Press, 1990; Bruno Palier and Claude Martin, “From ‘a Frozen Landscape’ to Structural Reforms: The Sequential Transformation of Bismarckian Welfare Systems,” Social Policy and Administration, 2007. ↩︎
  26. Palier and Martin, “From ‘a Frozen Landscape’ to Structural Reforms”; Anton Hemerijck and Werner Eichhorst, “Whatever Happened to the Bismarckian Welfare State? From Labor Shedding to Employment-Friendly Reforms,” IZA Discussion Paper 4085, 2009. ↩︎
  27. Palier and Martin, “From ‘a Frozen Landscape’ to Structural Reforms”; David Rueda, Social Democracy Inside Out, Oxford University Press, 2007. ↩︎
  28. Lane Kenworthy, Social Democratic Capitalism, Oxford University Press, 2020, ch. 7. ↩︎
  29. As Adam Smith put it in The Wealth of Nations (book 5, ch. 2, part 2), “The subjects of every state ought to contribute towards the support of the government, as nearly as possible, in proportion to their respective abilities; that is, in proportion to the revenue which they respectively enjoy under the protection of the state.” ↩︎
  30. This assumes high-income households don’t alter their behavior in response to the increase in the effective tax rate. ↩︎
  31. Lane Kenworthy, “Income Distribution,” The Good Society. ↩︎
  32. The amount paid by households in the bottom fifth is calculated as $1.8 trillion (the total tax revenue needed) multiplied by .02 (this group will account for 2% of the revenues) divided by 25.2 million (the number of households in this group) = $1,428. The calculation is analogous for the other four groups. ↩︎