By Bruce Bartlett
Bruce Bartlett is a former Treasury deputy assistant secretary for economic policy. His new book, The Benefit and the Burden: Tax Reform — Why We Need It and What It Will Take, has been published by Simon & Schuster.
In this article, Bartlett explains that President Obama’s endorsement of the “Buffett rule” to raise taxes on those with high incomes and Republican efforts to require dynamic scoring for tax bills will lead to a debate on what tax rate maximizes federal revenues. This is an issue that has been debated since the 1970s. Recent academic research shows that the top U.S. rate could rise substantially from its current level of 35 percent before the increase would have such disincentive effects that revenues would start to fall.
With the economy recovering and increasing attention being paid to the budget deficit, Republicans are finding it harder and harder to gain political traction on tax cuts. Although they continue to maintain that spending can easily be cut enough to finance even a big tax cut, they are quietly preparing an alternative strategy. They are moving to force the Joint Committee on Taxation and the Congressional Budget Office to adopt dynamic scoring, which would incorporate macroeconomic growth effects into revenue estimates. Republicans have long believed that incorporating those effects would greatly reduce the budgetary cost of tax cuts and make them easier to enact.1
Dynamic scoring got started with the so-called Laffer curve, supposedly drawn on a napkin in 1974 by University of Chicago business Professor Arthur Laffer for Donald Rumsfeld and Dick Cheney, both members of President Ford’s staff.2 The curve was popularized by Wall Street Journal editorial writer Jude Wanniski, who first described it in a 1975 article.3 He went on to develop it at greater length in his 1979 book, The Way the World Works.
At its core, the Laffer curve is unobjectionable. It shows simply that neither a 0 percent tax rate nor a 100 percent tax rate raises revenue; somewhere in between is a rate that maximizes revenue. The trick is to empirically estimate the revenue-maximizing rate based on the existing tax regime and economic conditions. It turns out that even supply-side economists have seldom found examples of tax rates that were so high that a rate cut would increase revenue.
In the 1970s Harvard economist Martin Feldstein and others argued that a cut in the capital gains rate would raise net revenue. However, that was mainly a short-term unlocking effect. A Treasury Department study later concluded that the cuts in the capital gains rate in 1978 and 1981 reduced long-term capital gains revenue and did not materially increase economic growth.4
When asked about the impact on revenues of an across-the-board rate cut, as proposed in the 1978 Kemp-Roth bill and later by President Reagan, Laffer declined to estimate whether that would raise revenue even though tax rates were substantially higher than they are now, with a top rate of 70 percent. The most Laffer would say was that the Kemp-Roth tax cut would self-finance by reducing spending for things like unemployment compensation as economic growth increased, raising private saving, reducing the value of tax shelters, and creating higher revenues at state and local levels.5
In 1978 economists Norman Ture and Michael Evans incorporated supply-side economics into their detailed revenue forecasts, and both concluded that the Kemp-Roth bill would never pay for itself. Ture estimated substantial revenues losses, net of feedback, even 10 years after enactment, when revenues would still be $53 billion (in 1977 dollars) below baseline.6 Evans’s figures were very similar, showing a $61 billion deficit increase in 1987.7
Contrary to popular belief, the Reagan administration never incorporated Laffer curve effects into its revenue estimates for the 1981 tax cut. All published estimates conformed to standard Treasury revenue-estimating methods and were almost identical to independent estimates done by the CBO.8
Treasury tried to empirically estimate the Laffer curve in 1984. It concluded that most tax cuts lose revenue. Rate cuts for those in the top bracket had the potential to raise revenue, but only in the long term. In the short term, revenues would fall. As Treasury explained:
Discussions of the Laffer Curve often presume that there is a single aggregate tax rate elasticity that applies to a nation. Thus they argue over whether a tax cut will increase or decrease revenues. In reality there is not a single tax rate and tax elasticity. Rather, there is a series of tax rates and elasticities that pertain to different income classes. Our estimates suggest that the income tax base is not very responsive to tax rate changes in the income categories occupied by most Americans. In this sense, they are highly consistent with estimates by other researchers indicating that aggregate tax elasticity is quite small.
In 1985 economist Lawrence Lindsey attempted to compute the revenue-maximizing tax rate. Given the 1982 tax structure, which had a top rate of 50 percent, Lindsey concluded that reducing the top rate to 43 percent would raise revenue, but that reducing any other rates would lose revenue.10
Not much was heard about the revenue-maximizing top rate for some years because tax reductions were the order of the day. But as the need to raise revenue — and perhaps legislate increases in the top tax rate, for both revenue and distributional reasons — has become pressing, there is once again interest in the subject.
An important contribution to the discussion happened in 2009. N. Gregory Mankiw, chair of the Council of Economic Advisers under George W. Bush and widely considered to be among the most conservative U.S. economists, coauthored a paper that explored optimal tax theory and concluded that the optimal marginal tax rate is between 48 and 50 percent.11
Also in 2009, economists Mathias Trabandt and Harald Uhlig examined revenue-optimizing tax rates for the United States and Europe. They found that the United States is well below the revenue-maximizing top rate of 63 percent, that taxes on labor could be increased by 30 percent before labor supply dropped enough to reduce revenues from further increases, and that taxes on capital could be increased by 6 percent.12
A 2010 paper by economists Anthony Atkinson and Andrew Leigh looked at five different Anglo-Saxon countries and found similar tax elasticities among high-income taxpayers. They concluded that the revenue-maximizing top rate is at least 63 percent and may be as high as 83 percent.13
Most recently, economists Peter Diamond and Emmanuel Saez concluded in a 2011 paper that the revenue-maximizing top tax rate is 73 percent — well above the current top rate of 42.5 percent.14
Informal surveys of top economists confirm that the top tax rate could increase substantially before the Laffer effect caused revenues to decline. One survey was taken by The Washington Post in 2010 and quoted University of Michigan economist Joel Slemrod as suggesting that the revenue-maximizing top rate is at least 60 percent:
The idea that we are on the wrong side [of the Laffer curve] has almost no support among academics who have looked at this. Evidence doesn’t suggest we’re anywhere near the other end of the Laffer Curve.
University of California, Berkeley, economist Brad DeLong and Dean Baker of the Center for Economic and Policy Research have said that the revenue-maximizing top rate is about 70 percent. Even conservative economic journalists Larry Kudlow of CNBC and Stephen Moore of the Wall Street Journal editorial page said that revenues would rise until the top rate hit at least 50 percent.16
Revenue Loss From Tax Cuts for the Top 1 Percent of Taxpayers Since 1986
(billions of dollars)
Actual Effective Effective
Year Rate (percent) Actual Revenues Rate of 33.1% Difference
1987 26.4 $91.6 $114.8 $23.2
1988 24.0 $113.8 $156.9 $43.1
1989 23.3 $109.2 $155.1 $45.9
1990 23.2 $112.3 $160.1 $47.8
1991 24.4 $111.3 $151.3 $40.0
1992 25.0 $131.2 $173.5 $42.3
1993 28.0 $145.8 $172.5 $26.7
1994 28.2 $154.3 $181.1 $26.8
1995 28.7 $178.0 $205.3 $27.3
1996 28.9 $212.6 $244.0 $31.4
1997 27.6 $241.2 $289.2 $48.0
1998 27.1 $274.0 $334.4 $60.4
1999 27.5 $317.4 $381.9 $64.5
2000 27.4 $366.9 $442.9 $76.0
2001 27.5 $300.9 $362.5 $61.6
2002 27.2 $268.6 $326.6 $58.0
2003 24.3 $256.3 $349.4 $93.1
2004 23.5 $306.9 $432.8 $125.9
2005 23.1 $368.1 $527.3 $159.2
2006 22.8 $408.4 $593.7 $185.3
2007 22.4 $450.9 $665.3 $214.4
2008 23.3 $392.1 $558.4 $166.3
2009 24.0 $318.0 $438.8 $120.8
Total $5,629.8 $7,417.8 $1,788.0
Source: Author's calculations based on IRS data.
I’m not sure how much we could raise the top rate before it would become counterproductive in terms of revenue. But I think it is revealing that as recently as 1986, during the Reagan administration, those in the top 1 percent of taxpayers, ranked by adjusted gross income, had an effective federal income tax rate of 33.1 percent when the top marginal rate was 50 percent. Their effective rate has been significantly lower every year since. Had they simply kept paying the same effective rate, the federal government would have reaped $1.8 trillion in aggregate additional revenue between 1987 and 2009, not counting interest.
Of course, it goes without saying that the optimal tax rate in terms of revenue is not necessarily the one that maximizes growth or is socially optimal. Personally, I would prefer not to have a top income tax rate exceeding 50 percent, because it is important psychologically and morally that people not be forced to give more than half of their income to the federal government. However, given the magnitude of our nation’s fiscal problem, a rate higher than that may be inevitable unless the United States adopts a VAT, carbon tax, or other broad-based tax to supplement existing revenue sources.FOOTNOTES
1 On February 2 the House passed H.R. 3582, the Pro-Growth Budgeting Act of 2012, Doc 2012-1555, 2012 TNT 17-22, which would force the JCT and CBO to do a dynamic score for major tax bills, but not for appropriations bills.
2 Arthur Laffer, “The Laffer Curve: Past, Present, and Future,” Heritage Foundation Report No. 1765 (June 1, 2004).
3 Jude Wanniski, “The Mundell-Laffer Hypothesis — A New View of the World Economy,” The Public Interest (Spring 1975), at 49-50.
4 Treasury report to Congress on the capital gains tax reductions of 1978 (1985).
5 Bruce Bartlett, The New American Economy 113 (2009).
6 House Ways and Means Committee, “Tax Reductions: Economists’ Comments on H.R. 8333 and S. 1860 (The Kemp-Roth Bills)” (1978), at 96.
7 House and Senate Budget committees, “Leading Economist’s Views of Kemp-Roth” (1978), at 76.
8 Bartlett, “The 1981 Tax Cut After 30 Years: What Happened to Revenues?” Tax Notes, Aug. 8, 2011, p. 627, Doc 2011-16766, 2011 TNT 152-7.
9 James Gwartney and James Long, “Income Tax Avoidance and an Empirical Estimation of the Laffer Curve,” Treasury’s Office of Economic Policy (July 1984), at 22.
10 Lawrence Lindsey, “Estimating the Revenue Maximizing Top Personal Tax Rate,” National Bureau of Economic Research Working Paper No. 1761 (Oct. 1985), at 18.
11 N. Gregory Mankiw et al., “Optimal Taxation in Theory and Practice,” 23 J. of Econ. Persp. 147, 158 (Fall 2009).
12 Mathias Trabandt and Harald Uhlig, “How Far Are We From the Slippery Slope? The Laffer Curve Revisited,” NBER Working Paper No. 15343 (Sept. 2009).
13 A.B. Atkinson and Andrew Leigh, “The Distribution of Top Incomes in Five Anglo-Saxon Countries Over the Twentieth Century,” Institute for the Study of Labor (IZA) Working Paper No. 4937 (May 2010), at 29.
14 Peter Diamond and Emmanuel Saez, “The Case for a Progressive Tax: From Basic Research to Policy Recommendations,” 25 J. of Econ. Persp. 165, 171 (Fall 2011).
15 Dylan Matthews, “Where Does the Laffer Curve Bend?” The Washington Post, Aug. 9, 2010.
Never in American history has the debate over income inequality so dominated the public square, with Democratic presidential candidates and congressional leaders calling for massive tax increases and federal expenditures to redistribute the nation’s income. Unfortunately, official measures of income inequality, the numbers being debated, are profoundly distorted by what the Census Bureau chooses to count as household income.
The published census data for 2017 portray the top quintile of households as having almost 17 times as much income as the bottom quintile. But this picture is false. The measure fails to account for the one-third of all household income paid in federal, state and local taxes. Since households in the top income quintile pay almost two-thirds of all taxes, ignoring the earned income lost to taxes substantially overstates inequality.
How Redistribution Works
Average earned and net income by quintile, 2017
Source: Calculations by authors based on official government data
The Census Bureau also fails to count $1.9 trillion in annual public transfer payments to American households. The bureau ignores transfer payments from some 95 federal programs such as
- Medicaid and
- food stamps,
which make up more than 40% of federal spending, along with dozens of state and local programs. Government transfers provide 89% of all resources available to the bottom income quintile of households and more than half of the total resources available to the second quintile.
In all, leaving out taxes and most transfers overstates inequality by more than 300%, as measured by the ratio of the top quintile’s income to the bottom quintile’s. More than 80% of all taxes are paid by the top two quintiles, and more than 70% of all government transfer payments go to the bottom two quintiles.
America’s system of data collection is among the most sophisticated in the world, but the Census Bureau’s decision not to count taxes as lost income and transfers as gained income grossly distorts its measure of the income distribution. As a result, the raging national debate over income inequality, the outcome of which could alter the foundations of our economic and political system, is based on faulty information.
The average bottom-quintile household earns only $4,908, while the average top-quintile one earns $295,904, or 60 times as much. But using official government data sources on taxes and all transfer payments to compute net income produces the more complete comparison displayed in the nearby chart.
The average bottom-quintile household receives $45,389 in government transfers. Private transfers from charitable and family sources provide another $3,313. The average household in the bottom quintile pays $2,709 in taxes, mostly sales, property and excise taxes. The net result is that the average household in the bottom quintile has $50,901 of available resources.
Government transfers go mostly to low-income households. The average bottom-quintile household and the average second-quintile household receive government transfers of some $17 and $4 respectively for every dollar of taxes they pay. The average middle-income household receives $17,850 in government transfers and pays an almost identical $17,737 in taxes, while the fourth and top quintiles of households receive government transfers of only 29 cents and 6 cents respectively for every dollar paid in taxes. (In the chart, transfers received minus taxes paid are shown as net government transfers for low-income households and net taxes for high income households.)
The average top-quintile household pays on average $109,125 in taxes and is left, after taxes and transfer payments, with only 3.8 times as much as the bottom quintile: $194,906 compared with $50,901. No matter how much income you think government in a free society should redistribute, it is much harder to argue that the bottom quintile is getting too little or the top quintile is getting too much when the ratio of net resources available to them is 3.8 to 1 rather than 60 to 1 (the ratio of what they earn) or the Census number of 17 to 1 (which excludes taxes and most transfers).
Today government redistributes sufficient resources to elevate the average household in the bottom quintile to a net income, after transfers and taxes, of $50,901—well within the range of American middle-class earnings. The average household in the second quintile is only slightly better off than the average bottom-quintile household. The average second-quintile household receives only 9.4% more, even though it earns more than six times as much income, it has more than twice the proportion of its prime working-age individuals employed, and they work twice as many hours a week on average. The average middle-income household is only 32% better off than the average bottom-quintile households despite earning more than 13 times as much, having 2.5 times as many of prime working-age individuals employed and working more than twice as many hours a week.
Antipoverty spending in the past 50 years has not only raised most of the households in the bottom quintile of earners into the middle class, but has also induced many low-income earners to stop working. In 1967, when funding for the War on Poverty started to flow, almost 70% of prime working-age adults in bottom-quintile households were employed. Over the next 50 years, that share fell to 36%. The second quintile, which historically had the highest labor-force participation rate among prime work-age adults, saw its labor-force participation rate fall from 90% to 85%, while the top three income quintiles all increased their work effort.
Any debate about further redistribution of income needs to be tethered to these facts. America already redistributes enough income to compress the income difference between the top and bottom quintiles from 60 to 1 in earned income down to 3.8 to 1 in income received. If 3.8 to 1 is too large an income differential, those who favor more redistribution need to explain to the bottom 60% of income-earning households why they should keep working when they could get almost as much from riding in the wagon as they get now from pulling it.