Discussion: (0 comments)
There are no comments available.
View related content: Health Care
For a complete listing of all On the Margin articles, please visit: www.aei.org/onthemargin/.
On November 7, 2009, the House passed H.R. 3962, the Affordable Health Care for America Act. The bill would expand Medicaid, grant subsidies to moderate-income households buying health insurance on newly established exchanges, and provide health insurance tax credits to some small businesses. These expenditures would be financed, in part, by a new 5.4 percent surtax on households with very high incomes, including married couples with incomes above $1 million.
The proposed millionaire surtax is politically attractive because its direct burden would fall on a very small group, roughly 0.3 percent of the population. Moreover, this group is extremely wealthy and could undoubtedly afford to pay additional taxes. For three reasons, however, the proposed surtax would be bad tax policy.
First, it would significantly increase marginal tax rates for the affected households, giving them greater incentives to reduce their taxable income through various avoidance strategies. Even with moderate responsiveness to incentives, the revenue generated by the surtax would be significantly smaller than the burden that it would impose on affected taxpayers.
Second, the surtax would significantly increase the marginal tax rate on saving and investment by the affected households, whether done through corporate or noncorporate firms. The impact would be magnified because these households, despite their small numbers, account for a large portion of national saving. The resulting drag on capital accumulation would lower real wages for workers throughout the economy.
Third, the proposed surtax reflects an unsustainable approach to tax and fiscal policy. As commentators across the political spectrum have recognized, the existing fiscal imbalance cannot be addressed without imposing sacrifices on a broad segment of the population. Any new spending programs, such as those in H.R. 3962, will impose additional burdens. By linking these programs to a tax imposed on only 0.3 percent of the population, the bill obscures that fiscal reality. If the programs in H.R. 3962 are worthwhile, they are worth paying for in an open and broad-based manner.
Section 551 of H.R. 3962 would add new section 59C to the code. The new section would impose a surtax equal to 5.4 percent of the excess of the taxpayer’s modified adjusted gross income (MAGI) above thresholds of $1 million for married couples filing joint returns and $500,000 for other taxpayers. MAGI would be adjusted gross income, as defined in section 62(a), plus foreign earned income excluded under section 911, minus the itemized deduction for investment interest expense allowed by section 163(d). The surtax in the House-passed bill is narrower than that in the bill reported by the House Ways and Means Committee on July 17, 2009; the latter bill would have extended the surtax down to incomes of $350,000 for married couples and $280,000 for other taxpayers, although reduced tax rates would have applied at the lower income levels.
The surtax would apply to tax years beginning on or after January 1, 2011. Because most of the new spending authorized by H.R. 3962 would not begin until January 1, 2013, the surtax would take effect two years earlier than most of the spending that it would finance. The Joint Committee on Taxation has estimated that the surtax would raise $460.5 billion of revenue in fiscal years 2011 through 2019, including $68.4 billion in fiscal 2019.
The surtax would differ in three respects from a simple increase in the individual income tax rate schedule. First, because the tax base would be MAGI instead of taxable income, itemized deductions, other than investment interest expense, could not be claimed against the surtax. A high-income household that earned an additional $1,000 of ordinary income and spent all of it on deductible charitable contributions would still incur a net tax increase of $54. Second, the surtax would apply to long-term capital gains and qualified dividends (gains and dividends) that are taxed at preferential rates under section 1(h), as well as to income taxed under the ordinary rate schedule. Third, the surtax would apply to taxpayers subject to the alternative minimum tax, as well as those subject to the regular income tax.
In 2007 roughly 495,000 returns, 0.35 percent of all (taxable and nontaxable) returns, reported AGI exceeding the thresholds of the proposed surtax. Because some households do not file returns, the group probably composed about 0.3 percent of the overall population. This small group earned a significant share of the nation’s income. They reported combined AGI of $1.47 trillion, 16.9 percent of the total AGI reported on all tax returns. The income share of this group has risen over time, reflecting an increase in income inequality in the United States.
Even without the surtax, these households paid a large amount of individual income tax, $325 billion, in 2007. They bore 29.2 percent of the nation’s individual income tax liability, significantly greater than their 16.9 percent share of AGI. Still, after paying $658,000 income tax on an AGI of $2,971,000, the average household in this group cleared a tidy $2,313,000. It seems safe to say that these households will not be driven into destitution if their taxes are further increased.
Over time, the surtax could spread to a larger group. Although many dollar values in the code are indexed to the 16-month-lagged change in the Consumer Price Index for All Urban Consumers (CPI-U), H.R. 3962 does not provide for inflation adjustment of the surtax thresholds. In the absence of further congressional action, inflation would slowly, but steadily, erode the real value of the thresholds. An extrapolation based on the inflation projections of the Congressional Budget Office suggests that their real value would be cut in half from 2011 to 2048.
Of course, Congress could counteract this trend by increasing the nominal thresholds from time to time, although doing so would diminish the surtax’s revenue yield. Congress has taken such action regarding some nominal values in the code, such as the AMT exemption amounts in section 55(d)(1). But, Congress has left other nominal values unchanged for extended periods. Interestingly, two $1 million values–the section 162(m)(1) limit on deductible executive compensation and the section 163(h)(3)(C)(ii) limit on deductible mortgage interest–have remained unchanged since 1994 and 1987, respectively. Also, the income thresholds at which Social Security benefits become partially taxable under section 86(c)(1) and the capital loss deduction limit imposed by section 1211(b) have been unchanged since 1984 and 1978, respectively.
Regardless of what Congress might do, the surtax would remain confined to a narrow group of wealthy taxpayers for the foreseeable future. At first glance, it may seem desirable to target this group for additional taxes. But closer examination reveals the flaws in this approach. To begin, it is useful to examine why the surtax’s revenue yield would fall short of the burden that it would impose on the affected taxpayers.
Revenue and Burden
A central tenet of public finance economics is that a tax typically imposes a greater dollar burden on a taxpayer than the revenue raised by the government. The additional burden on the taxpayer, beyond the revenue collected, is referred to as the excess burden, or deadweight loss, of the tax. Excess burden arises when the taxpayer changes his behavior to avoid the tax. Although the change in behavior is burdensome to the taxpayer, it does not result in revenue for the government.
For example, consider a consumer who, in the absence of tax, buys 100 widgets for $10 each. When a $2 tax raises the price to $12, the consumer chooses to buy only 80 widgets, so tax revenue is $160. That $160 payment is clearly a burden on the consumer. However, the consumer suffers an additional burden because she purchases 20 fewer widgets. To the consumer, each of those widgets was worth at least $10 (because she purchased them at the $10 price), but no more than $12 (because she did not purchase them at the $12 price). In the no-tax situation, therefore, the consumer received a net gain of at least zero, but no more than $2, by buying each of the 20 widgets at the $10 price. Assuming a $1 average net gain, the loss of the 20 widgets imposes a $20 excess burden above and beyond the $160 tax payment.
Excess burden is greater when the taxpayer’s behavior is more sensitive to tax incentives. When the taxpayer does not change behavior at all in response to tax incentives, there is no excess burden.
Some analyses of the behavioral effects of income taxation examine only the effect on hours worked and often find little impact. As Martin Feldstein observed, however, income taxes can induce people to reduce their taxable income through means other than a reduction in hours worked. People can reduce taxable income by holding tax-exempt municipal bonds rather than taxable bonds, receiving fringe benefits rather than cash wages, engaging in tax shelters, and spending more money on tax-deductible items. Economists could investigate each of these behavioral changes, one by one. Or, as Feldstein suggested, they can simply investigate the overall change in taxable income prompted by changes in tax rates.
Recent studies have therefore focused on the overall elasticity of taxable income, which roughly equals the percentage change in taxable income that results from a 1 percent change in (1 – t), where t is the tax rate. Suppose, for example, that the elasticity is 0.5, the estimate that I use below. Consider an increase in the marginal income tax rate from 25 percent to 26 percent, which reduces (1 – t) from 0.75 to 0.74, a decline of 1.33 percent. With a 0.5 elasticity, the rate increase reduces taxable income by roughly 0.67 percent (0.5 times 1.33 percent).
Two recent papers, one by Seth Giertz and one by Emmanuel Saez, Joel Slemrod, and Giertz, provide surveys of the numerous statistical studies that have used tax return data to estimate the elasticity of taxable income. As these papers describe, early estimates were very high, often well above 1. Recent estimates have been more modest, with considerable variation across studies. There is strong evidence that the elasticity is higher for high-income groups. A recent Tax Foundation analysis assumes that the elasticity for taxpayers with incomes above $100,000 is 0.6. The 0.6 value for high-income taxpayers appears to represent a reasonable middle ground, as some studies have estimated much higher values while others have estimated lower values.
I reduce that estimate slightly, however, to reflect that the surtax would be imposed on MAGI rather than taxable income. As the literature has emphasized, the elasticity tends to be higher when the base is narrower, because a narrower base offers more opportunities to avoid tax by changing behavior. For example, the relevant elasticity would be much higher for a tax on blue cars than for a tax on all cars, as the former tax can be avoided by simply changing the color of the car. One way in which taxpayers can reduce taxable income is by spending more on things that can be claimed as itemized deductions. That response would be largely unavailable in the context of the surtax, because itemized deductions, other than investment interest expense, would not be allowed. I therefore reduce the elasticity to 0.5, which I assume applies both to ordinary income and to gains and dividends.
The next step in the analysis is to specify the tax rates that the affected taxpayers would face, with and without the surtax.
Under current law and President Obama’s proposals, the top statutory individual income tax rate on ordinary income will be 39.6 percent in 2011 and thereafter. Further, under current law and the president’s proposals, the section 68 limitation on itemized deductions (the Pease provision) will be restored in 2011 and thereafter. Under the Pease provision, 3 cents of itemized deductions will be lost for each additional dollar of AGI. The Pease provision raises the top marginal tax rate, in the absence of the surtax, to 40.788 percent.
I assume a 5 percent state income tax. The section 164 itemized deduction for state income taxes provides a federal tax saving of 1.98 percent, which lowers the federal tax burden from 40.788 percent to 38.808 percent, which I round to 38.8 percent. The combined federal-state marginal tax rate is then 43.8 percent without the surtax.
Today, gains and dividends are taxed at a maximum rate of 15 percent. Under current law, starting in 2011, the top rate for long-term capital gains will revert to 20 percent and qualified dividends will be taxed as ordinary income. Obama has proposed, however, that a preferential rate of 20 percent be maintained for qualified dividends in 2011 and thereafter. I therefore assume that gains and dividends will be taxed at 20 percent. As with ordinary income, the Pease provision would add 1.188 percentage points to the tax rate and state income tax (net of the effect of federal deductibility) would add another 3.02 percentage points. Without the surtax, the effective federal-state tax rate would be 24.208 percent, which I round to 24.2 percent.
The surtax would increase marginal tax rates on both types of income by 5.4 percentage points. The rate on ordinary income would rise from 43.8 percent to 49.2 percent and the rate on gains and dividends would rise from 24.2 percent to 29.6 percent.
In 2007 the average married couple with AGI greater than $1 million had an AGI just below $3.5 million. I therefore consider a married couple that would have $3.5 million AGI without the surtax. Of that amount, I assume that $1.5 million is gains and dividends, because that income is roughly three-sevenths of AGI for households at these income levels.
The “static” revenue gain that the surtax would produce in the absence of behavioral changes is $135,000, which is 5.4 percent of the $2.5 million of income that exceeds the threshold. But behavioral responses would reduce the revenue yield.
For ordinary income, the surtax would reduce (1 – t) from 0.562 to 0.508. With a 0.5 elasticity, ordinary income would fall by 4.93 percent, from $2 million to $1,901,488. For gains and dividends, the surtax would reduce (1 – t) from 0.758 to 0.704. With a 0.5 elasticity, gains and dividends would fall by 3.63 percent, from $1.5 million to $1,445,583. In total, AGI would fall from $3.5 million to $3,347,071. The $152,929 income decline would include a loss of $98,512 of ordinary income and $54,417 of gains and dividends.
The net revenue impact would reflect two opposing effects. First, the federal government would collect an additional 5.4 percent tax on the $2,347,071 of remaining income above the threshold, a gain of $126,742. Second, 43.8 percent tax (38.8 percent federal and 5 percent state) would be lost on the $98,512 of ordinary income driven out by the surtax and 24.2 percent tax (19.2 percent federal and 5 percent state) would be lost on the $54,417 of gains and dividends driven out by the surtax. That second effect would represent a revenue loss of $56,317 ($48,671 federal and $7,646 state).
Combining the two effects yields a $70,425 net revenue gain, reflecting a $78,071 net federal gain and a $7,646 state loss. The net revenue gain would be only 52 percent of the static revenue gain.
The above analysis could be modified in several respects:
This example is only illustrative, but demonstrates the potential for high excess burdens from the surtax. Although the surtax would raise only $70,425 of revenue, it would impose a $130,902 total burden on the couple. The surtax would therefore impose an excess burden of $60,477, almost as large as the revenue gain.
As stated above, excess burden reflects behavioral changes induced by taxation. In this case, the surtax would cause the couple to forgo earning $98,512 of ordinary income, each dollar of which would have yielded a net value (to the couple and the government) of 43.8 cents to 49.2 cents, and $54,417 of gains and dividends, each dollar of which have yielded a net value of 24.2 cents to 29.6 cents.
Broad tax increases would have a much lower ratio of excess burden to revenue than a surtax narrowly targeted to high-income taxpayers. For example, even if the 0.5 elasticity assumed above were still applicable, an across-the-board increase in a flat-rate income tax, starting at a 25 percent tax rate, would cause an excess burden equal to only one-fifth of the revenue increase. There are two reasons why the surtax would have a higher ratio of excess burden to revenue than the hypothetical across-the-board increase. First, the taxpayers subject to the surtax already face high marginal tax rates on ordinary income, which magnifies the excess burden from any further increase. Second, the surtax would apply only to income above a threshold, which diminishes the revenue gain for any given increase in marginal tax rates. Of course, the actual difference in excess burden would be even more dramatic than this calculation suggests, because the statistical studies discussed above find that the elasticity is lower for people not in the highest income groups.
This analysis illustrates the perennial trade-off between efficiency and equity. Although the surtax would impose a high burden per dollar of revenue, that burden would fall on a very wealthy group. One could argue that the surtax is appropriate on equity grounds, despite its economic inefficiency. But any such conclusion would be premature, because the impact of the surtax on saving and capital accumulation has not yet been considered.
Impact on Saving and Capital Accumulation
The above analysis would be more or less the full story if the surtax applied only to labor income or consumption, because there would be no increase in the taxation of the returns to saving. In reality, however, the proposed surtax would also apply to the capital income of the affected taxpayers, which is a large part of the nation’s capital income.
It is sometimes argued that tax rates on high-income taxpayers have little impact on noncorporate investment because relatively few owners of noncorporate firms have high incomes. The relevant factor, however, is the volume of investment and income, not the number of firms or owners. Although returns with AGI greater than $1 million were only 0.27 percent of all returns in 2007, the Statistics of Income data reveal that they reported 29 percent of all interest income, 35 percent of all dividends, 37 percent of all passthrough business income net of losses, and 64 percent of all capital gains (including capital gains distributions) net of capital losses. Including unmarried taxpayers with AGI between $500,000 and $1 million would further increase these percentages. The surtax would therefore affect a large fraction of investment.
The impact on saving and investment is part of the overall response of taxable income to tax rates, the topic of the statistical studies considered in the previous section. In practice, however, those studies capture little of this impact, because they generally study the effects of tax rates with a relatively short lag, such as three years. The impact on wealth and capital builds up over a much longer period.
The impact of the tax system on investment incentives depends on the effective marginal tax rate, which is the proportional gap between the before-tax and after-tax rates of return. Because the surtax would apply to dividends, interest, capital gains, and profits of noncorporate firms, it would increase effective tax rates on a wide range of investments.
It is simplest to begin with an example of equity-financed investment in a passthrough firm. In the first year, the firm buys a $100 machine that produces $108 of output in the second year and then becomes worthless. Without taxes, the owners receive an 8 percent return. To isolate the effects of the surtax, I assume that the participants are above the surtax threshold.
It is sometimes claimed that higher tax rates will not reduce investment because investment is made with before-tax dollars. One analysis claims, “In addition, some criticisms of [higher tax rates] imply that the investments that small business owners make in staff or equipment are made out of after-tax income, with the income tax being applied to small-business receipts before such expenses are taken into account. That is incorrect. Business owners pay tax on the profit that a firm produces after such expenses are taken into account, not on the receipts the firm takes in.” While it is true that tax rate increases would not reduce investment incentives if investments were made with before-tax dollars, today’s tax system generally requires that investments be made with after-tax dollars.
Investments are made with before-tax dollars only if they are expensed. If the $100 machine in the example is expensed, the owners deduct $100 in the first year. With a 43.8 percent tax rate, expensing provides a $43.80 tax saving, so that the out-of-pocket cost of the investment is $56.20. Next year, the owners pay 43.8 percent tax on the $108 cash flow, yielding a $60.70 net payoff, which is 1.08 times the initial out-of-pocket cost. The after-tax return is 8 percent, which equals the after-tax return, so the effective tax rate is zero. Raising the tax rate to 49.2 percent does not change the picture. The out-of-pocket cost falls to $50.80 and the after-tax payoff falls to $54.86, leaving the after-tax return unchanged at 8 percent. With expensing, higher statutory tax rates do not impair investment incentives, as the effective tax rate remains equal to zero.
Expensing is not, however, typical in today’s tax system. Although many intangible investments are expensed and a limited amount of tangible personal property investment can be expensed by small firms under section 179, most investments must be depreciated, which means that they are made with after-tax dollars. Any treatment less generous than expensing generates positive effective tax rates. In the simple and approximately correct case in which tax depreciation allowances have the same present discounted value as allowances equal to inflation-adjusted economic depreciation, the effective tax rate equals the statutory tax rate.
If the $100 machine is depreciated, nothing is deducted in the first year and $100 is deducted in the second year, reflecting the machine’s complete depreciation at that time. The firm’s owners pay tax on $8 of income, $108 cash flow minus $100 depreciation, in the second year. The after-tax rate of return is reduced by a fraction equal to the tax rate, 43.8 percent without the surtax and 49.2 percent with the surtax. The effective tax rate equals the statutory tax rate on passthrough business income, which the surtax would increase.
Similar results apply to debt-financed investment by passthrough firms. Suppose that the firm finances the machine purchase by issuing $100 of debt that carries an 8 percent interest rate. The owners would have no net taxable income in the second year, because they would deduct $8 business interest expense along with the $100 depreciation. But the lender would pay tax on $8 of interest income. The effective tax rate on the investment equals the tax rate on interest income, which the surtax would increase.
Political rhetoric has tended to focus on how the surtax, and other proposed individual income tax rate increases, would affect investment by flow-through firms, particularly small businesses. When the surtax was proposed, the president of the U.S. Chamber of Commerce complained that the “real victims would be America’s small business owners” and a leading House Republican commented that the surtax would fall on the “small business men and women we are counting on to start hiring workers again.” As Amy Roden and I recently noted, political rhetoric and policy initiatives often focus on small business to the exclusion of big business, even though firms of all sizes contribute to the prosperity of the American economy.
In reality, the surtax would reduce investment incentives, not only for small (and big) passthrough firms, but also for corporations. The analysis for debt-financed corporate investment is the same as for debt-financed passthrough investment. Because the corporation deducts interest expense, the investment has no net impact on corporate or stockholder tax liability, but creates tax liability for the bondholder. The effective tax rate on the investment equals the tax rate on interest income, which the surtax would increase.
The surtax would have a similar, but somewhat smaller, impact on corporate investment financed by the issuance of new equity, which is taxed at both the corporate and stockholder levels. Dividends paid from the investment profits are taxed immediately to stockholders while undistributed profits generate capital gains that are taxed if and when the stockholder realizes them. The surtax would increase both the dividend and capital gains tax rates, but its impact on capital gains would be blunted because the tax is deferred until realization.
The surtax would have some, but a much smaller, impact on corporate investment financed from retained earnings. As the “new view” of dividend taxation emphasizes, the dividend tax does not impose a marginal burden on such investment; although the investment generates future taxable dividends, it is financed by a reduction in current taxable dividends. Such investment is subject to the corporate tax and the capital gains tax. The surtax would increase the latter, but deferral would again blunt the impact.
Taxing capital income is distortionary because it imposes a penalty on future consumption relative to current consumption. Economic analysis suggests that the optimal tax rate on capital income, even taking distributional and revenue concerns into account, is probably close to zero. The surtax would therefore move the tax system further away from the optimum.
Any reduction in saving caused by the surtax would have ramifications throughout the economy. Although part of the reduction would show up as smaller holdings of foreign assets, much of it would show up as a smaller U.S. capital stock, which would make workers less productive and drive down real wages. Statistical estimation of the impact of tax rates on saving is difficult, but simulations suggest that the impact could be significant. For example, simulations reported by Treasury’s Office of Tax Analysis in 2006 found that extending the 2003 dividend and capital gains tax cuts would increase long-run real gross national product by 0.2 percent to 0.5 percent, depending on the economic assumptions and how the extension was financed. As I noted in a previous analysis, the estimates in the Treasury study imply that real wages would rise by as much, or more, than the revenue loss from extending the tax cuts. The effects of raising the tax rate on dividends and gains would, of course, go in the opposite direction.
Given the effects on saving and capital accumulation, it is hard to defend the surtax as good policy. But, the other considerations discussed below provide further grounds for opposition.
Limits of Taxing the Wealthy
As I have noted elsewhere, the amount of revenue that can be raised from taxing only the top 2 percent or 3 percent of incomes is relatively modest. The point applies all the more strongly to the even smaller group that the proposed surtax would target.
The key issue concerns the long-run limits on taxing the very wealthy. Simple calculations show that this group cannot finance all, or much, of the long-run fiscal gap. To be sure, as I noted above, households that would be subject to the surtax earned 16.9 percent of the nation’s AGI in 2007. To begin, however, only the income above the $500,000 or $1 million threshold would be subject to the surtax, which reduces the surtax base to 12.2 percent of the nation’s AGI. Now, consider, for example, an increase of 20 percentage points in the marginal tax rates on this income. Even if such a steep increase in marginal tax rates had no impact on behavior, it would raise only 2.4 percent of national income. This amount is dwarfed by the size of the government’s long-run fiscal gap; in a baseline roughly corresponding to current policies, the CBO recently estimated that the deficit would grow to 14.6 percent of GDP in 2035.
The conclusion about the limits of taxing the very wealthy does not rest on my calculations. The conclusion has been reached by a wide range of analysts across the political spectrum:
In short, there is a broad consensus that taxes cannot be raised solely on the very rich, even among those who believe that taxes should be raised on that group. Fiscal burdens will eventually have to be imposed on the middle class and any new spending program will ultimately add to that burden. Linking a major spending program, such as healthcare reform, to taxes on a small group of wealthy people hides that reality.
Several commentators have recognized the need for broader financing of healthcare reform. Leonard Burman, former director of the Urban-Brookings Tax Policy Center, recently noted that Social Security and Medicare Part A are financed by broad-based payroll and self-employment taxes, giving all Americans a vested interest in the programs, and urged that the same philosophy apply to healthcare reform. William Gale of the Tax Policy Center similarly comments, “If health care is a benefit worth having, then it’s worth paying for. This [surtax] gives the impression that it’s worth having if someone else pays for it.” Thorndike observes:
When you ask a small minority to pick up the tab for the vast majority, you implicitly devalue the program. . . . An approach of something for nothing is not the basis for a lasting progressive revolution. . . . If Congress decides to finance a mass program with a class tax, it will have built a grand edifice on a shaky political foundation.
Although space does not permit a full description of all of the financing mechanisms included in H.R. 3962, it should be noted that some of the other mechanisms are also designed to avoid visible costs on any large segment of the population. For example, the Medicare reimbursement cuts are marketed as affecting providers rather than beneficiaries and the tax on medical devices is described as falling on producers rather than patients.
Of course, these failings should be placed in perspective. This would not be the first time that federal healthcare spending has been expanded without proper attention to financing. In 2003, Congress and President George W. Bush enacted a Medicare prescription drug benefit without any budgetary offsets. By providing offsets, however imperfect, Obama and congressional Democrats have chosen a more courageous and responsible path than their predecessors and they deserve considerable credit for doing so. Nevertheless, a financing mechanism broader and less distortionary than the surtax is necessary on both political and economic grounds.
The millionaire surtax in the House healthcare bill is politically attractive because it would apply to only a small group of very wealthy households. Unfortunately, it would raise marginal tax rates, spurring tax avoidance, hampering saving and investment and reducing real wages throughout the economy.
Attempting to impose the costs of a major new federal program on a mere 0.3 percent of the population is ill-conceived and unsustainable. If we decide that healthcare reform is worth doing, we should decide to pay for it in a broad-based manner.
Alan D. Viard is a resident scholar at AEI.
There are no comments available.
1150 17th Street, N.W. Washington, D.C. 20036
© 2016 American Enterprise Institute for Public Policy Research