The Last Best Hope For Progressivity in Tax

Edward Mccaffery

about 12 years ago

The current landscape of U.S. taxation reveals a critical moment for progressivity, as President Obama’s attempts to restore higher tax rates for the wealthiest Americans face significant challenges. Despite initial hopes for a more equitable tax system, the reality is that the proposed changes fall short of historical norms, leaving many progressives disillusioned.

Understanding the implications of these tax policies is vital, as they shape the future of income redistribution in America. The article delves into the historical context of tax rates and the ongoing struggle for a more just system, raising questions about the feasibility of true progressivity. What does this mean for the average citizen, and how might it affect the broader economic landscape?

Written by: Edward J. McCaffery and James R. Hines Jr.

1. Introduction

Pity President Obama. Pity tax. Pity the dreams of progressives everywhere.

Barack Obama’s presidential campaign gave believers in redistribution a brief shining moment of hope. Obama’s candidacy featured stirring calls for the richest Americans to give back and share more with their less fortunate fellow citizens. Yet even then, beneath the high- flying rhetoric, there was reason for despair. For there was little talk about any kind of significant return to progressivity in the tax system, the main policy instrument for effecting redistribution in America. Now, over a year into the Obama Administration, the President has been unable to enact the most progressive element in his tax program: a partial repeal of President George W. Bush’s 2001 and 2003 tax cuts, largely meaning a restoration of the two top marginal rate brackets under the federal income tax from 33 and 35 to 35 and 39.6%, respectively.[2] The President has consistently underscored that no one earning under $250,000 of income – that is, no one not among the top few percentage points of earners[3] would see a tax increase, and no one at all would see her rates raised above the level they reached by 1999.[4] And the President is at pains to stress that he is not actually even raising anyone’s tax rate, simply letting the Bush tax cuts expire for top earners.[5]

Yet the 39.6% top rate is far below one-half of the highest marginal rate in the less-than- century-long history of the contemporary income tax, a rate obtained in the midst of World War II, of 94%.[6] Tax rates persisted as high as 90% throughout the 1950s, until President Kennedy cut the top rate---to 70% in 1963. That rate in turn obtained until President Reagan, who, in 1981, first cut it to 50% before bringing it down, to 33 or 28% (depending on how one looked at it)[7] in 1986. President Obama is advocating raising the top rate on highest income earners by a mere 4.6% in absolute terms, resulting in a final top rate more than 54% lower, again in absolute terms, than the historical peak, and far below the rate that obtained for more than four decades, from 1941 to 1986.

This is hardly radical stuff. Yet the last days of the presidential campaign, featuring Senator John McCain’s whole-hearted embrace of “Joe the Plumber,” showed that even Obama’s extremely modest proposal was extremely controversial. McCain sneeringly called Obama “Redistributionist in Chief,”[8] even as Obama himself tried to distance himself from the word “redistribution” or any of its cognates.[9] Obama continues to insist that his proposals will not actually raise any tax rate; he is simply letting the Bush tax cuts expire to restore the status quo ante before he took office. Still the attacks on the marginal tax rate proposals come swiftly and furiously.[10] While a looming fiscal crisis makes a return to 1999, at least for the very top income earners, quite possible – it is too early to tell on that score --- attempts to interject more progressivity, as in the quickly aborted (if not yet dead) “surcharge” under health care reform, appear beyond the pale.[11]

What are we doing? Where are we headed? Are we forever set in a post-Reagan mindset when it comes to tax rates? What has happened to the argument for more steeply progressive marginal rates? Why does a country that seems to believe in redistribution not seek it more forcefully in its major policy instrument devoted to doing so?

There are many aspects of an acceptable answer to this puzzling question, involving politics, perception and more.[12] This Article adds one important piece from the teachings of neoclassical economics and traditional tax policy. It explains that the case for more progression in tax rates has been fatally hampered by the choice of tax base. This is, in part, a story of how academics, ordinary politics, popular understanding---and inertia---can lead a democracy to a place far removed from its initial hopes and dreams.

Simply put, there are compelling economic reasons for avoiding more steeply progressive rates under an income or a wage tax, such as we have in America today. President Reagan successfully made this case out over thirty years ago, and he had powerful academic and intellectual support for it in the writings of the Nobel Laureate James Mirrlees and others. Political rhetoric picked up and tracked the economic fundamentals. The journalist David Leonhardt, in a long profile of Obama’s economic philosophy published in the New York Times Magazine published shortly before the election, reflected the prevailing view:

When Reagan was elected, in 1980, tax rates on top incomes were so high that even liberal economists now say the economy was suffering. There simply wasn’t enough of an incentive for rich people to start new companies or expand existing ones, because so much of their profits would have gone to the federal government.[13]

By 1980, “even liberal economists” had come to fret over the effects of high tax rates on productive incentives. Add our names to the choir. But note that Mirrlees’ important academic analysis, Reagan’s political rhetoric and accomplishments, and Leonhardt’s journalistic conclusion all relate to taxing income. Here is where inertia plays its pivotal role: by failing to rethink initial premises in regards to the appropriate tax base, we have doomed the prospects for more progressive tax rates. In this Article, we argue that switching to a spending tax base holds out the possibility for a new analysis, and a return to more steeply progressive marginal tax rates. Hence our title, echoing Lincoln: we believe that a spending tax is the last best hope for progressivity in tax.

The simple explanation for this central claim looks to the pattern of incentive effects, especially on “rich people,” as both economists and commentators such as Leonhardt are inclined to focus. (Joe the Plumber, after all, at least pretended to be rich.). High marginal tax rates under an income tax fall on and hence deter the productive activities of work and savings. High marginal rates under a wage tax fall on and hence deter the productive activity of work alone. But high marginal rates under a spending tax fall on and hence deter high-end spending, which is arguably a social “bad,” and do not necessarily deter the social goods of work and savings, and may even promote more private savings. Hence it is possible to have higher marginal rates under a spending tax than under a wage or income tax with equivalent or lower efficiency costs.

This is a possible empirical result. In this Article, we present the analytic arguments for it and sketch out a research agenda that might verify it. The idea is that because one can escape or defer paying taxes under a progressive spending tax by saving, an activity with positive social externalities, the efficiency costs of high marginal rates under a spending tax can be mitigated. Unless people work only in order to be able to spend on themselves, and even then only if they fully internalize in their present labor supply decisions the ultimate tax they will pay – and we argue that each of these assumptions is unlikely to hold in the extreme - a spending tax can bear more progressive rates with less cost in efficiency or social wealth than can an income or wage tax. A progressive spending tax also holds out the possibility of sorting the rich or high ability into two groups, of elastic savers and inelastic spenders, which could yield welfare gains unavailable to income or wage taxes, which under present technologies can only sort the high ability into workers and non-workers. Finally, progressive spending taxes, unlike wage taxes, fall on consumption financed by windfall gains, as to which unexpected good fortune ex ante incentive effects are likely to be weak – and hence such taxes should have lower efficiency costs.

We know that our central claim will seem counterintuitive to many, perhaps most, readers. How can a spending tax be more progressive than an income tax? Since most income is spent, the two tax systems are obviously similar, indeed identical for the vast majority of American taxpayer-worker-spenders. Any differences follow from the fact that a spending tax base does not include income that is earned and not spent --- that is, savings. Since the rich do most of the saving, it would seem as if their incomes are the ones that are most exempted by a spending tax. At first blush it appears that a spending tax would have to be less progressive than an income tax.

This initial take on the matter, however, fails fully to factor in the effect of the choice of tax base on the pattern of feasible tax rates. If tax rates under a spending tax are considerably more progressive than those under an income tax – and there is ample reason to think this can be the case – then a spending tax can be the more progressive alternative. That is the project for this Article.

What about the comparison between a spending tax and a wage tax? While these tax systems are commonly lumped together in the economics and tax policy literatures as roughly equivalent variants of consumption taxes, and in fact have clear analytic similarities, there are nonetheless important practical differences. One especially important difference is that a wage tax does not attempt to tax unexpectedly high returns to capital investments, sometimes known as “windfalls.” It is not only the case that the rich save; it is also the case that the “ex post” rich those who turn out to be rich, looking backwards through time tend to be those whose capital investments earned extremely high, indeed unexpectedly high, returns. Most of Warren Buffett’s or Bill Gates’ income has come from capital specifically, the capital appreciation in the price of Berkshire Hathaway or Microsoft stock and hence would be exempt from wage taxation. Yet the same income would be taxed under a spending tax when consumed. A spending tax alone among feasible alternatives can collect significant tax revenue from the propertied, capitalist classes, in a way that does not necessarily undercut their incentives to become propertied in the first place.

The Article begins with an overview of the logic of tax, progresses to separate consideration of the base and rate issues, and then, in combining bases and rates together, turns to explain why more progressive rates under wage or income taxes are unlikely, but possible--- and possibly even desirable---under a spending tax. We end with a call for more research, and hope.

2. The Logic of Tax

Any tax is the product of a base, or what is being taxed, and a rate structure, or how much the base is being taxed. The government must choose both, a base-rate pair. Tax policy for decades and by some measures centuries has exhaustively considered the base question, where a debate has raged over income versus consumption taxes.[14] A far quieter debate has taken place on the rate structure, where proponents of progressivity—the norm of expecting the rich (or those with greater “ability to pay”) to pay more, in percent terms, than the not-rich—have been hanging on, fighting back political and to some extent intellectual arguments for flat-rate taxation.[15] The dominant framework for the analysis of tax rates has become the optimal income tax literature, most importantly owing to Mirrlees’s work starting in the early 1970s.[16] For the most part, the base and rate issues have not been well joined. Where they have been connected, a basic misunderstanding haunts the analysis: the case for consumption taxation, gaining adherents at an accelerating pace in the academy, has been unnecessarily linked, in the intellectual history of tax, with flat or proportionate tax rates.

While tax bases and tax rates are analytically distinct subject matters, logically separable, they are linked politically, rhetorically, and economically. Simply put, the choice of base affects the arguments for, and the effects of, the degree of progression in the tax rate structure, and vice versa. Our primary purpose in this Article is to press a positive, descriptive argument that, in general, the nature of the effects of tax rates depends on the tax base and, in particular, the case for significant increases in the progressivity of the marginal rate structure is stronger under a certain form of consumption tax, namely a “postpaid,” expenditure, or (all equivalently) spending tax, than it is under either an income or a wage tax.

3. Tax Bases

3.1. Income Tax

We begin with an income tax, the most commonly thought of comprehensive individual tax system. An income tax falls on all present period earnings, from “whatever source derived,” in the language of the XVIth Amendment to the U.S. Constitution and Section 61 of the U.S. ROBERT E. HALL AND ALVIN RABUSHKA, THE FLAT TAX (Palo Alto, CA: Hoover Institution Press, 1985; 2nd edition 2007). For counter-arguments, see Joseph Bankman and Barbara H. Fried, Why Not a Flat Tax?, Boston Review (Summer 1986).[16] James A. Mirrlees, An Exploration in the Theory of Optimum Income Taxation, 38 REVIEW OF ECONOMIC STUDIES 175 (1971).

Internal Revenue Code (“IRC”), that is, from labor or capital (and, arguably, from beneficent transfers as well). As Henry Simons put it in an influential definition, named after him and his predecessor Robert Murray Haig:

Income may be defined as the algebraic sum of the market value of the rights exercised in consumption plus the change in value of the store of property rights between the beginning and end of the period in question.[17]

We can restate this definition, in reality a simple accounting relation or tautology, as: Income = Consumption + Savings This tells us little more than the facts that inputs or incomes must equal outputs or outflows, or that all wealth (Income) must be either spent (Consumption) or not (Savings).

From the Haig-Simons definition we can see that savings are at a disadvantage compared to immediate consumption under an income tax. Wealth is taxed when it first comes into a household, typically via wages. For wealth that is not immediately consumed, a further tax is levied on the yield to the savings, as when interest is credited to a bank account, or stocks or real estate rise in value. As John Stuart Mill first pointed out in 1848, the inclusion of the yield to capital in the income tax base makes it a “double tax” on savings.[18]

3.2. Two Consumption Taxes

In part to avoid the income tax’s double-tax sting and its attendant distortions, critics have long advocated some form of consumption tax that would exempt savings from the tax base, on both economic efficiency and fairness grounds.[19] The efficiency case against taxing the return to capital is strong. An income tax creates a bias in favor of present consumption over deferred consumption, that is, savings. The bias of the tax widens over time. Compound interest implies that the tax system’s distortions loom larger as the time horizon lengthens, thereby making even a very low rate tax on the return to savings highly distortionary over a lifetime perspective. A thirty percent tax on the return to savings, for example, increases the cost of tomorrow’s consumption relative to today’s, but it much more dramatically increases the cost of consumption ten years from now compared to today’s.

But the case for consumption taxation is not just a matter of efficiency or traditional welfare economics analysis. Taxing savings at a time of low aggregate national and individual savings strikes many as an unwise idea, and one that is unfair to the vast numbers of middle-class Americans attempting to save for retirement, to buy a house, put their children through college, hedge against economic misfortune, and so on. It should not therefore be surprising to learn that the so-called income tax is replete with nominally pro-savings provisions, reflecting the ambivalence of policy.[20]

The case for consumption taxation has deep roots, going back to Thomas Hobbes in the seventeenth century and Adam Smith in the eighteenth, before moving on to Mill in the [19] Fisher and Fisher, supra; Kaldor, supra; Andrews 1974, supra. Kenneth L. Judd, Redistributive Taxation in a Simple Perfect Foresight Model, 28 JOURNAL OF PUBLIC ECONOMICS 59 (1985); and Christophe Chamley, Optimal Taxation of Capital Income in General Equilibrium with Infinite Lives, 54 ECONOMETRICA 607 (1986).[20] See our discussion of “hybrids” in Section 6, infra, and of the real-world tax system, in Section 9, infra. nineteenth and Irving Fisher, Nicholas Kaldor, and others in the twentieth. In contemporary times, however, the case for moving to a consumption tax has been held back by analytic confusion: It seems as if a standard consumption tax must be more regressive than an income tax. This is so both for reasons of the base, because a consumption tax is thought not to include the yield to savings, the virtually exclusive domain of the rich, at all, and of rates, because it is further thought that a certain form of consumption tax, in order to be a consumption tax at all, must feature flat rates.

To see this latter point, it helps to understand the two canonical types of consumption taxes. Tax policy theorists have noted the equivalence of two forms of consumption-based taxes under plausible assumptions, namely the “prepaid” or “yield-exempt” tax model and the “postpaid,” cash-flow, or expenditure tax.[21] We shall hereafter refer to these two forms of consumption tax by their most commonly known incarnations, wage and spending taxes, respectively. If tax rates do not change and there is no uncertainty, then levying a tax up-front, and never again, as does a wage tax, is equivalent to deferring the tax and levying it at the single time of ultimate private preclusive use, as does a spending tax.[22] This point is evident algebraically, which we set out in the notes,[23] and through a simple example.

Suppose that Ant, the iconic saver, earns $200, the tax rate is 50%, and the return to savings is 10%. Ant will save for two full years then consume. Under a wage tax, Ant’s $200 is reduced to $100 right away. This grows to $100 + $10, or $110, (P (1 + r)), after one year, and after two years. Ant spends away.

Under a spending tax, Ant gets to save the full $200. This sum grows to $200 + $20, or $220, after one year, and $220 + $22, or $242, after two years. When Ant goes to spend, the government comes to assert its 50% stake under the spending tax. This reduces Ant’s $242 to $121, just as under the wage tax model. The algebraic formulation above proves that this result is not simply an artifact of the numbers chosen for the example, but a fully general result, with t and r constant.

We can see the same idea more graphically. Return to the basic Haig-Simons definition of Income:

Income = Consumption + Savings

This basic identity identifies sources with uses. The income tax is a classic source-based tax, looking, in the language of the Constitution, to all “incomes, from whatever source derived.” Consider what the sources of income might be, in broad terms: payments for labor or for capital, the two great factors of production, one’s own or another’s, the latter cases entailing beneficent transfers. In short:

Since the annual rate of return (r) is certain, as is the tax rate, it does not matter when the government reduces potential consumption by a tax rate, t, leaving the taxpayer to consume (1 – t). Under the commutative principle of multiplication, which holds that ab = ba:

This identity holds as long as r and t are constant.

A wage tax gets to be a single-tax on savings by exempting capital as a source of income on the left-hand side of the equation, as it were; it, like an income tax, is source-based, but systematically ignores the returns to financial capital as a source. Think of the payroll tax or social security/medicare contribution system, which make no attempt to tax savings. A spending tax, on the other hand, exempts savings as a use, that is, on the right-hand side of the equation, as it were, following the simple logic of

Income – Savings = Consumption.

Think of a common sales tax, which does not apply to money saved in a bank account. Both wage and spending taxes are single taxes on a household’s flow of funds, whether saved or spent.

3.3. The Traditional Income versus Consumption Debate

The income-versus-consumption-tax debate has taken place in the shadows of Mill’s critique of the income tax as a double tax on savings or, equivalently, non-consumed wealth, and the analytic understanding just sketched out. Wage taxes exempt capital returns by design. For spending taxes to produce the same result, they must feature flat tax rates. This traditional view of tax has been unfortunate, because the systematic non taxation of the return to financial capital is only one reason for moving to a consumption tax, and not necessarily the best one. Our central claim in this Article is that restoring or increasing progressivity in the marginal rate structure is another reason to adopt a consumption tax, of the right sort --- a spending tax.[24] To see this point, it is important to understand the traditional-income-versus-consumption-tax debate, and then to get beyond it.[25]

It is easy enough to see that a wage tax avoids the double-tax sting of the income tax. Such taxes have, indeed, been the most popular form of tax reform proposal coming from conservative political factions of late: the most common “flat tax” plans, as from Hall-Rabushka, feature a two-bracket wage tax with no additional taxation of savings.[26] In order for the second form of consumption tax, the spending tax, to be equivalent to a wage tax and thus to avoid any taxation of savings, it must, as the Ant example illustrates, feature constant or relatively flat rates. Hence, the “Fair Tax,” a widely publicized proposal for a national retail sales tax,features a flat-rate spending tax.[27] But the analytics are not normative; an ought does not follow simply from the is. Whether or not we want a spending tax to be equivalent to a wage tax depends on why we want spending taxes, a decision which must be foundational to the inquiry. Hence we arrive at the following important general points that we set off in italics for emphasis, as they factor throughout our analysis:

If the principal aim of a consumption tax is to avoid any additional taxation of savings, the choices come down to wage or flat spending taxes, which are broadly equivalent.

But if we have another reason for moving to a consumption tax, particularly one looking to the rate structure, wage and spending taxes are no longer equivalent and more analysis must be done. A progressive spending tax emerges as a distinct option for comprehensive tax reform.

3.4. Three Types of Tax

Returning to our central project, the important point to see is that the traditional income- versus-consumption-tax debate, centered on the yield to savings, in essence eliminated an important base-rate pair from further analysis, the progressive spending tax. Those who want to tax the return to savings came to favor an income tax; those who thought it unfair or inefficient or both to ever tax savings favored a wage or a flat spending tax. For under progressive marginal rates, the automatic, analytic equivalence of wage and spending taxes disappears. The tax rate at the time of initial earnings may not be the same as that at the time of spending. Savings could be “subsidized” or “penalized” vis a vis an income or a wage tax.[28] This analytic fact led many to argue for flat or constant-rate spending taxes, because they thought it a norm not to tax capital or its yield at all--- they bought into Mill.[29] But this argument, typically generated by policymakers invoking conceptions of fairness—and hence parallel to, but distinct from, the efficiency-based arguments—presumes that the principal normative reason for choosing a consumption tax turns on the principled non-taxation of the yield to savings. It need not be so. Policymakers might desire a progressive tax on spending for reasons unrelated to the non-taxation of capital or its yield, or at least not pre-committed to such non-taxation.

Consider two distinct arguments.

One, a progressive spending tax might generate a welfare-maximizing outcome: it might be the “best” base-rate pair in a specifically welfarist sense. This could be so, as we explain further below, because individuals in the face of such a tax may continue to work and save, avoiding present period taxation but improving social welfare both by the positive externalities of their contributions to the capital stock and by the implicit deferred tax on the savings. Call this the welfarist argument.

Two, policymakers might consider it fair and appropriate to tax some but not all savings, and a progressive spending tax is a particularly good and principled way to do so. Call this the fairness argument.

We shall explain and explore these two arguments further below, and return to them especially in Section 10, when we discuss normative implications of our analysis and call for further research.

The critical descriptive, analytic point is that, with progressive marginal tax rates, there are three distinct choices for a comprehensive tax base: income, wage and spending. Each affects the taxation of savings differently. An income tax double taxes all saving, whatever their use. A wage tax exempts all savings, whatever their use. A spending tax splits the difference. To understand all this more fully, we need a richer understanding and vocabulary of different types of savings. We get to that in Section 5. We first leave base matters aside temporarily to develop a better vocabulary and understanding of tax rates, the other half of any base-rate pair, for progressivity is the main focus of the Article, and it is progressive rates that destroy the equivalence of wage and spending taxes and give us three not two choices for progressivity’s future.

4. Tax Rates

4.1. Vocabulary

Tax rates form a function over the relevant range: income, wages, or spending, depending on the base. These rates can be progressive, meaning increasing; flat or proportionate; or regressive, meaning decreasing—and, indeed, can be all three of these shapes over differing ranges of the base.

There are two different meanings of “tax rates.” The income tax in the United States and most developed nations around the world depends on a system of progressive marginal rates that work like a step function. Thus, and to simplify, for a family of four,[30] there might be a “0 bracket” reaching up to $20,000; a 15% bracket extending from $20,000 to $60,000; and a 30% bracket over that, as presented in

Table 1:

Table 1: Sample Marginal Rate Schedule

This means that the family’s “first” $20,000 is not taxed, and their “next” $40,000—the income that takes them from $20,000 to $60,000---is taxed, on the margin, at 15%. It does not mean that all of the family’s income is taxed at 15%. So, for example, a family making $40,000 would pay $3,000 in taxes under this simplified rate structure: 0 on their first $20,000, plus 15% of $40,000 – $20,000. An interesting aside is that virtually all “flat rate” tax plans that have been proposed in the United States are in fact “two rate” plans, hence featuring progressive marginal rates: the plans typically have an exempt level or “zero bracket,”followed by a positive tax bracket, of somewhere between 15 to 30%.[31]

Average or (equivalently) effective tax rates, in contrast, equal the total taxes paid by a taxpayer divided by her income or other base. Using the same simplified example from Table 1, a family of four would face an average tax rate of 0 on their first $20,000 of income. At $30,000 of income, the family would pay $1,500 in taxes (15% of $30,000 – 20,000), for an average tax rate of 5% (1,500/30,000); at $40,000, the family would pay $3,000 in taxes, as calculated above, for an average tax of 7.5% (3,000/40,000).

Under a system of progressive marginal rates, average tax rates, after the initial zero bracket range, keep increasing, asymptotically approaching the highest marginal rate as the base goes out to infinity. That is, again in the simple example, the taxpayer family will eventually face an average tax rate of practically 30% as their income increases, so that the “benefit” of the initial 0 and 15% rate brackets becomes a trivial part of the whole. As long as the marginal tax rate is higher than the average, the average is increasing—the higher marginal tax rate pulling it up.

4.2. Adding Demogrants to the Mix

While progressive marginal rates necessarily mean progressive average tax rates, the converse does not hold. Progressive average rates can also obtain with a system of proportionate or even declining marginal tax rates, as long as the marginal rate exceeds the average. This is an important aspect of optimal tax analysis, explained more fully below. The idea is easiest to understand together with a common feature of optimal tax models: the use of a lump-sum rebate or “demogrant.”[32]

Consider adding to the simple illustration using Table 1 a demogrant of $750 per person, $3,000 for our family of four. With no other changes, there would now be negative average tax rates: a family earning $10,000,for example, would pay a negative $3,000, because it would get the rebate without paying any tax, for an average rate of -30%. The family earning $40,000 would net out at 0, and so on. Table 2 summarizes, adding in a$100,000 household:

Table 2: Average and Marginal Tax Rates with Demogrant

Suppose finally that we changed the rate structure from Table 1 to be a “flat” 20%, but kept the demogrant of Table 2 in place. The family at $10,000 would net out at -1,000, a -10% average tax rate; at 40,000, the familywould pay a positive 5,000, for a 12.5% average tax rate; by $100,000, the family would pay $17,000, on net, for a 17% average. Table 3 summarizes:

Table 3: Average and Marginal Tax Rates with Demogrant and “Flat” Tax

Under the tax system represented in Table 3, there are progressive average tax rates without progressive marginal rates. Indeed, the average rate for each family in Table 3 has gone up, compared with Table 2,although this would not continue to be true for wealthier families. (The asymptote is now 20% not 30%). The insight that progressive marginal rates are not needed to obtain progressivity in average tax burdens is especially important to optimal tax, where a core finding of almost all models is declining marginal tax rates.

4.3. What Matters About Rates?

Most policy discussions of tax in a welfare economics tradition simply define progressivity as referring to average tax rates.[33] This follows simply enough if one is looking at a snapshot of distributive outcomes;measuring income (or wages or spending) pre and post-tax. In the fullest and most complete statement of optimal tax analysis, as Louis Kaplow has forcefully pointed out, all government tax and transfer programs would be netted out, just as demogrants and marginal tax rates are in the classic Mirrlees formulation, and as we did in Tables 2 and 3.[34]

In more technical philosophical terms, average rates point to end-state distributive justice: the resources people have after all government programs are considered. In the standard welfarist conception, marginal tax rates matter only for their incentive effects at the individual level---for their effects on efficiency---such that, in a very crude nutshell, the project for optimal tax, at least assuming a redistributive social welfare function, is to generate progressive average tax (net tax and transfer) rates while minimizing the marginal disincentives to work, especially among the most able. Ultimately, this division of labor between the two types of rates leads to declining marginal rates, often ending in zero on the highest able worker, and sometimes even going negative at the top, while average rates are constantly increasing, with demogrants doing the work to ensure the redistributive result.

Further investigation, however, suggests that marginal rates can matter, both to conceptions of fairness and to efficiency, once we consider changing the tax base along with the rate structure. The basic insight is simple: because optimal income tax has developed considering an income or wage tax, marginal rates have affected only the decisions to work and save. The disincentives generated by high rates lead to a socially inefficient substitution of present consumption for savings and leisure for labor. But if the base is allowed to vary, to include exclusions for charitable contributions, for example, as suggested by Peter Diamond,[35] or savings,as is our principal theme here, high marginal tax rates can push people to give to charity or to save, as well as or in lieu of adding leisure, and these activities---savings and philanthropy have very different equity and efficiency characteristics than does consumption or leisure. Both philanthropy and savings, for example, can be seen as public goods, which benefit all. In such cases, marginal rates, independent of their influence on average rates, can shape matters of concern to both efficiency and fairness advocates.

This is a different argument than one pressed by others, such as by Lawrence Zelenak and Kemper Moreland in the legal literature.[36] Zelenak and Moreland argue that marginal tax rates are relevant because they are related to average tax rates, and, in fact, are necessary to achieve any form of progressivity in tax once demogrants are ruled out. These are analytically correct claims: without some kind of demogrant, progressive marginal rates are for all practical purposes needed for progressive average rates.[37] But this is still an argument about average tax rates, given a political constraint---no demogrants. We agree that America is unlikely to engraft a significant new demogrant onto the existing government spending platform, as we discuss below. But our argument for the relevance of marginal tax rates is logically independent of their impact on average tax rates.

In brief, marginal incentives matter. Obtaining average rate progression by maintaining steep marginal rate progression under an income or wage tax means sacrificing efficiency--- welfare or wealth---to obtain equity. This is what politicians, commentators, and ordinary citizens have noticed, ever since Ronald Reagan’s presidency – and it is what haunts President Obama today.[38] Minimizing the tradeoff between efficiency and equity is why optimal income tax models accept declining marginal tax rates on the upper income. These declining rates can be justified on the Paretian principle that it is better for the most able to work than not to work. But then these declining marginal rates under a wage or income tax—reaching zero in many models for the most able[39] mean that society must sit by and watch as its most highly compensated citizens, at the margin, get richer while paying little or no additional taxes, and are free to spend what they earn. This pattern is what bothers Zelenak and Moreland, along with many if not most ordinary citizens.[40] The philosophers Liam Murphy and Thomas Nagel in their work on tax policy make a similar point, which we comment on below. But resolving the dispute by insisting on progressive marginal rates under an income tax still means choosing equity over efficiency. Changing the tax base changes the tradeoff.

4.4. A Short Exploration of Optimal Tax Theory

We have already introduced the idea of optimal tax, the dominant contemporary analytic framework for considering tax rates. It is time to explain the idea more fully.

4.4.1. The Basics

Mirrlees, beginning with a seminal paper in 1971, famously led the development of optimal income tax theory, building on the much earlier optimal commodity tax work of Frank Ramsey[41] Ramsey had demonstrated that an optimally efficient government should set differential commodity tax rates based on the demand for different goods, generally leading to the “inverse elasticity rule” whereby inelastically demanded goods should be most highly taxed. Mirrlees situated the optimal income tax problem in a welfare and information economics setting, analyzing the properties of income taxes designed to redistribute income in the most efficient possible manner, given the information limits facing governments.

Importantly, because there is only a single period in Mirrlees’ model---no time---there is no savings. All income comes from wages and is spent in the single period. This is a wage tax. Taxes generate revenue for the government that in the model are simply turned back over to the people via demogrants. Given a social welfare function with declining marginal utility of money income, a Mirrlees-style model can lead to large amount of redistribution. In a certain limiting case, where taxes have no distorting effects, the government would tax all wages and return demogrants to all to equalize after-tax wealth. But of course taxes do distort, and therein lies the rub.

The specific problem facing the government in a Mirrlees model is an information one: individuals know their abilities, which, in the model, represent their abilities to cash their native talents out into wages, but the government cannot observe this information. In the face of high marginal tax rates, taxpayers, as the agents to the government’s principal, might “shirk,” substituting leisure for labor, earning less, and there by appearing to be, in the government’s eyes, of lesser ability than they in fact are. This formulation is analogous to the perhaps more familiar language of “substitution effects:” Tax rates distort the allocation of resources, which ought, in standard welfare economics theory, to be set efficiently by the pre-tax price system, causing taxpayers to substitute untaxed leisure for taxed labor because of the fall in the after-tax price of labor.

The major implications of the Mirrlees model and its considerable progeny[42] are that, given a redistributive social welfare function and with certain restrictions on technology, skills, and individual utility functions, the marginal tax rate on labor earnings should (1) never be negative, (2) attain its peak in the middle of the income range and be declining at the upper ranges of income for workers with high wage-earning ability,and (3) reach zero on the last dollar earned by the highest ability wage-earner (though Mirrlees himself did not have this top rate of zero).[43] While there is great disagreement about the precise contours, there is little argument that marginal tax rates end up declining over the upper income (or ability) range. Matti Tuomala, a prominent proponent of the optimal tax tradition, has put it simply that

. . . one of the main conclusions to be drawn from the Mirrleesian optimal non-linear income tax model is that it is difficult (if at all possible) to find a convincing argument for a progressive marginal rate structure throughout.[44]

Mirrlees included simulations in his seminal 1971 paper that featured peak (that is, highest tax rates, typically on the lower-middle income class) and highest end (that is, tax rates for the most able or highest income)marginal tax rates of 26 and 16; 20 and 15; 28 and 19; 34 and 20; 39 and 21; and 60 and 49.[45] These peak-highest end rate pairs show not only the considerable range of possible outcomes under an optimal tax analysis(depending on the social welfare and individual utility functions used), but also the general pattern of peaking then falling. Later analysis pushed the case out to finding a zero rate---even, in the case of one model advanced by the Nobel laureate Joseph Stiglitz, a negative rate---for the highest earner/most able citizen.[46] In all optimal tax models, progressivity in average tax rates is achieved by means of demogrants,combined with the pattern of often increasing (over low to middle income ranges), but intermittently declining (especially over upper income ranges), marginal rates.

The intuitions behind the complex mathematics of optimal income tax models are simple enough to state. The pattern of tax rates once again reflects the tradeoff between the distorting effects of taxes on labor supply decisions, on the one hand, and the benefits of producing and then redistributing income via the tax system, on the other. While higher marginal tax rates at low income levels have the undesirable feature of discouraging work effort by low-ability taxpayers---one aspect of many of the models is that some citizens will rationally chose not to work[47]---they also offer the prospect of raising significant tax revenue from all taxpayers,including the high income or ability ones, for whom the higher rates are infra-marginal. That is, the high tax rates on low levels of income occur well “before” the high income/ability taxpayers make their marginal decisions about additional work effort. If a well educated lawyer can earn $250,000 a year, for example, her marginal work decisions---should she work those extra hours towards the end of the year to get a larger bonus, say---are unlikely to be affected by a high marginal rate on her first $50,000 of income. The government can then redistribute the resulting tax revenues to all citizens, including the poor, via demogrants.

High marginal tax rates at upper incomes and abilities, in contrast, while they also distort labor supply, here of high income/ability taxpayers, raise only modest amounts of revenue, because they apply only to a small fraction of the labor force. Unless the labor supply of the rich is unusually unresponsive to taxes, that is, inelastic---and the opposite is likely to be the case--- tax rates applying to upper-income levels attempt to tax what is a very elastic as well as small base. Hence optimal income taxation entails declining marginal tax rates over the upper ranges of the income distribution. Suppose that our well-paid lawyer from the prior paragraph faces a 90% rate on any earnings over her $250,000 pay. Now would she work those extra hours in December, say, giving up time with family and friends in order to keep 10 cents on the dollar? In the limiting case, it is better—for all, under the standard Paretian condition—for a worker to work than not to work simply for tax reasons. Hence the zero tax rate on the last dollar earned by the otherwise fully-deterred highest potential wage earner. Another way to express the same concept is that there is nothing for society to gain by imposing a positive marginal tax rate on the most able worker, and thereby distorting her labor supply, since such taxes yield no greater revenue than less distorting alternatives. It is always better to eliminate the taxation of the last dollar earned by the most affluent taxpayer, paying for this reduction by increasing the taxation of income just below that level, since doing so creates fewer labor market distortions.

4.4.2. Qualifications, Questions, and Caveats

That is the basic framework for optimal income tax analysis, which has been highly influential in the academy and beyond. But there are several important qualifications.

One, optimal tax models are extremely sensitive to changes in key assumptions and parameters, as the simulations from Mirrlees’s initial article itself suggest. The most important variable in a public policy setting is labor supply elasticity. This is typically found to be low for men, at least, suggesting that tax rates can be higher than some models would otherwise suggest.[48] (This raises puzzling questions about why we do not have better mechanisms to lower the taxable burden on women, especially married women, who typically register higher labor supply elasticities, but we leave that puzzle for another day.)[49] In the face of this empirical observation, the literature has generally shifted from the elasticity of labor income, per se, to the elasticity of taxable income.[50] Here, scholars have typically found more dramatic effects, suggesting that,while people may continue to work in the face of high tax rates, they devote more time and effort to shielding their income from tax. This can, of course, be a “bad” thing, if the shifts are to illegal non-compliance, or to socially distorting decisions, such as the substitution of untaxed capital income for taxable ordinary wages.[51] But the shift can also be a “good” thing, if the reason for the diminished labor effect of high marginal tax rates is that taxpayers are responding to socially beneficial incentives in the tax law—such as by making charitable contributions, as suggested by Diamond,[52] or via a generalized savings incentive, as developed as the central insight of this Article.

Two, and more particularly, there is considerable disagreement about how general the Mirrlees results are, and what the precise details should be—where, exactly, the marginal rate should begin to decline and where (if at all) it should reach zero. For example, Diamond in a 1998 article reports that the top marginal rate does not reach zero in his model with an unbounded distribution of individual types.[53] Later scholar shave argued that Diamond’s result follows from the very particular form of the utility function that Diamond (and some subsequent authors) adopted for modeling convenience.[54] The Nobel Laureate Joseph Stiglitzshowed in a 1982 article that the top income tax rate can be negative in a setting in which labor of different types is not perfectly substitutable.[55] There are yet other generalizations of the basic Mirrlees result, in a flourishing optimum income tax literature.[56]

Three, the optimal tax analysis in its classic incarnation depends crucially on demogrants. Without demogrants, average tax rates cannot be increasing over significant ranges with declining marginal tax rates. The difficulty is that demogrants are unlikely to happen, on any scale---and certainly not on the scale countenanced by the optimal tax models---any time soon, or perhaps ever. We nonetheless draw significantly different analytic conclusions from this apparent fact than have others, as we discuss below.

Four, it is worth noting that there is certain angst even within the optimal tax literature that important variables might be missing. Mirrlees himself, at the conclusion of his 1971 article, reflects that he would“hesitate to apply the conclusions regarding individuals of high skill,” because he surmises that, “apart from the possibilities of migration,” the highly able find that “their work is, up to a point, quite attractive,” andare thus inelastic.[57] But this is surmise on Mirrlees’ part --- explicitly wishful thinking --- and note his own qualifications, up to a point” and vis a vis the “possibilities of migration.” Mirrlees was clearly not thinking at the level of detail of substituting nontaxable for taxable income, where a high elasticity is plausible (that is, taxpayers are likely to highly respond to incentives to shelter income from taxes when tax rates are high).Mirrlees also concludes that “[t]he income tax is a much less effective tool for reducing inequalities than has often been thought” and ends his article sounding a plaintive plea for the “great desirability of finding some effective method of offsetting the unmerited favours that some of us receive from our genes and family advantages.”[58]

Finally, it bears stressing that the optimal income tax has been developed in the context of an income tax base, which is, indeed, a wage base in Mirrlees’ precise formulation, because there is no time and hence no savings. The central point of this Article is that the project of choosing optimal tax rates must be connected to the project of choosing a tax base.

5. Base-Rate Pairs: A New Look

We now combine the discussion of bases and rates to consider base-rate pairs. Because our project considers the case for progressivity in tax---specifically more steeply progressive marginal as well as average tax rates---we consider three pairs, progressive income, wage, and spending taxes. As explained above, the equivalence of wage and spending taxes, a key plank in the traditional income-versus-consumption-tax debate, holds only for regimes without uncertainty and then only for essentially flat taxes. The three bases differ under progressive rates. We must consider each, especially as to how they affect savings, the critical difference among them.[59] To review, an income double-taxes all savings, whatever their source and use, by design. A wage tax exempts all savings, again whatever their source and use, again by design. A progressive spending tax splits the difference, taxing some but not all uses of savings. To understand this point, and get a sense of how to characterize the savings that a progressive spending tax does and does not tax, we need a better understanding of the uses of savings.

5.1. Three Types of Savings

Consider in financial terms how most of us live out our lifetimes. As any parent knows full well, we spring forth into the world nearly fully formed as consumers: we cost money from the get-go. But (as any parent also knows) we do not earn anything for quite some time. When we do start earning, it is necessary to earn more than we spend (let us hope!), to pay off the debts of youth, including school loans, and then to set aside funds for retirement so that we do not have to work all the days of our lives. Our time on earth looks like one fairly steady consumption profile, from cradle to grave, financed by a lumpy period of labor market earnings concentrated in midlife.

If we lived as islands, onto ourselves, we would by stipulation have to balance the books on our own account, borrowing in youth, first paying off debts and later saving for retirement in mid-life, spending down in old age. Financial intermediaries such as banks and insurance companies would facilitate this process. In reality, many families work as more or less informal annuities markets, between generations.[60] Thus our parents pay for our spending in youth, and we pay for our children’s youths; we also stand ready to pay our parents back, should their needs exceed their resources in their old age, and so on.

Figure 1 is meant to be a very simple picture of this pattern. The curved line indicates earnings from work. The straight horizontal lines represent spending. The lower, solid horizontal line is a crude approximation of a fully self-financed taxpayer, whose lifetime spending equals, in present value terms, her lifetime earnings. The dotted straight line above that solid one represents a taxpayer who has been able to live “better” --- amore expensive lifestyle --- than her labor earnings alone, smoothed across time, would seem to allow.

In this perhaps untypical characterization of a typical life, note three broad uses of savings. One is to smooth out consumption profiles, within lifetimes or across individuals—to translate uneven labor market earnings into level consumption flows. We do this by borrowing in youth and saving for retirement in mid-life.[61] We can do this using third party financial intermediaries, or within the family. Economists call this life-cycle savings.[62]

A second use of savings is to provide for periods of emergency, such as heightened medical or educational needs, or times of low income due to un- or under-employment. These are exceptional needs, off the usual plan for living shown in Figure 1. Economists call this precautionary savings.[63] Such savings are significant in our analysis because provisions in the tax’s base may carve out the attendant uses for lower or even no taxation; we return to this theme below, as another application of the significance of viewing bases and rates in conjunction.

This then leaves as a third use of savings essentially all else, the analytic complement of smoothing: capital transactions can shift consumption profiles, up or down. An upward shift occurs when the fruits of our own or another’s savings (via beneficence) allow us to live a “better” lifestyle than we could on the basis of our own labor market earnings, alone, smoothed out over time. Suppose that we inherited wealth, or got lucky in the capital markets (received windfalls) and made a high return on our investments, like Warren Buffet or Bill Gates: our spending could increase. This is what the dotted line in Figure 1 illustrates. A down ward shift occurs when our own beneficence or bad fortune means that we will live at a lower lifestyle than we otherwise could, again on the basis of our smoothed out labor market earnings profile alone.

Economists tend not to have a handy phrase for this type of shifting transaction, in part because the issue has not arisen under the traditional income-versus-consumption-tax debate. Economists write instead about bequest savings, by which they mean private capital handed over to the next generation.[64] But this phrase does not necessarily get at the ex ante motivation for such savings, and ex ante motivations are key to understanding the possible incentive effects of various tax regimes. Individuals may be saving for lifecycle or precautionary reasons, and simply end up with leftover funds to pass on, what some economists call“accidental bequests.”[65]

Others, as we discuss below, may actively desire to leave bequests to their heirs, or to save for charity, or simply get a consumption value from owning stores of capital.

We will use both the standard economics vocabulary of savings and our own terminology of shifting and smoothing. The traditional breakdown of life-cycle/precautionary/bequest does not distinguish between smoothing and shifting savings within a lifetime. Smoothing reflects an ordinary return to savings, and the use of savings to even out uneven earnings typically from the labor market. Shifting refers to those extraordinary returns, as well as bequests and inheritances, that tend to characterize the lifestyles of the rich, like Warren Buffet and Bill Gates: these fortunate few are living better than they could be on the basis of their labor earnings alone, narrowly defined, spread over time at normal interest rates. Hence we use the vocabulary of smoothing/precautionary/shifting. These terms ultimately account for all uses of capital, distinguish between smoothing and shifting intra-generationally, and work perfectly well inter- generationally as well: capital passed between generations can be used to smooth out familial spending patterns,provide for emergencies, or enable later generations to live more expensive lifestyles than they otherwise could (that is, to shift up). Most importantly, smoothing transactions, compared to the alternative of not saving, lower the burden of taxation under progressive spending rates, whereas upwards shifts increase such burdens, as we continue to explore below.

5.2. Bases and Savings

Under progressive marginal rates, a tax is constantly increasing in the base: that is, the tax rate is rising on the implicit y axis in Figure 1. Now we can see, visually, that the three tax systems, income, wage, and spending, affect different patterns of work, savings, and spending differently.

Both income and wage taxes, as source- based taxes, apply to the curved earnings line in Figure 1. Thus both penalize taxpayers with uneven earnings as compared to steady ones--- highly educated professionals, for example, who must wait while their education progresses to earn back their keep, and then have a period of high-earnings concentrated in mid-life. Further, an income tax double taxes all savings, come what may.An ideal income tax would add to the tax on wages reflected in the curved line a second tax on the yield to the savings needed to smooth labor earnings out into retirement. An income tax would also double tax all precautionary savings, and burden taxpayers who had to borrow, as for their education, in their youth. And so on.

A wage tax also falls on all current labor effort. It has the same problem with “lumpy” earnings as does an income tax. But a wage tax simply ignores all capital transactions, again whatever their use. It does nothing to get at upward shifts, whereby certain taxpaying citizens can live a better lifestyle than their labor market earnings alone would seem to support.

A progressive spending tax, which is a uses- based tax, differentiates among uses of savings. Such a tax falls on the straight line in Figure 1. Smoothing savings under a progressive spending tax lower the burden of taxation, compared to non-savings, because they move material resources, income, from a period of high earnings into a period of a lower level of spending. So, too, precautionary savings---presuming that the precautionary use, such as for medical or educational needs, is tax-favored, somehow---also bring down the level of taxation. Thus these types of savings are taxed more like a wage, and less like an income, tax.Upward shifts, in contrast, increase the level of tax, measured off the baseline of the smoothed consumption pattern. The taxpayer living out the dotted line in Figure 1 is paying a higher effective tax, under a spending tax, than the taxpayer living out the solid line. It does not matter how, exactly, the spending occurred: through luck in the capital markets, successfully disguising labor as capital, someone else’s beneficence. All and only what matters is that there is higher material consumption. These material-lifestyle-enhancing types of savings are taxed like an income tax, not like a wage tax, which would ignore them. In other words, a progressive spending tax includes the yield to capital or savings in its base when this yield is used to finance a greater material lifestyle, but not when that yield is used to smooth out labor market earnings or for emergencies.

5.3. A Hint of Normativity

The analytics hint at a normative position. Perhaps societies are, indeed, of two minds about the taxation of savings. Societies sometimes want to tax the yield to financial capital, Mill be damned. But other times they do not: sympathetic to the Ants of the world, they want to avoid an anti-savings bias, a double-tax sting. Society may even want to encourage certain types of savings.

It makes intuitive sense that one would want to tax savings, in the spirit of redistribution and fairness, when the yield to capital allows individuals to live better, in material terms: the case of upward shifts. It similarly makes sense that one would not want to burden savings with a double-tax sting when an individual is simply moving around, in constant present-value terms, her labor market earnings, or is prudently providing for some emergency: the case of smoothing and precautionary savings.

Note that these are norms about the use of capital or savings: they are not norms about where the yield came from, whether it be from stocks or bonds, farms or small businesses, and so on. These are quite general norms, applying to most people in the course of their financial lives. We term the two norms the yield-to-capital norm, which holds that the return to capital is an element of value that ought to be taxed, and the ordinary-savings norm, which holds that savings in the ordinary course of a typical life, for retirement or certain urgent uses and needs, ought not to be double taxed.[66] When we put the two norms together, we in essence subtract out smoothing and precautionary savings, the domain of the ordinary-savings norm, from all of savings, the domain of the yield-to-capital norm, leaving what we have called shifting savings as the item being taxed. This would operate both intra- and inter-generationally: capital that finances enhanced lifestyles, whether it be own-generation or that of an heir, will bear a positive burden of taxation under progressive rates. A consistent progressive spending tax effects this result by design.

A final note on the two norms just articulated: The current tax system reflects just such norms, in its theoretical commitment to double-taxing all capital, by means of the choice of an income tax (the yield to capital norm), combined with the plethora of exceptions for smoothing and precautionary savings¸ as in the provisions for retirement accounts, or medical or educational savings accounts (the ordinary-savings norm).

6. The Limits of “Hybrids”

We pause in the ongoing analysis of tax base-rate pairs to consider the subject of “hybrid” income-consumption tax bases, now joined by the comments on the taxation of savings. Given that society may quite plausibly have mixed and even conflicting thoughts and moral intuitions about taxing savings, it may appear as if some kind of mixed solution, between the income tax’s double-tax and a consumption tax’s utter non-taxation of savings is the appropriate response for tax policy. Thus, indeed, the current tax has come to be known by tax policy experts as a “hybrid” income-consumption tax.[67]

The first difficulty is in sorting out precisely what a “hybrid” tax means. There are three possibilities.

One, and how the term is most commonly used in the tax policy literature, “hybrid” can refer to a single comprehensive tax---paradigmatically the current U.S. income tax---containing within itself aspects of both an income and a consumption tax. The income tax today has both wage and spending tax provisions---prepaid and postpaid consumption tax elements---designed to favor savings by avoiding the income tax’s double-tax burden. Traditional IRAs work on the spending tax model: no tax is paid now, when sums are contributed, only later, when the value is withdrawn, presumably to be spent.[68] The initial deduction for savings follows the simple logic of the Haig Simons definition of Consumption, as Income – Savings, just as the more general progressive spending tax does. In addition to IRAs, traditional 401(k) plans work this way. In contrast, “Roth”IRAs work on the wage tax model; there is no deduction up-front, but the accounts, properly maintained, are free from all further taxation.[69] Such treatment is the preferred means for a growing trend of tax-favored provisions, as in the recently introduced Roth 401(k)s, and the special educational savings plans blessed by IRC Section 529, medical savings accounts, and so on.

The trouble is that this type of hybrid, aside from being oxymoronic---an income tax that exempts the yield to savings is not an income tax---is also incoherent and counter-productive. There is no reason why a mish mosh of income, wage and spending tax provisions would result in any new savings at all.

Consider first an income-plus-spending tax, as with the traditional IRA structure under the income tax. An income tax does not include debt within its base: borrowing is literally outside the Haig-Simons definition of Income, as the asset of cash is offset by the liability of the obligation to repay, such that there is no change in one’s net wealth when she borrows.[70]

Traditional IRAs, which work on a spending-tax model, therefore need not generate any savings at all. A taxpayer can put money into an IRA with one hand, generating a tidy tax deduction, and borrow with the other hand: resulting in no net savings, but plenty of consumption.[71] Why might a taxpayer do such a thing? Presumably because she is myopic, eager to spend now and hesitant to save: the very traits leading supporters to propose traditional IRAs are the ones likely to undercut their efficacy. It is, after all, a fact of the matter that the era of IRAs, begun in the 1970s, has featured rising consumer debt and declining American savings rates.[72]

Consider next an income-plus-wage tax, as derives from the increasingly common Roth- style IRAs and savings accounts. These plans offer no current benefit for new savers, who get no cash-flow relief today, because there is no immediate deduction for the savings. This is thus no help at all to the middle classes struggling to save for ordinary-savings reasons, such as for their retirement or extraordinary medical or education needs. The Roth-style savings plans also have no mechanism to assure that “old” capital is not simply moved into the wage-tax-model accounts.[73]

In sum, hybrid income-consumption taxes breed complexity and confusion while offering little relief to new savers and no enduring incentives for new savings.

Two, “hybrid” might mean alternative comprehensive tax bases running side by side, as we in fact have in the United States with the payroll and income tax systems, or as Michael Graetz (and others) have proposed with a value-added-tax (VAT) replacing the income tax for those of relatively modest incomes.[74] We have no objections, in principle or otherwise, to any of a number of taxes raising revenue from most worker/taxpayers; for people who do not save, or do not save much, income equals consumption after all, and the choice of tax base and system for the masses might best be made out on transactions costs grounds.[75] This second sense of hybrid is not incoherent and counter-productive, as was the first and most common sense. But this Article is about finding the last best hope for progressivity in tax, if there is such a thing. A hybrid tax system that combines a VAT, sales, or wage tax for the masses is still left with the question of what kind of tax to have for high income/ability taxpayers. A progressive income or wage tax on these fortunate few will be difficult to obtain, for reasons we have pressed throughout.

This then leaves, three, a progressive spending tax as a hybrid: a possibility that few seem to have considered. Yet this is precisely what such a tax base-rate pair is, if we mean by a “hybrid” a tax that sometimes does, and sometimes does not, tax savings. A progressive spending tax, as the prior section illustrated, allows some types of savings---smoothing or life- cycle and precautionary---to lower taxes, whereas savings used to finance greater material lifestyles, via shifting, raise the burden of tax. This is, or can be, as we shall argue below, the most logical, principled and attractive “hybrid” to consider.

7. The Uneasy Case for Progressive Source-Based Taxes

It should be fairly clear by now why a progressive marginal rate source-based tax, income or wage, is swimming upstream. We can discern this on both optimal tax grounds, where we recall Mirrlees’ own lament that“[t]he income tax is a much less effective tool for reducing inequalities than has often been thought,”[76] and on not unrelated political-rhetorical grounds.

To begin, we can group together both wage and income taxes, and consider primarily the former. An income tax, after all, is a wage tax with a second tax on savings engrafted on to it. If the case for steeply progressive rates under a wage tax is weak, it is unlikely that a steeply progressive income tax would be appealing in theory. Indeed, the most common analysis of the taxation of savings in an optimal income tax framework suggests a zero rate of capital taxation.[77]

The most common flat-tax plan, owing to Hall-Rabushka, is for a flat “income” tax with no taxation of capital: a wage tax. In practice, as we explain further in Section 9, the so-called income tax in the United States is largely a wage tax in fact. Mirrlees himself, in modeling an “optimal income tax,” chose to make it into a wage tax, by having but a single period in his model. But a wage tax is limited in its ability to bear significant progressivity.

In a crude nutshell, we never need get very far beyond a popular instinct that the government should tax “bads,” not “goods.” When it comes to broad-based taxes such as the income tax or any likely alternative, the language of “goods” and “bads” is surely too simplistic and blunt, so we use the phrases socially productive and private-regarding in their stead. Work and savings are socially productive; they increase the size of the celebrated social pie. Since the tax system makes work and savings generally under-provided compared to a hypothetical no-tax state, marginal contributions to work and saving are to be preferred, on collective welfarist grounds, to their complements, leisure and consumption. By directly taxing work and savings, wage and income taxes discourage individuals from partaking in them. Because society might especially want the work effort of the most highly able, the case for declining marginal tax rates under optimal income or wage tax analysis is straightforward.

That simple fact limits how progressive, in marginal as well as average rate terms, a source-based tax can be. Note that Mirrlees’ own lament about the redistributive capabilities of an income tax came after he had modeled the problem using demogrants to transfer resources back to the poor. As we discuss more fully in Section 9, America does not now have a general program of cash grants back to all, including the poor, and is unlikely to engraft one onto a status quo where government spending already exceeds government revenue and has been notoriously “sticky” or hard to cut.[78] Even if we were to create a massive program of pure redistribution, it is apparent that it simply could not be funded strictly by tax increases on the most able.[79] These high ability sorts are too few, and their labor decisions too elastic, to bear the burden. As long as we have wage taxes, we are going to have relatively flat, and declining, marginal tax rates; as long as we have limited or no demogrants, the same goes for average tax rates, because we are left without the two best mechanisms for average rate progressivity, namely demogrants and marginal rate progression.

These are not simply abstract claims. At least since the time of Ronald Reagan’s presidency, a political and rhetorical argument against high tax rates on high ability taxpayers has been powerfully made in the public culture by such advocates as Martin Feldstein and, in the most recent presidential election campaign, Joe the Plumber. The progressivity of the U.S. income tax has fallen since 1980. Recall that the journalist David Leonhardt observed that “even liberal economists” had come to view the pre-Reagan marginal tax rates, of 70%---themselves a significant drop from the 90% bracket in place from World War II through 1963---as unreasonably high. As we noted in the opening section, the debate between Senators Obama and McCain, now being played out in the real-world politics of the Obama Administration, is over restoring the top marginal tax rate, on individuals earning over $250,000, to 39.6%, from its current 35%, and it is far from clear that this will happen as we write. Simply put, it is highly unlikely that we will see a return to significantly progressive tax rates under an income or wage tax.

8. The Easier Case for Progressive Spending Taxes

Changing from a source-based tax, namely an income or a wage tax, to a uses-based spending tax offers the last best hope for returning to significant marginal rate progressivity in tax. The reason for this is straightforward. Whereas wage and income taxes deter work, and work and savings, respectively, a progressive spending tax deters, in the first instance, only spending. If all people work only to be able to spend, and in the present period no less, there will be no difference between income, wage, and spending taxes: Income = Consumption + Savings, so if Savings = 0, all three taxes are the same.

But this is not generally the case: people, especially the high income and wealthy, do save. This means that they do not spend all that they can, non-consumption being the same thing, analytically, as savings.

The technical question for optimal spending tax analysis is what is the labor-supply elasticity of marginal spending tax rates.[80] In other words, what effect on work effort, today, will the prospect of high spending tax rates, tomorrow, have? This is, largely, a question for an ambitious new research agenda, one that has not generally been pursued in part because of the confusion in the traditional income-versus-consumption tax debate,[81] in part because of simple inertia – scholars study what we have. We sketch out this agenda more fully in Section 10; the present Section discusses a number of reasons why the labor supply effect of high marginal tax rates under a spending tax might be muted.

8.1. A Note on a Parallel

We want first to mark an important parallel to work by Peter Diamond, a prominent public finance economist and leading light in the optimal tax tradition. In a 2006 journal article, Diamond argued that optimal tax analysis could well support higher marginal tax rates under an income tax with a deduction for charitable contributions than under an income tax without the deduction.[82] Characteristically, Diamond’s analysis is mathematically complex, and subject to various particular specifications. But the core insights revolve around the ideas that, one, for the philanthropically-minded, the option of giving to charity dampens the effects of high marginal tax rates on labor, and, two, the philanthropic contributions themselves are public goods, benefiting the wider society---unlike the leisure that is the only behavioral response to a high wage tax rate under a purer income or wage tax.

Diamond’s paper concerns a modification to the tax base. The same argument could well apply to other base exclusions, such as lower or even zero rates on medical or educational uses under a spending tax. These uses correspond to the category of precautionary savings noted above. In short, if a taxpayer is saving in order to provide for future medical or educational needs, she will rationally calculate that she faces a lower, or even zero, rate of taxation on the labor supply effort that goes into those savings. Thus her anticipated tax rate is declining, even to zero, without a nominal declension in the rate structure: it is as if there is a “shadow” rate structure for medical and educational spending affecting her marginal work decisions.

Savings for medical emergencies or educational uses may be special and limited cases, and may also lack the general feature of public goods. Under a progressive spending tax, all savings has the structure of Diamond’s charitable contribution deduction. If high-ability taxpayers are motivated to continue saving under a spending tax, they will not be deterred from productive labor effort under high marginal spending tax rates, except to the extent that they incorporate future spending tax liabilities in current work decisions. The wider society benefits from their actions, both from the positive externalities of the private capital or savings and from the implicit deferred taxation on ultimate use of the funds.

It should be evident that the case for more steeply progressive rates under a spending tax turns on why people save. In particular, the technical question is what individuals’ ex ante incentives to work will be, for the labor effort comes today, in a prior period, whereas ultimate spending comes later, and the taxes along with it.

8.2. Why Do People Save? The Rational Story

The question for analyzing optimal spending tax rates is how the socially productive decision to work would be affected by high marginal tax rates on the private-regarding act of spending.[83]

Because a saver is, by definition, not spending all that she could in the present period, the ultimate tax rate she might face---as well as the conceptually different tax rate she might anticipate ultimately facing---depends on why she is saving. In this section, we consider various possible motives for savings consistent with a traditional “rational” actor model. Our main task for now is to sketch the possibilities.

Recall the economist’s triad of life-cycle, precautionary and bequest savings. Consider how each type of savings affects the labor supply decisions of rational taxpayers under a progressive spending tax.

A traditional life-cycle saver – what we would call a “smoother” --- will anticipate during her peak earning years that her tax rate at the time of ultimate spending will be lower than it is today, by the definition of smoothing. Recall Figure 1. The deterrent effect of the tax rate schedule will be dampened, down to her anticipated level of spending. This reflects a smoothing process that makes spending exhibit less annual variation than does labor income, such that average tax burdens under a spending tax will be lower than those under a wage or income tax, with a progressive tax imposed on an annual basis, given the same rate schedule. Thus the nominal marginal tax rates can increase without necessarily deterring present period labor --- no one need pay the higher marginal rates. Suppose, to make the intuition sharp, that a high tax rate,of 50% or more, kicked in at $250,000 of spending. A taxpayer with the opportunity to earn $300,000 a year would not be affected by that highest bracket if she planned on saving at least $50,000, perhaps to fund a constant material lifestyle at $250,000 of pre-tax spending through retirement, or if she were paying off student loans, which would be deductible under a consistent spending tax (the tax having come due when the borrowed funds were spent, that is, in youth[84]). The progressive spending tax might push her to save more, or might allow her to do what she wants to do anyway, all without affecting additional labor supply decisions.

Note, however, that a progressive spending tax could affect the rational taxpayer’s retirement decision. Picking up the prior example, after years of saving $50,000 or more, once her savings account was sufficiently large such that, on an annuitized basis, her lifetime spending needs were met, the purely rational, self-regarding life-cycle saver might stop working. But of course that is also true under a wage or income tax: once anticipated lifetime consumption is fully funded, the rational taxpayer lacking any other motive for amassing capital would cease to work. Bill Gates knows this. And retirement savings today are taxed under a single tax, wage or spending tax, model.

The rational precautionary saver need not be deterred at all from work effort by a progressive spending tax that exempted the precautionary uses from its base. This is parallel to the effect of the charitable contribution deduction under the Diamond analysis discussed above. In essence, anticipated precautionary expenditures that are exempt face a zero rate of taxation, without such a rate being explicitly in the rate schedules. This follows, again, from the combined consideration of bases and rates. There are of course complex and contestable policy questions as to whether private medical or educational expenditures should be fully or even partially excludible from a spending tax base, and we take no position on that set of arguments here. We simply note that a further reason to exempt certain expenditures in whole or in part from a spending tax base is that, to the extent taxpayers might rationally save for such needs, the effect of the exemptions can be to diminish the labor supply disincentive effects of progressive marginal spending tax rates.

But most savings seems to be something else, other than life-cycle or precautionary. Studies using this traditional economics vocabulary consistently show that 75% or more of private capital is passed on inter-generationally, in what economists call bequest savings.[85] This is simply a residuum category, and possibly less responsive to ex ante labor incentives than are other economic decisions considered by optimal spending tax analysis. Perhaps savings that get passed on were intended as life cycle or precautionary, but the rainy day never came, or death came too early: what some call accidental bequests.[86] In such cases, the analysis is as above, because the labor supply effect looks to the ex ante incentives: if a taxpayer thinks that she will need funds for lifecycle and precautionary needs, she will rationally factor in the tax consequences of such uses. Hence risk aversion might favor social welfare under a progressive spending tax, as some individuals might continue to save beyond a more risk-neutral perception of their need.

Given what we know of the wealthy, however, it seems unlikely that all or even most of bequest savings comes from miscalculations of own-generation life-cycle or precautionary needs: both annuities and insurance markets allow for fairly precise life-cycle and precautionary savings these days, and much evidence shows that the wealthy continue to save, not dissave, until the end of their lives.[87] What is going on? Inter-generational or bequest savers might be motivated to pass wealth on to their heirs. Then the question becomes what if any tax rates on the heirs do these dynastic savers internalize in their labor supply decisions today? (Another question, or set of questions, would look to the work incentive effects on these heirs from getting the bequest; note that under a progressive spending without estate tax, the bequest might come late in these heirs’ life, when their lifetime patterns of work and savings have largely been set.[88])

All of this is speculation, and thus forms part of the new research agenda we aim to be announcing, but it is possible that many wealthy benefactors will expect their heirs to live as they did: that is, somewhat frugally, leaving a bequest at the end of their days. Such individuals might even welcome a progressive spending tax, which would form a kind of social “spendthrift trust” for their heirs: some evidence to support this possibility is the fact that many wealthy individuals now privately create just such trusts, that limit access to wealth to set levels of income and certain urgent needs of the beneficiaries.[89]

Accidental and intentional bequests do not exhaust the potential reasons for “excess” savings that have some individuals leaving this earth with private capital. It is possible, perhaps even likely, that many wealthy individuals simply get a consumption value from owning capital itself: they enjoy having large stores of wealth, take comfort in it, enjoy making investment decisions, and so on. Some may even have ultimate philanthropic motives, as in the Diamond analysis. The critical point is that, from what evidence we now have at hand, many wealthy people seem motivated to save for reasons other than presently anticipated own-consumption, and these motivations might make them continue to work in the face of more steeply progressive marginal spending tax rates, which they – quite rationally – do not anticipate ever paying, themselves.

8.3 A Windfall Benefit?

There is another reason to believe that labor market earnings might not be as responsive to high marginal spending tax rates as they are to wage or income tax rates. It derives from uncertainty in the returns to labor effort as well as capital. Such uncertainty has multiple origins: individuals do not know exactly what effect today’s labor effort will have on ultimate labor compensation (in the form of bonuses, subsequent promotion and raises, and job satisfaction that encourages delayed retirement, and so on), and they do not know what rates of return they will receive on their capital, income that is earned today but invested for years rather than spent immediately. Individuals who are lucky, whose returns ultimately exceed their expectations--- like Bill Gates and Warren Buffet---tend to have the greatest spendable wealth, the highest spending levels, and therefore will face the highest spending tax rates. Since these high spending tax rates were unanticipated at the prior time of earning labor income – the good luck or windfall by definition being unexpected – it follows that the high spending tax rates will not have had the same discouraging effect on labor supply that they would have if known at the time of earning, as would by definition be the case under wage or income taxes. Lucky returns or windfalls of some sort are apt to be a major source of funding enhanced material lifestyles or upwards “shifts” in the perspective of Figure 1. Recall that a progressive spending, but not a wage tax, falls on such resources when consumed.

Uncertainty in the returns to labor supply and capital gives yet another reason–aside from the motivations for private savings and capital accumulation, canvassed above---why it can be possible to design a spending tax with sharply redistributive tax rates that nonetheless does not excessively discourage labor effort. In such a setting, a major function of a progressive spending tax system is to redistribute the unexpected portion of spending power – positive and negative – between laborers. Spending taxes offer a type of insurance against uncertain returns, and while the same is true of almost any tax system, spending taxes have the advantage that they partially insure all consumption risks. By contrast, a progressive wage tax partially insures against uncertainties in wages but not capital returns, and a progressive income tax partially insures against income uncertainties but does so at the cost of discouraging saving. Spending taxes uniquely offer the prospect of significantly redistributing unexpected returns without discouraging savings and without excessively discouraging labor supply ex ante.

8.4. A Possible Sort?

We have been discussing why people work and save, and more or less slipping into a standard academic conceit that all people are the same. Of course they are not. We have no doubt that people are what economists call heterogeneous, such that high-ability taxpayers—the main focus of both optimal tax analysis and any call for greater progressivity in tax today---have a mix of motives for their work, savings, and spending decisions. Here we note another possibly beneficial effect of a progressive spending tax: its intrinsic ability to treat two types of taxpayers differently.

Under a wage tax such as Mirrlees modeled, taxpayers face only one decision margin: to work or not. The fear that high-ability taxpayers may reduce their labor supply drove the analysis of declining marginal tax rates. Even if there were some inelastic high ability workers, who would continue to work no matter what tax rate they faced at the margin, the tax rates on high ability types would have to be low if there were a suitably high number of elastic taxpayers, such as, presumably, Joe the Plumber. In such a case, the inelastic high ability taxpayers would get a windfall, as it were. This problem has led some to look for “markers” of elasticity (or ability), such that an income or wage tax could better differentiate among high-ability types.[90] But there are practical, political, legal and moral questions with any such approach.

Under a progressive spending tax, in contrast, there are two decisions margins: to work or not, and to spend or save. This leads to a natural “sort” among high ability types, based on their propensities to spend currently.

Imagine that there are two types of high-ability earners. Type 1, captured in such contemporary bestsellers as The Millionaire Next Door,[91] is a frugal sort, inclined to save – a noble Ant to the end. Type 2, as described in such contemporary analysis as Robert Frank’s Luxury Fever,[92] is a Grasshopper type, addicted to spending his wealth in a showy fashion. Now, although, once again, more analysis is needed, it is at least possible---dare we say likely?---that the Ant is more elastic in her labor supply than the Grasshopper. The Ant is frugal, after all, and hence more likely to be concerned with prices and taxes. She might well stop working under high labor tax rates. The Grasshopper, in contrast, looks like a consumption addict: a lamentable personality type, to be sure, but just the kind of person that optimal tax advocates should want to tax. A progressive spending tax, by design, would fall more heavily on Grashopper than on Ant. Such a tax holds out the possibilities for sorting the rich and highly able into elastic and inelastic segments, based on their propensities to spend, allowing society to extract a higher tax burden from spenders while continuing to enjoy the benefits of the productive activity (and thrift) of the worker-savers.

8.5 Behavioral Models

The analysis up to this point in this Section, including that contained in the Diamond paper, has been predicated on a traditional “rational actor” model. There is nothing irrational about saving for one’s retirement, for precautionary needs, for one’s heirs, or simply to enjoy the psychic pleasures of owning independent stores of capital. There is no disputing tastes, after all, in traditional economics analysis.[93] Certainly a spend-it-all-and-die-broke mindset is not irrational. But a growing field of research, behavioral economics, has challenged the assumptions of the rational model. Much of this literature has focused on savings and other behavior with inter-temporal consequences. Richard Thaler and colleagues for example have famously argued for a “behavioral life cycle hypothesis” to supplant the rational lifecycle hypothesis of Modigliani and others.[94] Under a behavioral view, ordinary persons make a variety of “mistakes” in their savings behavior, on account of such heuristics and biases as myopia, time-inconsistent preferences, mental accounts, and so on.

What might be the relevance of behavioral insights to our analysis here? Once again, we hope to be opening up a research agenda. We believe it important that such an eganda have a behavioral component. Here we simply sketch some possibilities. The most common finding from behavioral economics on point is that people save too little.[95] This myopic tendency to live for the day has led to the types of savings vehicles within the income tax that make it a “hybrid,” as discussed above in Section 6, and as continue to feature in President Obama’s Budget proposals.[96] Yet myopia also gives reason to doubt the efficacy of these ad hocpatches.[97] The myopic tricked into opening an IRA can just as easily run up credit card debts or home equity loans. To put the matter generally and simply, a system such as the income tax that does not tax borrowing when incurred will face a difficult task in getting myopic people to save. A consistent spending tax, in contrast, eliminates the arbitrage opportunity, not by taxing the savings – which it systematically does not do – but rather by taxing the debt-financed consumption.[98]

There is some reason to hope that the systematic exemption of savings under a progressive spending tax will thus encourage greater and more enduring savings than the ad hoc approaches under current law.[99]There are other places in the analysis of a progressive spending tax where behavioral heuristics and biases might play a role. A persistent myopia, combined with self- serving optimism, may lead many people to highly and perhaps completely discount any tax on their savings. This possibly obtains today: on a casual survey of friends and colleagues, almost all give some indication of the amounts in their retirement accounts without discounting for the inchoate tax liability. This tendency could lead to a “money illusion” under a consistent spending tax that might, ironically, lead to less savings.[100] People will believe that they are wealthier, in after-tax terms, than they are: they will not be able to consume as much as they think they can. This “behavioral income effect,” as it were, of the savings accounts under a consistent progressive spending tax can then lead to lower savings.[101] On the other hand, the substitution effect cuts in the other direction, as spending faces an increasing marginal tax rate whereas savings, more so under a behavioral perspective, faces a low or even zero rate of taxation.[102] And the “money illusion” generated by the pre-tax accounts under a consistent spending tax ought to be less problematic than it is today, under the so called income tax, because there will not be any ability to borrow tax-free with the pre-tax accounts in mind.

Again, far more research is needed, but these are the kinds of questions that are worth asking. As a crude guess, we suspect that the optics of placing the tax on the act of spending will have beneficial effects, in that people will not be as deterred from working or savings because of the logical remoteness of the tax, but that at least the wealthy will be deterred from additional high- end spending which ought to be an effect desired by advocates of progressivity and redistribution. A progressive spending tax gets its redistribution, and seeks its relative equality, in the consumption space of material lifestyles, what arguably ought most to matter to advocates of redistribution in the first place.

9. The World as We Know It

Thus far we have discussed the relatively abstract analytics of tax, considering income, wage and spending tax bases and progressive marginal rate structures. We turn now to a consideration of the real world of tax today in the United States in particular.

Three strong themes emerge.

One, looking to the tax base, we see that we largely have a wage tax. This is so, both in regard to the so-called income tax alone---because this tax’s commitment to taxing the yield to capital is highly porous, at best and when we widen the perspective to consider also the payroll or social security/medicare tax system, a close second in total revenue to the income tax.[103] The payroll tax does not even attempt to be anything other than a wage tax.

Two, looking to the rate structure, we observe that it is highly compressed – indeed, it is quite close to being an optimal tax rate schedule, with low and declining rates on the most able and wealthy citizens. It would seem as if, despite its highly abstract quality and nearly four decades of qualifications, questions, and refinements, Mirrlees and his followers in the optimal income tax tradition have had a strong influence in contemporary tax systems. Perhaps as important, and related, the political and rhetorical arguments against high marginal tax rates on high earners have had enormous real-world traction, at least since the presidency of Ronald Reagan in the 1980s through the rise of Joe the Plumber.[104] The economics as mediated through political and popular processes has led to what might be called “everyday optimal tax theory,” borrowing and adjusting from Murphy and Nagel’s term of “everyday libertarianism.”[105]

Three, we do not have anything close to a lump-sum program of redistribution or “demogrants,” and we are all but certain never to have one.

Combining these themes leads us with a relatively flat-rate wage tax, with declining tax rates on the high ability, and limited redistribution today and little hope for more tomorrow.

The balance of this Section explores these claims and comments further.

9.1. Tax Today

The so-called income tax in the United States is in reality largely a wage tax, for structural and ad hoc reasons. The structural reasons relate to such deep-seated features of the income tax as the “realization requirement” and the non-taxation of debt, which in turn render such features as low tax rates on capital gains and corporate dividends virtually inevitable.[106] The ad-hoc reasons include the panoply of ostensibly pro-savings provisions, modeled along both wage and spending tax lines, such as traditional and Roth IRAs, Section 529 plans, and so on, considered above. When added to the increasingly important payroll tax system---which does not purport to be anything other than a wage tax---onto the income tax, the theme is clear enough.

Turning from base to rate issues, Table 4 combines current income and payroll rates for an individual in 2009:[107]

Table 4: Payroll and Income Tax Rates (approx), Single Person, 2009

This rate structure begins to look like an optimal income tax. It features marginal tax rates that peak in the middle income ranges and then gradually decline. Looking at just these two taxes, combined, for now,marginal tax rates peak at 43.3% over the $92,000-$106,000 range of incomes. Marginal rates then fall, to 30.9%, before rising a bit at the end, to wrap up at 37.9%.

This is well on its way to being an optimal income tax rate structure, as suggested by Mirrlees. Note that President Obama’s proposed changes would only raise the last two income tax brackets, the final one to39.6%, a combined 42.5% rate on income still below the middle class peak. And a fuller, richer understanding of the status quo suggests that the true picture is even closer to Mirrlees, and more dramatic in its peak and trough structure than this quick sketch. Consider three further factors.

One, the earned income tax credit (EITC) of IRC Section 32 adds to the “hump” effect in the above Table by placing a high marginal tax rate burden on the working poor. This provision is the key element in the“workfare” that largely replaced “welfare” under President Clinton’s reform, although the EITC dates back to the early 1970s.[108] The credit works like a negative tax, giving the working poor additional resources.The problem is, it is repaid by the near-poor in an effort to minimize the cost of the program by preventing upper-middle income taxpayers from receiving benefits. Thus a single parent with two children, having received a benefit of approximately $4,000 on her first $10,000 of labor market earnings, must pay this benefit back by facing an additional tax of some 21% on her earnings between approximately $20,000 and $40,000 dollars: putting her in a 50% marginal rate bracket over this range, adding the EITC phase-out onto the rates in Table 4.

Two, in addition to the phase-out range of the EITC, there is a host of other phase-outs of means-tested programs facing the working lower-middle income class. Daniel Shaviro showed in a paper published in 1999that marginal tax rates facing some working poor can equal or exceed 100 percent, an astonishing if unintended feature of overlapping programs, but one that has been well verified (and left largely unchanged)since then.[109]

Three, at high levels of income, the actual marginal taxes individuals face depend critically on sources of income and avoidance possibilities: the gap between “income” and “taxable” income generally grows with income.[110 ]Many wealthy taxpayers in fact face marginal rates of 15% or even 0% on the last dollars that they earn, because they are able to gain wealth in the form of taxable capital gains, at a 15% rate, or untaxed capital appreciation, respectively. Hedge fund managers fall into the former camp; Bill Gates, Warren Buffet and other billionaires into the latter.[111] This is, of course, a perfectly understandable—even satisfying---state of affairs to those steeped in the Mirrlees model.

Thus we conclude that the “real world” in the United States today looks roughly like a Mirrlees optimal income tax both in its base, that is, wages, and in its rate structure. The implementation is not precise, at least in part because there are no generally accepted precise findings from optimal tax itself. Some of the imprecision also no doubt derives from pragmatic and practical constraints: for example, and perhaps most important, there is a single rate structure that applies to all individuals, without excessive particularization. The only way to particularize taxes today is through the base, where taxpayers indeed often face decision margins---whether to give to charity, save, spend on medical needs, and so on: This continues to be the principal theme in this Article, namely, the need to consider the base and rate aspects of tax systems hand-in- hand. But individuals do not face particular rate structures based on their characteristics: the government does not attempt to tax I.Q., or look to genetic “markers” to individuate taxes,[112] taxes are collected on an annual basis using accounting information, and reflect other practical and realistic considerations.

Other features of the real rate structure seem to reflect pragmatic and political compromises. For example, a perceived popular political desire not to allow high-income taxpayers to pay nominally zero taxes led to the adoption of the AMT,[113] although Congress has never changed any of the structural elements of the income tax that allow those living off of stores of financial capital to pay little or no taxes by showing little or no taxable income.[114] Yet when all is said and done, what is remarkable is how much reality follows abstract theory. Or, as John Maynard Keynes put it, “practical men, who believe themselves to be quite exempt from any intellectual influence, are usually the slaves of some defunct economist”[115]---in this case the not-defunct James Mirrlees.

9.2. On Not Adding Demogrants to the Mix

We just noted that the United States tax system, in its base and its general rate structure, looks like an optimal tax model. For the system to fully conform to Mirrlees’ sketch of the ideal, however, there would also need to be demogrants, or lump-sum transfers to all. This step is necessary to bring about average rate progression and hence redistribution given the pattern of quickly rising then gradually declining marginal tax rates.

But we have no demogrants. As others have pointed out, the United States is highly unlikely ever to offer demogrants on anything like the scale contemplated by the standard optimal income tax model.[116]Certainly only very modest government support is currently available to Americans without other sources of income. On a quick impression, the EITC might look like a demogrant. But on closer inspection, it is far from it, and, in fact, significant evidence that America is unlikely ever to have a major demogrant-like program. The EITC, after all, is set as a percentage of earned income: it is not lump sum, but, in effect, a negative marginal tax rate. This was an important, consciously chosen design feature of the program. The significant expansion of the EITC under President Clinton formed the lynchpin of the “death of welfare as we knew it,” “making work pay,” and moved “welfare to workfare.” Far from the optimal tax result that demogrants plus high marginal tax rates on the poor deter the least able from working---a feature that Mirrlees himself found attractive, at least during times of high unemployment,[117] and which others, such as Louis Kaplow, are willing to accept as a perhaps inevitable byproduct of other desirable properties[118]---American social policy seems to have a strong principle of encouraging the poor to work.

There are also perfectly strong rational reasons to stay away from demogrants. Quite apart from a prevailing political orthodoxy that is skeptical of demogrants even if they did not trigger costs, lump-sum transfers are of course not costless. Money given unconditionally to citizens is money that the government either must raise with taxes or else divert from other spending priorities. But presumably the public likes the other spending programs now in place, many of which have redistributive and “social safety net” components to them. There is a tremendous “stickiness” in the status quo, a reluctance to cut spending programs already in place, a social fact that might relate to certain well-known cognitive biases and tendencies, such as the endowment effect.[119]

Even if there were a political will to provide significant demogrants, how would they be financed? Higher marginal tax rates on middle-income taxpayers, the most likely source of significant tax revenue, create unappetizing incentives and thereby likely greater efficiency losses than even the average dollar of tax revenue does currently.[120] Engrafting more taxes--- perhaps significantly more taxes---onto the status quo discussed above runs the risk of pushing infra-marginal rates, already above 100% for many, as we pointed out in the prior section, even higher. This would then run counter to the pro-work principle in American social policy. There can also be no iron-clad guarantee that a government composed of flesh and blood human beings, having found a means to raise revenues, would in fact remain committed to the spending program of the demogrant system: there is a great deal of evidence bred of experience, as with the supposed dedication of lottery proceeds to education, or in the so-called flypaper effect, that spending program scan go askew.[121] Further, in times of budgetary stringency, a demogrant would be the easiest expenditure to cut, or not to increase, and the reality of political and social life is that every epoch is an age of budgetary stringency---so it would be a frail social safety net on which to depend.

And so, although it does not have the status of analytic “truth,” it appears unrealistic that the United States would adopt a major demogrant program. It is important, therefore, to consider the significance of this absence of demogrants.

9.3 Progression without Demogrants

So it comes down to this: we have, roughly, an optimal tax base and rate structure with no demogrants and hence limited redistribution and progressivity in average tax rates. What to do?

Many scholars seem to assume that the optimal tax findings are all of a piece, such that it is somehow illegitimate or inappropriate to take arguments for and analysis of the rate structure and divorce these from the case for demogrants. In sum, without demogrants, optimal tax has nothing to say to the public political discourse. The philosophers Murphy and Nagel characteristically state this case in the strongest and most colorful language:

But there is one very important point to make about economists’ lessons on the distinction between ends and means. If we are told that lower marginal rates coupled with a demogrant would be better even from the point of view of a strongly egalitarian theory of justice than graduated rates with a high marginal rate at the top, that gives us absolutely no reason to abandon high marginal rates without introducing a demogrant. This is blindingly obvious. But in practice the point is frequently ignored. It is frequently claimed, for example by Joseph Stiglitz, that the conclusions of optimal tax theory were an influence on the trend to much lower marginal tax rates in the 1980s. This trend has been linked not with a greater role for tax transfers, but rather the reverse. No one concerned with welfare, not even utilitarians, can regard the growing inequality that has characterized the last two decades in the United States as an improvement from the point of view of justice. It is possible that, in its short-run practical consequences, economists’ interest in the behavioral effects of taxation has done more harm than good to the cause of social justice.[122]

It may be “blindingly obvious” to Murphy and Nagel that there is “absolutely no reason to abandon high marginal rates without introducing a demogrant,” but we fail to see it. The critical point is that the revenue raising and expenditure aspects of optimal tax theory are analytically separate. They may be, and are, in the hands of the optimal tax theorists themselves, normatively connected.That is, there is or can be a reason to make the move Murphy and Nagel suggest, and argue for high marginal tax rates if demogrants are off the table: that reason is a foundational commitment to redistribution, come what may, such that advocates would be willing to choose progressive marginal rates without demogrants if demogrants were ruled out. We comment further on this set of policy prescriptions below. For now we simply note that there is no necessary reason why the government or wider society must accept the social welfare function used by the theorist, or even the general social welfarist framework. Having a revenue need, derived from whatever source, the optimal tax framework teaches a policymaker how most efficiently to meet it. And so there is, indeed, a reason---a compelling one---why policymakers should choose a pattern of declining marginal tax rates under a wage tax, even without demogrants. That reason is efficiency or wealth-maximization. Optimal tax’s insight that a “high marginal rate at the top” can be socially inefficient—wasteful---has nothing logically to do with the use of demogrants or not.

Consider, to drive intuitions home, the social logic at least implicit in Murphy and Nagel’s case for supporting high progressive marginal rates under an income tax in a world without demogrants. In order for this position to be coherent, people must accept the welfarist, utilitarian or other argument for the degree of redistribution manifest in the optimal tax models: they want to redistribute. If they did not want to redistribute, they already disagree with Murphy and Nagel, and Mirrlees for that matter. But then by stipulation the people have rejected the demogrants, or the best means for achieving the ends of redistribution.Next, then, the people would have to support some inefficiency, or waste---perhaps some considerable waste, if the top marginal rate is high enough----in order to get more redistribution without the demogrants.[123] In sum,under the structure of this logic, people, committed to a significant level of redistribution or equity, are willing to pay a high price, in inefficiency, not simply to get more equity or redistribution, but to get it withoutusing the best means to do so, namely, demogrants. This is possible---there is no disputing tastes, as the saying goes---but certainly a bit odd. In a move familiar from the work of Louis Kaplow and Steven Shavell, aPareto improvement---of potentially large magnitude---would be possible, because the more efficient means of achieving the same ends could bring vast wealth to the table.[124]

Here is a different interpretation of the status quo, and the possible inchoate principles of the people. Society does not quite agree on any one social welfare function. It collectively chooses government programs,many of which are indeed redistributive, though redistributive in kind: education, national defense, environmental regulation, social security, medicare, public health, and so on. Having chosen these expenditure programs with at least an eye towards equity, society seeks to raise revenues in the most efficient way. It experiments with high marginal tax rates, which might actually work better during war times, when first imposed,[125]but, over time, such high tax rates distort labor supply decisions and lead to vast complexities and waste. Policy makers fret that people might stop working, migrate to other places, spend time and resources converting their potentially taxable income into untaxed, or lightly taxed, forms. In fact, they are right to have such concerns. Mirrlees and kin come along to give a formal gloss to these intuitions, and the people listen to and learn from the teaching of formal economists, even if they reject some of their more precise findings---the public does not want a top nominal tax rate of zero; we do not generally want to create incentives for people not to work; we do not want a massive tax and transfer program in a demogrant fashion engrafted onto what we are already spending money on. But voters do not want pure, spiteful, waste,either—they do not want the most able and productive citizens sitting idly on the beach. So the government adopts a modified form of optimal taxation, particularly its general rate structure. The polity develops,that is, an everyday optimal tax. Joe the Plumber emerges from the hustings.

The challenge is where to go next. How do we get more equity and redistribution, and more progressive marginal rates, which the public does seem to want, without abandoning major current public expenditure programs or instituting a massive new demogrant system, and without simply sacrificing wealth en masse on the altar of these preferences? Our answer—what we think is the last best hope for obtaining progressivity in tax---is to change the tax base.

10. Conclusions and Suggestions for Further Research

This final Section turns to normative considerations: why a progressive spending tax might, indeed, be a “good” policy option for the United States. It is not for scholars to dictate the good, of course. We simply sketch out how a progressive spending tax might be the optimal tax base- rate pair under a traditional welfarist conception, or how certain advocates of social contractarian or fairness-based perspective might come to see it as the best alternative. In all this, there are many empirical claims and issues that need to be developed: certain important questions have not yet been asked, or asked well and often enough, because a traditional analytic view of tax has impeded them.

We first sketch out four distinct paths or tax’s future, then we turn to the normative arguments for a progressive spending tax, and conclude with a call for more research.

10.1. Paths to a More Progressive Future

The prior Section stressed three themes: that America has essentially a wage tax base, a relatively flat, nearly “optimal” tax rate structure, and no demogrants. This combination left us with limited redistribution today and even more limited hopes for more tomorrow. Let us continue to assume, with Murphy and Nagel and many others,[126] that America right now is far from having obtained the redistributive goals and ambitions of optimal tax theory in the tradition of Mirrlees and others. We are a land of great inequality. How might it be transformed to a promised land of greater redistribution, in the spirit of optimal tax? There are four general approaches, looking at the tax and spending sides of the ledger, playing with different combinations of tax base, tax rates and demogrants.

One, the United States could eliminate or reduce all or most existing spending programs and substitute demogrants in their place. We humbly suggest once more that this is not going to happen. For better or worse, current spending programs are deeply entrenched. No American President or political party has succeeded in significantly reducing spending---the Republican Presidents Ronald Reagan and George W. Bush certainly did not---and there is precious little reason to count on significant reductions any time soon. Indeed, the chief fiscal obstacle confronting the federal government is to reduce the rate of increase in government spending, especially on social security and, even more so, medicare.

Two, the United States could engraft demogrants onto the existing spending platform. But as we discussed in the prior Section, this is not only nearly as infeasible as cutting existing spending programs, it is also unwise, in part because it would ratchet up revenue-raising demands far beyond the capacity of the system to bear.

Three, we could, without demogrants, change the marginal rate structure under the existing tax structure to create progressivity, perhaps even with high marginal rates at the top. This is the suggestion of Murphy and Nagel, Zelenak and Moreland, and at least a handful of liberal political commentators. It makes sense in theory, but we are skeptical in practice, for reasons we have detailed and built up throughout this Article. The efficiency losses from high marginal rates exist whether or not there are demogrants. Americans, since at least the 1980s, seem to have figured this out, as this quick survey of the political environment now and over the last several decades points out. Ignoring the lessons of optimal tax because the political system is not willing or able to adopt demogrants means being committed to accepting waste----burning money or psychic value----to achieve greater “equity.” The country might indeed be willing to do this, to some small extent, at the margin: perhaps President Obama will get his way, and the top marginal rate under the income tax will return to 39.6%. But that leaves the troubling question that, if the United States is not willing to generate equity or greater redistribution efficiently, why would it be willing to do so inefficiently? And could we go any further than those modest steps?

This leaves the fourth option: to change the way the country taxes, specifically the tax base. Some might consider the possibility posed this way fundamentally inconsistent. The Article argued throughout that the way things are reflects a certain widespread and popular acceptance of the way things are, and hence must be: this is the essence of “everyday optimal tax theory.” We apply this logic, after all, to existing spending programs, as we just discussed: taking not only the status quo spending programs “off the table” for reform options, but refusing to put a new “demogrant” program on that table. And we also apply this “best of all possible worlds” logic to the structure of marginal tax rates under an income or wage-based tax.

But we are not, after all, hard and fast skeptics for all reform. How can this be: how can we believe that the way things are reflect deeply entrenched policy preferences, positions, and practicalities, and yet there is room still for seemingly radical change?

Our answer lies in the different economic effects and political-rhetorical properties of a progressive spending tax, a public policy option that has not been well enough studied or considered, in significant part because it has not been well enough understood. Hence the descriptive task throughout most of this Article. Under a progressive spending tax, we the people might well be able to have our cake and eat it, too: progressivity without deleterious effects on the productive activities of work and savings. If the research bears out our hopes and suppositions, we submit that this would be a set of policy outcomes that many Americans would endorse.

We now turn to the two normative arguments sketched out above.

10.2. The Welfarist Argument

The goal of policy in a social welfare tradition is to maximize some specified index of aggregate individual well-being using a utilitarian calculus. The intuition that a progressive spending tax might score best under such an approach is that, of all base-rate pairs, such a tax might lead to the highest welfare in a classical welfare economics sense: that is, it gives a “better,” more utility-maximizing optimal tax result than did optimal income tax analysis for Mirrlees and followers. We have developed the components of this argument throughout the Article. Partly it follows the Diamond approach, which itself was an example of formal welfarist analysis. Diamond showed that an income tax with a charitable contribution deduction can support higher marginal rates than such a tax without the deduction, both because the deduction diminishes the labor supply disincentive of the tax rates and because the charitable contributions are themselves public goods. Our argument in this Article substitutes a generalized savings deduction for the charitable contribution deduction. Depending on why people save, how rational and risk averse they are, and what behavioral biases they might have, high spending tax rates need not deter ex ante labor supply decisions, and the overall private capital stock, a public good, will likely increase. Further, under a progressive spending tax, there is an intrinsic way differentially to tax inelastic high spenders, who will continue to work and pay the spending tax, on account of their inelasticity, while getting more savings with diminished work disincentive from elastic worker- savers, who would be deterred from additional work effort by a progressive wage or income tax. These were the hopes intimated in Section 8, though they depend on data analysis yet to be done. The idea is to model plausible ranges of optimal spending tax structures, and to compare them with the optimal income and wage tax structures produced thus far by scholars. Once again, we believe that there is at least hope for greater progressivity under a spending tax, bearing in mind that Mirrlees himself, the progenitor of the optimal income tax field, found his own efforts at generating progressive income tax structures rather disheartening.

10.3. A Fairness Argument

Aside from efficiency or wealth-maximizing, the concern of a welfarist approach, many might accept a progressive spending tax for its fairness properties. Perhaps the simplest way to sketch out such an argument is to look at the various treatments of savings, where the differences among the various progressive tax base-rate pairs are most stark.

An income tax, in theory, double taxes all savings. It is important to note that we have never come very close to an ideal income tax in the United States, as we noted in the prior section, both because it is practically difficult to tax many forms of savings---unrealized appreciation comes to mind—and because we seem to lack the will to double tax all savings--- the non-income taxation of most retirement savings within the nominal income tax comes to mind. A wage tax taxes no savings, by design. A progressive spending tax splits the difference, by design, and on principle, in particular taxing any unexpected returns to savings.

With these analytic facts before us, we ask the normative question: What should be done about taxing savings? Mill’s claim that the income tax is a double tax on savings is simply descriptive. It is true both within the income tax’s own base, where savers are penalized vis a vis spenders, and relative to a hypothetical no-tax world, where the income tax destroys the equivalence, in present value terms, between savers and spenders, Ants and Grasshoppers. Yet neither of these facts exert a strong pull on moral intuitions; it is hard to get from Mill’s is to any compelling ought.

A strong sense of a compelling normative, fairness-based argument in contrast can be gleaned from the near century of experience with the income tax in the United States. At the dawn of the creation of the comprehensive individual tax system, reformers wanted an income tax because it included the yield to savings, and thus would impose an added burden on financiers and the like.[127] Those were, however, simpler times. As the income tax expanded in both scale, becoming a higher burden and more steeply sloped in its rate progression, and scope, reaching the majority of earners in the U.S. and elsewhere, things changed.[128] Lawmakers began to have second thoughts about double-taxing the yield to savings, anywhere and everywhere. Exceptions to the income tax’s theoretical commitment to double-taxing savings have been piled on one another, whether by happenstance, inertia, deliberate policy plan, or mere mistake. The result is hybrid taxes, perched—typically, uneasily—between an income tax model, with its double tax, and a consumption tax, with its principled non taxation of savings. But the compromises to bring about this state of affairs have been effected without much normative or practical reflection, resulting in a tax system in which the lucky and the well-endowed—the capitalist class—can live well and consume away, tax-free. The system neither favors new savings nor effects a fair distribution of tax burdens across taxpayers; individuals who can live off the yield to capital quite simply need pay no tax.

On reflection, however---what we hope to have brought to bear in this Article---the seemingly divergent strands in contemporary tax systems are not random. Settled reflection reveals that ordinary moral intuitions may in fact reasonably reach different normative judgments about different uses of savings. How one uses her savings seems more important, from a political, moral and social point of view, than the sources of such savings, which can be morally arbitrary: luck in many cases. On the one hand, one is sympathetic to the noble Ant, especially when she is manifest as a middle-aged wage-earner, struggling to make ends meet while paying her taxes and setting aside some funds for her later retirement, or for medical or educational needs within her family. Why should we punish her, with a second tax, for her prudence? And so the country has tax-favored retirement, medical, and educational savings accounts. On the other hand, one is haunted by the specter of the socially privileged, such as a second or third- generation rich child, living well off the fruits of someone else’s prior capital accumulation. Surely this “trust fund baby” should be taxed more than the hard working Ant? Surely his income, in the form of rents, royalties, interest, dividends, and the like should count in the tax base, at least as much as the product of Noble Ant’s blood, sweat, and tears?

These simple insights and intuitions cash out into the two norms about savings we introduced above. The yield-to-savings norm holds that capital that enables a higher, better, lifestyle should bear a burden, one at least commensurate with normal wage earnings. The ordinary-savings norm holds that capital transactions (borrowing, saving, investing) that are simply used to move around uneven labor market earnings in time, allowing people to save for their retirement or for periods of high spending needs/low earnings, such as times of education or medical urgency, should not be double-taxed or otherwise discouraged and burdened. Fitting these norms into our preferred vocabulary for savings, smoothing effects the ordinary-savings norm, shifting the yield-to-capital norm. Indeed, shifting transactions are what are left once smoothing and precautionary uses have been subtracted out from the universe of all uses of savings. The net idea is that society ought not to burden smoothing transactions with a double tax, but that the yield to capital is an element of value that can properly be taxed when used to enable a “better,” more expensive lifestyle. A progressive spending tax gives this result.

We suggest that many might find this the “fair” set of norms and principles to govern the taxation of savings, regardless of the efficiency arguments canvassed above. But we also cling to the hope that a progressive spending tax could deliver greater equity and efficiency than the alternatives. To sustain that claim, however, considerably more research is needed

10.4. Our Last Words: A Research Agenda

We did not come to offer definitive answers. We came rather to pose important questions. We believe that a great many citizens are interested in obtaining more progressivity from the tax system. We also believe that income and wage taxes---and we believe that the income tax is well on its way to being a wage tax---are intrinsically limited in their abilities to bear any more progressivity than what already obtains. Indeed, this was an insight that Mirrlees had at the dawn of the optimal income tax movement, in the early 1970s. For both sound economics and not- unrelated political and rhetorical reasons, the current tax system offers only very limited hope for greater progression. See Joe the Plumber. Or see the Obama Administration’s modest proposals to increase progressivity in tax – and the utter failure to enact them to date.

We believe that the best---possibly the only---way out of the bind, for those who want more progressivity, is to change what we are taxing, the tax base. There are good reasons to believe that a progressive spending tax could support higher and more steeply sloped marginal rates than could an income or wage tax. The reasons are that many people in the face of progressive spending taxes will react by saving more rather than working less, an action that could be fully rational, depending on why they are saving and their anticipated tax rates on ultimate consumption, and/or consistent with certain behavioral biases. Society will benefit from the work effort, the increased savings, and the inchoate tax on ultimate consumption. Others, inelastic spenders---consumption addicts---may continue to work and spend, paying the high tax rates. Society will benefit from the greater tax revenue. Still others will see some considerable tax fall on their spending out of life’s good fortune, windfall gains having little or no ex ante deterrent effects on their labor supply. Meanwhile, tax rates on lower and middle class consumption can be lowered. A progressive spending tax holds out hope for greater efficiency and equity.

As of now, this is just hope: what we believe is the last best for hope for progressivity in tax. To make it more real, we need more and better analysis of why people save, under both plausible rational and behavioral models. We need to understand more precisely how work effort might respond to progressive marginal spending tax rates. We also need to understand how the various practical forms of implementing a progressive spending tax effect opportunities for evasion and non-compliance, and how these might undercut whatever gains could otherwise be expected. We suspect that a mixture of theoretical, applied, experimental and other types of research will be needed. This will all take some time.

It is our final hope in this Article that such time be taken, such work be done. It is important work, which has been obscured by misunderstandings about the analytics of taxation and a limited sense of the appropriate argument structure for consumption taxes. Putting aside metaphysical arguments about the double taxation of savings, yield-exemption, and “neutrality” between present and deferred consumption, and focusing instead on the end of redistribution, a spending tax might well be the last best hope for progressivity in tax. As we have said throughout, we came, not to save progressivity in tax itself, but to save the argument for others to have out. This is the fiscal agenda for the future.

[1] McCaffery is the Robert C. Packard Trustee Chair in Law and Professor of Law, Economics, and Political Science, University of Southern California; Visiting Professor of Law and Economics, California Institute of Technology; B.A., Yale College; J.D., Harvard Law School; M.A. (Economics) University of Southern California. Hines is the Richard A. Musgrave Collegiate Professor of Economics and Professor of Law, University of Michigan; Research Associate, National Bureau of Economic Research (NBER); B.A., Yale College; Ph.D. (Economics) Harvard University. Earlier versions of this article were presented at the Harvard Law School Seminar on Current Research Issues in Taxation, and workshops at UCLA, Northwestern, Columbia, Loyola (L.A.), Michigan, NYU and the University of California at Berkeley Law Schools; we thank participants at all of these sessions, including Ellen Aprill, Steve Bank, Neal Buchanan, Aaron Edlin, David Gamage, David Hasen, Mitchell Kane, Louis Kaplow, Ed Kleinbard, Wojciech Kopczuk, Kyle Logue, Katie Pratt, Eric Rakowski, Alex Raskolnikov, Chris Sanchirico, David Schizer, Jonathan Schwartz, Ted Seto, Dan Shaviro, Joel Slemrod, Kirk Stark, Eric Talley, Alvin Warren, David Weisbach and, especially, for his detailed commentary and advice, Jeff Strnad. We thank Nataline Viray-Fung for excellent research assistance, along with the reference librarians at USC Law School.

[2] For the recent budget proposal, which restates the President’s prior year and never-enacted proposal in regard to the marginal tax rate increases, see Jackie Calmes, In 3.8 Trillion Dollar Budget, Obama Pivots to Trim Future Deficits, N.Y .TIMES Monday Feb. 1, 2010, page A1. see also Tax Notes analysis, February 8, 2010. For illustrative campaign rhetoric, consider for example the following exchange in the Democratic presidential debate before “Super Tuesday” at the Kodak Theater in Los Angeles. Moderator Wolf Blitzer of CNN asked candidates Hilary Clinton and Obama to clarify their positions on income tax reform:

BLITZER: . . . I just want to be precise. When you let --- if you become president, either one of you --- let the Bush tax cuts lapse, there will be effectively tax increases on millions of Americans.

OBAMA: On wealthy Americans. CLINTON: That’s right.

BLITZER: And you are willing to go into . . . . . (CROSSTALK)

OBAMA: I’m not bashful about it. CLINTON: Absolutely. Absolutely.

OBAMA: I suspect a lot of this crowd --- it looks like a pretty well- dressed crowd, potentially will pay a little bit more. I will pay a little bit more . .

. .

CLINTON: But Wolf, it’s just really important to underscore here that we will go back to the tax rates we had before George Bush became president. And my memory is, people did really well during that time period.(APPLAUSE)

See Lynn Sweet: Clinton, Obama Hollywood debate, Kodak Theater Jan. 31, 2008, Transcript.Online at https://blogs.suntimes.com/sweet/2008/02/sweet_clinton_obama_hollywood.html(last visited June 2, 2008).

[3] Emmanual Saez suggests that $250,000 would cover 3-4% of earners. See <<https://elsa.berkeley.edu/~saez/saez- UStopincomes-2006prel.pdf>> That figure may be too high. Census publications indicate that the95th percentile of income earners in 2007 began at $177,000 of household income. See Carmen DeNavas-Walt, Bernadette D. Proctor and Jessica C. Smith, U.S. Census Bureau, Current Population Reports, P60-235, Income, Poverty, and Health Insurance Coverage in the United States: 2007, U.S. Government Printing Office, Washington, DC, 2008, at Table A-3.

[4] In March, 2009, President Obama announced the formation of a Tax Reform panel led by former Federal Reserve chairperson Paul Volcker. See https://news.yahoo.com/s/cq/20090326/pl_cq_politics/politics3084893, last visited April 6, 2009. The “constraints” imposed on the panel were that there could be no tax increases at all in 2009 and 2010, and that no one earning less than $250,000 see any tax increase. See Meg Shreve, White House Announces Formation of Tax Reform Panel, 122 Tax Notes 1539, March 30, 2009. The story of the possible taxation of bonuses received by AIG executives at 90% rates hardly belies our general theme. For one thing, this potential tax was clearly intended to send a special and harsh message. For another thing, President Obama, in part because of concerns over incentive effects moving forwarded, ultimately failed to support the idea, and it was quietly tabled. See Administration: Changes May Be Neededin 90 Percent Tax on AIG Bonuses, Foxnews.com, <<https://www.foxnews.co m /politics/first100days/2009/03/22/administration-changes-needed-percent-tax-aig- bonuses/>>, last visited April 12, 2009.

[5]See Obama to let Bush Tax Cuts Expire, Reuters News Service, Saturday Feb. 21, 2009, available at https://www.reuters.com/article/idUSTRE51K1ZF20090221, last visited Feb. 22, 2010.

[6] For historical background, as on tax rates, see GENE STEUERLE, CONTEMPORARY U.S. TAX POLICY (2008); JOEL SLEMROD AND JON BAKIJA, TAXING OURSELVES, 4TH ED. (2008); STEVEN BANK, KIRK STARK AND JOSEPH THORNDIKE, WAR AND TAXES (2008)

[7] The Tax Reform Act of 1986 introduced a top individual tax rate of 28% and many individual tax changes, including one that required high-income individuals to reduce their exemptions and deductions as their incomes rose over a range. Due to the reduction in exemptions and deductions, an individual’s total tax burden rose by $33 for every additional $100 of earnings within this range, which can therefore be interpreted as a marginal tax rate of 33%. See Steuerle, supra; Edward J. McCaffery, Cognitive Theory and Tax, 41 UCLA LAW REV. 1861 (1994).

[8] McCain: Obama “Redistributionist in Chief,” upi.com, Oct 28, 2008, <<https://www.u p i.com/Top_News/2008/10/28/McCain_Obama_redistributionist_in_chief/UPI-6282125212571/>>; see also Michael Cooper, “ Spreading the Wealth” as Both Accusation and Prescription, New York Times, Oct. 20, 2008, at page A20.

[9] See David Leonhardt, A Free-Market-Loving, Big-Spending, Fiscally Conservative Wealth Redistribution, N.Y. Times Magazine, Aug. 24, 2008.

[10] See Richard Wolf, Obama’s Budget Proposal Draws Rapid Fire from Legislators, USA Today, Monday Feb. 1, 2010, A1; Jeremy Scott, Snow, Republicans put Freeze on Congressional Tax Agenda, Tax Notes, Feb. 15, 2010, 813-814.

[11] Congress first proposed the surcharge in the Summer of 2009. See Wall Street Journal On-Line, The Small Business Surtax: The Obama Democrats Pick Income Redistribution Over Job Creation and EconomicGrowth, available on-line at < https://online.wsj.com/article/SB124753106668435899.html >, last visited Feb. 22, 2010. But there was widespread opposition, and Congress soon backed off. See Talk Radio News Service, Pelosi: Reducing Health Care Surcharge for Wealthy a Possibility, July 16, 2009, available at < https://talkradionews.com/2009/07/pelosi-reducing-health-care-surcharge-for-wealthy-a-possibility/ >, last visited Feb. 22, 2010. The idea appears not yet dead. See Obama’s Health Care Proposal, Kaiser HealthNews, Feb. 22, 2010, available online at < https://www.kaiserhealthnews.org/Stories/2010/February/22/Obama-Health-Care- Proposal.aspx>.

[12]See for example Larry M. Bartels, Homer Gets a Tax Cut: Inequality and Public Policy in the American Mind, 3 PERSPECTIVES ON POLITICS 15 (2005) (suggesting confusion and the salience of narrow self-interest over ideology); see more generally LARRY M. BARTELS, UNEQUAL DEMOCRACY: THE POLITICAL ECONOMY OF THE NEW GILDED AGE (2008) (accord); see also Jonathan Baron and Edward J. McCaffery, Masking Redistribution, in Edward J. McCaffery and Joel Slemrod, editors, BEHAVIORAL PUBLIC FINANCE (2006) (showing how redistributive effects of law change can be hidden and under-considered by subjects in Web-based experiments).

[13] See Leonhardt, supra, at page 9. See also Alan D. Viard, The Trouble with Taxing Those at the Top, American Enterprise Institute for Public Policy Research, Tax Policy Outloook, No. 2., June 2007.

[14] The fountainhead for the debate can be found in JOHN STUART MILL, PRINCIPLES OF POLITICAL ECONOMY, (Jonathan Riley, ed., Oxford University Press, 1998) (1848). Subsequent contributions include IRVING FISHER AND HERBERT W. FISHER, CONSTRUCTIVE INCOME TAXATION, A PROPOSAL FOR REFORM (New York: Harper, 1942); NICHOLAS KALDOR, AN EXPENDITURE TAX (London: Allen and Unwin, 1955). The classic argument in an American law review context was had out in William D. Andrews, A Consumption-Type or Cash Flow Personal Income Tax, 87 HARVARD LAW REVIEW 1113 (1974); Alvin C. Warren Fairness and a Consumption-Type or Cash Flow Personal Income Tax, 88 HARV. L. REV. 931 (1975); William D. Andrews, Fairness and the Personal Income Tax: A Reply to Professor Warren, 88 HARV. L. REV. 947 (1975); and Alvin C. Warren, Would A Consumption Tax Be Fairer Than an Income Tax?, 89 YALE L. J. 1081 (1980). For recent discussion, see Joseph Bankman and David A. Weisbach, The Superiority of an Ideal Consumption Tax over an Ideal Income Tax, 58 STAN. L. REV. 1413 (2006); Daniel Shaviro, Beyond the Pro-Consumption Tax Consensus, 59 STAN. L. REV. 745 (2007); and Joseph Bankman and David A. Weisbach, Consumption Taxation is Still Superior to Income Taxation, 59 STAN. L. REV. 789 (2007). For a summary of the intellectual history of the debate, with many citations, see Edward J. McCaffery, A New Understanding of Tax, 103 MICH. L. REV. 807 (2005).

[15] Here the classic citation is WALTER J. BLUM AND HARRY KALVEN, JR., THE UNEASY CASE FOR PROGRESSIVE TAXATION (Chicago: University of Chicago Press, 1953). See also Joseph Bankman and Thomas Griffith, Social Welfare and the Rate Structure: A New Look at Progressive Taxation, 75 California Law Review 1905 (1987); Harvey Rosen, Income Tax Progressivity: A Century-old Debate, Business Review, Federal Reserve Bank of Philadelphia, January/February 1990. For recent statements in favor of flat rate or proportionate taxation see

[17] HENRY C. SIMONS, PERSONAL INCOME TAXATION: THE DEFINITION OF INCOME AS A PROBLEM OF FISCAL POLICY 50 (1938).

[18] Mill 1848, supra.

[21] Andrews 1974, supra; DAVID F. BRADFORD AND THE U.S. TREASURY TAX POLICY STAFF, BLUEPRINTS FOR BASIC TAX REFORM, 2ND ED. (Arlington, VA: Tax Analysts, 1984), DAVID F. BRADFORD, UNTANGLING THE INCOME TAX (Cambridge, MA: Harvard University Press, 1986).

[22] Andrews 1974, supra; McCaffery, A New Understanding, supra; Edward J. McCaffery, Three Views of Tax, 18 Canadian J. of Law & Jurisprudence 1553 (2005).

[23] Consider the formula for the future value (FV) of a present amount (P), invested over n periods at a rate of return r:

[24] We say the “right” sort not for normative reasons----that a spending tax is “better” than a wage tax but for the analytic reason that the case for progressive marginal rates under a spending tax is stronger than under a wage or income tax: the central positive claim of this Article. We discuss in the concluding Section 10 whether this might lead to a normative argument for a progressive spending tax.

[25] An income tax is, by design, a tax on wages combined with a “double tax” on savings. As Mill pointed out, the saver is left worse off than the immediate consumer under an income tax, and worse off than she would be under a consumption tax. The above equations can help us to see this, with a second tax, t, levied on the return to savings, or r, under an ideal income tax. The inequality below adds an income tax, on the left, to the wage and spending taxes, on the right:

The addition of the second (1 – t) on the left---note that there are two t s for the income tax---reduces the after-tax return to savings under an income tax below 100 percent, and thereby makes the income tax less favorable to savers than either a wage or a spending tax, at constant rates.

[26] HALL AND RABUSKHA, supra; SLEMROD AND BAKIJA , supra; EDWARD J. MCCAFFERY, FAIR NOT FLAT: HOW TO MAKE THE TAX SYSTEM BETTER AND SIMPLER (2002).

[27] See generally <<https://www.fairtax.org >>; NEAL BOORTZ, THE FAIRTAX: THE TRUTH: ANSWERING THE CRITICS (2008); Laurence J. Kotlikoff and David Rapson, Comparing Average and Marginal Tax Rates under ther Fair Tax and the Current Systems of Taxation, in John W. Diamond and George R. Zodrow, editors, FUNDAMENTAL TAX REFORM: ISSUES, CHIOCES AND IMPLICATIONS (2008).

[28] See e.g., Anne L. Alstott, The Uneasy liberal Case Against Income and Wealth Transfer Taxation: A Response to Professor McCaffery, 51 Tax L. Review 363 (1996); McCaffery, A New Understanding, supra, at footnote 15, and sources cited therein.

[29] Others seem to take this as a semantic matter: namely, that a consumption tax must not include the yield to savings. Being semantics, one could have this any way one wants, as long as terms are consistently defined and used. In the interest of keeping semantics close to ordinary language, we define a postpaid consumption or expenditure—a spending—tax as a “consumption” tax, even if its rate structure means that the yield to capital will sometimes be taxed, by design.

[30] The U.S. income tax is not completely transparent in its “zero bracket,” which does not appear directly in I.R.C. § 1, which generally sets out the marginal rate structure. A family’s range of nontaxable income depends on whether or not the family itemizes (as most families do not), the number of personal credits including dependents, and whether the family is subject to the alternative minimum tax (AMT).

[31] HALL AND RABUSKA, supra; MCCAFFERY, FAIR NOT FLAT, supra.

[32] The word “demogrant” combines “demo,” or people, and “grant,” in this case money. True demogrants must be “lump sum,” that is, calculated without regard to income. Otherwise they easily collapse into combinations of grants and marginal taxes, since grant amounts that decline with income have the same features as income taxes. Put another (equivalent) way, lump sum payments (or taxes) have no incentive effects. See Bankman and Griffith, supra.

[33] SLEMROD AND BAKIJA, supra; Bankman and Griffith, supra; LIAM MURPHY AND THOMAS NAGEL, THE MYTH OF OWNERSHIP: TAXES AND JUSTICE (New York: Oxford University Press, 2005), LOUIS KAPLOW, THETHEORY OF TAXATION AND PUBLIC ECONOMICS (Princeton, NJ: Princeton University Press, 2008), but cf. Lawrence Zelenak and Kemper Moreland, Can the Graduated Income Tax Survive Optimal Tax Analysis?, 53 TAXLAW REVIEW 51 (1999).

[34] Louis Kaplow, Public Goods and the Distribution of Income, 50 EUROPEAN ECONOMIC REVIEW 1627 (2006).

[35] Peter A. Diamond, Optimal Tax Treatment of Private Contributions for Public Goods with and without Warm Glow Preferences, 90 JOURNAL OF PUBLIC ECONOMICS 897 (2006).

[36] Zelenak and Moreland, supra. Zelenak and Moreland do a nice job of setting out the centrality of average tax rate in other analysis, particularly Bankman and Griffith’s, supra. They then list as the first reason why marginal rates might still matter the political infeasibility of demogrants. We discuss this argument below, in the context of Murphy and Nagel’s similar argument. Zelenak and Moreland’s other reasons why progressive marginal rates might matter are that envy might figure into the social welfare function; taxation might serve as insurance against wage uncertainty; high income taxpayers might be more inelastic than generally assumed (a qualification also noted by Mirrlees in 1971); the distribution of abilities in society differs from the usual assumptions (there is a “thick” upper tail); or there are “winner take all” markets. These qualifications typically fit the standard models and are indeed often made in the literature; they also all come in the context of an income or wage tax. Our argument, in contrast, is that society might have reasonable concerns over the patterns of behavior of its richest citizens, especially a pattern of unchecked, high-end consumption. We develop these arguments infra.

[37] Technically, average rates will constantly---monotonically---increase as long as marginal rates are above average rates; this could obtain with some intervals of declining marginal rates. In the rate tables presented above, for example, imagine that the top marginal rate fell from 30 to 25% before the average tax rate had obtained 25%; we would have at least episodically declining marginal rates while steadily increasing average ones.

[38] See Leonhardt, supra.

[39] Indeed, the Nobel laureate Joseph Stiglitz even has an optimal tax model wherein marginal tax rates on the most able are negative. See Joseph E. Stiglitz, Self-selection and Pareto Efficient Taxation, 17 JOURNAL OF PUBLIC ECONOMICS 213 (1982).

[40] For example, Warren Buffett famously noted that his own average tax rate was considerably lower than those of his much less well compensated employees, including his secretaries, and conjectured that the same was true of every other wealthy investor. See David Ellis, Buffett Talks Tax Reform with Sen. Clinton, CNNMoney.com, June 27, 2007.

[41] See also Peter A. Diamond and James A. Mirrlees, Optimal Taxation and Public Production II: Tax Rules, 61 AMERICAN ECONOMIC REVIEW 261 (1971); Bankman and Griffith supra; and Alan J. Auerbach and James R.Hines Jr., Taxation and Economic Efficiency, in ALAN J. AUERBACH AND MARTIN FELDSTEIN EDS., HANDBOOK OF PUBLIC ECONOMICS, vol. 3 (Amsterdam: North-Holland, 2002).

[42] Auerbach and Hines supra review major contributions in the Mirrlees tradition.

[43] The zero top marginal rate was first identified by Efraim Sadka, On Income Distribution, Incentive Effects, and Optimal Income Taxation, 43 REVIEW OF ECONOMIC STUDIES 261 (1976), and Jesus K. Seade, On the Shape of Optimal Tax Schedules, 7 JOURNAL OF PUBLIC ECONOMICS 203 (1977).

[44] MATTI TUOMALA, OPTIMAL INCOME TAX AND REDISTRIBUTION (1990), at 14. See also Kaplow 2006 and 2008, supra; Auerbach and Hines supra.

[45] Mirrlees 1971, supra.

[46] Seade, 1977, supra; Stiglitz, 1982, supra.

[47] Mirrlees 1971, supra; Emmanuel Saez, Optimal Income Transfer Programs: Intensive versus Extensive Labor Supply Responses, 117 Q.J. E. 1039 (2002); Kaplow 2006 and 2008, supra.

[48] See, for example, Alan B. Krueger and Bruce D. Meyer, Labor Supply Effects of Social Insurance, in ALAN J. AUERBACH AND MARTIN FELDSTEIN EDS., HANDBOOK OF PUBLIC ECONOMICS, VOL. 4 (Amsterdam: North-Holland, 2002). One interesting quirk of the literature is that most of the studies of labor supply elasticities look at “uncompensated” elasticities: that is, how the variable, labor supply, responds to a change in the price, taxes affecting the after-tax wages that are the real price of work. In standard welfare economics, however, economists look to compensated elasticities. The difference arises because any tax has two effects.One, the substitution effect, alluded to above in the text, arises because the tax affects the price, and thus leads to a substitution away from higher priced goods or activities. But taxes also take money away from taxpayers, leading to the second effect, the income one. Because individuals need income, they might work more in the face of a tax. A “compensated” elasticity corrects for the income effect, to look at the distortion caused by the substitution effect alone. An uncompensated elasticity, in contrast, looks at the real behavior, the amalgam of income and substitution effects. This can often be low, because the two effects roughly cancel each other out, although there can be a high utility loss from the tax. Interestingly, although a strong case can be made that economists “should” look to compensated elasticities, the government, simply as an institution concerned with real revenues, would typically look to uncompensated ones. Hence this may, in fact, be the more relevant variable for “everyday optimal tax” theory.

[49] See EDWARD J. MCCAFFERY, TAXING WOMEN (1997 and 1999 (paperback ed.)).

[50] Martin S. Feldstein, The Effect of Marginal Tax Rates on Taxable Income: A Panel Study of the 1986 Tax Reform Act, 103 J. POL. ECON. 468 (1995); Martin S. Feldstein, Tax Avoidance and the Deadweight Loss of the Income Tax, 81 REVIEW OF ECONOMICS AND STATISTICS 674 (1999); Raj Chetty, Is Taxable Income Sufficient to Calculate Deadweight Loss? The Implications of Evasion and Avoidance, AMERICAN ECONOMIC JOURNAL:ECONOMIC POLICY, forthcoming.

[51] See Joel Slemrod, Optimal Tax and Optimal Tax Systems, 4 J. ECON. PERSP. 157-78 (1990); Wojciech Kopczuk, Tax Bases, Tax Rates, and the Elasticity of Reported Income, 89 J. PUB. ECON. 2093-2119 (2005); Chetty, supra.

[52 ]Diamond 2006, supra.

[53] Peter Diamond, Optimal Taxation: An Example with a U-Shaped Pattern of Optimal Marginal Tax Rates, 88 AM. ECON. REV. 83 (1998)

[54] M. Dahan and M. Strawczynski, “Optimal taxation: An example with a U-shaped pattern of optimal marginal tax rates: Comment,” 90 American Economic Review 681 (2000).

[55] Stiglitz 1982, supra.

[56] See Auerbach and Hines 2002, supra, and Kaplow 2008, supra, for surveys.

[57] Mirrlees, supra, at 207-08.

[58] Mirrlees, supra, at 208.

[59] The tax bases under progressive marginal rates also differ in how they affect uneven flows: progressive income and wage taxes burden “lumpy” or uneven earnings vis a vis even earnings, whereas progressive spending taxes are neutral as to the timing of earnings but burden “lumpy” or uneven spending. See McCaffery, A New Understanding of Tax, supra.

[60] See, e.g., Lawrence J. Kotliokoff and Avia Spivak, The Family as an Incomplete Annuities Market, 89 J. POL. ECON. 372-91 (1981); Lawrence J. Kotlikoff andf Lawrence Summers, The Role of Intergenerational Transfers in Aggregate Capital Accuulation, 89 J. POL. ECON. 706-32 (1981).

[61] Note that a spending tax consistently subtracts savings, under the simple formula (a rearrangement of the Haig- Simons definition), Consumption = Income – Savings. This means that a spending tax will include debt within its base, as a form of negative savings. Debt that is used to save ends up being a “wash:” an inclusion qua debt minus a deduction qua savings. But debt that is used to finance spending is taxed currently.This may sound odd, but need not: consider a basic sales tax. You pay this tax when you buy an item on credit, as with a credit card; you do not pay the tax again when you pay off your balance. So too debt to finance personal spending is within a spending tax base; repayments of principal of debt, which in fact reflect positive savings, are not. Students and others often consider that this means a progressive spending tax will hurt them, but, in fact, allowing the tax to come due in the period of spending lowers average tax rates across a lifetime for students, under progressive rates, compared to the income tax treatment of ignoring both debt and its repayment. Note also that borrowing under a progressive spending tax works like retirement savings in reverse; it pulls down the level being taxed. For more detail, see McCaffery, A New Understanding, supra, and Three Views of Tax, supra.

[62] Francisco Modigliani and Richard Brumberg, Utility Analysis and the Consumption Function: An Interpretation of Cross-Section Data, in KENNETH K. KURIHARA, ED., POST-KEYNESIAN ECONOMICS (1954); Francisco Modigliani, The Role of Intergenerational Transfers and Life Cycle Saving in the Accumulation of Wealth, 2 J. ECON. PERSP. 15-40 (1988).

[63] See generally LAURENCE J. KOTLIKOFF, ED., WHAT DETERMINES SAVINGS? (1989).

[64] Id.

[65] See generally Wojciech Kopczuk and J. Lupton, To Leave or Not to Leave: An Empirical Investigation of the Distribution of Bequest Motives, 74 (1) REVIEW OF ECONOMIC STUDIES 207–235 (2007); Wojciech Kopczuk, Economics of Estate Taxation: Review of Theory and Evidence, Draft on file and available on-line, February 6, 2009.

[66] McCaffery, New Understanding, supra.

[67] See Andrews, 1974, supra; Edward J. McCaffery, Tax Policy Under a Hybrid Income-Consumption Tax, 70 Tex. L. Rev. 1145 (1992); Michael Knoll, Designing a Hybrid Income-Consumption Tax, 45 UCLA LAW REV. 99(1994).

[68] IRC § 408.

[69] IRC § 408A.

[70] See McCaffery, A New Understanding, supra, at 827-29, 878-880.

[71] See Edward J. McCaffery, Good Hybrids/Bad Hybrids, Tax Notes 1699-1705, June 27, 2005; Edward J. McCaffery, Behavioral Economics and Fundamental Tax Reform, in JOHN W. DIAMOND AND GEORGE R.ZODROW, EDITORS, FUNDAMENTAL TAX REFORM: ISSUES, CHOICES AND IMPLICATIONS (2008). See also Roger Gordon et al., Do We Now Collect Any Revenue from Taxing Capital Income?, 89 J. PUB. ECON 981 (2004).

[72] Elizabeth Bell, Adam Carasso and C. Eugene Steuerle, Retirement Savings Incentives and Personal Savings, 105 TAX NOTES 1689 (2004).

[73] See Gordon et al., supra; McCaffery , Behavioral Economics and Fundamental Tax Reform, supra.

[74] Michael J. Graetz, 100 Million Unnecessary Returns: A Fresh Start for the U.S. Tax System, 112 YALE LAW JOURNAL 261 (2002); MICHAEL J. GRAETZ, 100 MILLION UNNECESSARY RETURNS: A SIMPLE, FAIR, AND COMPETITIVE TAX PLAN FOR THE UNITED STATES (New Haven, CT: Yale University Press, 2007).

[75] Slemrod, Optimal Tax and Optimal Tax Systems, supra.

[76] Mirrlees, supra.

[77] Judd 1985 supra and Chamley 1986 supra, as recently generalized by Mikhail Golosov, Narayana Kocherlakota and Aleh Tsyvinski, Optimal Indirect and Capital Taxation, 70 REVIEW OF ECONOMIC STUDIES 569(2003) and Mikhail Golosov, Aleh Tsyvinski and Ivan Werning, New Dynamic Public Finance: A User’s Guide, in DARON ACEMOGLU, KENNETH ROGOFF, AND MICHAEL WOODFORD EDS., NBER MACROECONOMICSANNUAL 2006 (Cambridge, MA: MIT Press, 2006). To show the under-determinacy of optimal tax analysis, as well as its sometimes starkly counterintuitive findings, note that a recent trend, in dynamic optimal tax analysis, has been to support a confiscatory tax on savings. The intuition behind this result is that it is easier for those with stores of financial capital to adjust their labor supplies in response to steep wage or income tax rates, because if necessary they can resort to living off the savings. To prevent tax avoidance of this form, and thereby permit the government to impose a broader range of tax rates, certain dynamic optimal tax models would thus confiscate private capital and discourage subsequent capital accumulation. However attractive its properties in an optimal tax framework, this is not a politically viable option in the present United States. We thank Jeff Strnad for discussing the literature and analysis with us.

[78] James R. Hines Jr. and Richard H. Thaler, The Flypaper Effect, 9 JOURNAL OF ECONOMIC PERSPECTIVES 217 (1995).

[79] Feldstein 1995 supra.

[80] There would also be capital supply elasticities to the marginal spending tax rate to consider, to pick up the effects of the savings-spending decision.

[81] We are aware of just one draft asking a related question in an experimental context. See Torner Blumkin, Bradley J. Ruffle, and Yosef Ganun, Are Income and Consumption Taxes Ever Really Equivalent? Evidence from a Real- Effort Experiment with Real Goods, Department of Economics, Ben-Gurion University, January 2009 (draft on file with authors).

[82] Diamond 2006, supra.

[83] We understand the usual objection that “consumption is good,” the so-called engine of our economy. See McCaffery, A New Understanding, supra, and FAIR NOT FLAT, supra. We offer only two quick responses here. One, savings is also good, and savings is the analytic complement of consumption: savings is non-consumption, as the Haig-Simons definition helps us to see. Two, the hope for at least the welfarist argument for a progressive spending tax is that the change from the status quo be both revenue-neutral and preserve at least as much total spending. The idea is that the behavioral responses of the rich or high ability---both towards more savings and for those who continue to spend and pay the tax---will enable a welfare gain that will lead to lower spending tax rates (compared to the stats quo ante) and hence more spending among the lower and middle classes. Put another way, the hope for a progressive spending tax is that the wealthy save more, in part so that the not-wealthy can spend more.

[84] See McCaffery, A New Understanding, supra.

[85] Kotlikoff and Summers, supra; B. Douglas Bernheim, Taxation and Saving, in ALAN J. AUERBACH AND MARTIN FELDSTEIN EDS., HANDBOOK OF PUBLIC ECONOMICS, VOL. 3 (Amsterdam: North-Holland, 2002).

[86] See Kopczhuk, supra.

[87] Michael D. Hurd, Savings of the Elderly and Desired Bequests, 77 AM. ECON. REV. 298 (1987).

[88] One of us has argued separately that a consistent progressive spending tax needs no separate gift and estate tax, for the simple reason that “dead men don’t spend.” See e.g., Edward J. McCaffery, Tax reform to Die For, WALL ST. JOURNAL, Nov. 21, 2003. A consistent spending tax of the sort considered in the text makes ultimate spending the measure for taxation; death or any other time of wealth transfer would preserve the private savings in the capital stock and would not trigger a tax: spending by the heirs or donee would. In essence, this is a “carryover basis” regime for bequests as for gifts (cf. present law IRC § 1014 (bequests)and § 1015 (gifts)), although there is no “basis” under a spending tax, as savings have not been taxed. So heirs take wealth with a basis of zero, in effect. See McCaffery, Fair Not Flat, supra; Uneasy Case, supra. This feature of the analysis is logically separate, so we take no position on an estate tax vel n o n here.

[89] See Max M. Shanzenbach and Robert H. Sitkoff, Jurisdictional Competition for Trust Funds: An Empirical Investigation of Perpetuities and Taxes, 115 YALE L. J.356 (2005).

[90] Kyle D. Logue and Joel Slemrod, “Genes as Tags: The tax implications of widely available genetic information,” University of Michigan Law and Economics, Olin Working Paper No. 07-021, November 2007 (also check NTJ).

[91] THOMAS J. STANLEY AND WILLIAM D. DANKO, THE MILLIONAIRE NEXT DOOR: THE SURPRISING SECRETS OF AMERICA’S WEALTHY (1998).

[92] ROBERT H. FRANK, LUXURY FEVER: MONEY AND HAPPINESS IN AN ERA OF EXCESS (2000).

[93] See Gary Becker and George Stigler, De Gustibus Non Est Disputandum, 67 AMERICAN ECONOMIC REVIEW 76 (1977).

[94] Hersh M Shefrin and Richard H. Thaler, The Behavioral Life Cycle Hypothesis, 26 ECONOMIC INQUIRY 609 (1988).

[95] See e.g., James J. Choi et al., Saving for Retirement on the Path of Least Resistance, in MCCAFFERY AND SLEMROD, EDS., BEHAVIORAL PUBLIC FINANCE, supra; Bernheim 2004 supra.

[96] See Tax Notes analysis of 2009 and 2020 Budget Proposals, supra.

[97] McCaffery, Behavioral Economics and Fundamental Tax Reform, supra.

[98] See Larry Zelenak, Debt-Financed Consumption and a Hybrid Income-Consumption Tax, Tax Law Review (forthcoming, 2009).

[99] See generally Section 6, supra.

[100] Technically, “money illusion” refers to the tendency to view economic matters in nominal rather than real terms, that is, crudely, to ignore the effects of time and inflation on real purchasing power. See Eldar Shafir, Peter Diamond and Amos Tversky, Money Illusion, 112 QJE 341 (1997). We use it, somewhat idiosyncratically, to refer to what we believe is the common tendency to view pretax retirement savings accounts as if they were after-tax ones.

[101] See note on income versus substitution effects, footnote supra.

[102] We say “behavioral” here again because of what we take to be a common tendency to ignore or overly discount an ultimate tax; this would be related to common heuristics and biases such as myopia, present-bias, and hyperbolic discounting.

[103] In 2006, the most recent year for which final statistics are available, individual income taxes accounted for 43.4 percent of all federal revenues; payroll taxes for 34.8. See Budget of the United States: Historical Tables for Fiscal 2008, Table 2.2., available online at <<https://www. g poacc e ss.gov/usbudget/fy08/sheets/hist02z2.xls>>. This meant that payroll taxes raised just over 80 percent as much as personal income taxes. See also Andrew Mitrusi and James Poterba, The Distribution of Payroll and Income Tax Burdens, 1979-1999, 53 Nat. Tax J. 765 (2000). Statistics, Poterba and Mitrusi

[104] See STEURLE, CONTEMPORARY U.S. TAX POLICY, supra; Bartels, supra; Leonhardt, supra; Viard, supra.

[105] MURPHY AND NAGEL, supra.

[106] McCaffery, A New Understanding, supra; Gordon, supra.

[107] Since benefit formulas tie ultimate Social Security benefits to payroll tax payments, it follows that payroll tax burdens differ in kind from income tax burdens for which the government does not offer individual taxpayers anything in return. Nevertheless, the highly incomplete matching of benefits to an individual’s tax payments during the working lifetime give payroll taxes an important tax aspect.

[108] See generally Anne L. Alstott, The Earned Income Tax Credit and the Limitations of Tax-Based Welfare Reform, 108 HARV. L. REV. 533 (1995); Edward J. McCaffery, The Burdens of Benefits, 44 VILL. L. REV. 445 (1999).

[109] Daniel N. Shaviro, Effective Marginal Tax Rates on Low-Income Households, EMPLOYMENT POLICIES INSTITUTE PAPER, February, 1999; see also Stephen D. Holt and Jennifer L. Romlich, Marginal Tax Rates Facing Low- and Moderate-Income Workers Who Participate in Means-Tested Transfer Programs, 60 NAT. TAX J. 253 (2007); Kaplow, 2006, supra.

[110] Slemrod, supra; Kopczhuk, supra.

[111] On the former problem – the ability of sophisticated financiers to receive compensation taxed at favorable capital gains rates – see Victor Fliescher, Taxing Blackstone, 61 TAX LAW REV. 89 (2008).

[112] Slemrod and Loque, supra.

[113] See e.g. Michael J. Graetz and Emil M. Sunley, Minimum Taxes and Comprehensive Tax Reform, in HENRY J. AARON, HARVEY GALPER, AND JOSEPH A. PECHMAN, EDS., UNEASY COMPROMISE: PROBLEMS OF A HYBRID INCOME- CONSUMPTION TAX, Washington: Brookings Institution (1988) pp. 385-429.

[114] McCaffery, A New Understanding, supra.

[115] JOHN MAYNARD KEYNES, THE GENERAL THEORY OF EMPLOYMENT, INTEREST, AND MONEY (London: Macmillan, 1936).

[116] Zelenak and Moreland, supra; MURPHY AND NAGEL, supra; Kaplow 2006 and 2008, supra.

[117] Mirrlees 1971, supra.

[118] Kaplow 2006, supra

[119] Jonathan Baron and Edward J. McCaffery, Starving the Beast: The Psychology of Budget Deficits in ELIZABETH GARRETT, ELIZABETH GRADDY AND HOWELL JACKSON, EDS., FISCAL CHALLENGES: ANINTERDISCIPLINARY APPROACH TO BUDGET POLICY (2008).

[120] Martin Feldstein, The Effects of Taxes on Efficiency and Growth, NBER Working Paper No. 12201, 2006.

[121] Hines and Thaler 1995 supra; CHARLES T. CLOTFELTER AND PHILLIP J. COOK, SELLING HOPE: STATE LOTTERIES IN AMERICA (Cambridge, MA: Harvard University Press, 1989).

[122] MURPHY AND NAGEL, supra, at 138-39.

[123] Zelenak and Moreland, supra, give other reasons for supporting high marginal tax rates without demogrants. But, aside from their main argument, that, without demogrants, high progressive marginal rates are the only means for achieving progressive average rates as well---the argument that is “blindingly obvious” to Murphy and Nagel--- Zelenak and Moreland’s other arguments have nothing, logically, to do with the absence of demogrants, and thus are refinements that can be made in the traditional optimal tax literature. Thus, e.g., if the high ability are inelastic, marginal rates can be high “at the top” even with demogrants.The fact that optimal tax models can but typically do not make these other assumptions is relevant.

[124] LOUIS KAPLOW AND STEVEN SHAVELL, FAIRNESS VERSUS WELFARE (Cambridge, MA: Harvard University Press, 2002).

[125] BANK ET AL, WAR AND TAXES, supra.

[126] MURPHY AND NAGEL, supra; see also recent editorial commentary from Paul Krugman, Robert Reich, Edward Wolff, and Robert Frank. Kaplow (2006 and 2008) is rather more sanguine, and we share much of hisoptimism. But again, the purpose of our Article is to show the way, as a positive matter, towards more redistribution or progressivity in tax, not to make the logically separate normative arguments that such movement is necessary or desirable.

[127] For some among many sources of good political history of the income tax, see SHELDON D. POLLACK, THE FAILURE OF U.S. TAX POLICY (1996); ROBERT STANLEY, DIMENSIONS OF LAW IN THE SERVICE OF ORDER:ORIGINS OF THE FEDERAL INCOME TAX (1993); and Steven A. Bank, The Progressive Consumption Tax Revisted 101 Mich. L. REV. 2238 (2003) (reviewing MCCAFFERY, FAIR NOT FLAT, supra).

[128] Carolyn C. Jones, Class Tax to Mass Tax: The Role of Propaganda in the Expansion of the Income Tax During World War II, 37 BUFF. L. REV. 685 (1988/89).

Previously published by the University of Southern California - Law School, January 2008

Table of Contents

1. Introduction

2. The Logic of Tax

3. Tax Bases

3.1. Income Tax

3.2. Two Consumption Taxes

3.3. The Traditional Income versus Consumption Debate

3.4. Three Types of Tax

4. Tax Rates

4.1. Vocabulary

4.3. What Matters About Rates?

4.4. A Short Exploration of Optimal Tax Theory

4.4.1. The Basics

4.4.2. Qualifications, Questions, and Caveats

5. Base-Rate Pairs: A New Look

5.1. Three Types of Savings

5.2. Bases and Savings

5.3. A Hint of Normativity

6. The Limits of “Hybrids”

7. The Uneasy Case for Progressive Source-Based Taxes

8. The Easier Case for Progressive Spending Taxes

8.1. A Note on a Parallel

8.2. Why Do People Save? The Rational Story

8.3 A Windfall Benefit?

8.4. A Possible Sort?

8.5 Behavioral Models

9. The World as We Know It

9.1. Tax Today

9.2. On Not Adding Demogrants to the Mix

9.3 Progression without Demogrants

10. Conclusions and Suggestions for Further Research

10.1. Paths to a More Progressive Future

10.2. The Welfarist Argument

10.3. A Fairness Argument

10.4. Our Last Words: A Research Agenda

Continue Reading

\ \ E\ \ Edward Mccaffery \ \ 🇺🇸United States\ \ Robert C. Packard Trustee Chair in Law and Political Science, USC Gould School of Law, Los Angeles, USA](/content/emccaffery/index.html)