LENGTH: 11404 words

SYMPOSIUM ON WEALTH TAXES PART I: Commentary Inequality, Wealth, and Endowment

DANIEL N. SHAVIRO *



* Professor of Law, New York University School of Law. I am grateful to Liam Murphy and Eric Rakowski for helpful comments on an earlier draft.

SUMMARY:
... In his contribution to this issue, Eric Rakowski asks whether "the unequal possession of wealth itself, originating in people's choices to labor and save at different times and in different ways, itself grounds a valid claim to redistribution." ... Rakowski does not subject income or consumption taxes to the burden of justification that he sets for a wealth tax. ... Each prefers the point achieved by the individual with the higher wage rate and budget line. ... Under both inequality measures, a high wage-rate individual may have less wealth, income, or market consumption than a low wage-rate individual simply by reason of having different tastes that led to a different commodity choice as between leisure and the goods earned through work. ... The decision to defer consumption and thus have greater income and wealth down the road, however, can be conceptualized in much the same way as the decision to choose work over leisure. ... An endowment measure helps show that, if two individuals with high earning spouses and the same job opportunities make different choices, the earnings difference between them may provide a poor measure of their relative positions. ... Work aversion may suggest that the beachcomber would suffer extra disutility from the income effect of having to work at least enough to pay the tax. ...  

TEXT:
 [*397] 

I. Introduction
 
In his contribution to this issue, Eric Rakowski asks whether "the unequal possession of wealth itself, originating in people's choices to labor and save at different times and in different ways, itself grounds a valid claim to redistribution." n1 Only if wealth "ideally is the right basis, or part of the right basis, for assessing people's just contributions to the cost of government or for calculating people's redistributive obligations" n2 will he consider it. Arguments of mere "expediency," such as that a wealth tax might be a "proper vehicle for accomplishing any redistribution that justice requires on account of people's unequal capacities ... or well-being," n3 lie beyond the scope of his analysis.

In short, rather than being a proxy for taxing something else, wealth must be the very thing we want to tax. n4 And a well-founded skepticism that it could be such is already built into the above-quoted observation that wealth differences "originate in people's choices to labor and save at different times and in different ways." n5 How could someone's private choices in this regard, which do not in any obvious way harm others and may even benefit them via the opportunity to engage in mutually beneficial transactions, be used to "ground[ ] a valid claim to redistribution?" n6

To my way of thinking, once the analysis is framed this way, Rakowski's conclusion - that a wealth tax cannot be justified on the first-best ground that wealth is the very thing we want to tax - is preordained, yet not particularly significant. Little that we tax, or as a practical  [*398]  matter could ever hope to tax without abandoning distributional considerations, could meet the standard that he sets. True, wealth taxes burden decisions to work and save. So do income taxes. Consumption taxes burden decisions to work. Estate and gift taxes burden decisions to work, save, and engage in gratuitous transfers. All of these taxes, therefore, penalize decisions that are neither bad in themselves nor, at least so obviously as pollution, sources of external burden on others. n7

Rakowski does not subject income or consumption taxes to the burden of justification that he sets for a wealth tax. Indeed, he asserts that "they alone among administrable taxes can plausibly base people's contributions on fairly comprehensive measures of what matters to us materially." n8 (Doesn't wealth also come to mind as a "fairly comprehensive measure of what matters to us materially?") He treats the normative status of those two taxes - or at least whichever of them wins the longstanding income versus consumption tax tournament - as sufficiently well-established to support his considering at some length the possibility that a wealth tax might be justified as an indirect means of perfecting one or the other of them. It should be plain, however, that if "people's choices to labor and save at different times and in different ways" do not by themselves ground valid claims to redistribution, then all three tax bases are equally unacceptable, at least from the ideal as opposed to the expedient standpoint.

Rakowski's critique of wealth taxation therefore has somewhat of a straw-man character concerning whether an explicit federal wealth tax should be added to the U.S. fiscal system. (I say "explicit" because, as he notes, "under certain assumptions, a tax on capital income is equivalent to a tax on the value of that capital itself." n9) Nevertheless, his critique serves a valuable purpose through its extendability. It helps remind us that neither "wealth" nor "income" nor "consumption" plausibly can be what we really want to tax. Rather, the defense of any of these bases must lie in its capacity to provide a crude proxy for something else that is relevant to distributive justice but cannot be directly observed.

What is this other underlying thing? While unlikely to be the subject of universal agreement, it has something to do with inequality.  [*399]  After all, to the extent benefit taxation is unfeasible, there presumably would be no objection to raising all government revenues through a uniform head tax but for the idea that those who are better-off should bear greater burdens. Inequality therefore plays an important role in a variety of views of distributive justice, although under any it rests at least one turtle from the bottom. n10 The move from a description of who is better-off under some metric to the claim that tax burdens should vary by reason of the differences that this metric identifies requires motivation.

Tax bases such as wealth, income, and consumption obviously respond to this aim of discerning who is better-off, but can be defended only as evidentiary of something else. After all, if I turn down a $ 10,000 consulting fee because I would rather do something else with my time, then, while I appear to be worse-off under all three measures than if I had accepted the offer, I presumably have done what I thought most personally advantageous.

Writers in the income tax literature have often recognized the conceptual need for a lower-lying distributional "turtle" than the Haig-Simons (or any competing) income definition n11 itself. The true but unobservable underlying measure that, for reasons lying at least one more turtle down, income tries to implement often goes by the name of "ability" or "ability to pay." The spirit, however, in which this hypothetical measure typically is discussed (or, rather, deliberately not discussed) was well illustrated by Henry Simons, when he argued that attempts to poke too far behind the supposed objectivity of an income definition "lead[ ] directly back into the utter darkness of 'ability' or 'faculty' or, as it were, into a rambling, uncharted course pointed only by fickle sentiments." n12

Hard-nosed views of this sort notwithstanding, I argue here that consideration of the presumed underlying measure of inequality for which wealth, income, and consumption are merely rough proxies is crucial to understanding questions such as whether to adopt an explicit federal wealth tax. Section II attempts, as a descriptive matter, to flesh out the underlying measure, which might be called "endowment," "ability," or "wage rate." Section III considers its possible significance  [*400]  under two leading distributional approaches: utilitarianism or weighted welfarism, and the "liberalism" or "liberal egalitarianism" that has been much discussed in recent legal literature about taxation. n13 In so doing, it argues that conventional rejections of endowment taxation as an orienting idea, usually explained on the ground that it would require enslaving a beachcomber who could have been a Wall Street lawyer, are in key respects confused. Section IV provides a brief conclusion.

II.
 
"Endowment" as an Underlying MeasureofInequality

A. Inequality Measures Based on Welfare orOpportunity
 
The idea that some people are better-off than others is familiar and uncontroversial. Wealth, income, and consumption, while apparently meant (at least in part) to provide a measure of such differences, are flawed. In effect, all three treat the decision not to work as making one worse off, and wealth or income taxation treats the decision not to save as making one worse off, even though these decisions may be rationally self-interested. My aim in this Section is to sketch out the rough contours of an underlying measure that at least comes closer to bedrock (substituting this more hopeful metaphor for that of the unending turtles).

My aim, at this point, is only to sketch out why one might say "A is better off than B," without yet considering what relevance this statement might have to tax-transfer policy. An obvious initial problem is that ways of judging inequality may differ, and that once one gets to views of distributive justice, it may turn out, from some perspective, that one measure rather than another ought to have been used. As it happens, however, a bit of "cheating," in the sense of looking ahead slightly, reveals this problem to be less serious, at least for very crude and general purposes, than initially it might appear.

One obvious measure of inequality would focus on a presumed psychic state that might be called welfare, subjective well-being, or utility. This is the measure most familiar from economics, both descriptively (people are assumed to try to optimize it given their preferences)  [*401]  and normatively (through the use of social welfare functions that depend on the welfare of individuals). One perhaps could question the extent to which the concept is meaningful, since evolution appears not to have equipped us with comprehensive, single-metric "hedometers" or utile counters that are readable even by ourselves. And one certainly could challenge the usual way of looking at it, which is that, the more preferences you satisfy, the greater your subjective well-being. n14 Nonetheless, the concept evidently has enough intuitive appeal for acceptance of its rough descriptive meaningfulness to be fairly widespread.

Rakowski, for example, mainly disparages welfare-based views normatively, rather than on the ground that they rely on something that is not descriptively meaningful. n15 He also complains that the empirical difficulties associated with the use of a welfare-based or other consequentialist standard can make "arguing [under it] that a wealth tax would not be part of the optimal tax scheme ... like fencing with the fog." n16 Still, he mentions what might lead to an alternative measure of inequality by suggesting that distributive justice depends on people's "non-welfare-based entitlements to resources, opportunities, or capabilities in consequence of their choices and of their natural and social fortune." n17 Elsewhere, he has more fully sketched out a theory of justice that he calls "equality of fortune," which holds that, in general, "no one should have less valuable resources and opportunities available to him than anyone else, simply in virtue of some chance occurrence the risk of which he did not choose to incur." n18 Other liberal or liberal egalitarian writers have sketched out broadly similar ideas. n19

The descriptive difference between the inequality measure that such views suggest and one that is welfare-based can be illustrated by comparing a miserable millionaire to a happy beggar, where the difference in their material fortunes results purely from the millionaire's having richer parents and greater innate business ability. By stipulation, the happy beggar has greater subjective welfare, but his opportunities, at  [*402]  least in a material sense, were not as great due to an accident of birth. One need not subscribe to the liberal normative view in order to accept the descriptive meaningfulness of a measure of inequality that is based on people's opportunities or broadly-defined resources.

While the welfare and opportunity-based inequality measures can be coherent descriptively, their differences potentially pose difficulty, at the normative stage, for discourse about distribution between welfarists and liberal egalitarians. For example, even if both groups are willing to use some proxy such as "wealth" or "income" for the thing they care about distributionally, the interpretation that they want to give the proxy may differ. Nonetheless, I would argue that, in describing measures of inequality (leaving aside for now the question of what to do about them), these differences to a considerable extent can be ignored. The reason for the substantial reconcilability of the welfare and opportunity-based measures of inequality relates to the economists' credo that increased choice is generally advantageous, n20 and to the view that, so far as one can tell, two people who make different selections from among the same choices are likely to be equally well-off. If one assumes that greater choice is advantageous, but that we cannot assess how well people have chosen given their preferences and have no other information about utility, then the two alternative measures should produce the same rankings. n21

The point can be made in terms of consumer sovereignty, defined as an evidentiary inference about people's choices. It may seem reasonable to assume that people generally have the best information about their own preferences and the best incentives to act on the basis of these preferences. Thus, it often may be sensible to assume that people make the best choices for themselves given what they know. There are of course exceptions - say, compulsive gamblers or people who undersave due to a lack of self-discipline - but in general, it would be arrogant and fatuous to assume that everyone should make a given choice in a particular way, and thus erred if she chose, say, leisure over work.

 [*403]  That being so, suppose one observes people with the same opportunities making different choices in a case where one has no reason to question consumer sovereignty. One person works more than another due to different tastes regarding work versus leisure affinity or market versus nonmarket consumption. Or one person saves more than another due to differences in consumption preference. Either way, without more information, there may be no basis for concluding that one is better-off than the other. Identical opportunities may be treated as implying the achievement of identical welfare, if only out of ignorance about the welfare actually achieved in each case.

A simple hypothetical may help to illustrate how the two measures of inequality may operate similarly. Imagine a world where there are only two goods, leisure and food. People can freely allocate their time between enjoying leisure and working to earn food. For each of the two goods, they always prefer a larger quantity to a smaller one but experience declining marginal utility (unit n + 1 always yields less satisfaction than unit n).

People in this world differ from each other in just two respects. First, some have higher wage rates than others; that is, they can earn more food per unit of effort or time. Second, under identical circumstances, they may vary in their taste for an extra unit of leisure versus one of food. These tastes can be portrayed graphically by "indifference curves" that show what combinations of the two goods a given individual regards as equally valuable. Suppose, for example, that I am indifferent between receiving (1) 10 units of food plus 12 hours of leisure and (2) 15 units of food plus 11 hours of leisure. These two combinations, along with all others that I regard as of identical value, can be depicted as lying on the same indifference curve. The same amount of one of these goods plus less (more) of the other lies on a lower (higher) indifference curve.

In this world, each individual has a "budget line," depicting the largest food-leisure bundles that she can procure. For example, if I have 24 hours a day to allocate as I like and I can earn one unit of food per hour, I can choose 24 hours of leisure and no food, 24 units of food and no leisure, or various intermediate combinations (such as 16 hours of leisure and 8 units of food, or 14 of leisure and 10 of food). Given wage rate variations, people face different budget lines. For any quantity of leisure below 24 hours, a person with a higher wage rate can get more food than one with a lower wage rate.

Among the bundles available to an individual at her budget line, she picks the one that lies on the highest indifference curve. This basic setup is depicted in Figure 1, where an individual who can earn one unit of food per hour picks B, the point of tangency between budget  [*404]  line AA[su'1'] and U<1>, the highest indifference curve that touches AA[su'1']. (All attainable points on U<0> are less desirable, and none on U<2> are attainable.) Someone who faced the same budget line but had different tastes, portrayed by a distinct set of indifference curves, would have picked a different point along AA[su'1'].

Figure 1
 
[mg f:'[data.t.tax.53-3]tax30401.eps',w24.,d15.2] Suppose two individuals face the same budget line but have distinct indifference curves, leading one to work more and end up with more food but less leisure. We cannot say which of them is better-off. Each prefers the point where she is to that where the other individual is, and indeed chose it for that reason.

Now suppose instead that the individuals have the same indifference curves but different budget lines because one of them has a higher wage rate. Each prefers the point achieved by the individual with the higher wage rate and budget line. Moreover, if one assumes that they have the same utility functions for each good (not just the same preferred rate of trade between the goods) and that nothing else differentiates their utility levels, one can say that the individual with the higher budget line is better-off. Admittedly, these assumptions go far beyond anything we can really hope to know. They may be defended, however, as a reasonable best guess given our ignorance. n22

 [*405]  In short, under the stated assumptions, the higher an individual's wage rate and resulting budget line, the greater the utility one can assume she experiences. Wage rates and budget lines therefore provide a rough measure of inequality if the standard is well-being. What is more, an opportunity standard should lead to the same set of equality rankings under these circumstances as one based on well-being. If the wage rate is defined as exogenous (beyond the individual's control), then, in Rakowski's terms, anyone with a low wage rate has "less valuable opportunities available to him ... simply in virtue of some chance occurrence the risk of which he did not choose to incur." n23

Under both inequality measures, a high wage-rate individual may have less wealth, income, or market consumption than a low wage-rate individual simply by reason of having different tastes that led to a different commodity choice as between leisure and the goods earned through work. When this happens, the high wage-rate individual will misleadingly appear to be worse-off, rather than better-off, if (given the difficulty of observing wage rates) we rely instead on a wealth, income, or consumption measure of inequality. Figure 2 provides an illustrative example.

Figure 2
 
[mg f:'[data.t.tax.53-3]tax30402.eps',w25.5,d16.1] In this example, Andrea has a higher wage rate than Brian, thus giving her a budget line of AA[su'1'] to his AB[su'1']. She works so much less than he does, however, that she ends up at point X, compared to his point Y. (Perhaps, owing solely to differences in taste, she is a law  [*406]  professor and he is a practitioner?) She therefore has lower earnings, yet, presumptively is better-off given that, if she had liked, she could have matched his earnings and still had more leisure, or alternatively matched his leisure and had more earnings. (I ignore for now the "lumpiness" of some work decisions, which may prevent people from fine-tuning their work levels precisely as they would like.) A tax on earnings, however, would likely burden Brian more than Andrea, although this depends on how they respond to it. Under either inequality measure, therefore, a tax on earnings seems likely to increase inequality between Andrea and Brian, by reason of its greater adverse impact on the individual who already was worse-off in its absence.

B. From Simple Hypotheticals to the RealWorld
 
Despite the above hypotheticals' extreme reductionism, they capture something important about the real world. People have different opportunities to garner material resources, and they take advantage of these opportunities in different degrees. Or, to break it down further still, the outcomes that people experience reflect both choice and nonchoice. This distinction may matter either because choice is morally important (as under a liberal egalitarian view) or because it is evidence of preference (as under a welfarist view).

The notion of an exogenous wage rate is meant to capture the unchosen element. Wage rates differ significantly from, say, a lawyer's hourly billing rate, however, both because they express a relationship between earnings and effort, not just time on the clock, and because the lawyer who charges, say, $ 400 an hour did not start out at that level, but is partly earning a return on her past efforts to acquire skills and connections. "Endowment" or "ability" are synonyms for wage rate that may not similarly imply a misleading equivalence to a skilled professional's hourly billing rate.

Terms such as endowment or ability, however, should not be interpreted too smugly as a statement about personal worth. They merely describe some aspect of the available interactions between a given individual and the world in which she finds herself. Intellect may be a part of ability, along with sound judgment, self-discipline, and the other aspects of what some call "emotional intelligence." n24 Ability to earn also may depend, however, on good looks, the exercise of parental influence, possessing white skin if customers are racists, having  [*407]  widely-shared bad taste as an artist or performer, and a willingness to stoop to fraud or violence when they pay off. n25

While the wage rate hypotheticals capture something important about the world, there is much (to put it mildly) that they leave out. Just to name a few things:

(1) They ignore the element of time that is ultimately crucial to assessing wealth taxation. People's income and wealth depend not only on what they earn through work effort, but also on when they consume and on their investment choices with respect to saving. The decision to defer consumption and thus have greater income and wealth down the road, however, can be conceptualized in much the same way as the decision to choose work over leisure.

(2) They treat wage rates, once drawn, as leading to determinate outcomes that are risk-free, thus begging the question of to what extent, say, Bill Gates was skillful rather than lucky. Risk, after one has drawn a wage rate, complicates the analysis (although, as I have discussed elsewhere, the "social insurance" issues it raises are quite similar to those posed by endowment risk, n26 it is beyond my present scope).

(3) They use a model of rational consumer choice in a frictionless world (for example, without lumpy labor supply decisions) that is in some tension with reality.

(4) They ignore (or dump into the undifferentiated leisure category) all the nonmaterial things that may affect an individual's life, such as the quality of her personal relationships, interests, and health, or personal characteristics such as (in Rakowski's colorful expression) "a puckish sense of humor, a lovable disposition, or a melting smile." n27 These would not be part of endowment.

These limitations suggest care in generalizing from the wage rate analysis, but do not require modifying it in any particular way. Holding them all to one side, I argue that wage rate or endowment or ability is the best available measure of inequality under either a welfarist or liberal egalitarian standard. Thus, if one cannot make any constructive use of taste differences, the beachcomber who could have been a Wall Street lawyer is ideally grouped for purposes of measuring inequality with the individual who actually is a Wall Street lawyer, not the one for whom beachcombing was the only option. n28 Tax bases  [*408]  such as wealth, income, and consumption, insofar as they are designed to adjust tax burdens in response to inequality, can be justified only as rough proxies for endowment or ability, used in its stead because they are more observable.

I have not yet considered, however, why the observation that one individual is better-off than another might suggest differentiating their tax burdens. We normally think of this in terms of the tax bases rather than endowment. It is worth asking briefly, however, why endowment (if observable) might matter under a welfarist or liberal egalitarian view, as a way of shedding light on the distributional issues that arise under real world tax systems.

III. Differing Endowment Levels and DistributiveJustice

A. An Endowment Tax and"Slavery"
 
Previous writers, when aware of the endowment idea, have tended to dismiss it out of hand as a distributional measure. Rakowski, whose treatment is more thoughtful than most, notes the prohibitive practical difficulty of measuring endowment, and mentions as well "an aversion to forcing fortunate people to work if they would rather not turn their talents to materially productive activities." n29 Actually levying an endowment tax, he says, "would effectively enslave the able, by forcing them to put their highly taxed talents to some lucrative employ, on pain of sitting in a debtors' prison, however unpalatable the person found richly compensated work." n30

More sweeping expressions of this concern abound in the literature, such as in the following selection from Mark Kelman:


 
[An endowment tax] would violate the simple libertarian principle that the state should not require people, directly or indirectly, to engage in particular activities .... The widespread principle that people ought to be taxed only when they voluntarily convert property rights into marketable form may seem simply a matter of neutrality, a reluctance to force people to make nontax decisions for tax reasons. But  [*409]  this principle, in more refined form, may really be a political
recognitionofabasichumanresistancetocommoditization. n31


 
These arguments can be challenged on several grounds. An initial question is what motivates this particular use of the "simple libertarian principle." Libertarianism in distributive justice often is relied on to make the case against progressive redistribution n32 - a position not congenial to Kelman. n33 Why use it in one place and not the other?

One also could challenge Kelman's definition of neutrality. To people who have a work-leisure choice, he characterizes as "neutral" the decision to tax one choice but not the other. Is a speeding ticket similarly neutral between speeding and not speeding?

Finally, consider the issue of "commoditization." Kelman (writing in 1979) expands on this point by saying that the beachcomber's "refusal to treat potentially marketable resources as commodities represents a desirable anticapitalist strain in a market-obsessed culture." n34 He is mistaken, however, if he thinks that the beachcomber is not, like people in general, making a deliberate commodity choice. If she chooses beachcombing over Wall Street lawyering, that must be what she finds the best use of her talents. Entering the explicit labor market and working for a wage just means that one has reduced self-production (including leisure) to engage in exchange transactions with other people, presumably due to what an international trade economist would call comparative advantage. Deciding whether to enter the explicit labor market is little different in principle from deciding whether to trade baseball cards with a fellow collector. Why should one think that people who eschew trades are somehow nobler, happier, or for that matter less calculating optimizers than those who decide that a trade has something to offer them?

There is of course a vast literature about commoditization or commodification, alienation, and the like, that may underlie Kelman's apparent empirical claim about the perils of more extensive and explicit exchange. The 20 years since he wrote the above passage, however, have brought greater understanding of the empowerment, rather than diminishment, that thinking of one's labor as a valuable and valued commodity can bring. In practical terms, today's main "beachcombers," in the sense of people who really stand poised at the border between  [*410]  entering the explicit labor market and sticking to home production, are households' secondary earners, such as the wife of a working husband in a traditional family. Surely Kelman does not want to keep married women home too, as valiant anticapitalist holdouts in our sadly market-obsessed culture?

Rakowski steers clear of such excursions in criticizing the endowment tax approach. He emphasizes instead a consideration that I would give the label "ease of payment." Given that exchange transactions in the labor market typically involve cash, the goverment can more easily commandeer a portion of the Wall Street lawyer's return to work than any of the beachcomber's return to leisure. So one suspects that paying tax in practice will be more onerous for the beachcomber. Now, this may be both true and relevant, but it is at the level of an administrative problem - albeit a very important one - in implementing a tax system. It has no bearing on the question of how to think about inequality, or who is presumptively better off than whom, as an input to the determination of what sort of system is desirable.

Suppose that Wall Street lawyers were paid in yogurt that spoiled within five minutes if not eaten. And suppose that a percentage of the beachcomber's imputed return from beachcombing could be transferred to other people, just like a portion of a cash wage, without requiring her to leave the beach or impairing her enjoyment of what was left. These "technological" changes in compensation transferability would reverse the character of the administrative problem. Now one would think that Wall Street lawyers uniquely could not pay the tax unless they were dragged in chains to federal office buildings after hours to mop the floors, and one would associate slavery with taxing lawyers rather than beachcombers.

Ease of payment concerns and measurement problems clearly affect the desirability of attempting to implement endowment taxation. Given their irrelevance to measuring pretax inequality, however, they should not affect endowment's standing as an inequality measure more fundamental than income, consumption, and wealth, potentially helping to orient real world policy choices in cases where the administrative problems are not insuperable. For a close analogy, consider the measurement and liquidity arguments that may counsel not imposing current income taxation on unrealized gain from the appreciation of property. Even if these administrative arguments are considered compelling, they do not rebut the view that "economic income" lies closer to bedrock than "realized income." Such a view might motivate, say, denying deductions for investment interest and capital losses that we suspect are being arbitraged against unrealized gain in the taxpayer's portfolio.

 [*411]  Only one more step is needed for the plausibility of "income" or "consumption," defined in terms of market transactions, to collapse altogether as coherent fundamental measures of inequality. Unrealized gain might be considered an example of imputed income, or considered not directly evidenced by an observable market transaction yet fitting comfortably within the accepted scope of economic income. n35 The same holds for, say, the imputed rent enjoyed by a homeowner who forgoes the money she could have earned by charging someone else to live in her home. Implicitly paying oneself the market value of one's labor by choosing housework or leisure over a cash wage could be called imputed and economic income (and imputed consumption) by the same logic, but regularly is assumed not to be within the contours of "income" or "consumption" as they might even hypothetically be defined for tax purposes.

Why do accepted notions of economic income (and consumption) reflect the use of imputation in some but not all of the cases in which there has been no observable market transaction? Presumably, the decision where to draw the line reflects such considerations as the following: ease of payment, measurability, the degree of ready comparability and/or substitutability (at least in the mind of the observer) between actual and implicit transactions, and a sense of what extensions are politically plausible or at least minimally explainable to nonspecialists. Clearly, however, no one really thinks an observable market transaction is needed for conceptual includability, despite all the rhetoric to the contrary. Thus, definitions of "income" and "consumption" that are largely, but not entirely, based on market transactions, even if ultimately good descriptions of what we actually might choose to tax given considerations such as ease of payment, are not sufficiently coherent, well-defined, or focused on relative well-being to serve as "ideal" inequality markers.

To help show the advantages of an endowment concept in assessing relative well-being, consider again stay-at-home spouses who are potential secondary earners for their households. The very fact that stay-at-home spouses are valuable producers n36 makes the policy of adding a tax burden if they enter the explicit labor market both questionable  [*412]  distributionally and likely to promote inefficient tax avoidance behavior in the form of staying home. An endowment measure helps show that, if two individuals with high earning spouses and the same job opportunities make different choices, the earnings difference between them may provide a poor measure of their relative positions. n37

A further implication of the secondary earner problem is also worth noting here. Whenever the government imposes a tax burden on a household that has two or more potential earners, it may force or at least induce explicit labor market participation by members of the household who otherwise would have preferred to stay home. Suppose, for example, that a tax increase on the husband's earnings in a traditional family "forces" the wife to get a job and hire childcare providers for the time she now will be spending at an office. (Otherwise, suppose the family will face a choice between defaulting on its home mortgage and becoming homeless, or else being unable to buy adequate food.) As I discuss further below, no one seems to think that this scenario suggests a "libertarian constraint" on taxing households with two or more potential earners, and some commentators may even welcome the possible effect of discouraging traditional family structures.

So far, this Section has aimed just to show that endowment may provide an important orienting principle for the distributional component of tax-transfer policy notwithstanding that we probably do not want to force talented beachcombers to put away their surfboards, don jackets and ties, and start reviewing loan documents or arranging interest rate swaps. To evaluate endowment more definitely, one must specify why the inequalities that it helps identify are pertinent to tax-transfer policy to begin with. Here the answer may vary, depending on whether one adopts a welfarist or a liberal egalitarian perspective (the two that I consider).

B. Taxing Endowment From a WelfaristPerspective
 
The optimal income tax literature inspired by the work of James Mirrlees n38 generally treats endowment as the desirable thing to tax. (This reflects the assumption that endowment is roughly correlated with utility and that utility otherwise cannot be observed.) The available tax instrument is typically an earnings tax, but this reflects the  [*413]  assumption that only earnings (which reflect both wage rate and effort) are observable. An earnings tax redistributes from people with high wage rates to those with low wage rates, but also discourages work effort. From a welfarist perspective, where the desirability of a state of affairs depends purely on people's well-being in it, n39 the tax rate is set to optimize the value of the progressive redistribution that can be accomplished relative to the cost of deterring work effort through the tax.

These models, however, abstract from the liquidity or ease of payment problem that underlies the hypothetical of the beachcomber who could have chosen to be a Wall Street lawyer. This omission reflects an extreme reductionism that is needed to make the models analytically tractable. In particular, the models generally leave out differences in taste that might induce two individuals with the same wage rate to make different commodity choices, such as those that result from choosing Wall Street versus the beach. Thus, everyone with the same wage rate ends up with the same earnings, n40 works enough to be able to pay the tax, and suffers the same utility loss from having to pay the tax.

Under a utilitarian calculus in which social welfare depends purely on the sum of people's utilities, the motive for progressive redistribution comes mainly from the assumption of declining marginal utility. n41 An extra dollar is worth more to a low-wage-rate individual than to one with a high wage rate because the former has fewer dollars to spend. (For simplicity, I assume that a higher wage rate implies greater earnings; this would not be the case, although the analysis of declining marginal utility might not change much, if money were an inferior good n42 as endowment increased.) The choice of a "weighted" welfarist alternative to utilitarianism, or one in which the welfare of worse off individuals counts for more in the social welfare calculus  [*414]  than that of better off individuals would add an additional motive for progressive redistribution.

Under either a utilitarian or a weighted welfarist approach, the acknowledgment of taste differences, and the consequent possibility that two individuals with the same endowment will end up as a Wall Street lawyer and a beachcomber respectively, complicate the analysis in several ways. Presumably, the beachcomber has greater work aversion than the Wall Street lawyer and/or places less value on the market consumption that earnings put within her reach. Work aversion may suggest that the beachcomber would suffer extra disutility from the income effect of having to work at least enough to pay the tax. And this effect may get a lot worse with the realistic assumption that many labor market decisions are "lumpy;" for example, it is hard to earn $ 400 per hour on Wall Street without years of training followed by a full-time commitment. By contrast, the beachcomber's lesser interest in market consumption might suggest that she would suffer less rather than more disutility than the lawyer from paying the tax. Taking her work decision as given, every dollar of tax paid deprives her of market consumption opportunities that we already have inferred from her behavior are generally of lesser interest or value to her.

In practice, it is safe to surmise that few if any welfarists will consider overall social welfare likely to be increased by dragging off talented beachcombers in chains to labor in the corporate vineyards of recapitalizations and public stock offerings. Note, however, that weighted welfarists initially may place less weight than utilitarians on the disutility to high endowment beachcombers of having to work a bit to pay the tax. This presumably shifts to placing more weight on the beachcomber's disutility once she has become worse off than other people due to the compelled work effort.

Labor market lumpiness may significantly extend welfarist reluctance to follow through on the endowment standard. Suppose, for example, that law professors who deliberately have passed up higher paying jobs in law practice would be greatly inconvenienced by having to pay the same tax as their practitioner peers. That is, they would have to give up some of the more limited consumption opportunities that they did value a lot or else give up on serious academic effort. This problem may reflect a lumpy labor market, in which they find it hard to generate high billings with just a few hours of scattered effort here and there. Lumpiness suggests the possibility that, as between two individuals with the same endowment, the value of an extra dollar may be greater to the one who has chosen lower earnings, because she cannot finetune her own choices to equalize the marginal utility of her last earned dollar versus her last unit of leisure.

 [*415]  Now consider a politically more prominent version of the beachcomber and law professor problems: the tax treatment of secondary earners. For the wives of high earning husbands, it is easy to see that taxing their market work but not their home production will tend both to discourage labor market participation and to disfavor two-earner households relative to one-earner households with similar opportunities. One possible way to mitigate these undesirable effects of the current system is to lower the tax burden on a secondary earner's market work (for example, by permitting income splitting by couples, which may lower the marginal tax rate on extra earnings if the rate structure is progressive). n43 An opposite response would be to tax the home production of nonworking spouses, perhaps via an imputed measure, as Nancy Staudt has proposed. n44

Even as adjusted to raise the same overall revenue, the two approaches presumably would result in different tax burden allocations between households. Obviously, this is not the place to sketch out a detailed welfarist analysis of how to choose between them. I note, however, that problems of this sort are considered in contemporary debate about tax policy even when income imputation to stay-at-home spouses is not among the options on the table. Debate about the proper relative treatment of different types of households, such as single individuals versus married couples, often turns on how to compare need or well-being in the two settings. The fact that a couple has two sets of hands rather than one routinely is taken into account, no less than the fact that it has two mouths to feed rather than one. n45 "Simple libertarian principles" n46 apparently do not get in the way.

The well-recognized consequences of imposing a tax burden on a one-earner couple include the possibility of an income effect inducing the previously stay-at-home spouse to go to work. Sometimes, especially among those nostalgic for the traditional family, this is cause for complaint. n47 For others who are hostile to this nostalgia and its possibly sexist underpinnings, this complaint is itself cause for complaint. n48 Both sides appear to agree, however, that this possible consequence of  [*416]  the income effect does not trump all other considerations (as the beachcomber example might suggest), but rather belongs in the same overall stew of relevant considerations as any other consequence of a given household taxation scheme. Some commentators may even like the income effect on labor supply by married women, on the ground that their increased labor market participation has positive external effects on other women or both sexes.

In sum, from a welfarist perspective, the normative significance of endowment is complicated. There is no clear implication that one would want to impose a straightforward endowment tax even if measuring endowment were possible. This reflects that the normative significance of inequality is complicated from a welfarist perspective. Despite the implication from declining marginal utility (and weighted welfarism to its subscribers) that one generally should favor transfers from high-endowment to low-endowment individuals, the measure ceases to provide clearcut guidance to distribution policy once one allows for taste differences and lumpy labor markets.

The fact that endowment does not have more decisive implications for distribution policy raises the possibility that one might be willing, in practice, to stick with a proxy measure such as wealth, income, or consumption even if endowment were directly observable. Yet, the proxy measures still would be functioning as stand-ins, rather than being used because they mattered for their own sake. Their ultimate justification still would have to lie in terms of their relationship to endowment plus the other relevant factors (such as work aversion, materialism, and the effect of lumpy work decisions on the marginal utility of a dollar). And as a matter of analytical clarity, endowment remains closer to bedrock than the proxy standards as a tool for measuring inequality, as distinct from deciding what to do about it.

C. Taxing Endowment From a Liberal EgalitarianPerspective
 
Turning to liberal egalitarianism, a couple of methodological problems arise. One is that this approach is generally more open-ended than welfarism, or at least open-ended in a different way. Welfarism purports to specify exactly what matters - utility - along with a mathematical relationship between individual utilities and social welfare. (To be sure, the specification is not just empirically but analytically problematic, since "utility" is unobservable and may not literally exist.) Liberal egalitarian frameworks tend to be less definite in specifying what matters and in describing how in principle one should  [*417]  choose between conflicting aims that have the same rank. n49 In addition, while welfarists may differ about such matters as interpersonal weighting and even how to think about utility, n50 one might say of liberal egalitarians (even accepting the existence of such a group) that each has a distinct theory, and thus that generalization about their views is dangerous.

Nonetheless, some leading names and roughly shared principles come to mind. John Rawls, Ronald Dworkin, and Bruce Ackerman have been named as liberal egalitarians, n51 and one could mention as well the likes of Thomas Nagel, Amartya Sen, Thomas Scanlon, and Rakowski himself. The core ideas of these thinkers that are most pertinent here are that rights matter independently of consequences, and that (as Rakowski puts it) distributive justice is concerned, not with the welfare content of outcomes, but with people's "non-welfare-based entitlements to resources, opportunities, or capabilities in consequence of their choices and of their natural and social fortune." n52 This claim rests on what Liam Murphy calls the "choices thesis," under which "the distinction between wealth or income people have acquired through their own 'effort or voluntary choices' and wealth or income obtained 'purely by chance' is of foundational importance to distributive justice." n53

The choices thesis maps nicely onto the distinction noted by welfarists between endowment on the one hand and effort or commodity choice on the other as the determinants of earnings. n54 Once again, there is an argument for redistribution from high endowment to low endowment individuals, but it does not extend to earnings differences that result from differences in effort or commodity choice. To be sure, the two schools' arguments stand on very different grounds,  [*418]  and one thus could specify circumstances under which their implications would sharply diverge. For example, showing that a given high endowment individual attaches the same marginal utility to a dollar as a low endowment individual would eliminate the utilitarian's, but not the liberal egalitarian's main reason for supporting progressive redistribution. And the utilitarian would be concerned about the empirical consequences of discouraging effort through the earnings tax (leading to deadweight loss), rather than about depriving effort of an inherently fair reward. Still, under circumstances of limited information, where one assumes but cannot measure declining marginal utility and predicts that reducing the reward to effort will discourage it but is uncertain by how much, the differences in underlying concern may lack clear implications.

What about differences in taste? A welfarist who as an empirical matter generally accepts consumer sovereignty will tend to think that differences in commodity choice provide little if any information about relative endowments or the utility consequences of raising or lowering a given individual's budget line. Differences in taste, therefore, have no immediate pertinence to distributional arguments, although with added information they can become pertinent, as suggested by my discussion above of differences in materialism and work aversion. n55 As for liberal egalitarians, they tend to "place[ ] a high value on letting people pursue their own visions of the good," n56 albeit only within the limits of what seems an acceptable "rational plan" n57 as determined through some "objective" specification of proper goods rather than "exclusively in terms of preference satisfaction." n58 This may suggest not redistributing on the basis of differences in taste - to much the same end as the welfarist view if information is limited in the right way.

Also worth mentioning is a discussion in the philosophical literature of "expensive tastes," or those that require a greater expenditure of scarce resources to satisfy. Rakowski elsewhere has criticized some welfarist views for suggesting that "the allotment of someone who cultivated expensive tastes - for flashy cars, posh restaurants, designer clothes - be increased" at the expense of those with cheaper tastes. n59  [*419]  He regards this as unfair, and argues under the choices thesis that deliberately cultivated expensive tastes should not be rewarded.

Once again, however, there is an overlap with welfarist views if information is sufficiently limited. For example - illustrating anew the overlap between considering choice morally important and caring about it for incentive reasons - welfarists naturally would be concerned about the consequences of rewarding expensive tastes. Externalization of the cost of developing expensive tastes creates a divergence between the individual and the optimal social incentive regarding such development. In addition, the fact that one individual has more expensive tastes than another tells too little about their utility functions to show which of them would get more benefit from a given quantum of cash or change of budget line. n60

How does all this come out with regard to designing the tax-transfer system? Of greatest practical import, the broad outlines of a liberal egalitarian's view of wealth, income, or consumption taxation should resemble those of a welfarist. Under either view, the tax bases involve a trade-off between accomplishing good redistribution based on endowment and bad redistribution based on effort or commodity choice. The choice of tax base, tax rates, and other design features must turn on complicated trade-offs between competing considerations, and undoubtedly depends in part on elusive empirical information. Rawls called these "questions of political judgment and not part of a theory of justice," n61 presumably meaning that they could not be resolved just by abstract reflection, rather than that his theory could not help to guide their resolution.

Given this conclusion, Rakowski's gibe that considering whether a wealth tax would be "part of the optimal tax scheme under one or another consequentialist theory ... would be like fencing with the fog" n62 comes home to roost under a liberal egalitarian view as well. Neither that view nor perhaps any that treats inequality as relevant to distributive justice can hope to resolve at a broad level of abstraction the desirability of a particular tax instrument (the wealth tax) that  [*420]  modifies the existing tax burden on work plus saving to fall somewhat more on those who save.

If endowment were measurable, one could imagine liberal egalitarians being more committed than welfarists to basing tax burdens on it. After all, some version of it appears to be the very thing they want to redistribute under the choices thesis, and they may be inclined to disregard the utility information that may modify its use under welfarism. Thus, to them, unlike to welfarists, endowment is not merely a convenient tool for getting at something else. To be sure, liberal egalitarians may generally share Rakowski's concern about the hypothetical beachcomber who is dragged off in golden chains to work on Wall Street. Liberal egalitarian discussions of this problem, however, generally repeat Kelman's unexplained discontinuity between outright compulsion (like forcing someone to work to pay the tax) and mere incentive or inducement (like penalizing the decision to work). Although compulsion is merely a strong case of inducement, suggesting that they differ only in degree, not in kind, the inclination is apparently strong to put them in separate categories without sufficient explanation or defense. n63

IV. Conclusion
 
"Can wealth taxes be justified?," Rakowski asks in the title of his article. He shows that, by a sufficiently demanding standard, they cannot. By the standard that he sets, however, income or consumption taxation, along with any other practical real world tax that is designed to take account of inequality, also cannot be justified.

By a more realistic standard, a wealth tax plainly can or might or could be justified, whether or not it actually is justified. From a variety of normative perspectives, however, a lot of empirical input is likely to be necessary. Among the most important empirical issues are how the adoption of a distinct wealth tax is likely to affect the tax system's political equilibrium and the overall costs of tax planning, administration, and compliance.

In any event, the answer depends on what the overall fiscal system, rather than just one piece of it, would look like with and without a separate instrument called a federal wealth tax. The current tax, even without such an instrument, burdens the very decisions to work and  [*421]  save that it would target, on grounds that I have argued are best rationalized as responding to inequalities that can better be conceptualized in terms of endowment. A more meaningful way to pose the federal wealth tax question would be to ask whether, in one way or another, we should increase the tax burden on saving, while keeping the tax burden on work approximately constant (assuming no change in the overall tax levels that would have applied over time) and perhaps affecting the overall level of redistribution from high earners to low earners (depending on how the enactment affected the tax system's political equilibrium). This is an important grab-bag of issues, but ill-suited for resolution at a purely abstract level.



FOOTNOTES:
n1. Eric Rakowski, Can Wealth Taxes Be Justified?, 53 Tax L. Rev. 263, 285 (2000) (emphasis omitted) [hereinafter Wealth Taxes].

n2. Id. at 279.

n3. Id. at 279 (emphasis omitted), 285.

n4. Rakowski accepts the possibility that wealth might serve as a placeholder for some other underlying "imbalances in people's fortunes" that justice abhors, but argues that, if this is so, "those inequalities themselves should serve as triggers for redistributive taxation, unless some principle or practical impediment blocks reliance on them." Id. at 285.

n5. Id.

n6. Id.

n7. For a sophisticated argument that work, saving, and market consumption actually do cause external harm that is akin to pollution, see Robert H. Frank, Choosing the Right Pond: Human Behavior and the Quest for Status 230-32 (1985) (arguing that, because of people's concern about relative status, greater consumption by one individual results in harmful "positional externalities," whereby others must now work and consume more in order to achieve the same level of satisfaction).

n8. Rakowski, Wealth Taxes, note 1, at 347.

n9. Id. at 286 (footnote omitted).

n10. I refer to the old story of the woman who claimed that the earth rests on the back of a turtle and, when asked what the turtle rests on, responded that it was "turtles all the way down." On the origin of this story, see Roger C. Cramton, Demystifying Legal Scholarship, 75 Geo. L.J. 1, 2 n.4 (1986).

n11. See Henry C. Simons, Personal Income Taxation 50 (1938) (defining personal income as "the algebraic sum of (1) the market value of rights exercised in consumption and (2) the change in the value of the store of property rights between the beginning and the end of the period in question").

n12. Id. at 31.

n13. Recent discussions of liberalism or liberal egalitarianism by legal scholars writing on tax policy include not only Rakowski, note 1, but also Bruce Ackerman & Anne Alstott, The Stakeholder Society (1999); Anne L. Alstott, The Uneasy Liberal Case Against Income and Wealth Transfer Taxation: A Response to Professor McCaffery, 51 Tax L. Rev. 363 (1996); Edward J. McCaffery, The Political Liberal Case Against the Estate Tax, 23 Phil. & Pub. Aff. 281 (1994); Edward J. McCaffery, The Uneasy Case for Wealth Transfer Taxation, 104 Yale L.J. 283 (1994) [hereinafter Uneasy Case]; Liam B. Murphy, Liberty, Equality, Well-being: Rakowski on Wealth Transfer Taxation, 51 Tax L. Rev. 473 (1996); Eric Rakowski, Transferring Wealth Liberally, 51 Tax L. Rev. 419 (1996) [hereinafter Transferring Wealth].

n14. See, e.g., Robert H. Frank, Luxury Fever: Why Money Fails to Satisfy in an Era of Excess 71-74 (1999) (discussing sociological evidence that increasing average income levels in recent decades have not brought significant and lasting increases in subjective well-being).

n15. Rakowski, Wealth Taxes, note 1, at 280-83.

n16. Id. at 283.

n17. Id. (footnote omitted).

n18. Eric Rakowski, Equal Justice 1 (1991) (footnote omitted).

n19. See, e.g., Ackerman & Alstott, note 13, at 24 (advocating equality of "opportunities, not outcomes"); John Rawls, A Theory of Justice 73 (1971) (discussing equality of opportunity); Richard J. Arneson, Equality and Equal Opportunity for Welfare, 56 Phil. Stud. 77 (1989) (arguing for equality in people's opportunities to achieve welfare); Ronald Dworkin, What Is Equality? Part 2: Equality of Resources, 10 Phil. & Pub. Aff. 283 (1981) (arguing for equality in the resources, including ability, available to people).

n20. See, e.g., Peter Diamond, Administrative Costs and Equilibrium Charges With Individual Accounts 18 (NBER Working Paper No. 7050, 1999).

n21. The assumption that one cannot assess the comparative well-being of two people who make different selections from among the same choices may seem unduly stringent if we think of people we know personally. The information that we glean from talking with or otherwise observing them may encourage us to think that one is better-off than the other (due either to choosing better or to a better fit between available choices and her preferences). From the standpoint of a tax administrator in a mass society, however, the assumption is quite reasonable. Is there any general reason to think, say, that people who choose to work more are generally better-off than those who have the same choices but, presumably due to differing preferences, choose to work less?

n22. The standard defense of this assumption that different people should be assumed to experience utility with equal intensity comes from Abba P. Lerner, The Economics of Control 29-32 (1944). Lerner argued that, if we are entirely ignorant of whose utility is more intense but believe that people generally experience declining marginal utility, making any differential intensity assumption (with a 50% chance of being wrong) reduces expected utility.

n23. Rakowski, Equal Justice, note 18, at 1.

n24. See Daniel Goleman, Emotional Intelligence (1995).

n25. To simplify the discussion, I ignore the inheritance of property, which increases one's consumption opportunities but generally without regard to one's own work effort, and of course is a poor fit with the term "ability."

n26. See Daniel N. Shaviro, Making Sense of Social Security Reform (forthcoming 2000).

n27. Rakowski, Wealth Taxes, note 1, at 267 n.10.

n28. Under a welfarist view, this statement would need modification insofar as one could evaluate the welfare consequences of having tastes (such as for engaging in a particular type of artistic production) that have differing payoffs in the state of the world in which an individual finds herself. A liberal egalitarian view also might suggest caring about taste differences for which individuals were not themselves responsible.

n29. Rakowski, Wealth Taxes, note 1, at 353.

n30. Id. at 267 n.10.

n31. Mark G. Kelman, Personal Deductions Revisited: Why They Fit Poorly in an "Ideal" Income Tax and Why They Fit Worse in a Far From Ideal World, 31 Stan. L. Rev. 831, 842 (1979) (footnote omitted).

n32. See, e.g., Robert Nozick, Anarchy, State, and Utopia 167-74 (1974) (arguing that progressive redistribution violates appropriate respect for property rights).

n33. See Kelman, note 31, at 880.

n34. Id.

n35. At one time, however, many thought of realization as a necessary component of economic income. See Eisner v. Macomber, 252 U.S. 189 (1920); Edwin R.A. Seligman, Are Stock Dividends Income?, 9 Am. Econ. Rev. 517 (1919). For a discussion of the withering attack on Seligman's exclusion of unrealized gain from economic income, see Simons, note 11, at 85-89.

n36. A stay-at-home spouse need not be engaged in housework, childcare, and the like to be a valuable producer. Leisure in the colloquial sense (such as reading books or watching television) is no less a way of producing utility than performing chores that someone in the household values.

n37. For simplicity, I ignore here the imposition of tax on the household rather than individual level, along with the difference between employment prospects at a given point in time and underlying endowment.

n38. See, e.g., J.A. Mirrlees, An Exploration in the Theory of Optimum Income Taxation, 38 Rev. Econ. Stud. 175 (1971).

n39. See, e.g., Amartya Sen, Inequality Reexamined 43 n.14 (1992) (defining welfarism as the view that "states of affairs must be judged by the individual utilities in that state").

n40. Risk after drawing a wage rate commonly is left out of the models as well. For a model in which people face income risk after they have made work decisions, see Hal R. Varian, Redistributive Taxation as Social Insurance, 14 J. Pub. Econ. 49 (1980).

n41. In addition, the welfare effects of observing that others are better off or worse off than oneself may be relevant. For example, in a straight utilitarian account, envy of the better off might provide an added argument for progressive redistribution, as might the inclination to pull rank on those worse off if the pain this inflicted typically was greater than the pleasure it brought.

n42. An inferior good commonly is defined as one for which "demand decreases as income increases." Harvey S. Rosen, Public Finance 553 (5[su'th'] ed. 1998). A standard example might be hamburger, if people start replacing it in their consumption baskets with steak as they become more affluent. By extension, one could call income an inferior good relative to endowment if an increase in endowment reduces demand for it.

n43. Other possibilities include ignoring marital or household status for tax purposes (thus requiring separate filing with each spouse including his or her own earnings), allowing a secondary earner deduction, widening the tax brackets for married as compared to single taxpayers, and allowing a deduction for costs (such as childcare expenses) incurred when both spouses work.

n44. Nancy C. Staudt, Taxing Housework, 84 Geo. L.J. 1571, 1624-27 (1996).

n45. See, e.g., Joseph A. Pechman, Federal Tax Policy 107 (5[su'th'] ed. 1987).

n46. Kelman, note 31, at 842.

n47. See, e.g., Christian Coalition, Contract With the American Family 54 (1995) (blaming tax increases since the 1970's for forcing married women to work) (cited in Edward J. McCaffery, Taxing Women 207-09 (1997) [hereinafter Taxing Women]).

n48. See, e.g., McCaffery, Taxing Women, note 47, at 207-14.

n49. Priority rules in some cases can make a liberal egalitarian view more definite in its verdicts than welfarism. For example, suppose that a given liberal egalitarian categorically rules out torturing an innocent individual to benefit millions of others, thus obviating the need for a determination of the overall welfare effects. See Thomas Nagel, Equality and Partiality 148 (1991).

n50. Consider hedonistic versus preference-satisfaction versions of utilitarianism, or John Stuart Mill's controversial suggested distinction between "higher" and "lower" pleasures. See Amartya Sen & Bernard Williams, Introduction to Utilitarianism and Beyond 11 (Amartya Sen & Bernard Williams eds., 1982); J.J.C. Smart, An Outline of a System of Utilitarian Ethics, in Utilitarianism: For and Against 12-13 (J.J.C. Smart & Bernard Williams eds., 1973).

n51. See McCaffery, Uneasy Case, note 13, at 290.

n52. Rakowski, Wealth Taxes, note 1, at 283 (footnote omitted).

n53. Murphy, note 13, at 474. As a terminological matter, Murphy warns against permitting the terms "liberal" and "egalitarian" to be appropriated by subscribers to the choices thesis, noting that many people who consider themselves liberal and egalitarian in sentiment find the thesis dubious. Id. at 473-75.

n54. To simplify the discussion, I again ignore risk or variance in the determination of earnings apart from that in the endowment lottery.

n55. See Section III.B.

n56. Alstott, note 13, at 379.

n57. Rawls, note 19, at 395.

n58. Rakowski, Transferring Wealth, note 13, at 434 (citing Derek Parfit on the "objective" list of goods).

n59. Rakowski, Equal Justice, note 18, at 41. Rakowski specifically criticizes "egalitarian welfarism," or the view that everyone's welfare should be equalized, as distinct from the aim of maximizing total welfare by some weighted or unweighted metric. To a utilitarian or weighted welfarist, expensive tastes likewise could have the redistributive import that Rakowski criticizes, although this depends on how they affect relevant marginal utilities.

n60. Suppose that, as between two individuals, neither is in general a more intense consumer or utility monster, but one develops more expensive tastes than the other - say, a taste for expensive wine rather than cheap beer. If consumption opportunities are lumpy - you can get a bottle of beer for $ 5 but no bottle of good wine for less than $ 20 (and no good wine by the glass) - one might imagine that going from $ 0 to $ 5 would benefit the beer drinker more, while going from $ 15 to $ 20 would benefit the wine drinker more. More information is needed, however, to say who benefits more by going from zero to $ 20, and for an otherwise unspecified $ 5 shift, it may be impossible to tell.

n61. Rawls, note 19, at 279. Less well supported or coherently defended was an earlier throwaway statement to the effect that "a proportional [consumption] tax may be part of the best tax scheme." Id. at 278.

n62. Rakowski, Wealth Taxes, note 1, at 283.

n63. In calling inducement merely a lesser case of compulsion, I should not be taken to assume that neutrality (such as leaving workers with precisely their market return) has special moral standing. As Murphy has noted, such a view would require "implicit moralization of the (suitably idealized) market." Murphy, note 13, at 492. Just as compulsion shades into mere inducement, so greater inducement (relative to what the market happens to offer) shades into lesser inducement and then into the opposite inducement.