Copyright (c) 1999 Chapman Law Review
Chapman Law Review
2 Chap. L. Rev. 157
SYMPOSIUM: FEDERAL TAX POLICY IN THE NEW MILLENNIUM: The
Case for Proportionate Taxation
* Professor of Law and Deane Johnson Faculty Scholar, Stanford University. I am
grateful to Joe Bankman, David Bradford, Marvin Chirelstein, Noel Cunningham,
Richard Epstein, Tom Grey, Mark Kelman, Lewis Kornhauser, Liam Murphy, Tom
Nagel, Dan Shaviro, and Deborah Schenk for helpful comments on an earlier draft.
... I am also not arguing in favor of progressivity, regressivity, or any
other rate structure on fairness grounds. ... A true public good is one that
exhibits two characteristics. ... It has been suggested that a degressive tax,
or even a tax rate structure that declines slightly at the top, may well be
optimal if combined with substan tial lump-sum transfers (so-calleddemogrants).
... The case for any tax rate structure that is tied to the level of income,
be it pro gressive, proportionate or even regressive, thus depends on show ing
that income levels are morally relevant in setting differential tax burdens.
... We have lived with a progressive tax rate structure for so long that we
are all accustomed to think regressivity unthinkable. ... If it did here - if
the political landscape changed enough to make a regressive tax rate structure,
or at the extreme even a head tax, a politically plausible alternative - then
there is every reason to expect that the odd convergence on a flat tax as
"the fairest tax of all" would dissipate. ...
Over the past half century, the view that the only fair tax is a flat-rate tax
has attracted support from a surprising range of political philosophers and
pundits, including Friedrich Hayek,
n1 Ludwig von Mises,
n2 Milton Friedman,
n3 John Rawls,
n4 and Richard Epstein,
n5 along with Walter Blum and Harry Kalven in their fa mous 1952 essay.
n6 Notwithstanding trenchant critiques, the view not only has persisted, but
(over the past decade) has gained enor mous popularity in political and
academic circles. In their 1995 book, The Flat Tax, Robert Hall and Alvin
Rabushka defended their version of a flat tax with the Grimmsian pronouncement
that a flat-rate tax is
"the fairest tax of all."
n7 In the same year, Richard Epstein declared,
"from the Lockean perspective, a strong case can be made that ... the flat tax
is the only acceptable [tax]."
n8 Similar pronouncements can be found throughout the current political debates,
as well as more serious philosophical considera tions of the just distribution
of tax burdens. There is some non trivial chance that, over the next few years,
that sentiment could carry the day in Congress.
It is, of course, impossible to evaluate these extravagant claims on behalf of
proportionality except in reference to a theory of fairness. It goes without
saying that these odd bedfellows can
[*158] not all have the same theory in mind. What follows is an attempt to sketch out
plausible intuitions of fairness on which these claims might rest, and ask
whether any of them supports it. I conclude not.
Let me be clear what I am not arguing. I am not arguing that some version of a
flat-rate tax scheme cannot be defended as a sensible policy solution on any
other grounds, including adminis trative convenience, political compromise
between warring consid erations, or our best guess as to how to achieve a
distributive end resting on firmer philosophical grounds. Perhaps it can be so
de fended. I mean only to try to dislodge the apparently intractable notion
that it deserves to be adopted because it is
"fair" in itself, or because it is an obvious instantiation of some other fairness
prin ciple. I am also not arguing in favor of progressivity, regressivity, or
any other rate structure on fairness grounds. The deeper moral is that no
sensible theory of distributive justice would fix on rate structures themselves
as fair or unfair. Rate structures are just a means to operationalize other
prior, moral commitments about the proper role of government. The case for any
particular rate structure must stand or fall on how well it realizes those
prior commitments. An examination of plausible, prior moral commit ments,
however, suggests that, of all possible rate structures, pro portionality may
be the hardest to derive from any coherent theory of fairness.
In the first part of the paper, I consider the case for propor tionate taxation
in the context of two broad views of governmental power in fiscal matters. The
first, an essentially libertarian view of the state, takes the proper role of
government to be limited to solving the collective action problems that prevent
private actors from spontaneously reaching optimal outcomes through voluntary
agreements. While tolerating whatever minimal redistribution might occur as a
consequence of solving collective action problems, the libertarian view
generally rules out income redistribution as a motive for state action. One
recent proponent of that view has located that prohibition in the Constitution,
through an expansive interpretation of the Takings Clause.
n9 Most have simply deduced it from natural law principles.
n10 In tax theory, the clear corollary to libertarianism is the so-called
"benefits" theory of taxation. I start with libertarianism, because it is the premise
from which most of those who advocate flat-rate taxes on fairness grounds seem
to be starting.
The second, a social welfarist view of the state, accepts that income
redistribution for the express purpose of improving the
[*159] welfare of the less well-off in society is a legitimate part of the state's
job, although there is obviously much disagreement about the optimal extent and
form of redistribution. While there are no clear corollaries to the social
welfarist view of the state in tax the ory - something that I think is not
"ability to pay" and
"minimum sacrifice" arguments undoubtedly reflect that view in inchoate form.
Other views of the legitimate role of the state in fiscal affairs are, of
course, possible, but most that are offered in tax debates, when pushed hard
enough, end up being some version of one of these two views.
Both of the foregoing views evaluate taxation in the context of a comprehensive
system of government (re)distribution. In the last part of the paper, I take up
a number of defenses of a flat-rate tax viewed in isolation from the uses to
which tax revenues are put. These include the widespread view that a flat-rate
tax vindi cates some important notion of equality; that it leaves individu als'
choices among various activities undistorted; and that it limits the (unfair)
expropriation of the wealthiest classes by the majority. Finally, I consider
two other positive explanations for the popular convergence on proportionate
taxation. The first is that its apparent properties of mathematical simplicity
and cer tainty have made it a Schelling-like focal point solution to the
problem of appropriate tax rates. The second is that its popular ity,
particularly among libertarians, is a product of political fram ing - of the
fact that, at least in recent history, no one would have considered a
regressive rate structure for our broad-based income tax to be a politically
viable alternative. Were that fact to change, it seems quite likely that many
who support proportionate taxa tion on libertarian grounds would find
regressivity, on reconsider ation, a more plausible expression of libertarian
II. Libertarian (Benefits) Theory
In contemporary debate, the most vociferous advocates of a flat-rate tax -
n13 and Gauthier,
n14 to name some prominent examples - have been animated by liberta-
[*160] rian impulses.
n15 The same is true of earlier defenders.
n16 In such a view, the just limits of state power derive from an essentially
Hobbesian and Lockean social contractarian view of the state, in which the
state exists solely to provide services that, for a variety of reasons of
market failure, cannot be provided optimally by pri vate, voluntary agreement.
As has been noted, that view implies a quid pro quo relationship between the
taxpayer and the state, in which taxes function as the shadow price for goods
or services that the state provides.
n17 Hayek provided the classic argument for de riving proportionate rates from a
benefits theory of taxation:
"a person who commands more of the resources of society will also gain
proportionately more from what the government has contrib uted" to the provision of those resources, and taxation ought to be levied in
proportion to the benefits so provided.
The first big problem in assessing the case for proportionate taxation under a
social contractarian view of the state is that al most none of the proponents
of proportionality on (vaguely) liber tarian grounds, from Adam Smith on down,
have in fact supported proportionate taxation. A true flat-rate tax would tax
all income (or consumption, as the case may be) starting with the first dollar,
[*161] at the same rate. Instead, they have supported a so-called degres sive version
of a progressive tax, in which the first x dollars of income or consumption,
sufficient to cover basic needs, is taxed at a zero rate, and all income or
consumption above that is taxed at the same positive rate.
The political advantages of a degressive tax over a proportion ate tax are
obvious, and hardly lost on its supporters.
n20 There is little public support for a true proportionate tax that contains no
exemption for basic income. It is hard to overstate, however, the difficulties
that that concession entails for those whose opposition to any greater degree
of progressivity via a graduated rate struc ture is based on the fact that such
progressivity is motivated by purely redistributive concerns. If all of us
earning income above the exemption level have an obligation to pick up the tab
for gov ernment services provided to the poor, simply because they are poor and
we are not, why stop there? Why not tax the rich for government services at an
even higher rate than the middle class, simply because the rich have more money
than everyone else and hence can better afford to defray the cost? Or, for that
matter, why not raise the exemption level to, say, $ 30,000 for a family of
four, raise the flat rate tax from, say, 19% to 25%, and use the excess
proceeds to finance a guaranteed minimum income to all?
n21 At that point,
"proportionate" taxation would start looking mighty attractive to the sorts of welfare state
liberals who would reject the libertarian premises of benefits taxation out of
hand. Surely Frank Taussig was right in declaring many years ago that
"the demand for the exemption of the lowest tier of incomes results from the
same state of mind as the advocacy of progressive taxa
[*162] tion [through graduated rates]."
n22 If there is some principled basis for absorbing redistributive motives into
benefits theory to the ex tent of basic needs, but no further, libertarians
have yet to articu late it.
n23 Until such time as they do, those few hearty libertarians who have rejected an
exemption for basic income as inconsistent with conventional libertarianism
seem to have the better of their libertarian fellow travellers, and are to be
commended, from the point of view of principle, at least, for sticking to their
libertarian guns here.
For the balance of the discussion, I set that difficulty aside, to turn to the
central question. If, as benefits theorists argue, taxes ought to function as
the shadow price for goods or services that the state provides, how do we set
that price? The logical place to begin is with a determination of what a
well-functioning private market would have charged for those goods or services.
As Musgrave stated many years ago,
"since the relation is one of exchange, the rules of the public household are
taken to be more or less the same as those of the market."
n25 What then are the rules of the (private) market - or more precisely, which of
the rules of the private mar ket do we think ought to be transported to the
public household, and once transported, what do those rules imply about
appropri ate tax rates?
The uncontroversial starting point for analyzing the fairness of the market is
the requirement of strict paretianism: the re
[*163] quirement that every trade leave no side worse off, and at least one side
better off. This requirement is assumed to be met auto matically by any
voluntary market transaction. Strict paretian ism, however, is a minimally
exacting standard, which gives little guidance in resolving the central
problems in assigning the tax costs of providing public benefits.
First, the pareto principle provides no guidance on the level of aggregation at
"betteroffness" from being in society is to be judged. If the level of aggregation is great
enough - at the ex treme, for example, if we test the pareto superiority of the
bene fits/taxation deal each American is offered at the level of the decision
whether to exit this society or not - then there is almost no redistributive
mischief that cannot be justified. At the other extreme, if we insist that each
government program satisfy strict paretianism, then almost no program can be
justified, since there will always be individuals under any feasible tax scheme
who are paying more for a particular public good, such as a road through the
middle of Kansas, than the benefit they derive from it.
In practice, libertarian benefits theorists seem to employ a mixed criterion,
along the following lines. First, strict paretian ism is used to judge the
aggregate deal each of us has gotten from being in society rather than out of
n26 This standard will almost always be met if we are comparing any plausible
configuration of tax burdens in America circa 1999 to our likely individual
posi tions in some hypothetical Crusoeian state of nature, and, indeed, is met
automatically as long as there is a meaningful opportunity for exit from
society. Second, a potential pareto (that is, Kaldor- Hicks) test is used to
judge expenditures at the maximally disag gregated level. Thus, each particular
road through Kansas must be justified by showing that the social gains from
building it out weigh the social costs entailed. However, one need not show
that each citizen derives benefits from that particular road at least equal to
her share of the social costs of building it.
n27 Finally, strict
[*164] paretianism is used to test the tax/benefits deal over some repre sentative
class of incremental public goods being considered. That is to say, in a just
state, over a representative period, each individ ual in the state will derive
benefits from government services that are at least equal to the cost she bears
for them. This formulation is just a guess, however, since most proponents of a
benefits the ory of taxation are unclear or inconsistent on the point.
Second, the pareto principle requires only that tax prices be set so that no
one is made worse off by the compulsory purchase of public goods. It says
nothing about the proper distribution of sur plus above that - that is, about
distribution of the gains from trade.
In private competitive markets, both of these problems - the level of
aggregation on which to test pareto improvements and the just division of gains
from trade - are automatically resolved through equilibrium prices. When
dealing with public goods, they are not. Since in the latter case we are
dealing with hypothetical and not actual markets, and moreover hypothetical
markets in which the government operates as a monopoly supplier who can compel
payment at dictated prices, we have enormous theoretical latitude in
constructing the hypothetical exchange between gov ernment as supplier and
citizens as consumers. In particular, we have enormous latitude in resolving
the two questions above: the level of aggregation at which to bundle goods or
services in judg ing the pareto fairness of our tax scheme, and the just
division of surplus value. Depending upon the resolution of both these ques
tions, there is an enormous range of acceptable distributions of tax burdens to
finance public goods, within the constraints of paretianism.
At one extreme, imagine the state as monopolist with perfect information about
people's reservation prices for public goods, and hence the ability to
perfectly price discriminate in setting tax prices for those goods. Consistent
with the pareto principle, the state could tax each person in society at a rate
that appropriates 100% of the surplus value of public goods. On a broad view of
pub lic goods or benefits that should be counted in setting tax rates,
including not merely traditional public goods or services, but all the benefits
of being in our organized society rather than on a Crusoeian island or in
war-torn Rwanda, this broad power to ex propriate surplus value would authorize
something close to a con fiscatory tax on almost everyone.
[*165] At the other extreme, suppose that the benefits that can be counted are
limited to identifiable costly goods or services pro vided by government, and
suppose further that the government must act as a supplier in a private
competitive market would act; that is, it is compelled to charge a uniform per
unit price for all goods consumed. The most likely outcome of benefits theory
on those suppositions is a highly regressive tax.
For the balance of the discussion, I will focus on the second question: how
prices for public goods should be set, within the con straints of strict
paretianism, and hence how the surplus value generated by public goods should
be divided. As the discussion that follows makes clear, neither the libertarian
literature ap proaching the benefits question from the perspective of distribu
tive justice, nor the public economics literature approaching it from the
perspective of allocative efficiency, gives a determinate answer to that
A. The Lessons of the Private Market
There is a fatal ambiguity running through the benefits the ory literature
about how benefits ought to be measured. This am biguity is reflected in
Hayek's statement that
"a person who commands more of the resources of society will also gain propor
tionately more from what the government has contributed" to the provision of those resources.
n28 Does Hayek mean by
"gain ... more" that the rich consume a proportionately greater quantity of public goods, or
that the rich derive proportionately greater utility from public goods?
Taking the private competitive market as the benchmark, one is pushed to the
former interpretation. In a purely competitive market, all consumers face the
same per unit price, set by the equilibrium price for the good. Suppliers
cannot price-discrimi nate based on consumers' different reservation prices for
the goods, because any effort to set a higher price for any consumer will be
undercut by another supplier. As a consequence, the rela tive prices that
consumers pay for a given good depend solely on the relative quantities of the
good they consume. Whatever con sumer or producer surplus is generated by
trades at that price lies where it happens to fall. Assuming a conventional
downward- sloping demand curve and an upward-sloping supply curve, equi librium
prices, reflecting the marginal utility or cost of the last unit consumed, will
leave each consumer or producer with surplus from all but the last unit
consumed or sold. The precise amount of the surplus depends on the fortuitous
shape of individual and ag gregate demand and supply curves.
[*166] Under this interpretation, benefits theory leads to proportion ate taxation if
and only if the quantity of public goods that people consume is proportionate
to income. As even proponents of a pro portionate tax concede, this premise is
n29 The premise is doubtful for many public goods, such as roads, fire pro
tection, and garbage collection. It is clearly wrong for others, in cluding
true public goods (for example, clean air, defense, and broadcast spectra), one
of the defining characteristics of which is that they are jointly consumed,
with the result that everyone con sumes the same quantity (or has the same
quantity available for consumption). In the end, such a measure of benefits is
much more likely to lead to a highly regressive tax.
If, on the other hand, what Hayek and others mean by
"gains more" is that people ought to be taxed in proportion to the utility that they derive
from public goods, the implications for appropri ate tax rates are much less
clear. Indeed, the question is probably completely unresolvable, both because
of the impossibility of mea suring the relative utility levels each income
class derives from public goods, and because of conceptual problems in deciding
what one ought to be measuring. If, for example, Bill Gates's immense wellbeing
is a joint product of publicly provided goods and a host of other factors such
as social opportunities, talents, and hard work, and each of these factors is a
but-for cause of his wellbeing, then how much of his wellbeing gets credited to
publicly provided goods, and how much to other sources?
The choice between these two interpretations of
"gains more" boils down to the question of what is so great about market-based
distribution, beyond its guarantee of strict paretianism. In partic ular, is
there any normative reason to preserve (or mimic) the haphazard distribution of
gains from trade that happens to result from uniform, equilibrium pricing in
competitive markets, when we move from the private market to the shadow market
for public goods? Here, we encounter a wide range of views from the liberta
rian camp that don't lead to any obvious conclusion.
Most libertarians treat as sacrosanct the division of surplus that happens to
result from market prices, with exceptions often made for privately
appropriated, scarce natural resources and other monopoly goods. There is
little agreement among libertari ans, however, as to why market prices are
[*167] many of the arguments offered for not disrupting the actual, hap hazard
assignment of surplus in a well-functioning competitive market do not translate
well to public goods. Still others have ex plicitly treated some or all of the
gains from trade, or rents, as morally up for grabs. Among contemporary
libertarians, David Gauthier is a leading exponent of that view.
n31 Hobbes was as well, maintaining in effect that all individual gains from
moving from a Hobbesian state of nature to civilization are expropriable by an
all-powerful sovereign. If we are to read Hayek's statement above in the second
way, to mean that it is permissible to tax people in accordance with the
utility they derive from public goods rather than the quantity they consume,
then he, and all others who sub scribe to that view, ought to be counted in the
Hobbesian camp as well.
When dealing with private, competitive markets, most de fenders of the free
market would argue that it is unnecessary to resolve these internecine
disagreements about the justice of mar ket-based distribution. However morally
arbitrary the market's division of gains from trade might be, the efficiency
costs of the government's trying to alter it through price regulation or other
means are simply too great to justify the effort. As one commenta tor has
"how the gains from trade are distributed [by the market] is determined
arbitrarily, but since this distributional is sue is resolved as a by-product
of a process benefitting all parties, it need not become a bone of contention."
n32 When dealing with publicly provided goods, however, that response is not
available. The government must set shadow prices in the first instance. As a
monopoly supplier with the power to coerce payment through tax
[*168] ation, it could price-discriminate if it wished to, without being un dercut by
competitors. Finally, unlike in the competitive market, price discrimination
may well be dictated by efficiency considera tions, at least in the case of
true public goods. I turn to the last point now.
B. Pricing True Public Goods
A true public good is one that exhibits two characteristics. It is jointly
consumed, and exhibits nonexcludibility, meaning that it is either
technologically infeasible to exclude people from use (the case, for example,
with clean air or defense), or inefficient to do so because the incremental
cost of not excluding is very low or zero, and/or the cost of monitoring use in
order to exclude very high, such as with roads or TV airwaves. Because of the
problem of nonexcludibility, a private market cannot provide such goods. From a
libertarian point of view, true public goods are the only goods that ought to
be publicly supplied; that is, the sole justifica tion for coercive government
is to correct market failure that would preclude optimal coordination in the
absence of coercion. How such goods would or should be priced in the absence of
a func tioning market is therefore a critical question for any benefits the
orist, and a notoriously difficult one.
An extensive literature in public economics over the past 80 years has been
devoted to devising a theoretical solution to the problem of optimal output and
optimal pricing of true public goods. Since the literature is almost
exclusively concerned with efficiency questions - that is, how to determine the
optimal output of public goods from an efficiency point of view - in one sense
it is all beside the point here. In a couple of other senses, however, it is
not. First, the implicit or explicit indifference of the public eco nomics
literature in this area to the distributional consequences of optimal pricing
schemes undercuts any claim that there is some broadly compelling view about
the just distribution of surplus. Second, many of the optimal pricing solutions
proposed on effi ciency grounds have determinate distributive outcomes that are
hard to square with anything resembling a proportionate tax scheme. That means,
at the very least, that benefits theorists who are drawn to proportionality on
fairness grounds may have to be willing to pay a theoretical price on
1. Lindahl/Bowen/Samuelson Solutions
The central problem in determining optimal output of true public goods derives
from the problem of nonexcludibility: be cause we cannot exclude anyone from
using true public goods, we cannot rely on a decentralized price system
automatically to re
[*169] veal true demand for the goods.
n33 Eighty years ago, Erik Lindahl proposed a theoretical solution to this
problem, subsequently re worked in a general equilibrium framework by Paul
Samuelson. His solution remains the starting point for any analysis of public
The mechanics of Lindahl-type solutions are somewhat com plicated, but the
basic intuition behind them can be stated simply. To achieve an efficient
allocation of all resources, we want every consumer to equate her marginal rate
of substitution between all commodities, whether privately or publicly
supplied. In a compet itive market, where everyone faces the same price for
goods, that equilibrium is reached by each consumer varying the quantity of
each good she purchases until the marginal utility of each good equals its set
price. In true public goods, where everyone must consume, or have available for
consumption, the same quantity of the good, that equilibrium is reached by
varying the price each consumer pays for the good, until the price equals the
marginal utility she derives from that set quantity of the public good.
The hypothetical mechanics for reaching a Lindahl equilib rium quantity and
price are as follows. Each citizen, in effect, sub mits bids, revealing the
marginal price she would be willing to pay for a given public good at each
plausible quantity of that good.
n35 A combined social demand curve for that good is created by adding up the
marginal unit price bids from all citizens at each quantity level.
n36 The optimal level of supply is the equilibrium point where the combined demand
curve, equal to the combined marginal prices for a given quantity of goods,
intersects with the supply schedule for that good, equal to marginal cost of
Once the equilibrium quantity is determined, each citizen is assigned a tax
share to pay for the cost of the goods. That tax share is equal to the marginal
price she bid at what turned out to
[*170] be the equilibrium quantity, times that equilibrium quantity. The result is a
set of personalized prices, such that, at those specified prices, everyone
demands the same level of each public good.
What are the distributive effects of Lindahl tax shares? They mimic the effects
of the competitive market in one respect: they divide the inframarginal surplus
from public goods in a haphazard fashion. Each citizen's Lindahl tax price for
a public good reflects the utility of the marginal unit value of that good at
equilibrium output, times the total agreed-upon output. Assuming a down ward
sloping demand curve, each citizen will therefore realize in framarginal
surplus, i.e. gains from trade, on all but the last unit consumed. As in
competitive markets, however, the amount of that surplus is arbitrarily fixed
by the shape of the individual's demand curve for that good relative to the
n38 In another respect, however, Lindahl tax prices differ from conven tional
prices in a competitive market: Lindahl tax prices are pro portionate to the
marginal utility each individual derives from a given quantity of a public
good, rather than to the quantity of that good consumed. In contrast, in the
competitive market, total prices spent on a good would be proportionate to
quantity con sumed, since, facing a constant price, consumers vary their quan
tity until the last unit purchased generates marginal utility equal to that
Just as it is unclear what Hayek's
"total utility" metric for tax shares may imply about tax rates, it is also unclear what Lin
"marginal utility" metric for tax shares implies about tax rates. For some true public goods,
marginal utility seems unlikely to be positively correlated with income,
suggesting a highly re gressive tax rate. For others, it might be positively
correlated with income, but whether that measure would lead to proportion ality
seems much more doubtful.
2. Demand-Revealing Price Structures
As numerous commentators have noted, Lindahl-type solu tions, which tie the
ultimate tax shares to stated preferences, all have a central flaw. Since the
government cannot exclude individ
[*171] uals from using true public goods, whatever their stated prefer ences, such
solutions - which, in effect, require each individual bidder to put her money
where her mouth is - create an incentive for individuals to understate their
preferences in the bidding pro cess, thereby leading to the suboptimal
production of public goods.
Over the past 35 years, a number of public economists have proposed ingenious
modifications to Lindahl's basic scheme to solve this central problem.
n40 The modifications take different forms, but most share one key feature: the
tax share that people ultimately will pay for each public good is decoupled
from their stated preferences for that good, in order to induce true revelation
of preferences. In most models, that result is accomplished by a two-level
pricing system: (1) a first-level tax is imposed to cover the full costs of
producing a public good at a given level, with the tax shares assigned
independent of each person's true demand schedule for that good; and (2) a
supplementary tax is imposed, the sole purpose of which is to force the true
revelation of prefer ences. This latter tax is not used to pay for the public
good in question.
n41 The level of this supplementary tax in most schemes is loosely correlated with
preferences, since it may, and in some cases certainly will, impose some cost
on each individual to move the outcome in her preferred direction.
n42 To that extent, the sup plementary tax may reflect a loose quid pro quo for
benefits re ceived. The correlation, however, is loose at best. In the case of
the leading schemes, once a large number of voters is introduced, the tax
What does the foregoing imply about the appropriate tax rates to finance true
public goods? The critical feature of all of these proposals is that each
individual's tax share must be unre lated to that individual's revealed demand
for the good, and hence the level of utility she derives from consumption. This
require ment rules out, at least in strict form, the one criterion for as
signing tax shares, quid pro quo on an individual basis, that a benefit theory
insists upon. Many of these schemes contemplate that the first-level tax will
be apportioned among individuals on a
[*172] per capita basis, a solution that would obviously be highly regres sive. For
some schemes, the per capita charge is a necessary fea ture, while for others,
it is not. In the latter cases, it is theoretically possible that, once optimal
output, and hence full costs, are determined, one could distribute that tax
burden by in come class or other measurable and taxable voter characteristics
in a fashion that approximates what Lindahl taxes for that group would look
n43 That decision, however, is independent of the efficiency concerns that are
driving the models, and would be dic tated by wholly external distributive
concerns, which require in dependent justification.
"Tale of Two Pies"
These and all other attempts to solve the problem of optimal output and
pricing in the true public goods context are concerned only with efficiency.
Thus, they take the background distribution of wealth as just, and disregard
whatever incidental distributive effects their solution might entail. In recent
years, some promi nent libertarians have directly addressed the question of the
just distribution of surplus from public goods and the tax costs of fi nancing
them. Two notable recent examples are Richard Ep stein's Takings
n44 and David Gauthier's Morals by Agreement. Both arguments attempt to develop a
theory of the just distribu tion of surplus value created by the existence of
an organized soci ety, and both purport to derive proportionate taxation from
that theory. Many of the intuitions that drive these two pieces seem to be
widely shared in libertarian defenses of proportionate taxation, including
Hayek's quoted above.
I have explored both arguments in greater length elsewhere.
n46 In the interests of economy, I will limit discussion here to Ep
[*173] stein's, the more conventional of the two. Epstein, consistent with
conventional libertarian premises, limits the state's taxing powers to raising
funds to finance true public goods, explicitly precluding any spending
motivated by redistributive intent.
n47 That con straint, however, leaves open the question of how precisely that
revenue is to be raised, and hence how any surplus value gener ated by the
existence of the minimal state - with surplus implic itly defined here as the
total value of society net of the tax costs of providing it - is to be
distributed among citizens.
For Epstein, all roads, whether constitutional, Lockean rights theory or
utilitarian, lead to the same two-part answer to the question,
"'Who gets the surplus?'"
n48 First, the surplus should be divided in proportion to the value of each
"private hold ings," which Epstein defines as the property she would have held in a state of nature
(SON), under just (Lockean) principles of ac quisition.
n49 Second, such proportional division of surplus will nat urally, if not
definitionally, lead to proportionate taxation.
Both halves of the argument have some surface plausibility. At least so one is
led to conclude by the long list of people who have subscribed to them. The
more one ponders them, however, the less plausible they seem. I take up each
half in turn.
1. Why Should the Surplus Value Created by the Existence of Organized Society
Be Distributed in Proportion to the Value of Rights Individuals Have Brought
As noted above, Epstein purports to defend this proposition from three
different vantage points: constitutionalism, utilitarian ism, and a Lockean,
natural rights, perspective. For present pur poses, I want to set the first two
to the side, and examine the
[*174] question only from the third; the perch shared by most of his fel low
Locke's state, like Hobbes's, Epstein argues, gets its legiti macy from the
strict pareto-superiority of the deal it offers its members: give up the formal
freedom you possessed when locked in your war of each against all, in return
for constraints of law that will leave all of you at least as well off, and
some of you better off, than you were in the SON. For Hobbes, that guarantee of
strict paretianism defined the full extent of the sovereign's obliga tions to
its citizens. For Locke, however, it merely set a floor on those obligations.
While Hobbes's sovereign was free to appropri ate virtually all the surplus
value of civilization, Locke's was not:
"in the Lockean world, ... the sovereign is to be fully con strained, so that
the lives, liberties, and estates of the citizens may be preserved. The
tangible measure of that constraint is that principled ... limits are placed
upon the appropriation of surplus by the sovereign."
As Epstein acknowledges, Locke himself did not supply those principled limits.
Epstein does so in his stead, concluding that the surplus
"should be divided among all citizens, pro rata in accord ance with their
n52 Why? Epstein, along with nu
[*175] merous others with libertarian leanings going back to Adam Smith, appears to
derive that conclusion from an unexamined analogy to private partnerships.
n53 The general line of argument goes as follows. Think of society as an n-person
joint venture, in which we all bring to the table our SON assets, which we
agree to pool in this collective venture called the state in return for our
aliquot share of the returns to investment, or surplus, that our cooperation
generates. If this were a private partnership between two persons, one putting
up $ 60,000 of capital and one putting up $ 40,000, it is perfectly clear that
the two partners would and should agree to split the profits from their venture
60/40, in ac cordance with their contributions. There is no reason for that con
clusion to change when we move from a two-person private partnership to an
n-person public one.
There are a number of serious problems with the analogy.
n54 The first is: how do we value SON assets for purposes of calculat ing what
would be a fair (pro rata) return on them? As Epstein concedes, his two-pie
"presupposes that we have a very clear sense of what counts as individual rights."
n55 Much of the criticism of Epstein's Takings, along with Nozick's Anarchy,
State, and Utopia and other libertarian schemes, has sought to argue that such
"clear sense" is impossible to come by, and that what Epstein et al. take to be the
boundaries simply reflects conven tional Anglo-American property arrangements
that would be im possible to deduce from any deep-seated theory of natural
rights. I want to set those objections aside for the moment, and suppose
[*176] that we could come up with some formal boundaries on such natu ral rights. The
question is: what are they worth in the SON?
Epstein concedes that the value of rights in the SON may be low, due to
uncertainty and insecurity.
n56 That concession, how ever, is hardly damaging to his argument. The real
problem is, how would we determine the value in the SON to begin with? Rather
than hazarding an answer - a daunting thought experi ment - Epstein simply
assumes that the value of rights in the SON are a scaled-down version of the
value of rights in society. Thus, if Brains is earning 100 times more than
Brawn in America circa 1999, we are to assume that the use value of Brains's
human capital in the SON was 100 times higher than the use value of Brawn's. As
discussed below, this assumption that people's en dowments retain the same
relative values in moving from SON to society is critical to Epstein's argument
for proportionate taxation; it also seems completely unfounded.
The second problem is, what portion of the surplus value gen erated by
civilization is subject to division by collective decision making? In the
typical private joint venture envisioned in the 60/40 partnership example
above, the answer is relatively clear. First, the boundaries between the
partnership's activities and the separate activities of each partner are
typically clearly demar cated, and hence that portion of social wealth
attributable to the returns to the partnership's activities clearly demarcated
as well. In committing their respective $ 60,000 and $ 40,000 to their joint
venture, each partner implicitly commits to dividing those returns by some
collectively agreed-to decision rule.
The answer to the question, however, is far from clear in the context of the
Lockean social contract. What exactly are the boundaries of the partnership we
are embarked upon here? Does it extend only to the creation and operations of
the formal state, or does it extend to social organization in any form? Epstein
and others appear to make the latter assumption, in assuming that the entire
(aggregate) increase in wealth we all realized in moving from the SON to
America circa 1999 is subject to division by col lective decisionmaking. That
broad view, of course, implies that the enormous gains society bestows on those
whose natural tal ents have little use value on their Crusoeian island are all
up for grabs. One could take this view of our implicit (Lockean) social
contract, pursuant to which all the gains that, say, Wayne Gret sky realizes by
moving from being Wayne Gretsky alone on a de sert island, thinking of
inventing a game called hockey if he could ever find ice, eleven other players,
and an audience to pay to watch, to being Wayne Gretsky in late
[*177] earning $ 20 million a year, are thrown into a common pool for di vision in
accordance with some norm of just distribution. This is certainly a plausible
moral view, with a number of respectable ad herents.
n57 But it seems like an odd concession to collectivist ethics for libertarians to
make. One would expect them to insist that a smaller portion of the gains to
Gretsky from civilization is fair game for redivision among all citizens.
n58 Limiting what is up for grabs to that portion of private gains directly
attributable to the provision of specific, costly, public goods, is one
possibility - a view that implicitly limits the scope of the social contract to
the opera tions of the formal state. Nozick and others have suggested an other:
that the only thing that should be up for grabs is that portion of income that
exceeds what the best endowed could have gotten, not by staying in the SON, but
instead by seceding from the existing state and forming a new state whose
membership was limited to the best endowed.
Third, assuming that America circa 1999 represents one great joint venture,
with each participant entitled to some share of the aggregate returns to
civilization, how are those returns to be di vided among the participants? A
pro rata division, in accordance with the value of assets contributed, is an
obvious solution only if one is dealing with inputs that have an opportunity
cost (expected return outside the partnership) equal to their value inside the
partnership. That condition is presumably met in the 60/40 part nership example
above, if we assume the partners' only contribu tions are (fungible) cash,
which is receiving a marginal (competitive) return. In such a case, the
constraints of the mar ket - the opportunity cost of capital - dictate a pro
rata division of the partnership's income, whatever justice might require, and
most people would take that result to be consonant with what jus tice in fact
But in the Lockean social contract, the value of each person's assets is (by
hypothesis) greater if exploited within the joint ven ture of civilization than
it is outside of that joint venture in the SON. That is to say, the Lockean
social contract resembles an n- person multilateral monopoly rather than a
competitive equilib rium. Imagine, for example, that in our two-person
partnership, partner A contributes a really good idea, which requires $ 100,000
in ready cash to exploit, and partner B contributes the $ 100,000 - the only
person, as it turns out, that was willing to come up on the
[*178] spot with the required cash. Now, there are gains from coopera tion in excess
of the returns available to either partner from her next best available
opportunity. To that extent, they are, with re spect to each other, locked in a
bilateral monopoly. How should returns from our grand joint venture be divided
The problem presents one instance of the allocation-of-com mon-costs problem
that has been subject to extensive analysis in the game theory literature. The
question is, how should the com mon costs of a value-enhancing cooperative
venture be allocated among the cooperating parties? While a detailed treatment
of the issue is beyond the scope of this paper, a brief overview may be helpful
in framing the problem.
In game-theoretic analyses, the possible solutions ("core") of the common-costs problem are taken to include all allocations that give
each player in the cooperative game a payoff at least as great as the greater
of (1) what she could have secured through a non cooperative strategy, and (2)
what she could have achieved as a member of the most profitable coalition of
players that could se cede from the group to pursue their own cooperative
n60 Clearly, condition (1) requires that each person get a return from
civilization (net of taxes) at least equal to the value of her assets in the
SON - that the choice be individually rational. (This, of course, correlates
with the minimal pareto condition imposed at the maximum level of aggregation.)
Condition (2) adds the addi tional requirement of group rationality; it just
restates in a slightly different form Nozick's secession thought experiment, de
n61 Nozick is concerned with defining the extent of private gains that are up for
grabs at all; condition (2) is con cerned with defining how such gains should
be divided. But the two are just different paths to the same conclusion: the
best en dowed group in society (and by extension each lesser endowed group in
turn) will claim for itself, at a minimum, that portion of the gains from
social cooperation that it could have achieved by seceding from America circa
1999 and forming a new state whose membership is limited to the best endowed.
What if anything do the above conditions imply about the ap propriate division
of surplus from society? First, the strict re quirement of individual or group
rationality holds as a descriptive matter - that is, predicts the actual
outcomes in allocating com mon costs - only where exit from the group is
[*179] costless. To the extent that it is costly for the best endowed mem bers of
society (and by extension any other subgroup) to secede from America circa 1999
to form their more perfect union, the gains they could have realized in that
more perfect union are, as a purely positive matter, expropriable by the less
well-endowed ma jority. This, of course, describes to a considerable extent the
situa tion facing the very rich in this society.
n62 In such a case, what, if any, normative force does condition (2) carry? Are we
required to abide by the distributive results the best endowed could have ob
tained for themselves were exit costless, even though it is not, and if so, why?
Assuming that we are, in fact, morally required to abide by condition (2) (the
results of the Nozickean thought experiment), what does it imply about the
appropriate division of surplus from society? As Kenneth Arrow has argued, if
there are no increasing returns to scale for society and no externalities, in a
large enough economy where no individual is large on the scale of the economy,
"the core shrinks to the competitive equilibrium."
n63 That is to say, condition (2) leads to whatever division of surplus happens to
re sult from market prices, and there is, as Arrow states, no problem of social
justice left at all.
n64 In such a world, taxation is presuma bly relegated to a pure benefits tax,
with rates set to mimic (as far as possible) the prices the market would itself
set for publicly pro vided goods or services.
If there are increasing returns to scale from organized society, generating
significant gains from nonsecession for any subgroup, including the
best-endowed, the range of solutions in the core is less clear.
n65 Contrary to Epstein's, Gauthier's, and others' appar ent assumption, however,
there is unlikely to be any unique solu tion to the problem, and an allocation
that assigns surplus (net of taxes) in proportion to the value of SON assets is
not necessarily even in the core of the game.
2. Would Pro Rata Division of the Surplus Value of the
Social Contract, in Proportion to the Value of
Individual Rights in a SON, Lead to Proportionate Taxation to Finance the Cost
of Public Goods?
Epstein, along with numerous others, presumes the answer to the above question
is yes. In fact, the answer is no, unless two conditions are met: everyone must
derive utility from the creation of the minimal state in proportion to the
value of his or her rights in an SON, and the utility each person derives must
all be re flected in income (or whatever other tax base we use). It is ex
tremely unlikely that either of these conditions will be met, let alone both.
A simple example will illustrate the problem. Suppose we have a two-person SON
society. Person A, a hunter, has rights, in the form of human capital, with a
market value of $ 10. Person B, a gatherer, has rights, in the form of human
capital, with a mar ket value of $ 20. The two enter into a compact to create a
Nozick ean/Lockean minimal state, pursuant to which they forswear Hobbesian
aggression against each other, agree to hire a merce nary police force and army
to enforce their Hobbesian contract against mutual aggression and to defend
themselves jointly against outside aggression, and agree to create a variety of
pure public goods, including a lighthouse, a TV broadcasting system, and roads.
All of these goods and services together cost $ 100 to supply.
Due to the advances that civilization has brought, person A is now able to give
up hunting and become a taxi driver for visiting tourists. She spends her spare
time watching television. The to tal subjective value to her of her rights in
civilization is equal to $ 230, of which $ 140 is attributable to income
generated by driving a cab and $ 90 to the pleasures of watching TV. Person B,
in the meantime, becomes a television technician. He works all the time,
generating a total subjective value to him of $ 260, equal to the total market
earnings he realizes.
Thus, in our simplified example, the total costs of civilization are equal to $
100, and the total benefits of civilization are $ 460 ($ 490 total utility in
civilization, minus $ 30 total utility in the SON), yielding a net surplus
value from civilization of $ 360. How is the $ 100 cost of producing that $ 460
benefit to be divided through the tax system? Epstein would argue it should be
di vided such that A ends up with one-third of the total $ 360 surplus value,
or $ 120, and B ends up with the remaining $ 240. But, as the chart below
indicates, that result would mean that A must bear the entire $ 100 cost of
public goods, and B nothing. In con
[*181] trast, a proportionate tax on earned income (which in this exam ple totals $
400) would take 25% of the earned income from each of the two, producing a $ 35
tax bill for A (.25 x $ 140) and a $ 65 tax bill for B (.25 x $ 290).
Epstein's Tale of Two Pies
[tdm'After-tax share of surplus',ql [tcgp1,m'$ 90 imputed income)',qc,vu1
[tcgp1,m'Gatherer Turned',qc,vu1] [tu3;3] AB HunterTurnedGathererTurned
TaxiDriverTVRepairman [tu3;3] SON(wealth) $ 10 $ 20 LS(wealth) $ 230
$ 90imputedincome) $ 260
(income) Epsteintaxallocation $ 100 $ 0 After-taxwealth $ 130 $ 260
NetofrightsinSON -$ 10 -$ 20 After-taxshareofsurplus $ 120 $ 240 Tax allocation
under 25% proportionate tax on income $ 35 $ 65 [tu3;3]
"proportionate division of the surplus" rule clearly produces results that are wildly removed from a propor tionate
tax on income. In this particular example, it produces a tax scheme in which
100% of the tax burden falls on the person with the lower earned income. While
that particular result is an artifact of the numbers chosen, the more general
conclusion is not. Indeed, only by the most implausible coincidence would
Epstein's rule ever produce something approximating proportionality.
To drive the point home, imagine we expand the example to a four-person SON
society, in which, in addition to A and B above, we also have C, a wily but
puny hunter, whose brains compensate only partially for her absence of brawn,
leaving her slightly worse off than A; and D, a 7 foot 6 inch gatherer whose
freakish height is on the whole a disadvantage for his chosen profession in the
SON economy, leaving him slightly worse off than B. In the Hobbesian minimal
state circa 1999, C becomes a well-paid law professor, the income from which
far outstrips A's or B's, and D becomes a basketball superstar, the income from
which dwarfs A's, B's and C's. The relative gains to C and D - whose distin
guishing talents were completely worthless in the SON - from en tering into
civilized society are wildly disproportionate to their relative positions in
the SON. As a consequence, a tax designed to return C and D to their same SON
positions relative to A and B must necessarily fall in a confiscatory manner on
C and D.
III. Social Welfarist/Consequential Theories of DistributiveJustice
The intuition that tax burdens should be set in a fashion that minimizes the
aggregate burden they entail has long held appeal. For the latter part of the
nineteenth century and first half of the twentieth, that intuition was most
frequently embodied in the no tions that tax burdens ought to be allocated in
"ability to pay," or to
"minimize sacrifice." For reasons thoroughly rehearsed elsewhere, neither
"ability to pay" nor
"minimum sacri fice" theories have moral coherence as freestanding principles of distributive
n67 They are incomplete intuitions of a deeper, unarticulated (social welfarist)
theory of distributive justice, in which the state has an obligation to use its
fiscal powers to further the aggregate welfare of society, aggregated in
accordance with some implicit or explicit social welfare function.
Viewed in that larger context, the social welfarist case for any tax structure
judged in isolation is inherently unsatisfactory, for reasons that critics of
"equal sacrifice" theory on the right and the left have long recognized. Once we admit that the
redistribution of income for social welfarist ends is a legit imate role for
the government, it is morally incoherent to isolate the tax side from the
transfer/expenditure side of the fiscal ledger. As Wicksell commented many
years ago, there is no point in achieving
"a just part in an unjust whole."
n68 It is operationally incoherent to isolate the tax side of fiscal affairs as
well, for the simple reason that we can undo any tax distribution on the trans
fer side. Take the following, simple example. Suppose we have a two-person
society in which Poor earns $ 1000 and Rich earns $ 10,000. Suppose we impose a
flat tax of 45% on both Poor and Rich, thereby reducing Poor's after-tax income
to $ 550 and Rich's to $ 5500. Suppose we then take the resulting $ 4950 in tax
reve nues, and distribute them in-kind and in cash entirely to Poor. One would
be hard-put to disagree if Rich is heard to complain that the net result of all
this fiscal legerdemain is a confiscatory tax on Rich for the purpose of
achieving absolute income equality between Rich and Poor.
[*183] What an optimal tax-and-transfer scheme would look like under a plausible
range of social welfare functions giving priority to the needs of the worst-off
has been much debated. The central problem, of course, is how to balance
welfare gains (assuming a declining marginal utility of income) from
redistributing income to the poor, against welfare losses from the
distortionary effects of taxes required to finance the redistribution. It has
been suggested that a degressive tax, or even a tax rate structure that
declines slightly at the top, may well be optimal if combined with substan tial
lump-sum transfers (so-calleddemogrants).
n70 The intuition behind flattening rates at the top is that the last, or
marginal, incremental decision to work or save is most sensitive to tax rates.
In a world with perfect information, each person would have her own optimal tax
schedule, with rates set very high on in framarginal earnings, and declining,
ultimately to zero, on the last (marginal) dollar earned. In our actual world,
where we can not observe people's true preferences, a degressive tax structure
might best approximate that ideal result. More recent work has cast doubt on
the efficiency arguments for a degressive tax, sug gesting that, at least in
some contexts, in particular, in winner- take-all markets and contexts where
work and savings decisions are driven by a desire for relative status, steeply
graduated rates may well be more efficient than flat rates.
n71 The larger point, how ever, remains clear. The case for any tax structure on
social welfarist grounds is purely derivative of prior moral commitments to a
particular social welfare function, and empirical hunches about the combination
of tax and transfer schemes that will best effect it.
Notwithstanding the foregoing, many thoughtful people, pro ceeding on social
welfarist premises, have found it irresistible to assess the desirability of a
given tax arrangement in isolation from the transfer side of fiscal affairs.
This approach is perhaps an inevitable occupational hazard for professional tax
types, whose habitual frame of reference and claims to expertise both push
toward that partial view. It may also, in some cases, sensi bly reflect
political exigencies. If a decisionmaker is limited, as a political matter, to
affecting the distribution of tax burdens, then it is perfectly sensible for
her to take all other variables as given.
[*184] The tendency for social welfarists to assess the fairness of tax bur dens,
divorced from the government's other fiscal functions, ex tends far beyond the
foregoing situations, however, to include those who are trying to work out a
just scheme of income distribu tion de novo. Perhaps the most recent striking
example is Rawls.
In A Theory of Justice, Rawls suggests that the fairest tax scheme for raising
"provide for the public goods and make the transfer payments necessary to
satisfy the difference princi ple" would be a proportional consumption tax.
n72 Rawls offers the suggestion in passing, as a part of a hodgepodge of proposed
fiscal arrangements to effect a just state, and it seems doubtful that he has
thought through the implications of any of these institutional arrangements
n73 At any rate, one suspects he would think better of many of them on further
[*185] In defense of a proportional rate structure - the element of the scheme of
relevance here - Rawls notes only that
"it treats everyone in a uniform way (... assuming that income is fairly earned)."
n74 At the outset, Rawls's argument runs smack into the same difficulty faced by
most libertarian defenders of a flat-rate tax: the tax rate structure Rawls
supports is, in fact, not a propor tionate one, but a degressive one, since
Rawls, like most other pro ponents of proportionality, would permit the
"usual exemptions for dependents."
n75 Whatever other merits such a scheme might have, it surely lacks the virtue of
"uniformity" as that word is used here. Setting that difficulty aside, the form of
uniformity or equal ity that Rawls implicitly champions here, equal percentages
of in come taken, is hardly an obvious solution to the question,
"equality of what?," a matter to be returned to shortly. That point is underscored by noting that
the difference principle Rawls would use in setting transfers adopts a quite
different notion of uniform ity: uniformity of absolute income levels after tax
and transfer, subject to the exception for inequalities that help the least
That last observation brings us to what is surely the greatest oddity of
Rawls's championing of a flat-rate tax. In the context of the overall Rawlsian
scheme, the right not to have one's income taxed at a higher marginal rate than
one's neighbor's stands as an island of deontological rights swamped by a sea
of redistribution. As the above example of Rich and Poor illustrates,
n76 if the pro ceeds of a Rawlsian tax are used to finance the transfers to the
least well off required by the difference principle, the fact that the rich are
"uniformly" with the poor on the tax side should be cold comfort to them, and a matter of
total moral indifference to a Rawlsian.
Indeed, the same criticism Rawls rightly makes at
"tradi tional criteria" for fair taxation, ability to pay and benefits theory, can be made of Rawls's
proposed tax scheme: it does not
[*186] sufficiently comprehensive point of view" of the government's role in securing a just distribution of income.
As a final oddity in Rawls's defense of proportionality, Rawls suggests that
flat rates are obviously fair only in an already just world. Given the
injustice of existing institutions, Rawls argues, there may be a role for
"even steeply progressive income taxes."
"injustice," Rawls appears to mean here a world in which not all income is
n79 It is hard to figure out what this essentially Nozickean principle of
rectification is doing in a Rawl sian world. Given Rawls's view that talents,
including a taste for hard work, are undeserved, and hence the income derived
from those talents undeserved, what could it mean to Rawls for a per son to
earn income fairly, such that it was not up for grabs under a steeply
progressive income tax?
In the end, the impossibility of making any logical sense, in the context of
his larger aims, of Rawls's positions on tax policy in general, and rate
structures in particular, makes those positions of more than common interest -
not in justifying those positions, but in appreciating the breadth and strength
of the irrational pull they exert. If Rawls, of all people, could think a flat
tax a just tax because
"it treats everyone in a uniform way," surely there is some powerful instinct at work here that resists logic. What
IV. Equality of Tax Burdens as a Percentage of Income as a DefaultPosition
Rawls may be the most surprising expositor of the view that a flat tax is the
fairest tax of all, but he is hardly alone. Many peo ple believe (1) that one
can assess the fairness of a tax system in isolation from government
expenditures or any other government policies affecting the distribution of
wealth, and (2) that viewing our tax system from such a perspective, a
proportionate tax scheme is the self-evidently fairest one, because, in the
words of Rawls,
"it treats everyone in a uniform [that is, equal] way."
n80 In deed, for many, flat rates (or more precisely a degressive rate structure)
appear to be the only feature of the tax system that is mandated on fairness
[*187] Everyone is in favor of equality. The question, of course, is
"equality of what?" Why this particular version of equality, taking an equal percentage of
everyone's income, has had such wide spread and persistent appeal is something
of a mystery. It has history on its side, of course, in the biblical tradition
of tithing, but that merely pushes the mystery one step back.
To take the case for proportionality in its most sympathetic posture, let us
assume that the beneficiaries of public expendi tures are absolutely impossible
to trace, thereby precluding any benefits-theory-based defense for any rate
structure. We can in voke here Blum and Kalven's helpful suggestion that we
treat the collection of taxes
"as though it were only a common disaster - as though the tax money once
collected were thrown into the sea."
n82 Let us also assume that the state has no legitimate, social welfarist interest
in redistribution, and that we take the back ground distribution of pretax
incomes as just. In this limited con text, in which the fairness of tax burdens
is to be judged completely in isolation, can we say anything at all about their
For most people, the default assumption in such a case, where taxes are, in
Blum and Kalven's words,
"a necessary evil falling upon a distribution of money ... which is otherwise
n83 is that the tax burden ought to be shared equally by all taxpayers. This view
is obviously reflected in the longstanding argument, go ing back at least to
"equal sacrifice" in taxation.
n84 That argument, however, simply raises the question adumbrated above, equality
[*188] Now, we are plunged into a long-running dispute, well-docu mented elsewhere,
and I do not propose to rehash it at length here.
n85 In brief, commentators have disagreed about the proper meaning of both
equality and sacrifice, in interpreting
"equality of sacrifice." Two principal measures of equality have been pro posed: equal absolute
sacrifice, and equal proportionate sacrifice (in proportion, that is, either to
income or to any substitute taxa ble base).
n86 Two principal measures of sacrifice have been pro posed: that sacrifice should
be measured by the utility of dollars relinquished, and that it should be
measured by dollars relinquished.
The resulting four-cell matrix obviously yields four different interpretations
"equality of sacrifice." Of the four, only one - equal proportionate sacrifice measured by dollars
relinquished - unambiguously (indeed, definitionally) leads to a flat-rate tax.
Equal proportionate sacrifice measured by the utility of dollars re linquished
under most plausible utility schedules for money would yield a progressive rate
structure. Equal absolute sacrifice mea sured by dollars relinquished obviously
yields a highly regressive head tax. Equal absolute sacrifice measured by the
utility of dol lars relinquished is the most ambiguous: under differing but
plausible assumptions about the declining marginal utility of in come, it could
imply a regressive, proportionate, or progressive rate structure.
What, then, explains the apparently widely shared intuition that the equality
principle should be interpreted to require equal proportionate sacrifice,
measured by dollars relinquished? The choice to use dollars, rather than the
utility of dollars, as a mea sure of sacrifice, does not seem hard to defend.
Indeed, the con trary choice, adopted by numerous tax theorists over the last
century, is much harder to understand.
n88 We customarily mea sure the price exacted for goods or services by nominal
dollars paid, not by the subjective disutility to the payor of relinquishing
those dollars. Opting for the latter measure here - a measure
[*189] that, under conventional assumptions of the diminishing marginal utility of
wealth, is inversely correlated with initial wealth - is hard to explain except
as an indirect way to smuggle in the very redistributive objectives that we
ruled out of bounds at the start of this inquiry, when we took the background
pretax distribution of incomes as just.
The former choice, however - the choice of proportionate over equal absolute
sacrifice - seems much harder to defend. Probably the most famous, albeit
indirect, brief for proportionate taxation, Blum and Kalven's
"Uneasy Case," has only two things to say on behalf of the fairness of flat rates in the
course of its 100-page assault on graduated rates. The first is that
"as a principle of justice [proportionate sacrifice] is intuitively attractive."
n90 The second is that the
"virtue of the proportionate sacrifice formula is that it remains neutral as to
the relative distribution of satisfac tions among taxpayers."
n91 Virtually all defenses of proportionality ultimately boil down to some variant
of the former
"I know fair ness when I see it" claim, or the latter tautology. In the tautologi cal vein, for example,
consider Hall and Rabushka's argument that
"the principle of equity embodied in the flat tax is that every taxpayer pays
taxes in direct proportion to his income."
n92 Or con sider Hayek's: while progression represents
"discrimination against the wealthy,"
"raises no problem of a sep arate rule applying only to a minority."
Can we do no better than this? The usual starting point for implementing
equality-based norms is that people who are identi cal in relevant respects
should be treated identically. The case for
[*190] any tax rate structure that is tied to the level of income, be it pro
gressive, proportionate or even regressive, thus depends on show ing that
income levels are morally relevant in setting differential tax burdens. In the
case of benefits theory or social welfarist ap proaches, the claim for moral
relevance of income levels is obvious. In the case of benefits theory, income
is taken as a proxy for bene fit levels; in the case of social welfarist
approaches, whether em bodied in inchoate form in
"ability to pay" arguments or full-blown optimal tax analyses, income (or capacity to generate
income, as reflected in actual income) is taken as the prime measure of wel
fare. If the justice of tax burdens is really to be judged in isolation from
either the distribution of government benefits or the desired end-state
distribution of income, however, it is difficult to see why income levels would
ever be relevant in measuring equal treat ment. The far more plausible measure
of equality under those conditions would seem to be equal treatment of
individuals, a premise that would seem to lead to a head tax. Indeed, Blum and
Kalven argue as much in another context, in attacking Mill's
"equimarginal sacrifice" as a plausible interpretation of equality. If we are really confining
discussion to how we should allocate a burden, and are ignoring benefits and
distributive concerns, Blum and Kalven argue, the principle of equimarginal
"seems not a little absurd.... It is strange indeed to have [two men] share a
common burden by putting all of it on the wealthier man."
n95 It is no less absurd in principle, although less extreme in degree, to put
more of it on the wealthier man than the poorer one just because he is
wealthier. At least, it is hard to see how the absurdity could ever be removed,
except by reference to either benefits theory or social welfarist concerns.
V. Taxes Should Be Levied so as to Leave People's ChoicesUndistorted
A number of commentators with libertarian sympathies have argued for
proportionality on the ground that it would leave peo ple's choices undistorted
from what they would have been in a no- tax world.
n96 Minimizing tax-induced distortion in behavior is, of course, the goal of much
of public economics, on efficiency grounds. For Epstein, Gauthier and others,
that ideal appears to have liber tarian roots, of a Randian/personalist sort.
As Epstein put it:
The creation of a system of government should strive not to re duce the scope
of permissible individual choice from what it was before. Accordingly, no
person or group should be able to use the tax system to change the pattern of
preferences of other in
[*191] dividuals. If A ranks a set of (noncoercive) alternatives 1 to n before
taxation, then A's ranking of those alternatives should remain 1 to n, without
alteration, after the tax is imposed.
This statement is hardly a universally shared interpretation of the
requirements of libertarianism. It treats individual liberty as residing in
undistorted individual choice among actions, rather than, for example, the
utility any particular choice generates. Thus, nothing in Epstein's example
would preclude a near-confis catory tax on earnings from any source, provided
only that it did not reverse rank orderings of choices. Be that as it may,
contrary to Epstein's assumption, the requirement that we preserve the rank
orderings of choices - including both the choice among kinds of work and the
choice between work and leisure - does not lead in the first instance to either
proportionate taxation or, what Epstein erroneously takes to be the same thing,
a tax system that leaves the relative wealth of taxpayers constant.
n98 It leads to a highly regressive head tax, endowments tax, or some other form
of pure lump-sum taxation. As a second-best alternative, where lump- sum
taxation is politically infeasible or distributionally undesir able, it leads
to a Ramsey-type optimal tax on labor or capital, in which tax rates are set in
inverse proportion to the elasticity of supply, in order to have the tax burden
fall, as far as possible, on suppliers' rents. As discussed above, what
precisely such an opti mal tax scheme implies about the rate structure in a
broad-based income tax has been much debated, without any definitive conclu
sion. A case can be made for a degressive, or even a regressive rate structure
at the top of the income scale, on efficiency grounds.
n99 But the choice to opt for any broad-based income tax in preference to lump-sum
taxation in the first instance must be driven by distributional concerns that
are hard to reconcile with the Randian libertarianism driving Epstein's
argument here, or the more conventional, property rights-based libertarianism
re flected in benefits theory.
VI. Political Economy Arguments for a FlatTax
Proponents of proportionate taxation have offered a variety of political
economy arguments on behalf of flat rates. Perhaps the most prominent is an
argument, chiefly associated with Hayek but espoused by many others as well,
that proportionality, as a rule of
[*192] general application, precludes the majority from imposing upon
"a minority a rule which it does not apply to itself."
n100 Such an impo sition, Hayek argues,
"is an infringement of a principle much more fundamental than democracy itself,
a principle on which the justi fication of democracy rests."
Hayek's argument for proportionality here, of course, depends on the fact that
in raising rates on the minority rich, the majority also raises rates on
themselves. Whatever merits the argument might have as an empirical or
normative matter, however, the property Hayek seizes on here is hardly unique
to proportionate taxation. It is even more true of a head tax, in which any
rate increase imposed on the minority rich will fall on the majority even more
heavily than in a proportionate scheme, a fact Epstein inadvertently surfaces,
in mistakenly defending a flat tax on the grounds that
"[a] rule that says you must pay a dollar for the dol lar that you wish to exact
from your neighbor" is an important constraint on political intrigue.
n102 Indeed, the same property holds for any rate structure, including a highly
progressive one, in which the rates of each income group are a positively
correlated arithmetical function of the rates imposed on others. With little
ingenuity, one could accommodate virtually any level of progres sion within
There is, however, a further problem with Hayek's argument. It presumes, as
Epstein argues, that
"the flat tax gives the gov ernment only one degree of freedom: what is the
level of the reve nues and, hence, of [the tax rate]?"
n103 That conclusion is true, however, only if a flat tax, once enacted, cannot be
repealed. If it can - and unless constitutionalized, it formally can as easily
as the progressive rate structure it has hypothetically replaced - then at any
time after a flat-rate structure has been enacted, the government has not one
degree of freedom, but as many as it ever had, since enactment of a flat-rate
structure does not itself pre clude the bottom 51% of the population from
sticking it to the top 49% at any time, simply by voting to abandon flat rates
in favor of progressive ones. The Hayekian argument, to make any sense, must
therefore depend upon a further assumption: that if one can persuade people to
go for a flat-rate structure to begin with, one has a higher than (politically)
customary chance of persuading
[*193] people to adhere to it going forward. On what could such an as sumption be
VII. Flat Rates as a Schelling-Like FocalPoint
One possible answer lies in the suggestion that the broad ap peal of
proportionate taxation is attributable less to any easily ex plicable moral
theory than to its operating as a Schelling-like focal point for people; that
is, a solution that is psychologically promi nent because of its apparent
mathematical certainty, along with its apparent properties of equality.
n104 I say apparent mathemati cal certainty and simplicity, because one can achieve
equal or greater simplicity with other schemes - a head tax, for example - and
equal mathematical certainty from any mathematically deter minate function that
correlates the tax rates of different income groups. But that logical quibble
is in a sense beside the point here - the point being that, for whatever
psychological or histori cal reasons, this particular relationship among tax
burdens has a strong pull on the popular imagination.
There is much, particularly in Hayek's writings, to support the focal-point
explanation. It may explain why people as diver gent in their political
commitments as Rawls and Hayek might fix on such a solution in good faith to
begin with, as unselfconscious participants in a Schelling-like convergence. It
also explains why proponents of less progressivity than now exists would fix on
it as a strategic matter, as a political solution that is both obtainable and
sustainable. Hayek himself seems to concede as much, when he acknowledges that
he is seeking a principle
"which has [a] pros pect of being accepted and which would effectively prevent
those temptations inherent in progressive taxation from getting out of hand," and rejects as an alternative solution
"setting an upper limit which progression is not to exceed."
n105 The problem with such a solution, notes Hayek, is that
"such a percentage figure would be as arbitrary as the principle of progression
and would be as readily altered when the need for additional revenue seemed to
n106 Of course, as noted above, a flat-rate scheme could be altered at will as
well, to revert back to a progressive rate structure. But Hayek has perhaps got
hold of a psychological truth here: a flat-rate structure has a psychological
prominence that may make it easier to sell to voters in the first instance than
many other possible rate structures that are less progressive than
[*194] the current one, and may make it more likely, once enacted, to stick.
VIII. Political Framing
Finally, the convergence on proportionate taxation seems to be, in significant
part, a product of political framing. Benefits the ory, at least under the
strict construction of benefits that seems most congenial to libertarian
premises, would almost certainly lead to a highly regressive tax structure.
Most opponents of progressivity on fairness grounds specifically reject that
outcome as politically unpalatable and hence infeasible. Blum and Kalven, for
example, take a regressive tax off the table, as they took off the table a tax
that does not exempt basic income, not because it is unpersuasive, but because
"it is so clear no one today favors any tax because it is regressive ...[that] a
regressive tax on income is not a serious [policy] alternative."
n107 Richard Epstein takes a head tax off the table without any justification
That decision permits opponents of a graduated rate struc ture to narrow and
restate their position as follows:
"on what grounds is a progressive tax on income to be preferred to a propor
tionate tax on income?"
n109 - or more precisely, for the overwhelm ing majority of flat-rate proponents
who also support an exemption for basic income,
"on what grounds is a progressive tax on all incomes over a minimum subsistence
exemption to be preferred to a proportionate tax on all incomes over a minimum
n110 Once regressivity is off the table as a viable alternative, along with a true
proportionate tax, a degres sive tax becomes the best available alternative to
those who are hostile to the degree of income redistribution effected at the
top of the income scale by a graduated rate structure. Indeed, the whole
structure of Blum and Kalven's implicit argument for proportion ality depends
on getting regressive taxes off the table, because al most all of the arguments
they make against progressivity, to the extent they are persuasive, would more
naturally lead to a highly regressive tax structure than a degressive one. (It
might also be noted that once a regressive tax and a true flat tax are off the
table, proponents of a degressive, rather than graduated, rate structure have
narrowed the choice to two alternatives that are so close to each other in so
many respects that it seems implausible that the differences between them will
be dispositive under most theories of distributive justice.)
[*195] We have lived with a progressive tax rate structure for so long that we are
all accustomed to think regressivity unthinkable. As Thatcherite England makes
clear, however, the unthinkable can become thinkable. If it did here - if the
political landscape changed enough to make a regressive tax rate structure, or
at the extreme even a head tax, a politically plausible alternative - then
there is every reason to expect that the odd convergence on a flat tax as
"the fairest tax of all" would dissipate. Whatever Rawls and other social welfarist participants in
that convergence might make of their inexplicable enthusiasm on second thought,
one would surely expect defection in the libertarian ranks. For liber tarians,
a regressive tax seems the more likely logical outcome of their philosophical
n1. Friedrich A. Hayek, The Constitution of Liberty (1960).
n2. Ludwig von Mises, Human Action (1949).
n3. Milton Friedman, Capitalism and Freedom (1962).
n4. John Rawls, A Theory of Justice (1971).
n5. Richard A. Epstein, Takings (1985) [hereinafter Takings]; Richard A. Epstein,
Taxation in a Lockean World, in Philosophy and Law 49 (Jules Coleman
& Ellen Frankel Paul eds., 1987) [hereinafter Taxation].
n6. Walter J. Blum
& Harry Kalven, Jr., The Uneasy Case for Progressive Taxation,
19 U. Chi. L. Rev. 417 (1952).
n7. Robert E. Hall
& Alvin Rabushka, The Flat Tax 3 (2d ed. 1995).
n8. Taxation, supra note 5, at 68. See also id. at 70:
"The flat tax ... is an indispensa ble part of the Lockean program of taxation."
n9. See Takings, supra note 5, at 99-100, 295-303.
n10. See, e.g., Hayek, supra note 1, at 315-18; Taxation, supra note 5.
n11. See supra note 3.
n12. See supra note 1.
n13. See supra note 5.
n14. David Gauthier, Morals By Agreement (1986).
n15. For other, recent expositions of a social contractarian justification for a
flat tax, see James A. Dorn, Introduction: The Principles and Politics of Tax
Reform, 5 Cato J. 361 (1985); Richard E. Wagner, Normative and Positive
Foundations of Tax Reform, 5 Cato J. 385 (1985). A number of other
libertarians, while not explicitly defending proportionate taxation, limit
their attacks on redistributive ("discriminatory") taxation to progressivity, thereby, at least by implication, treating
proportionate taxation as nondiscriminatory, and hence fair. See, e.g., von
Mises, supra note 2, at 803-05, 851-54; Harvey Lutz, Guideposts to a Free
Economy 73-82 (1945).
n16. For a summary of the proponents of proportionate taxation through the early
part of the 20th century, see Edwin R. Seligman, Progressive Taxation in Theory
and Pra ctice 148-84 (1908). Notable early proponents included Hobbes, Locke
and (somewhat more ambiguously) Adam Smith. For Smith's famous support of
proportionate taxation on fair ness grounds, see Adam Smith, An Inquiry into
the Nature and Causes of the Wealth of Nations 945 (Regnery 1998). For the
contrary impulse, in the context of a tax on house rents, see id. at 966,
"it is not very unreasonable that the rich should con tribute to the public
expence, not only in proportion to their revenue, but something more than in
n17. See Richard A. Musgrave, The Theory of Public Finance 62 (1959).
n18. Hayek, supra note 1, at 316. For similar statements, see Taxation, supra note
5, at 74; Friedman, supra note 3, at 175; Antonio de Viti de Marco, First
Principles of Pu blic Finance (Edith Paolo Marget trans., 1936), cited in
Musgrave, supra note 17, at 72-73; Smith, supra note 16, at 945. Among
proponents of a flat-rate tax on (vaguely) libertarian grounds, Blum and Kalven
strongly dissent from a
"benefits taxation" justification for flat rates, on both empirical and normative grounds. They
conclude, as an empirical matter, that the distribution of benefits among
income classes is so indeterminate that it is point less to defend either
progressivity or proportionality on these grounds. If forced to guess, they
would guess that all citizens benefit approximately alike, leading to a head
tax rather than a flat tax. Blum
& Kalven, supra note 6, at 455. On normative grounds, they adopt the view of
"to assert that individuals receive significantly different benefits from living
in a particular society is in effect to assert that there is something
seriously wrong with that society." Id. The latter objection seems silly, if it refers to unequal levels of
utility derived from government-provided services (why should everyone value
all government services the same?), as opposed to unequal access to those
n19. Under the Hall/Rabushka version of a flat tax, for example, exemption amounts
range from $ 9500 for a single person to $ 25,500 for a family of four; amounts
in excess of that are taxed at a 19% rate. Hall
& Rabushka, supra note 7, at 144.
n20. For Blum and Kalven's astonishingly bare concession to exemptions, see Blum
& Kalven, supra note 6, at 420:
"It is almost unanimously agreed that some exemption keyed to at least a minimum
subsistence standard of living is desirable. Since such an exemption will
necessarily result in some degree of progression among taxpayers above the
exemption level, and since this degree of progression appears inescapable, the
real issue is whether any added degree of progression can be justified." See also 2 Friedrich A. Hayek, Law, Legislation and Liberty 87 (1976),
endorsing in a slightly different context a uniform minimum income
"to all those who, for any reason, are unable to earn in the market an adequate
maintenance"; Smith, supra note 16, at 992-94 (opposing taxes on the wages of laborers),
999-1003 (opposing taxes on
n21. For a proposal along these lines, see Joseph Bankman
& Thomas Griffith, Social Welfare and the Rate Structure: A New Look at
75 Cal. L. Rev. 1905, 1950-55 (1987). Those who oppose progressivity via graduated rates, but support progres sivity
of a degressive sort, have long seen the potential for the latter to approach
the results of the former if exemption levels get high enough. See, e.g., Blum
& Kalven, supra note 6, at 513. They have erroneously assumed, however, that the
two could not converge, because they have ignored the government's
redistributive arsenals on the transfer side. See id. See also infra note 69.
n22. Frank Taussig, Principles of Economics 499 (1911), quoted in Blum
& Kalven, supra note 6, at 509 n.229.
n23. The familiar argument along these lines, going back at least to Adam Smith, is
that there is no point in taxing the poor, because (1) we can't get blood from
stone, and/or (2) we will only end up supporting the poor anyway through some
kind of welfare system. The second argument misses the boat entirely, since it
assumes an obligation to support the poor through affirmative transfers - an
obligation that most true libertarians would reject. The first argument has
some force, but does not explain our failure to exclude the
"dead beat" poor from those public benefits that are not true public goods, and hence from
which exclusion is possible, such as public schools and health care.
Hayek has sought to distinguish a degressive version of a progressive tax from
a grad uated rate version, on the ground that the former constitutes
"a majority's taxing itself to assist a minority," while the latter constitutes the much more egregious decision of a
"ma jority [to] impose upon a minority whatever burden it regards as right." Hayek, supra note 1, at 314. For similar sentiments, see sources quoted in
& Kalven, supra note 7, at 435. I take up this political process argument in
another context below. See notes 98- 101, infra, and accompanying text. For
now, I note only that the argument responds to those, like Hayek, who are most
concerned with class warfare qua classes, and is not an adequate response to
the more conventional concerns of libertarians that the rich are enti tled to
be protected from expropriation not only by the poor, but also by their fellow,
more beneficent rich, on behalf of the poor.
n24. See, for example, Takings, supra note 5, at 297, supporting a tax that takes
"a constant percentage of net income ... from the first dollar of net income to
the last." See also Robert Nozick, Anarchy, State and Utopia 169 (1974) (equating
taxation of wages with forced labor); Jeffrey A. Schoenblum, Tax Fairness or
Unfairness? A Consideration of the Philosophical Bases for Unequal Taxation of
12 Am. J. Tax Pol'y 221, 270- 71 (1995) (advocating a head tax as the only fair tax).
n25. Musgrave, supra note 17, at 62.
n26. See, e.g., Taxation, supra note 5, at 53. Both Nozick's
"liability rule" compensation scheme for private protective agencies banished by the
ultraminimal state, and his justifi cation of private appropriation out of the
commons whenever nonowners are better off with a private property regime than
they would be in a state of nature, rely on strict pareto compensation for
rights lost in the creation of the Lockean state. See Nozick, supra note 24, at
78-84, 174-77. For a similar argument justifying private appropriation of the
com mons with a liability rule of compensation, see Gauthier, supra note 14, at
n27. Epstein's position on this question is unclear, or at least inconsistent. He
invokes strict paretianism at the individual program level to bar
"the redistribution of income through the tax system." Taxation, supra note 5, at 68. At some points, he appears to revert implicitly
to a Kaldor-Hicks criterion to judge the acceptability of the remaining
programs. Takings, supra note 5, at 201. At others, however, he suggests he
will require strict paretianism for every government program. (See Taxation,
supra note 5, at 55-56, arguing that the system should
"try to insure that every public expenditure is worth more to every party taxed
than the revenues that are lost."). See also Richard Epstein, Ba rgaining with the State, 82-86 (1993). For a
similar insistence on strict paretianism at the level of every individual
program ("interaction"), see Gauthier, supra note 14, at 258. Given that no government expenditure
could pass this test, it seems unlikely that either of the foregoing authors
really means to require it.
n28. Hayek, supra note 1, at 316 (emphasis added).
"not clearly inappropriate assumption" is the best that Milton Friedman can do. Friedman, supra note 3, at 175.
n30. Some libertarians, such as Nozick and Hayek, appear to regard market prices as
sacrosanct not because such prices effect a just division of the surplus, but
because they are a necessary byproduct of freedoms independently worth
protecting. In Nozick's case, the relevant freedom is the freedom of the buyer
to give away her money to whomsoever she wishes. See Barbara Fried, Wilt
Chamberlain Revisited: Nozick's
"Justice in Transfer" and the Problem of Market-Based Distribution, 24 Philosophy and Public Affairs
226, 233 n.20 (1995). In the case of Hayek, it is the freedom of the seller of
goods or services to
"use his knowledge for his own purposes." Hayek, supra note 20, at 69.
Others have defended market-based distribution as just in itself, because it
rewards all individuals in accordance with their marginal product - that is,
with the value they have bestowed on others. For a discussion of John Bates
Clark, a prime expositor of this view, and his critics, see Barbara H. Fried,
The Progressive Assault on Laissez Faire 131-45 (1998).
n31. Gauthier's position on rents is at least confusing, if not confused. At one
point, he defines the rents that are up for grabs to include any excess paid to
a factor of production over the next best price offered - a measure that would
confiscate the incremental scarcity rents paid in moving from the second best
to best offer, but not those inherent in the second best offer. Gauthier, supra
note 14, at 272-73. At other times, he argues that everyone is entitled to, and
only to, their marginal product in a competitive market with no negative
externalities - a measure that would confiscate all scarcity rents in
noncompetitive mar kets but not inframarginal rents in competitive markets. Id.
at 90, 91-92, 95-98. Still else where, however, he suggests that everyone is
entitled only to the
"full cost of supply for the factor services she provides" - a standard, equivalent to reservation price, that would elimi nate a right
to inframarginal rents as well. Id. at 97-98. Without some clearer view of why
Gauthier believes that any portion of the market return is sacrosanct, it is
difficult to choose among these conflicting measures. The reasons Gauthier
suggests, however - some combination of optimization and a desire not to
distort choices among activities on liberta rian grounds (see id. at 93-94) -
push towards the last view.
n32. Dennis C. Mueller, Public Choice II 37-38 (1989).
n33. For a classic statement of problem, see Paul A. Samuelson, The Pure Theory of
Pub lic Expenditure, 36 Rev. Econ.
& Stat. 387, 389 (1954), reprinted in Kenneth J. Arrow
& Tibor Scitovsky, Readings in Welfare Economics 179, 182 (1969). See also
Mueller, supra note 32, at 123.
n34. For an overview of Lindahl's solution and the variants on it proposed by Bowen
and others, see Mueller, supra note 32, at 43-50; Anthony B. Atkinson
& Joseph E. Stiglitz, Lectures on Public Economics 487-89 (1980); and Musgrave,
supra note 17, at 74-80. For Samuelson's general equilibrium version of the
Lindahl solution, see Samuelson, supra note 33.
n35. The marginal price at any given quantity represents the marginal utility of
the last unit of that good at that quantity. In Lindahl's formulation, the bids
reflect the percentage of the cost at any specified quantity that each
individual is willing to pay. In Bowen's formulation, the bids reflect the
absolute dollars at a specified quantity that each individ ual is willing to
n36. Thus, while in private goods contexts we add up demand horizontally to
ascertain the total quantity demanded at a given price, in public goods
contexts we add up demand vertically to ascertain the total amount that all
individuals are willing to pay for a given quantity of public goods. This
difference simply reflects the difference in what we take as given in the two
contexts - price v. quantity.
n37. See Atkinson
& Stiglitz, supra note 34, at 509-12. The solution parallels the opti mal
solution for pricing jointly produced products, where the shared fixed costs of
produc tion are allocated between the products in proportion not to their
respective variable costs, but in proportion to respective demand. In the
public goods context, rather than facing two products with shared fixed costs,
we face one product with shared benefits. The solution is the same, however:
the costs of the product are allocated among consumers who share the benefits
in proportion to their demand schedule for the product.
n38. See T. Nicolaus Tideman, Ethical Foundations of the Demand-Revealing Process,
29 Public Choice 71, 74 (1977) ("if cost shares exactly equalled benefit shares ... redistribu tion would be
confined to that induced by the relatively arbitrary sharing of gains from
trade under Lindahl taxes").
n39. See Musgrave, supra note 17, at 80; Samuelson, supra note 33, at 888-89, 182:
"it is in the selfish interest of each person to give false signals, to pretend
to have less interest in a given collective consumption activity than he really
has." For a summary of the other reasons to anticipate that Lindahl's strategy would
not achieve an optimal solution, see Musgrave, supra note 17, at 79.
n40. Most of these demand-revealing solutions are traceable back to Theodore
Groves, Incentives in Teams, 41 Econometrica 617 (1973).
n41. In some models, the supplementary tax is not paid in dollars at all, but
instead in scarce voting points; in others, it is paid in dollars, but the
dollars are wasted, or at least not returned in any systematic fashion to the
individual voters (if they were, that would undermine motives for honest
n42. See Mueller, supra note 32, at 145.
n43. For such a suggestion, see Tideman, supra note 38, at 74. The logic here is
that, while taxpayers might still have a mild incentive to depress their income
levels to avoid Lindahl taxes, they have no incentive to lie about their true
preferences, since their indi vidual, truthful revelation has virtually no
effect on aggregate output, and hence on the absolute amount of the Lindahl
share their income class will bear.
n44. Epstein's arguments in Takings (supra note 5) are prefigured in an earlier
piece, Taxation in a Lockean World (supra note 5).
n45. See supra note 28 and accompanying text.
n46. Barbara Fried, Why Do Libertarians Love Proportionate Taxation? Epstein's Tale
of Two Pies and Other Puzzles (draft manuscript) (copy on file with Chapman Law
Review). Gauthier's convoluted theory of just distribution is not easy to
summarize. In brief, he argues that the surplus value from society ought to be
divided among all the cooperators who produced it in accordance with his
"minimax relative concession" (MRC). MRC requires that surplus be divided among members of a society so as
to equate the ratio of (1) the total amount of surplus that each person
actually gets from any cooperative ven ture in which he is a participant, to
(2) the largest amount of surplus he could have plausi bly claimed from that
cooperative venture. Gauthier's verbal formulation is different, but amounts
mathematically to the same thing. Gauthier, supra note 14, at 136-40. The larg
est amount he could have plausibly claimed, in turn, is the amount left over
after fellow cooperators' costs are covered; that is, after cooperators are
left no worse off than under noncooperation. Id. at 137-39. As Gauthier
ultimately acknowledges, this formula dictates an equal division of surplus
among all participants in each cooperative venture, assuming each makes the
largest plausible claim he could. Id. at 151-54, 277. In Gauthier's view, this
equal division may be deviated from, under a Rawlsian minimax principle, where
all will be benefitted by the deviation. What it implies about the division of
surplus at the societal level is much less clear. Gauthier argues that it
logically leads to proportionate taxation. For reasons I have explored
elsewhere, that argument must rest on assumptions about the distributive
effects of pretax prices that are wildly implausible on their face, and
impossible to reconcile with other aspects of Gauthier's argument. See id.
In the end, proportionate taxation is such an implausible expression of
Gauthier's view of distributive justice that his enthusiasm for it, like
Rawls's, is interesting chiefly as an illustration of the strength of the
irrational pull proportionality exerts on the imagination.
n47. Takings, supra note 5, at 163-64.
n48. Id. at 162.
n49. Id. at 3-5, 163-64. See also Taxation, supra note 5, at 52-53,
"noting that the base line against which 'being better off' is measured gives to
each person only his natural rights to liberty and property, and not the fruits
acquired by theft, aggression, or deception." For Gauthier's similar conclusion, see Gauthier, supra note 14, at 254,
arguing that each share in the joint gains from society should be
"proportioned to the contributions of its recipients."
n50. Epstein's idiosyncratic constitutional argument for dividing surplus in
proportion to rights in a state of nature, in brief, is as follows: (1) the
just compensation clause ex tends to taxation as well as conventional takings;
"public use" requirement in the just compensation clause mandates proportionality between
the benefits and burdens of any public taking; and (3) the requirement of
proportionality mandates that all surplus value created by the state be
distributed in accordance with the value of assets people bring into the state.
His utilitarian argument for pro rata division of the surplus, in brief, is
that such division
"does not skew the incentives of private parties in the choice between public
and private control over human affairs. For example, if each person received an
equal por tion of the general gain, there would be an incentive for persons
with smaller shares to force matters into the public arena, where they would be
relative gainers." Takings, supra note 5, at 163. The utilitarian argument fails for the same
reasons that the Lockean one does: the private market does not reward
individuals in proportion to the value of their rights in the state of nature,
and such pro rata division of the surplus would not, in any event, give rise to
a proportionate tax.
n52. Id. For those familiar with Takings, that conclusion is presented in the pie
chart of two concentric circles that opens the book. Id. at 3-6. The view that
the surplus value created by a Hobbesian state ought to be divided pro rata in
accordance with individuals' SON holdings is fundamental to Epstein's view of a
just society. See id. at 3, 5:
"The im plicit normative limit upon the use of political power is that it should
preserve the relative entitlements among the members of the group, both in the
formation of the social order and in its ongoing operation.... Each gain from
public action therefore is uniquely assigned to some individual, so that none
is left to the state, transcending its citizens." Epstein asserts it largely as fiat in Takings. In Bargaining with the State,
he defends it with a number of arguments, including that a vastly unequal
division of surplus will be an unstable solution and a pro rata division a
stable one, an argument reminiscent of Gauthier's game-theoretic arguments for
pro rata division; that it minimizes the chance that some party will be left in
a pareto-inferior position by the taxes-for-civilization deal, and that it
"leads to the automatic proration of the surplus among all contenders." Epstein, supra note 27, at 90-91, 95-97. The first two arguments rest on
empirical premises that seem doubtful, or at the very least in need of some
Some notion of equal proportionate division of surplus is fundamental to
Gauthier's view of a just society as well, although the precise notion is less
clear. As noted below, Gauthier suggests in passing the standard solution by
analogy to partnerships - returns- in-proportion-to-investment - proposed by
Epstein and others. But the version he develops in depth in Morals by
Agreement, under the rubric
"minimax relative concession" (MRC), rejects that model entirely. For further discussion of these matters,
see Fried, supra note 46.
n53. See Smith, supra note 16, at 945:
"The subjects of every state ought to contribute towards the support of the
government ... in proportion to the revenue which they respec tively enjoy
under the protection of the state. The expense of government to the individuals
of a great nation, is like the expence [sic] of management to the joint tenants
of a great estate, who are all obliged to contribute in proportion to their
respective interests in the estate." Among contemporary libertarians, see, for example, Gauthier, supra note 14, at
140-41, 152, 254.
"The ratio between the benefit [each member of society] receives and the
contribution she makes is, so far as possible, constant, the same for all." Id. at 152. Gauthier's actual scheme for dividing the surplus value generated
by society, however, abandons that principle entirely. See supra note 46.
n54. For these purposes, I accept the conventional libertarian premise that
individuals are entitled to what they have acquired by Lockean principles in
the SON. To the extent one rejects that premise, because, for example, one does
not regard as just the natural distribution of talents, or the distribution of
resources by grab in the SON, SON entitle ments lack justificatory
significance, and as a consequence, arguments for distributing sur plus pro
rata in accordance with those entitlements will lack justificatory weight as
n55. Takings, supra note 5, at 5.
n56. Id. at 3.
n57. See, e.g., Rawls, supra note 4, at 520-29; Kenneth J. Arrow, Social Choice and
Justice 188 (1983).
n58. This position is, of course, just a variant of the point made above,
concerning the level of aggregation at which to test the pareto-superiority of
n59. Nozick, supra note 24, at 192-93. For a similar argument implicit in his
parable of the greens and the purples, see Gauthier, supra note 14, at 284-86.
n60. For a discussion of this
"stand alone property," see Lewis Kornhauser, Fair Division of Settlements: A Comment on Silver and
84 Va. L. Rev. 1561, 1568-72 (1998); Herve Moulin, Axioms of Cooperative Decisionmaking, 89-95 (1988). I am
indebted to Lewis Kornhauser for pointing out the relevance of the standard
allocation-of-common- costs analysis to the problem at hand.
n61. See supra note 59 and accompanying text.
n62. Given the inherent administrative limitations of an income tax, which does not
tax leisure, the rich do of course have a less costly means of exit available
to them if tax rates get high enough: to trade off productive labor for
(untaxed) leisure. As a practical matter, that form of exit is likely to
constrain political choice far more than the more extreme sort of exit
n63. Arrow, supra note 57, at 188.
n65. For the thoughtful argument that there are, in a deep sense, such increasing
re turns to scale from organized society, see Arrow, supra note 58, at 188. I
am indebted to Marvin Chirelstein for bringing this passage to my attention.
n66. For a proof of the latter proposition in the context of joint litigation
costs, see Lewis Kornhauser, Control of Conflicts of Interest in Class Action
Suits, 41 Public Choice 145 (1983).
n67. See Musgrave, supra note 17, at 90-115.
n68. Knut Wicksell, Finanztheoretische Untersuchungen und das Steuerwesen
Schweden's 143 (Jena 1896) quoted in translation in Musgrave, supra note 17, at
72. See also Edgeworth's and Pigou's observations to the same end, both cited
in Musgrave, supra note 17, at 111; discussion in Fried, supra note 30, at
n69. The potential for undoing any tax distribution on the transfer side has not
been lost on the right, although proponents of proportionality on libertarian
grounds probably have underestimated the extent to which it undercuts any
fairness arguments for proportional tax rates judged in isolation. See, e.g.,
Hayek, supra note 1, at 307 (recognizing the poten tial for effecting radical
redistribution through use of proportionate tax and differential transfers to
the lower classes, but assuming that potential is limited in practice by the
inflexibility of in-kind transfers as a tool for income redistribution). Of
course, once one admits the possibility of straight cash transfers, that
problem goes away.
n70. See Bankman
& Griffith, supra note 21, at 1955-58.
n71. For a very suggestive treatment of these issues, see Robert H. Frank and
Philip J. Cook, The Winner-Take-All Society (1995), and Robert H. Frank,
Progressive Taxation and the Incentive Problem, Working Paper no. 98-4 (Off.
Tax Pol'y Res., U. of Mich. Bus. Sch., 1998). For a much earlier treatment of
the relative status problem along the same lines in a slightly different
context, see A.C. Pigou, A Study in Public Finance 90 (3d rev. ed. 1947).
n72. Rawls, supra note 4, at 278-79.
n73. Following Musgrave, Rawls divides the government into four functions: the
alloca tive, stabilization, transfer, and distribution branches. For our
purposes, only the latter two are relevant. The transfer branch Rawls defines
as concerned with guaranteeing a social minimum, provided either piecemeal by
family allowances, special payments for sick ness, and employment, or more
systematically by a negative income tax. Rawls, supra note 4, at 275. He
subdivides the distribution branch into two functions: (1) regulating
intergenerational transfers, through gift and estate taxes and restrictions on
rights of be quests, in order to
"prevent concentrations of power detrimental to the fair value of political
liberty and fair equality of opportunity," id. at 277; and (2) raising tax revenues necessary to
"provide for the public goods and make the transfer payments necessary to
satisfy the difference principle," id. at 278.
There are a number of peculiarities here. First, it would seem that the
guaranteed minimum income that is the focus of the transfer branch should be
subsumed under trans fers necessary to satisfy the difference principle, the
task of the second leg of the distribu tion branch. Rawls himself seems to
concede as much, in stating that
"once the difference principle is accepted, however, it follows that the minimum
is to be set at that point which, taking wages into account, maximizes the
expectations of the least advantaged group." Id. at 285.
Second, it is not clear why the first and second legs of the distribution
branch should be treated as distinct. Why single out intergenerational
transfers as a unique violation of the just state that need to be dealt with
through their own institutions, rather than simply treating them as one of many
possible sources of inequality that all require governmental correction in
accordance with the difference principle? One customary reason for distin
guishing intergenerational gratuitous transfers from other sources of wealth,
such as labor income and returns to one's own lifetime savings, is because the
former is regarded as uniquely
"unearned" and hence undeserved. See Fried, supra note 30, at 97-99; Eric Rakowski, Equal
Justice 158-62 (1991). Whatever the problems with this view - and they have
been much debated over the centuries - the view is clearly not one congenial to
Rawls. Rawls, consistent with his broad view of undeserved privilege,
explicitly states that
"the unequal inheritance of wealth is no more inherently unjust than the unequal
inheritance of intelligence," and that
"as far as possible inequalities founded on either should satisfy the difference
principle." Rawls, supra note 4, at 278. This statement suggests that wealth from
inheritance should be treated like wealth from any other source, presumably all
under the second distribution branch. The second danger that Rawls suggests is
posed by inher ited wealth--that when it creates inequalities in wealth that
exceed a certain limit, it threatens
"the fair value of political liberty and fair equality of opportunity" in society, id. at 277--likewise suggests no ground for separating inherited
wealth from any other source of wealth. If the concentration of wealth is what
matters, the source should not.
Finally, Rawls's hostility to intergenerational transfers is hard to reconcile
with the solicitude he shows to savers in supporting a consumption tax in place
of an income tax. The salient difference between a consumption tax and an
income tax is the treatment of savings: a consumption tax exempts from taxation
at least some forms of capital income currently taxed under our income tax. See
& Barbara H. Fried, Winners and Losers in the Shift to a Consumption Tax,
86 Geo. L.J. 539, 540-42 (1998). In defense of a consumption tax base, Rawls appeals to the
"common sense precepts of justice," pre cisely the ones that appealed to Hobbes some three hundred years ago, that
"it imposes a levy according to how much a person takes out of the common store
of goods and not accord ing to how much he contributes (assuming here that
income is fairly earned)." Rawls, supra note 4, at 278. Whatever the merits of this position, the
statement is hard to recon cile with Rawls's position on intergenerational
transfers. Given that a significant portion of lifetime savings, at least among
the wealthy, is destined for intergenerational transfer through inter vivos
gifts and bequests, the choice to protect the return to savings, via a
consumption tax, during a taxpayer's lifetime, while burdening the same
decision heavily upon death, seems to require some justification.
n74. Rawls, supra note 4, at 278-79.
n75. Id. at 278.
n76. See supra note 69 and accompanying text.
n77. Rawls, supra note 4, at 280.
n78. Id. at 279.
n79. Or so I infer from his repeated insistence on this qualification on his
support for a flat-rate consumption tax in the preceding passage. Id. at
278-79. Rawls doesn't state ex plicitly what injustices he has in mind here.
n81. Epstein, for example, treats all other fundamental choices in tax system,
including the choice of a taxable base, double taxation at the corporate level,
the realization require ment, the home mortgage interest deduction, and the
preferential treatment of capital gains, as normatively discretionary calls, on
the empirical assumption that the resolution of each of these questions will
have little systematic impact on the distribution of surplus. Takings, supra
note 5, at 300-02. That assumption is almost certainly wrong with respect to
many of these items, given the highly skewed distribution of capital income in
this coun try. The only other tax policy choice attracting constitutional
scrutiny from Epstein is in dustry-specific tax breaks like percentage
depletion for oil and gas.
Others, of course, notably including Rawls, have argued that the choice of tax
base (consumption versus income) raises fundamental issues of fairness. That
argument is con spicuously absent, however, from most contemporary defenses of
the so-called Hall/Rabushka flat tax championed by Steve Forbes and adopted as
the model for the Armey/Shelby plan now pending before Congress. The
Hall/Rabushka plan, along with most so- called flat tax schemes, adopts a
consumption tax base in place of an income tax base, in addition to shifting
from a graduated to a degressive rate structure. Notwithstanding that fact, as
Larry Zelenak argues at length in his piece for this symposium, most proponents
of the plans on fairness grounds are conspicuously silent about the former
change. Lawrence Zelenak, The Selling of the Flat Tax: The Dubious Link Between
Base and Rate,
2 Chap. L. Rev. 197 (1999). Hall and Rabushka themselves, for example, completely ignore the con sumption
tax feature of their proposal in their 30-page defense of the fairness of their
flat tax. Hall
& Rabushka, supra note 7, at 23-51.
& Kalven, supra note 6, at 517.
n83. Id. at 460.
n84. For typical expressions of that view, see Taxation, supra note 5, at 73-74
(where the benefits and other costs of government are hard to assess,
"a test of equal treatment across taxpayers becomes the next best alternative"); Blum
& Kalven, supra note 6, at 460 (if taxes are a necessary evil falling on a
distribution of income otherwise just, the object is
"to leave all taxpayers equally 'worse off' after taxes").
n85. See Blum
& Kalven, supra note 6, at 455-65; Musgrave, supra note 17, at 90-98; Fried,
supra note 30, at 153-54.
n86. The third measure proposed by Mill, equimarginal sacrifice, was rightly
dismissed by friends and foes alike as nothing more than a stand-in for
utilitarian concerns. See Fried, supra note 30, at 154-55.
n87. See Musgrave, supra note 17, at 99-100; Blum
& Kalven, supra note 6, at 458-59. David Gauthier, to take one recent example,
assumes that the measure (which he supports in a convoluted fashion) would
result in a flat-rate tax. Gauthier, supra note 14, at 272. There is clearly no
more empirical, or even strong intuitive, support for that conclusion, however,
than for the alternative Friedmanesque assumption, which Gauthier rejects, that
the quantity of public goods people consume is proportionate to their income.
Id. at 271.
n88. See, e.g., Antonio de Viti de Marco, cited in Musgrave, supra note 17, at 73;
Fried, supra note 30, at 154-55, 301 nn.261-62. For a more recent defense of
utility as a measure of sacrifice, see Gauthier, supra note 14, at 271-72.
n89. Gauthier attempts to justify a utility measure of sacrifice as not
redistributive in intent here by reference to his notion of
"minimax relative concession" (MRC). Gauthier, supra note 14, at 271-72, and note 46, supra. In brief, MRC
requires an equal division of the joint gains from cooperation, including gains
from the cooperative provision of public goods. Gauthier argues that a head tax
will violate the requirements of MRC, because under a head tax
"a person with fewer resources will lose a greater proportion of the gain he
would realize were he to obtain the good at no [tax] cost." Id. at 272. Presumably, this result follows in Gauthier's mind from the
declining marginal utility of income: since dol lars are worth more to the
poor, a head tax strips from them more absolute utility, and hence a greater
absolute share of the surplus. But if the utility of tax dollars given up to
purchase public goods is greater, so also, presumptively, is the utility of
public goods re tained, net of taxes (that is, the poor's share of the surplus
from cooperation less their share of the tax costs). If so, Gauthier's
requirement that surplus be divided equally, in utility terms, seems more
plausibly to lead to a steeply regressive tax than a flat tax. At any rate, it
requires some odd assumptions about the shape of marginal utility curves over
the rele vant ranges to conclude otherwise. In the end, Gauthier's inclination
to measure gain by utility, like his inclination to equalize the division of
surplus under MRC, seems hard to explain except as an expression of egalitarian
impulses that are uneasily reconciled with the libertarian premises of his
& Kalven, supra note 6, at 460.
& Rabushka, supra note 7, at 27.
n93. Hayek, supra note 1, at 313.
n94. Id. at 314-15.
& Kalven, supra note 6, at 470-71.
n96. See Taxation, supra note 5, at 74; Gauthier, supra note 14, at 272-73.
n97. Taxation, supra note 5, at 55. For similar comments from Gauthier, see
Gauthier, supra note 14, at 272-76.
"If tax neutrality [in this sense] could be perfectly achieved, the laws would
act as a prism which magnified equally all preexisting endowments. The nature
of private activi ties would not change, nor would the relative endowments of
private persons." Taxation, supra note 5, at 56. The former conclusion is true; the latter is
n99. See supra note 70 and accompanying text.
n100. Hayek, supra note 1, at 314. For similar sentiments, see sources cited in Blum
& Kalven, supra note 6, at 435 n.60; Taxation, supra note 5, at 70.
n101. Hayek, supra note 1, at 314. Epstein trumpets the same property of
proportionality for a different reason, arguing that by limiting political
choices to one tax rate that applies to all, proportionality reduces the
opportunities for, and hence transactions costs of, fac tional fighting.
Taxation, supra note 5, at 53, 57.
n102. Taxation, supra note 5, at 70.
n104. See Bankman
& Griffith, supra note 21, at 1914. The certainty of a proportionate rate
structure is a recurrent argument offered in its favor. See, e.g., Hayek, supra
note 1, at 313-14; Blum
& Kalven, supra note 6, at 430-35, 511; Lutz, supra note 15, at 70, 73-76.
n105. Hayek, supra note 1, at 323.
& Kalven, supra note 6, at 419.
n108. Taxation, supra note 5, at 68:
"(I ignore here capitation taxes that call for all per sons to pay a fixed
amount of taxes, regardless of income.)."
& Kalven, supra note 6, at 419.
n110. Id. at 420.