Journal of Public Economics 92 (2008) 17771786

Contents lists available at ScienceDirect

Journal of Public Economics

j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / e c o n b a s e

Should public policy respond to positional externalities?

Robert H. Frank
Johnson Graduate School of Management, Cornell University, Ithaca, NY 14853 USA

a r t i c l e i n f o a b s t r a c t

Article history: A nice suit is one that compares favorably with those worn by others in the same local
Received 31 March 2007 environment. More generally, a positional good is one whose utility depends strongly on how it
Received in revised form 14 February 2008 compares with others in the same category. 1
Accepted 6 March 2008
A positional externality occurs when new purchases alter the relevant context within which an
Available online 14 March 2008
existing positional good is evaluated. 2 For example, if some job candidates begin wearing
expensive custom-tailored suits, a side effect of their action is that other candidates become less
Keywords:
likely to make favorable impressions on interviewers. From any individual job seeker's point of
Relative income
view, the best response might be to match the higher expenditures of others, lest her chances of
Positional externalities
Happiness landing the job fall. But this outcome may be inefcient, since when all spend more, each
Consumption taxation candidate's probability of success remains unchanged. All may agree that some form of
collective restraint on expenditure would be useful.
In such cases, however, it is often impractical to negotiate private solutions. Do positional
externalities then become legitimate objects of public policy concern? In attempting to answer
this question, I employ the classical libertarian criterion put forth by John Stuart Mill 3, who
wrote the state may not legitimately constrain any citizen's freedom of action except to prevent
harm to others. I argue that many positional externalities appear to meet Mill's test, causing not
just negative feelings but also large and tangible economic costs to others who are ill-equipped
to avoid them. I also discuss an unintrusive policy remedy for positional externalities, one
modeled after the use of efuent charges to curb environmental pollution.
The paper is organized as follows. Section 1 notes the deep similarity between the conditions
that give rise to positional arms races and those that give rise to conventional military arms
races. Section 2 follows with a review of evidence concerning the strength of concerns about
relative position. Section 3 describes some of the tangible economic costs that people
experience as a result of positional externalities arising from such concerns. Section 4 takes up
the question of whether collective action directed against positional externalities is consistent
with respect for individual rights. Section 5 describes how a progressive consumption tax could
neutralize many of the most costly effects of positional externalities.
2008 Elsevier B.V. All rights reserved.

1. Positional arms races

Before addressing whether the harm caused by positional externalities should be mitigated by legislation, it is useful to review
the conditions under which such externalities can be expected to cause harm in the rst place. Those conditions are precisely
analogous to the ones that make military arms races between equally matched rival nations wasteful.

E-mail address: rhf3@cornell.edu.

1
[Hirsch, Fred. Social Limits to Growth, Cambridge, Mass.: Harvard University Press, 1976.].
2
[Frank, Robert H. Positional externalities, in Richard Zeckhauser, ed., Strategy and Choice: Essays in Honor of Thomas C. Schelling, Cambridge, MA: MIT Press,
1991: 2547.].
3
[Mill, John Stuart, 1982. On Liberty. Viking, New York.].

0047-2727/$ see front matter 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.jpubeco.2008.03.001
1778 R.H. Frank / Journal of Public Economics 92 (2008) 17771786

Consider two such nations, each of which must apportion its limited resources between the purchase of arms and non-military
goods. The utility from expenditures on weapons depends heavily on the relative stocks of armaments in the two nations. If one
had a signicantly smaller stock than its rival, its political independence would be jeopardized. And since estimates of a rival's arms
stocks are inherently imperfect, each nation has an incentive to err on the high side in its own military spending, thereby to provide
a margin of safety.
Under the circumstances, we expect equilibrium expenditures on armaments to be inefciently high from the collective
vantage point. That is, we expect that citizens of each nation would count themselves better off if both nations spent less on arms
and more on other goods. A necessary and sufcient condition for this inefciency is that relative position matter more for
armaments than for other goodsa condition that surely holds in practice. Thus, although the utility that people derive from the
consumption of non-military goods may depend on the corresponding level of consumption in other countries, that dependence is
much weaker than for armaments. Cross-national consumption comparisons tend to be less important in part because most people
rarely travel abroad, but mainly because the consequences of having fewer toasters and televisions are less severe than the
consequences of being less well armed.
If the dependence of utility on context were equally strong for both military and non-military goods, there would be no military
arms race. One nation might try to gain military advantage by building more weapons but would quickly discover that the cost of
doing so was to fall behind in non-military consumption. If the cost of the latter were as high as the benet of the former, there
would be no incentive to build too many weapons.
But because context matters more for armaments domain than for non-military consumption, an inefcient military arms race
is the inevitable result. If technology permits the monitoring of a rival's armaments, each nation can achieve a superior outcome by
entering into an agreement that limits spending on arms.
I sketch this argument in detail to emphasize that each element of it is completely uncontroversial. No libertarian would think to
object to a military arms control agreement on the grounds that it limited each side's freedom to spend as much as it pleased on arms.
Since that was precisely the objective each sought by entering into the agreement, such an objection would be absurd on its face.
Of course, there are many other reasons that might justify rejecting an arms control agreement. One might fear, for example, that
the verication technology is incapable of preventing rivals from cheating. Or an asymmetry in economic position might lead one side
to prefer an arms race on the grounds that it could better withstand the resulting nancial burden than its rival. But for equally matched
nations equipped with effective monitoring technologies, the mutual benets of military arms control agreements are indisputable.
The issues are essentially the same with positional arms races. To help dene terms, I pose two simple thought experiments. In
each case, you must choose between two worlds that are identical in every respect except one. The rst choice is between world A,
in which you and your neighbors will live in 4000-square-foot houses while residents of other neighborhoods live in 6000-square-
foot houses; and world B, in which you and your neighbors live in 3000-square-foot houses, others in 2000-square-foot houses.
Once you choose, your position on the local housing scale will persist.
According to the standard neoclassical economic model of choice, which holds that utility depends only on the absolute amount
of consumption, the uniquely correct choice is world A. For if absolute house size is all that matters, A is indeed a better world for
all, since everyone has a larger house there than the largest house in world B.
In fact, however, most people say they would pick B, where their absolute house size is smaller and their relative house size is
larger. Even those who say they would pick A seem to recognize why someone might be more satised with a 3000-square-foot
house in B than with a substantially larger one in A.
In the second thought experiment, your choice is between world C, in which your annual probability of dying from workplace
injuries would be 2 in 100,000, the corresponding probability for others 1 in 100,000; and world D, in which those probabilities
would be 4 in 100,000 for you and 8 in 100,000 for others. This time most people pick C, choosing greater absolute safety at the
expense of lower relative safety.
As I use the terms, positional goods denote goods for which the link between context and evaluation is strongest, and
nonpositional goods denote those for which this link is weakest. In terms of the two thought experiments, housing is thus a
positional good, safety a nonpositional good. The point is not that absolute house size and relative safety are of no concern. Rather,
it is that positional concerns weigh more heavily in the rst domain than in the second.
For the moment, let us take the modal responses to the two thought experiments at face value. That is, let us suppose that
although context matters for evaluations in both domains, it matters more for housing. This implies that the opportunity to
exchange riskier working conditions for higher wages will seem more attractive to individual workers than to society as a whole.
By accepting a riskier job at higher pay, each worker would rationally expect two sources of reward: The utility from having a house
that is larger in absolute terms and the utility from having a relatively larger house. From the collective vantage point, however, the
second source of reward is illusory, for if others also accept riskier jobs to buy larger houses, the distribution of relative house size
remains unchanged.4 (Alternatively, if others stand pat, the one individual's gain from having a relatively larger house will be offset
by the losses of those whose houses are now relatively smaller.)
Assuming that all workers accept riskier jobs for higher pay, people discover when the dust settles that the gain in absolute
house size alone was insufcient to compensate for the additional risk exposure required to get it. Yet failure to buy a larger house
when others do is not an attractive option for the individual, either. As in the familiar stadium metaphor, all stand to get a better
view, but when all stand no one sees better than when all were seated.

4
For a more formal treatment of this point, see Frank, 1985b.
 R.H. Frank / Journal of Public Economics 92 (2008) 17771786 1779

Fig. 1. Average happiness vs. average income over time in Japan. Source: R. Veenhoven, Happiness in Nations, International Monetary Fund, 1993.

From these simple examples, it should be clear positional arms races can lead to welfare losses for reasons precisely analogous to
those we saw in the case of military arms races. I hope it is also clear that if the affected parties had both the means and the desire
to implement an agreement that called for spending more on safety and less on housing, it would be absurd for a libertarian to
object in principle.

2. Concerns about relative position

Most of us were taught from an early age not to worry about how our possessions compare with those of others. This sensible
advice stems from the observation that since there will always be others with more and better things, focusing closely on such
comparisons cannot help but generate reasons to feel unhappy.
Yet virtually all consumption occurs in a social context, often multiple contexts, and the plain fact is that almost everyone's
choices are inuenced to at least some degree by the spending of others. Many economists appear reluctant to take seriously the
concerns that might lead people to choose the relatively large, but absolutely small, house in the thought experiment discussed
earlier. On its face, this is a curious position for a profession whose practitioners are quick to endorse Jeremy Bentham's dictum that
a taste for poetry is no better than a taste for pushpins. If most people say they would prefer World B, a genuine commitment to
consumer sovereignty would appear to rule out any categorical claim that World A is necessarily best for all.
In any event, the choice pattern in the thought experiment is consistent with the hypothesis that, in some domains, people
appear willing to sacrice absolute consumption for improved relative position. The same hypothesis has received strong support
in the empirical literature on human happiness and well-being.5
For several decades, behavioral scientists, mainly psychologists, have been actively attempting to measure human subjective
well-being. They employ a variety of instruments, ranging from simple surveys that ask people to report how happy they are to
sophisticated electroencephalographic assays that measure the frequency and amplitude of the electrical waves emanating from
different sites in the brain. Individual readings from the various measures turn out to be remarkably consistent with one another,
and consistent as well with a variety of other evidence regarding human well-being. People who say they are happy or who are
revealed as happy by the various other measures, for example, are also more likely to be rated as happy by their friends. And all of
the happiness measures are strongly correlated with observable behaviors that we associate with well-being. Those who identify
themselves as happy in surveys, for example, are more likely to initiate social contacts with friends and more likely to help others in
need. They are less likely to suffer from psychosomatic illnessesdigestive disorders, headaches, vascular stress, and other stress
disorders. They are less likely to be absent from work, or to get involved in disputes at work. And they are less likely to attempt
suicidethe ultimate behavioral measure of unhappiness. In brief, the happiness literature has identied measures of human
subjective well-being that are consistent, valid, and reliable.6
What factors inuence happiness? Much of the interpersonal variation in happiness appears hereditary, but environmental
factors also seem to matter. People who have many close friends, for example, tend to be signicantly happier than others, and also
to live longer.
What about income? Studies show that when incomes rise for everybody, measures of well-being do not change much.
Consider the example of Japan, which was a very poor country in 1960. In the ensuing decades, its per-capita income rose many-
fold, yet the average happiness level reported by the Japanese remained essentially unchanged (See Fig. 1).
The pattern shown in Fig. 1, which consistently shows up in other countries as well,7 poses an apparent challenge for
conventional economic models. If getting more income does not make people happier, why do they go to such lengths to get more
income? Why, for example, do legal associates work 100 h a week hoping to become partners in law rms? Why do tobacco
company CEOs endure the public humiliation of testifying before Congress that nicotine is not addictive?

5
For comprehensive surveys of this literature, see Layard, 2005; and Kahneman et al., 1999.
6
For a brief survey of the studies that support these claims, see Frank, 1999, chapter 5.
7
For a survey of more recent data, see Layard, 2005.
1780 R.H. Frank / Journal of Public Economics 92 (2008) 17771786

Fig. 2. Income vs. satisfaction in the US, 19814. Source: Diener, Ed, Ed Sandvik, Larry Seidlitz, and Marissa Diener. The Relationship Between Income and
Subjective Well-Being: Relative or Absolute? Social Indicators Research, 28, 1993: 195223.

It turns out that if we measure the incomehappiness relationship in a second way, income matters very much indeed. Consider
Fig. 2, which shows this relationship for the United States during a brief period during the 1980s. When we plot average happiness
vs. average income for clusters of people in a given country at a given time, as in the diagram, rich people are in fact substantially
happier than poor people.
The patterns portrayed in Figs. 1 and 2 are consistent with the view that relative income is a far better predictor of
happiness than absolute income.8 Kahneman et al. (2006) argue that traditional happiness surveys may overstate the
inuence of income on happiness, because people may focus on their material well-being in attempting to answer the
survey question. They show that income's inuence is smaller when the dependent variable is reported mood during
experience sampling rather than the response to a survey question. But even on that alternative measure, people with
annual incomes under $20,000 reported being in a bad mood more than 50% more often than people with incomes over
$100,000.
That relative income matters is often seen as a regrettable human frailty. Yet, as Adam Smith argued in The Wealth of Nations,
community consumption standards inevitably dictate which goods are viewed as essential, or necessaries:
By necessaries I understand not only the commodities which are indispensably necessary for the support of life, but
whatever the custom of the country renders it indecent for creditable people, even of the lowest order, to be without. A
linen shirt, for example, is, strictly speaking, not a necessary of life. The Greeks and Romans lived, I suppose, very
comfortably though they had no linen. But in the present times, through the greater part of Europe, a creditable day-
labourer would be ashamed to appear in public without a linen shirt, the want of which would be supposed to denote that
disgraceful degree of poverty which, it is presumed, nobody can well fall into without extreme bad conduct. Custom, in the
same manner, has rendered leather shoes a necessary of life in England. The poorest creditable person of either sex would
be ashamed to appear in public without them.9

The same issues play out in contemporary societies, the only difference being that the standards that now dene necessity have
escalated sharply since Adam Smith's day. In the late 1990s, for example, the New York Times's Chicago bureau chief, Dirk Johnson,
led a detailed report after having spent several days with the Williams family in Dixon, Illinois, a wealthy community west of
Chicago. Although the family's annual income was slightly above the Federal poverty threshold, they lived in a trailer park just
outside the city limits and were conspicuously disadvantaged by local standards. Some excerpts from Johnson's account of several
days spent with Wendy Williams, the family's 13-year-old daughter:
Watching classmates strut past in designer clothes, Wendy Williams sat silently on the yellow school bus, wearing a cheap
belt and rummage-sale slacks. One boy stopped and yanked his thumb, demanding her seat.

Move it, trailer girl, he sneered.

8
On this point, see especially Easterlin, 1974, 1995, and Clark and Oswald, 1996.
9
Chapter 2, Part II, Article 4. [http://www.online-literature.com/view.php/wealth_nations/34?term=by%20necessaries%20i%20understand] (Smith, 1937).
 R.H. Frank / Journal of Public Economics 92 (2008) 17771786 1781

It has never been easy to live on the wrong side of the tracks. But in the economically robust 1990's, with sprawling new
houses and three-car garages sprouting like cornstalks on the Midwestern prairie, the sting that comes with scarcity gets
rubbed with an extra bit of salt.

To be without money, in so many ways, is to be left out.

I told this girl: That's a really awesome shirt. Where did you get it? said Wendy, explaining that she knew it was out of
her price range, but that she wanted to join the small talk. And she looked at me and laughed and said, Why would you
want to know?

A lanky, soft-spoken girl with large brown eyes, Wendy pursed her lips to hide a slight overbite that got her the nickname
Rabbit, a humiliation she once begged her mother and father to avoid by sending her to an orthodontist.

For struggling parents, keenly aware that adolescents agonize over the social pecking order, the styles of the moment and
the face in the mirror, there is no small sense of failure in telling a child that she cannot have what her classmates take for
granted.

Do you know what it's like? asked Wendy's mother, Veronica Williams, to have your daughter come home and say,
Mom, the kids say my clothes are tacky, and then walk off with her head hanging low.10

By the standards of Adam Smith's day, the absolute quality of Wendy Williams' clothing would not have been objectionable. Nor
would her overbite have been viewed as a marker of social position, because not even the children of the wealthy had access to the
kinds of cosmetic dentistry we now take for granted. Although the best response available to someone in Wendy Williams's
position may be to attempt to ignore the taunts of her classmates, such attempts are rarely completely successful. The
psychological injuries suffered may be diminished, but not completely eliminated. I will return below to the question of what
status such injuries ought to have in public policy analysis.

3. Positional externalities impose tangible economic costs

Failure to meet community consumption standards has adverse psychological consequences, but it also entails more
tangible losses, even for families who are completely unconcerned about relative consumption per se. Consider, for example, a
middle-income family's decision about how much to spend on a house. The cost of sending a child to a school of average quality
is closely linked to the price of the average house in the community. In the United States this is true in part because of the direct
link between local school budgets and local property tax revenues. But in the light of evidence that any given student's
achievement level rises with the average socio-economic status of his or her classmates, property values and school quality will
be positively linked even in countries in which school budgets are largely independent of local property values.11 The upshot is
that individual families can send their children to better schools if they are willing to stretch their budgets and spend a little
more on housing.
We may safely assume that most middle-income families aspire to send their children to schools of at least average
quality. Indeed, parents who felt completely at ease with the prospect of their children attending below-average schools
would be judged harshly in most communities. Yet the notion of a good school is inherently context-dependent. No matter
how much everyone spends on housing, only half of all children can attend top-half schools. The decision of how much to
spend on housing thus confronts many middle-income families with a painful dilemma. They can either send their children to
a school of average quality by purchasing a house that is larger and more expensive than they can comfortably afford; or they
can buy a smaller house that is within their budget and send their children to a below-average school.12 To see why so
many families might nd the former option more compelling, we need not assume that they are strongly prone to envy or
jealousy.
Similarly, we need not invoke envy to explain why people might perceive opportunities to get ahead by spending more than
others on clothing. First impressions count for a lot during job interviews, but looking good is also a relative concept. It means
looking better than others who are competing for the same job. If others are spending more, you must spend more as well, or else
be prepared to live with reduced odds of landing the job you want.
A similar logic governs the decision of how much to spend on gifts. Suppose you have been invited to a professional associate's
home for dinner and want to bring a bottle of wine for your host. What should you bring? John Brecher and Dorothy Gaiter, whose
unpretentious, value-oriented wine column appears each Friday in The Wall Street Journal's weekend section, devoted a recent

10
Johnson, 1998, A1.
11
See, for example, Hoxby, 2000.
12
House size varies both within and across neighborhoods. The neighborhoods that have better schools have bigger houses, on average. The most ambitious
parents would choose the smallest house in whatever neighborhood they selected.
1782 R.H. Frank / Journal of Public Economics 92 (2008) 17771786

column to precisely this question. Ask a respected wine merchant to suggest an unusual wine, one that your host is unlikely to
have tried before, they sensibly recommended. And plan on spending about $30.
Why should you spend so much, given that many wines available today for under $10 are far better than the wines drunk by
kings of France in centuries past? In part because you have an interest not only in how the wine tastes, but also in how your gift will
be interpreted. Giving a wine that has become inexpensive by today's standards might be read as a statement that you view your
relationship with your host as unimportant. So unless you really don't care about the relationship, the extra $20 is probably worth
spending. Here again, we need not invoke envy to explain why the spending of others might induce people to spend more than
they can comfortably afford on gifts.

4. The limits of laissez faire

As noted at the outset, the libertarian tradition holds that the state may restrict someone's liberty of action only to prevent harm
to others. Positional externalities clearly harm others. Parents who save less or work longer hours or take greater safety risks in
order to be able to afford a house in a better school district may have no conscious intention to impose costs on other parents. Nor is
it likely that those who buy expensive interview suits wish to impose additional costs on other job candidates. Yet such actions do
impose costs on others, forcing them to spend much more than would otherwise be necessary to achieve ordinary life objectives.
But preventing harm to others is only a necessary condition for government intervention, not a sufcient condition. Indeed,
liberal societies routinely permit actions that cause harm to others, on the grounds that even greater harm would result from
attempts to prevent those actions. For example, most liberal societies protect the right to free speech, even though certain forms of
speech are manifestly harmful to others. The right to free speech, however, is never absolute. If the prospective harm of
unrestricted speech is too great, we impose restrictions, as when we prohibit yelling re in a crowded theater if there is no re.
The question of whether to restrict actions that cause harm to others requires a careful weighing of the harm caused by the
actions themselves against the harm that would be caused by restricting the freedom to act. Ronald Coase won the Nobel Prize in
economics in 1991 largely on the strength of his contribution to our way of thinking about this delicate balancing act.13 My aim in
this section is to explore how Coase's insights might help inform our thinking about public policy toward positional externalities.
Before Coase, it was common for policy discussions of externalities to be couched in terms of perpetrators and victims. A factory
that created noise, for example, was a perpetrator, and an adjacent physician whose practice suffered as a result was a victim.
Coase's insight was that externalities like noise or smoke are purely reciprocal phenomena. The factory's noise harms the doctor,
yes; but to invoke the doctor's injury as grounds for prohibiting the noise would harm the factory owner. It is not the factory
owner's intent to harm the doctor. Nor is it the doctor's intent to impede the workings of the factory. Their proximity to one
another and the nature of their specic activities creates a mutual problem to be solved. Both the factory owner and the doctor,
Coase argued, have a shared interest in nding the least costly solution to this problem. He concluded that if they were able to
negotiate freely with one another, they would resolve it efciently, regardless of whether the government held the factory owner
liable for noise damages.
But the deeper message of Coase's paper was a different onenamely, that when negotiation is impractical, the state should
assign the burden of adjusting to an externality to whichever party can accomplish it at lower cost. For example, if the doctor and
the factory owner cannot negotiate, the government should hold the factory liable for noise damage if it is less costly for the factory
to reduce its noise than it is for the doctor to move to a more sheltered location. But if the reverse is true, then the government
should not hold the factory liable. That posture, Coase reasoned, would induce the doctor to relocate, which under the
circumstances would be the efcient solution.
What does Coase's reasoning have to say about the problem of consumption externalities? To help frame ideas, I begin with a
simple example whose details echo many of the issues that arise in the analysis of free speech. Suppose some people enjoy wearing
purple shirts and others are offended by the sight of them. Suppose further that those who like them would collectively pay up to
$1000 for the privilege of wearing them, while those who dislike them would pay up to $2000 to avoid having to see them. If
negotiation between the two groups were practical, opponents could compensate those who favor purple shirts for agreeing not to
wear them.
But in this case, private negotiation is surely impractical. The essence of the difculty is nicely captured in a New Yorker cartoon
that depicts an accordion player on a subway platform with a sign on his donation cup saying, Will not play Lady of Spain, 25
cents. If one could demand a compensation payment for agreeing not to wear a purple shirt, there would be an elastic supply of
volunteers for such payments.
Does the fact that negotiation is impractical mean that the state should prohibit the wearing of purple shirts? Since the shirts
generate larger costs than benets, by assumption, a prohibition might appear to be the efcient outcome.
But that reasoning alone is insufcient. If it were, the laws of free speech would prohibit most personal insults, which typically
impose greater costs on those insulted than benets to those who insult them. In the free speech case, the right to insult others may
be defended on at least two other grounds that appear relevant to the purple-shirt example. First, the right to speak freely appears
to have general value above and beyond the value one might assign to speaking freely on any given occasion. By the same token, the
right to choose clothing according to one's whims of the moment may have general value beyond what one would be willing to pay
for the right to wear a purple shirt.

13
Coase, 1960.
 R.H. Frank / Journal of Public Economics 92 (2008) 17771786 1783

A second concern is the need to anticipate the extent to which affected parties might be able to adjust over time to the injuries
in question. In a climate in which personal insults are permissible as a matter of right, for example, potential victims of insults
might gradually learn to tune them out, or perhaps even alter their behavior and social circumstances to limit their frequency.
Similarly, in a climate that defends the individual's right to wear colors of her own choosing, those who are initially offended by
purple shirts might eventually get used to them, or they might tailor their environments to limit exposure to them.
Defending any given right entails both costs and benets. In the end, questions regarding which specic rights to defend are
thus quintessentially practical ones.14 Having the right to insult others and the right to choose the color of one's clothing permits
one to cause harm to others. These rights nonetheless appear defensible because the alternatives seem likely to generate even
greater harm. It is in this spirit that I frame my discussion about the extent to which we should defend the individual's right to
spend her income in whatever ways she chooses, even in the face of evidence that some forms of spending impose costs on others.
The basic question is whether society should attempt to create incentives to shift the composition of expenditure in favor of
nonpositional goods. For illustrative purposes, I focus again on the important category of housing expenditures, although virtually
the same issues would arise for other goods with strong positional components. Are the losses to those whose expenditures on
housing would be restricted by the incentive change likely to outweigh the gains to those who favor the expenditure shift?
If the only costs associated with positional externalities were psychological, members of the latter group might in time nd
means of adjusting to the discomfort they initially experience from living in houses that appear diminished by the increased
housing expenditures of others. But even if the capacity to adjust to psychological costs were limited, it might still be best to place
the burden of adjustment on those who experience these costs, since the alternative might be interpreted as a tacit invitation to
complain.
As noted earlier, however, positional externalities in the housing market also entail far more tangible costs, most notably that
failure to keep up with community spending patterns means having to send one's children to schools of below-average quality. The
scope for accommodation to such costs seems far more limited. For many families, the best response will be to match the housing
expenditures of other families with similar incomes. To do so they may have to accept riskier jobs, save less, work longer hours,
carry more debt, and commute longer distances, all of which are associated with reduced levels of subjective well-being.15
What about the costs to those induced by the policy shift to spend less than they wanted to on housing? In light of evidence
that, beyond some point, the capacity of a house to confer utility depends predominantly on its relative size, there would be little
effect on measured levels of subjective well-being if the rate at which house size has been growing were to diminish. Of course,
there might be more general costs associated with the very fact of having their spending options restricted. These costs would
depend on the specic instruments by which the restrictions were implemented. A regulatory commission assigned to
micromanage individual building plans might be a costly intrusion. But expenditures might also be restrained in other, less
intrusive ways, a point to which I now turn.

5. Taxes and positional externalities

Although it took several decades, economists nally persuaded policy analysts and legislators to move away from command-
and-control environmental regulation in favor of policies that reduce incentives to pollute. Instead of prohibiting rms from
polluting, or requiring specic pollution-control technologies, the new approach is simply to tax rms on the basis of how much
they pollute (or, equivalently, to require them to purchase efuent permits). As economists had long predicted, the costs of any
given level of pollution abatement are far lower under incentive-based systems than under prescriptive regulation.16
The problem of excessive environmental pollution is caused by an incentive gap much like the one that gives rise to positional
externalities. Consumption taxes may thus be an efcient way of limiting the costs of positional externalities, for essentially the
same reasons that efuent taxes are an efcient way of limiting the costs of pollution.
In a world of complete information and perfect government, the solution would be to set a different tax rate for every good in
accordance with the strength of the positional externalities it generates. But although researchers have begun to estimate the
differences in the extent to which context inuences demands for specic categories of goods,17 existing knowledge is far too
fragmentary to support such an ambitious approach.
Even if we knew much more about these magnitudes, however, it would be politically costly to establish a separate tax rate for
every good. Lobbyists would inundate legislators with studies purporting to show why their particular client's product or service
was nonpositional and therefore entitled to tax-exempt status.
I have argued elsewhere that a simpler, more promising, approach would be to abandon the current progressive income tax in
favor of a more steeply progressive general consumption tax.18 This approach rests on the observation that positional concerns are
stronger for luxuries than necessities. There are obvious pitfalls in trying to identify specic goods as luxuries. But given that luxury
is an inherently context-dependent phenomenon, it is uncontroversial to say that the marginal expenditures of those who spend
most are most likely to be spent on luxuries. A steeply progressive consumption tax is thus a luxury tax that sidesteps the need to
identify specic goods as luxuries.

14
Holmes and Sunstein, 2000.
15
For a survey of this evidence, see Frank, 1999.
16
For an excellent survey, see Dorris, 1996.
17
See, for example, Solnick and Hemenway, 1998 and 2005; Alpizar et al., 2005: and Kerwin et al., 2007.
18
See Frank, 1985a, 1999 and 2007. Other authors have also discussed tax remedies for positional externalities. See, for example, Boskin and Sheshinski, 1978;
Ng, 1987; Ireland, 1998; and Layard, 2005.
1784 R.H. Frank / Journal of Public Economics 92 (2008) 17771786

Implementing a progressive consumption tax would be straightforward. Taxpayers would report their incomes to the tax
authorities just as they do now. They would also report how much they had saved during the year, much as they do now in
order to exempt money deposited in retirement accounts. People would then pay tax on their taxable consumption, which is
just the difference between their income and their annual savings, less a standard deduction. Rates at the margin would rise
with taxable consumption. If the tax were revenue neutral, marginal rates at the top would be signicantly higher than current
marginal tax rates on income. If policy makers were looking to generate additional revenue, marginal rates would have to be
higher still.
Proposals to generate additional income tax revenue by raising top marginal rates invariably summon concern about possible
negative effects on the incentive to save and invest. Under a progressive consumption tax, by contrast, people's incentives would be
to save and invest more, even if top marginal tax rates on consumption were extremely high.
To illustrate how the tax would change spending incentives, consider a taxpayer whose marginal rate under the current income
tax is 0.33 and whose marginal rate under a progressive consumption tax would be rc. Under each tax regime, suppose that this
taxpayer forgoes an extra dollar of consumption for the length of time it takes for money in a savings account to double in value.
How much extra future consumption will his sacrice support in each case? Under the current income tax, his dollar of forgone
consumption generates a bank deposit of $1, which becomes $2 on the date in question. When he withdraws the $2, he must pay
$0.33 in income tax on his one dollar of interest income. So $1 of forgone consumption today translates into $1.67 of future
consumption under the current income tax.
Under the consumption tax, by contrast, forgoing $1 of consumption today would result in a $ rc reduction in current tax
liability, and so would support a current bank deposit of $(1 + rc). At withdrawal time, this deposit will have grown to $(2 + 2rc). To
nd CF, the amount of future consumption this deposit will support, we solve CF + rcCF = $(2 + 2rc) for CF = $2. Giving up a dollar of
current consumption thus supports only $1.67 of future consumption under the current income tax, but $2 under the progressive
consumption tax.
A progressive consumption tax would also alter other important relative prices that affect savings. In particular, it would
lower the marginal costs of self-insuring against lost earning power and of leaving bequests. Although traditional consumption
theories suggest that people consume most or all of their earnings before they die, many in fact leave estates whose
magnitudes are far larger than could reasonably be attributed to uncertainty regarding the time of death. People are reluctant
to spend down their assets for multiple reasons, but two in particular stand out. One is to hedge against the possibility that
becoming disabled would rob them of their earning power. Another is to leave bequests to heirs and charities. Moral hazard
and adverse selection make private savings more attractive than commercial insurance as a hedge against lost earning power. A
steeply progressive consumption tax would lower the cost of self-insuring. And it would also lower the cost of leaving
bequests.
A separate channel through which such a tax would limit current consumption is by directly constraining the expenditures of
high-income individuals who now consume most or all of their after-tax incomes. So for multiple reasons, a progressive
consumption tax can be expected to stimulate higher savings.
If the tax affected spending directly for the reasons given, it would also affect spending indirectly. Each individual's spending,
after all, constitutes part of the frame of reference that inuences what others spend. And given the apparent importance of
context, the indirect effects of a progressive consumption tax promise to be considerably larger than the direct effects.
Thus, for example, if people at the top save more and spend less on mansions, that will shift the frame of reference that
inuences the housing expenditures of those just below the top. So they, too, will spend less on housing, and so on all the way
down the income ladder.19
By all available evidence, that would be a good thing. The aggregate household savings rate in the United States was negative
during both 2005 and 2006. Americans spent more than they earned during full calendar years for the rst time since The Great
Depression. Liberals and conservatives alike agree that our failure to save has had damaging macroeconomic consequences, that
we would all be better off if we all spent less and saved and invested more. But no individual has the power to alter the aggregate
savings rates.20
The advantage of the tax approach in both the environmental and consumption domains is its exibility. In the environmental
domain, rms for which pollution reduction is most expensive may nd it in their interest to continue to pollute even after the
imposition of an efuent tax. Similarly, those families for whom consumption reductions would be especially difcult may respond
to higher tax rates by expending additional effort in order to maintain their previous spending levels. But in both the
environmental and consumption domains, the relevant externalities depend more on overall activity levels than on the activity
levels of particular individuals or rms. And just as the imposition of efuent charges mitigates pollution damage by leading most
rms to curtail pollution, a progressive consumption tax would mitigate the costs of positional externalities by increasing the
incentive to save.
Of course, both the UK and the US already have progressive tax systems. Absent more detailed knowledge about the strength of
positional concerns, what reason is there to believe that current levels of progressivity haven't already neutralized the ill effects of
those concerns?

19
Hopkins and Kornienko, 2004, who ignore the effects of such expenditure cascades, argue that welfare losses from positional competition are a declining
function of income inequality.
20
By increasing the savings rate, progressive consumption tax would increase future incomes and, beyond some point, future consumption. Families would
eventually have to meet a higher consumption than they do now, but they would also have the higher incomes necessary to meet that standard.
 R.H. Frank / Journal of Public Economics 92 (2008) 17771786 1785

Despite the gaps in current knowledge about positional concerns, recent experience provides some insight into how switching
to a steeply progressive consumption tax might affect social welfare. Compared to the current income tax, such a tax would reduce
high-end consumption and increase public spending. Private spending reductions would be concentrated in the categories that
afuent consumers consider least urgent. Political imperfections notwithstanding, governments would spend much of the
resulting tax revenue on the public services that voters value most. The practical question is thus whether the social welfare loss
from cutting the least urgent high-end private consumption categories would outweigh the welfare gain from increasing the most
highly valued public services. To be sure, it is possible to imagine a society so poor in private consumption and so rich in public
consumption that such a switch would reduce welfare. But what about societies with high levels of income inequality, like the US
and the UK in recent years?
In the absence of detailed empirical evidence, a plausible conjecture is that the rst expenditures that high-end consumers
would reduce in response to a steeply progressive consumption tax are the same ones they have recently been increasing in
response to their growing incomes. In the US, some of the most spectacular increases in high-end consumption in recent years have
occurred in housing and the events families use to mark special occasions. By all accounts, such expenditures are hyper-positional.
To celebrate his daughter's recent birthday, for example, David H. Brooks, the CEO of a company that supplies body armor to
the American military in Iraq, invited 150 of her closest friends to the Rainbow Room atop Rockefeller Center in Manhattan,
where they were serenaded by 50 Cent, Don Henley, Stevie Nicks, and other luminaries in a celebration reported to have cost
$10 million. Facing high marginal tax rates on consumption, Brooks and other similarly situated parents would surely spend
less on coming-of-age parties for their children. If they did, the standards that dene a special occasion in their circle would
shift accordingly. Could anyone argue with a straight face that these changes would constitute a signicant welfare loss for the
children involved?
The situation is similar with respect to high-end housing expenditures. Consider a family that spends $10 million a year and is
deciding whether to add a $2 million wing to its mansion. If the top marginal tax rate on consumption were 100% the project would
cost $4 million. The additional tax payment would reduce the federal decit by $2 million. Alternatively, the family could scale
back, building only a $1 million addition. Then it would pay $1 million in additional tax and could deposit $2 million in savings. The
federal decit would fall by $1 million, and the additional savings would stimulate investment, promoting growth. Either way, no
real sacrice would be required of the wealthy family, because when all build larger houses, the result is merely to redene what
constitutes acceptable housing.
What about the welfare impact of the public services made possible by additional revenue from a steeply progressive
consumption tax? The legislator's obligation is to spend every tax dollar in the way that best serves the public interest. Public
choice theorists are quick to point out that legislators fail to meet this obligation. But even allowing for the fact that some of the
extra revenue would be spent wastefully, much of the rest would pay for useful things. In the US, for example, budget decits
recently led the Bush administration to cut the Energy Department's program for helping lock down loose nuclear materials in the
former Soviet Union. These materials are currently guarded in poorly armored facilities by soldiers who drink heavily and are
paid only intermittently. Terrorists who acquired these materials could easily smuggle them into this country because we do not
inspect most of the cargo containers that enter the nation's ports. A progressive consumption tax would make such outcomes less
likely.
My point in describing these examples is that limited empirical knowledge does not always prevent us from drawing
reasonable inferences about the likely welfare effects of specic tax policy changes. Most economists would agree that the welfare
cost to wealthy families of having smaller mansions and less expensive coming-of-age parties would be smaller than their benet
from having improved security. If so, the benet of improved security to the non-wealthy would be pure gravy.
These observations call into question the prevailing assumption that tax policy confronts us with an agonizing tradeoff between
equity and efciency. If positional externalities inuence spending patterns in the ways suggested by available evidence, higher
marginal tax rates on top earners would appear justied not only on grounds of equity, but also on grounds of narrow economic
efciency.

6. Concluding remarks

If higher tax rates on top earners would mean not only a more equitable economy but also a more efcient one, why do many
economists continue to push for changes in the opposite direction? One possibility is that even though they recognize that many
people care about relative position as an empirical matter, they believe that such concerns are simply not a legitimate basis for
public policy. They might be concerned, for example, that to implement policies for curtailing positional externalities could
encourage people to give freer reign to destructive emotions such as envy and resentment. Or they might be concerned that
recognizing positional externalities as legitimate public policy concerns might provoke attempts to implement sweeping new
economic and social regulations. Both concerns merit careful attention.
As for the rst, society does in fact have a strong interest in maintaining traditional efforts to keep envy and other destructive
emotions in check. After all, as John Rawls and others have argued persuasively, absolute equality is neither an attainable nor even
a desirable social goal. It is easy to imagine that if society were to abandon its efforts to discourage envy, ruthless intolerance of
even the most minor forms of inequality might emerge.
Envy, however, is not the central issue. Positional externalities would continue to impose large and tangible economic costs
even if everyone were completely free of envy. Context inuences our assessment of material living standards for the same kinds of
practical reasons that context inuences our assessments of temperature and distance. When context matters, there is simply no
1786 R.H. Frank / Journal of Public Economics 92 (2008) 17771786

presumption that individual spending decisions give rise to the greatest good for all. To acknowledge the obvious importance of
context is in no way to encourage people to give free reign to feelings of envy.
It is also easy to see why some might fear that treating positional externalities on a par with other externalities might
provoke a wave of intrusive economic and social regulation. After all, positional externalities are widespread, and our early
experience with environmental regulation provides ample grounds for concern about government interventions that do more
harm than good. Yet we have no reason to believe that acknowledging the legitimacy of positional externalities would doom us
to protracted prescriptive regulation. With the lessons of environmental regulation fresh in memory, the same pitfalls might be
avoided entirely.
The progressive income tax, which is already part of our existing policy arsenal, provides an instrument for attacking positional
externalities in the same way that we now attack environmental externalities with efuent charges. A progressive consumption
tax would be an even better instrument for this purpose. We already confront the question of how steeply progressive our tax
system should be. On the available evidence, increasing its progressivity relative to the current structure would change
expenditure patterns in ways that would result in greater well-being for consumers across the entire income scale. Yet the current
policy debate is driven by an intellectual framework that continues to insist, contrary to all evidence, that relative consumption
does not matter.

References

Alpizar, F., Carlsson, F., Johansson-Stenman, O., 2005. How much do we care about absolute versus relative income and consumption? Journal of Economic Behavior
and Organization 56, 405421.
Boskin, Michael, Sheshinski, Eytan, 1978. Optimal redistributive taxation when individual welfare depends on relative income. Quarterly Journal of Economics
92 (4), 589601.
Clark, Andrew, Oswald, Andrew, 1996. Satisfaction and comparison income. Journal of Public Economics 61, 359381.
Coase, Ronald, 1960. The problem of social cost. Journal of Law and Economics 3, 144171.
Diener, Ed, Sandvik, Ed, Seidlitz, Larry, Diener, Marissa, 1993. The relationship between income and subjective well-being: relative or absolute? Social Indicators
Research 28, 195223.
Dorris, Gary W., 1996. Redesigning Regulatory Policy: a case study in urban smog, PhD dissertation. Cornell University.
Easterlin, Richard, 1974. Does economic growth improve the human lot? In: David, Paul, Reder, Melvin (Eds.), Nations and Households in Economic Growth: Essays
in Honor of Moses Abramovitz. Academic Press, New York.
Easterlin, Richard, 1995. Will raising the incomes of all increase the happiness of all. Journal of Economic Behavior and Organization 27, 3547.
Frank, Robert H., 1985a. Choosing the Right Pond. Oxford University Press, New York.
Frank, Robert H., 1985b. The demand for unobservable and other nonpositional goods. American Economic Review 75, 101116 March.
Frank, Robert H., 1991. Positional externalities. In: Zeckhauser, Richard (Ed.), Strategy and Choice: Essays in Honor of Thomas C. Schelling. MIT Press, Cambridge,
MA, pp. 2547.
Frank, Robert H., 1999. Luxury Fever. The Free Press, New York.
Frank, Robert H., 2007. Falling Behind: How Rising Inequality Harms the Middle Class. University of California Press, Berkeley, CA.
Hirsch, Fred, 1976. Social Limits to Growth. Harvard University Press, Cambridge, Mass.
Holmes, Stephen, Sunstein, Cass, 2000. The Cost of Rights: Why Liberty Depends on Taxes. W. W. Norton, New York.
Hopkins, Ed, Kornienko, Tatiana, 2004. Running to keep in the same place: consumer choice as a game of status, with Ed Hopkins. American Economic Review 94,
10851107.
Hoxby, Caroline, 2000. Peer effects in the classroom: learning from gender and race variation. NBER Working Paper No. W7867 August,.
Ireland, Norman, 1998. Status seeking, income taxation, and efciency. Journal of Public Economics 70, 99113.
Johnson, Dirk, 1998. When Money Is Everything, Except Hers. New York Times, A1 October 28.
Kahneman, Daniel, Diener, Ed, Schwartz, Norbert (Eds.), 1999. Understanding Well-Being: Scientic Perspectives on Enjoyment and Suffering. Russell Sage,
New York.
Kahneman, Daniel, Krueger, Alan B., Schkade, David, Schwartz, Norbert, Stone, Arthur B., 2006. Would you be happier if you were richer? A focusing illusion.
Science 312, 19081910 June 30.
Kerwin, Charles, Hurst, Erik, Roussanov, Nick, 2007. Conspicuous consumption and race. University of Chicago, mimeo, http://faculty.chicagogsb.edu/erik.hurst/
research/race_consumption_april2007_applications.pdf.
Layard, Richard, 2005. Happiness: Lessons from a New Science. Penguin, London.
Mill, John Stuart, 1982. On Liberty. Viking, New York.
Ng, Yew Kwang, 1987. Diamonds are a government's best friend: burden-free taxes on goods valued for their values. American Economic Review 77 (1), 186191.
Smith, Adam, 1937. The Wealth of Nations. Random House, New York.
Solnick, S.J., Hemenway, D., 1998. Is more always better?: a survey on positional concerns. Journal of Economic Behavior and Organization 37, 373383.
Solnick, S.J., Hemenway, D., 2005. Are positional concerns stronger in some domains than in others? American Economic Review, Papers and Proceedings 95,
147151.
Veenhoven, Ruut, 1993. Happiness in Nations. International Monetary Fund.