Conspicuous Consumption and Happiness
Rainer Winkelmann∗
January 2010
Abstract
Traditional tools of economic welfare analysis cannot quantify the envy-related
welfare loss from conspicuous consumption. The happiness approach offers itself as an
alternative for estimating such consumption externalities. The approach is developed in
the context of luxury car consumption (Ferraris and Porsches) in Switzerland. Results
from detailed panel data suggest that the prevalence of luxury cars in the area has
a negative impact on individuals happiness, with the biggest impacts arising for the
poor.
JEL Classification: D12, D62, I31
Keywords: Ferrari, status, consumption externality, Swiss Household Panel
∗
Address for correspondence: University of Zurich, Socioeconomic Institute, Zuerichbergstr. 14, CH-8032
Zurich, Switzerland, email: [email protected].
1
Introduction
There is ample evidence that the well-being of an individual depends to a large degree on
comparisons with others. Status concerns in particular are an important feature of our
social existence. Higher ranked individuals are on average healthier (Wilkinson, 2000) and
live longer (Oswald and Rablen, 2008) than lower ranked ones. They also report higher levels
of happiness in survey questions (Di Tella at al., 2007). Relatedly, actions of others affect
the “frame of reference” for own decisions (Frank, 1991). As Karl Marx (1847) famously
noted, “A house may be large or small; as long as the neighboring houses are likewise small,
it satisfies all social requirement for a residence. But let there arise next to the little house
a palace, and the little house shrinks to a hut.”
For the economist, such positional concerns and frame of reference effects are important,
as they can lead to inefficient market outcomes. For example, resources spent on improving
one’s status are wasteful from a societal point of view, as one person’s gain is another person’s
loss. As Layard (1980) puts it “For, though individuals are willing to make sacrifices to
improve their individual position, the net result of status-motivated action will be no increase
in status satisfaction but an increase in sacrifice.” (p. 738). Perhaps the best studied example
of such wasteful competition occurs in the presence of relative consumption effects, i.e. if an
individual’s utility depends not only on the level of her own consumption but also on how
that level compares with the consumption of others.
One cause of such interdependence has been explored by Veblen (1899) who, referring to
the behavior of the nouveau riche during the Second Industrial Revolution in Britain, coined
the term of “conspicuous consumption”. He argued that this consumption “is evidence of
wealth, and thus becomes honorific, and . . . failure to consume a mark of demerit.”
Veblen thus refers to consumption that is intently used as a signal for status. The selection
1
of goods that can serve as signal depends on context. Sometimes, the term of conspicuous
consumption is applied more broadly, for all consumption activities, not necessarily only
those serving as status signals, as long as they fulfill two requirements: First, the consumption
must literally be “visible” to outsiders. Second, the consumed good(s) must be positional,
in the sense that utility depends at least in part on relative rather than only absolute
consumption.
While there have been a number of theoretical papers on consumption externalities (e.g.
Layard, 1980, Dupor and Liu, 2003, Arrow and Dasgupta, 2009), showing the specific conditions under which overconsumption and thus inefficiencies do or do not arise, direct empirical
evidence on conspicuous consumption is relatively scarce.
Traditionally, economists have analyzed consumption externalities based on revealed preferences and behavior (Neumark and Postlewaite, 1998, Kuhn et al. 2008). For example,
under certain assumptions regarding preferences, consumption comparisons will shift the
marginal rate of substitution between consumption of goods and leisure, thereby changing
observed choices. Specifically, an increase in reference consumption will lead to an outward
shift in labor supply and an increase in own consumption, an effect known as “Keeping up
with the Joneses” (Frank, 1999). However, as shown in the next section, negative welfare
effects may arise even in the absence of such behavioral distortions, for example if conspicuous consumption triggers feelings of envy or jealousy. Dupor and Liu (2003) remark “that
nonstandard empirical approaches may then be required” (p. 428) in order to determine the
welfare loss from overconsumption.
Exactly such a “non-standard” analysis is the objective of this paper. I propose to use
information on what people say when asked whether they have a good and worthwhile life,
i.e., survey information on “subjective well-being”, or “happiness”, in order to determine the
welfare loss due to conspicuous consumption. The paper thus contributes to the economic
2
happiness literature, where “happiness” is taken as a proxy for “utility”, or “welfare”.
Recent evidence suggests that happiness can be measured with a reasonable degree of reliability, sufficient to be able to compare means over time or across regions, or to use multiple
regression analysis to understand the factors that affect individual happiness (Krueger and
Schkade, 2008). Moreover, happiness taps into key components of quality of life. For example, many studies have found a positive correlation between happiness and longevity. One
of the mechanisms seems to be that happiness protects the healthy from becoming sick, by
lowering stress, which in turn enhances the immune system (Veenhoven, 2008). Studies have
also shown that individuals who report higher levels of satisfaction with their lives are also
rated as happier by their relatives and friends, tend to smile more during social interactions,
have higher pre-frontal brain activity (the part of the brain associated with positive states),
and are more likely to recall positive life events (Layard, 2005). It is still contentious whether
and to what extent happiness represents “decision utility”, i.e., the utility that economists
think of when talking about “utility maximization”. Regardless of this unresolved issue,
happiness surveys provide direct evidence on the welfare implications of behaviors, policies
and institutions in general, and conspicuous consumption in particular.
In principle, there are a number of questions that could be asked: What is the effect of
conspicuous consumption on own happiness? What is the effect of conspicuous consumption on others’ happiness? What determines conspicuous consumption? What forms of
consumption do contribute most to overall happiness? The objective of this paper is to
provide evidence on the second question, i.e. to quantify the welfare loss due to the negative externality generated by conspicuous consumption. In contrast to prior related research
(e.g. Clark and Oswald, 1996), where such positional externalities were studied based on
constructed reference incomes (begging the question whether such income differences are at
all “conspicuous” for the individual), the approach here focuses on a specific instance of
3
conspicuous consumption, the purchase and display of luxury sports cars.
Two brands are considered, Ferrari and Porsche. Is their consumption conspicuous?
Although people, when asked about attributes that would affect their own decision of buying
(any) car, seem to downplay the role of status, a large majority thinks that status concerns
are important for others (Johannsson-Stenman and Martinsson, 2006). Regarding Ferraris
and Porsches, since other desirable attributes, such as “environmentally friendly”, can hardly
justify their purchase, pure demonstration of status is a plausible prime motive.
Data are for Switzerland, for the years 2002 and 2007. Happiness and other relevant
socio-demographic characteristics (most importantly current household income) are taken
from the Swiss Household Panel. For each person, it is possible to match the number of
new luxury car registrations (per 1000 population) at the municipal, the regional, and the
cantonal level. There is no information whether respondents actually acquired such a car
themselves, but the chances for this event appear minimal. The empirical question is then
whether happiness is lower, the higher the local or regional density of new luxury sport
cars. Ordered logit models are used to quantify the well-being loss due to conspicuous
consumption.
2
Background
2.1
Modeling conspicuous car consumption
The following model, an adaptation of Dupor and Liu (2003), is a useful framework for
thinking about conspicuous car consumption. Let c denote own consumption of car attributes
(such as horsepower, or looks) and c̄ the average attributes of cars in the population. The
utility function of identical consumers can be written as
U (c, c̄, l)
4
where l is individual labor. We assume that people enjoy horsepower (∂U/∂c = U1 > 0) and
that horsepower consumption is conspicuous and generates envy (∂U/∂c̄ = U2 < 0). People
have to work in order to be able to afford luxury cars. The budget constraint is c ≤ f (l),
where f (l) denotes the technology. To maximize utility
U (c, c̄, g(c))
where g(c) = f −1 (c) = l, individuals have to trade off the marginal utility from consumption
and the disutility from additional work that is required to support the increased consumption.
Specifically, the first-order condition for such a maximum is
U1 (c, c̄, g(c)) + g 0 (c)U3 (c, c̄, g(c)) = 0
(1)
A symmetric equilibrium is then characterized by the equilibrium condition
U1 (c, c, g(c)) + g 0 (c)U3 (c, c, g(c)) = 0
This equilibrium is inefficient, since individuals do not account for the negative externality
U2 (c, c, g(c)) they impose on others. The social optimun requires that
U1 (c, c, g(c)) + U2 (c, c, g(c)) + g 0 (c)U3 (c, c, g(c)) = 0
(2)
Assuming concavity of U (c, c, g(c)), the optimal car consumption c∗ that solves (2) is less than
the equilibrium consumption ĉ, an instance of overconsumption. A government imposing a
“luxury” tax can remove this inefficiency.
The underlying assumption of identical consumers is admittedly unrealistic, as every
person would then buy cars with similar attributes, whereas in reality Ferraris and Porsches
are only bought by few people. Changing the model in order to allow for heterogeneity in
endowments (whereby “rich” people would have a greater demand for c) would not change
the essential mechanisms leading to overconsumption. Arrow and Dasgupta (2009) consider
5
a dynamic version of the model. Making leisure conspicuous as well would change the model
but there is no evidence to that effect (Solnick and Hemenway, 1998).
An important takeaway from this model is that the marginal rate of substitution between
leisure and consumption, U3 (c, c̄, g(c))/U1 (c, c̄, g(c)) does not need to depend on c̄, as is the
case for example if the utility function is additively separable in c̄. In this case, envy would
not affect individual decisions but nevertheless lead to a welfare loss. Suppose, however, that
U (c, c̄, g(c)) can be measured empirically. Then it becomes possible to determine the effect
of c̄ on U using statistical methods.
2.2
Empirical literature
There is, to the best of my knowledge, no paper yet on the effects of conspicuous car consumption on happiness. More generally, the literature on consumption and happiness is still
in its infancy, although there is some evidence, mostly from lab experiments in psychology,
that the kind of consumption one engages in matters for happiness. Progress in this area is
delayed by the absence of suitable large scale datasets collected on representative samples,
since the usual happiness surveys rarely contain sufficiently detailed consumption information. One exception is the Health and Retirement Survey which has been used in a recent
study by DeLeire and Kalil (2009). They find, for example, that the interaction between
consumption activities and social connectedness is an important determinant of happiness.
As to the incidence of conspicuous consumption, Charles, Hurst and Roussanov (2009) show
that Blacks and Hispanics devote larger shares of their expenditure bundles to visible goods
(clothing, jewelry, and cars) than do comparable Whites.
A substantial fraction of evidence on consumption externalities comes from discrete choice
experiments in which people are confronted with (hypothetical) choices and asked for their
preferred option. The claim that people care for relative rather than absolute consumption
6
levels is often supported by answers to “parallel universe” questions, where one has to choose
between two fictitious societies. Consider the following example: In society A, everyone lives
in a house with 3,000 square feet of floor space, whereas in society B the own house has
4,000 square feet while neighbors have 5,000 square feet. Frank (1999, 2004) contents that
most people prefer society A. Solnick and Hemenway (1998) have conducted such discrete
choice experiments with real populations. Interestingly, they find that leisure appears to be
inconspicuous – perhaps in part because it is not so easily publicly observable.
The relativity of consumption experiences was also investigated in field experiments.
An example is provided by Hsee et al. (2008) who report on evidence from data collected
among randomly selected respondents from 31 large cities in China. Participants were asked
about their happiness with their jewelry possessions. The researchers found that residents
of cities with more expensive jewelry were not any happier on average about their jewelry
than residents of cities with less expensive jewelry (although within each city, people who
owned more expensive jewelry reported greater happiness about their jewelry).
One way to distinguish between conspicuous and inconspicuous activities is the nature
of the consumption experience: is it inherently evaluable (such as ambient temperature,
sleep, having social company etc.) or inherently inevaluable (the size of a diamond, the
horsepower of a car)? (Hsee et al., 2008). Frame of reference effects can also stem from
past own consumption. There is evidence that hedonic adaptation plays a major role in
some consumption activities, such as the size of the house, but not in others, such as a
short commute (Stutzer and Frey, 2008). One factor in adaptation is uncertainty: unstable
or uncertain consumption experiences create longer lasting happiness than their stable and
certain equivalents (Kahneman and Thaler 1991, Hsee et al. 2008).
The most comprehensive evidence on the effect of consumption externalities on happiness so far is indirect, as it documents the pervasive role of relative income for happiness.
7
Examples are studies by Clark and Oswald (1996), Luttmer (2005) and Dynan and Ravina
(2007). The two main challenges in this literature are, first, to determine an appropriate
reference income and second, to ensure that the measured effect is not confounded by other
factors, such as costs of living. The existence of relative income effects are suggestive of
negative consequences of conspicuous consumption, but other explanations exist, and this
approach cannot determine the welfare loss generated by such consumption.
3
Data
The paper combines data from a number of sources: information on individual happiness
and income is obtained from the Swiss Household Panel; car registration data were provided
by the Federal Roads Office; and a cost of living index was constructed from data obtained
from an internet property platform as well as the Swiss Labor Force Survey.
The Swiss Household Panel is comparable in structure and scope to other European
panel household surveys, such as the German Socio-Economic Panel or the British Household Panel. It was initiated in 1999 and has been repeated annually since. Importantly, it
collects information on the following question: In general, how satisfied are you with your life
if 0 means “not at all satisfied” and 10 means “completely satisfied”? The answer is henceforth referred to as measurement of “happiness”. In addition, a number of variables can be
extracted that are used as typical controls in happiness regressions. These include household
income, household size, age, marital status, employment status, gender and language region.
The dataset provides no direct information on cars owned by respondents, nor on individuals’ perceptions of (conspicuous) car consumption by others. Therefore, it is necessary to
merge this information from external sources. Data on registrations of Ferraris and Porsches
for each ZIP code were obtained from the Federal Roads Office. The merge is done on a
8
geographic basis, the assumption being that Ferraris and Porsches are most visible around
the place of residence of the owner.
Three types of “neighborhoods” are considered, the same municipality, the same region
or the same canton. Switzerland is composed of about 2,900 municipalities. According
to the population census of 2000, they range in size from 24 inhabitants (Corippo in the
Verzasca Vally in Ticino) to 368,875 inhabitants (Zurich). A total of 1,053 municipalities
are represented among participants of the Swiss Household Panel. A regional division of
Switzerland, consisting of 106 so-called spatial mobility (mobilité spatiale, or MS) regions,
was developed by the Swiss Federal Statistical Office for the original purpose of analyzing
spatial mobility between comparable micro regions (SFSO, 2005). Finally, Switzerland is a
confederation of 26 cantons whose role and political functions pretty closely resembles that
of the States within the U.S.; Again, population sizes vary considerably, from a mere 15,199
in the canton of Appenzell Innerrhoden to 1.3 million in the canton of Zurich.
Exposure to conspicuous car consumption is defined as the number of newly registered
Ferrari and Porsche cars per year per 1000 population in the municipality, region or canton
the individual lives in. This information is available for the years 2001 and 2007. In 2001,
the number varied between 0 and 7 at the municipal level, between 0 and 2 at the regional
level, and between 0.09 (canton of Berne) and 0.65 (canton of Zug) at the cantonal level.
Over the whole of Switzerland, a total of 1,324 Ferrari and Porsche cars was newly registered
during the year 2001, corresponding to a ratio of about 0.24 per 1000 population.
− − − − −− Figure 1 about here − − − − − − −
By 2007, the nationwide number of new registrations of Ferraris and Porsches had increased by 72 percent to 2,281. The “least conspicuous” canton (with a rate of 0.08) was now
the canton of Glarus. In Zug, again by far the canton with the highest density of such new
9
sport cars, the number had almost doubled to 1.27. It has to be pointed out that the number
of newly registered cars may not be the best measure for conspicuous car consumption, as
it ignores the stock of existing Ferraris and Porsches. Unfortunately, such stock data are
not kept on file by the Federal Roads Office. In a steady state, the stock is proportional
to the flow. A problem arises however if registrations are sufficiently rare, since a zero flow
(during a year) is a poor proxy for the accumulated stock. This censoring problem is most
severe when considering municipal level data, which renders the use of data at this level of
disaggregation potentially less informative than the use of data for regions or cantons.
Local cost of living indicators were constructed using information on rentals paid for
private use apartments and houses. There are two principal sources for such information.
The first is the largest online property website in Switzerland, homegate.ch, which made
available all ads that had appeared during 2002 and 2007. The strength of this datasource
is the sheer number of observations. There were 90,669 relevant ads in 2002, and 165,757
in 2007, which gives sufficient information to compute price indices, adjusting prices for the
number of rooms and the property type, at the municipal level. The disadvantage of this
datasource is that current market rentals are often substantially above rentals paid by longterm tenants. The indices for the regional and cantonal level are therefore based on the Swiss
Labor Force Survey, with annually about 50,000 participants, that provides information on
actual rents.
After merging the SHP data with the two indicators for conspicuous consumption and
cost of living, and restricting the sample to those aged between 20 and 70 with non-missing
responses on the key variables, there are 8,827 observations 3,823 for the year 2002 and 5,004
for the year 2007. The increase in the sample size is explained by a refreshment sample drawn
in 2004. The panel is highly unbalanced, as less than one third of all persons (2065) are
included in both years.
10
4
Empirical Models and Results
The effect of conspicuous consumption on happiness is estimated by exploiting regional
variation in the density of Ferraris and Porsches.
Are those who live in a municipal-
ity/region/canton with relatively more such luxury cars less happy than others, and if so,
how large is the effect of conspicuous consumption on well-being on others? We would like
to estimate the effect “all else equal”, and therefore use regression analysis to account for potential confounders. Among them, income is perhaps the most important variable to include,
since it is, first, a proxy for the own consumption level, and, second, a natural monetary
comparison scale in order to gauge the magnitude of the consumption externalities. Other
controls are the usual individual social demographic variables, then local variables such as
cost of living and population size, and finally regional fixed effects.
Since the dependent variable, happiness, is ordinal, taking integer values from 0 to 10,
the regression is specified as an ordered logit model and estimated by maximum likelihood.
In such models, coefficients indicate the relative change in the odds of reporting a happiness
greater or equal to j, rather than less than j, where j is any value between 1 and 10. A
positive coefficient thus means that a person is more likely to be very happy if the associated
regressor increases. Rather than reporting marginal effects, it is more natural in the present
context to directly monetize the implied welfare loss (if any) from conspicuous consumption
by relating the estimated effect to the income coefficient.
Table 1 displays the basic ordered logit results. The first three columns show the conspicuous consumption effects for the three regional aggregation levels (municipality, region,
canton) in a bare bone model, with logarithmic income as the only additional control variable. The estimated coefficient of income on happiness is around 0.5 in all three equations.
Since income enters the model on a logarithmic scale, this coefficient tells us that a doubling
11
of income (a change in logarithmic income by 0.69), moves the linear index of the ordered
logit model by an estimated +0.35. The effect is highly significant, and it can be taken as
point of comparison for assessing the magnitude of conspicuous consumption effects. The
modelling of the Ferrari and Porsche density (FP) differs between the municipal model on
one hand, and the regional and cantonal models on the other. In the former case, 37% of all
values are zero. Thus, rather than using only FP, a second variable is included that indicates
whether or not there are any FPs, thus allowing for a discontinuity of the effect at zero.
− − − − −− Table 1 about here − − − − − − −
In all three models, a significant negative effect of a higher FP can be detected. At the
level of the municipality, zero FP is predicted to increase the linear index by about 0.2, more
than half of what the effect of a doubling income would be. The linear term is statistically
insignificant at the municipal level. However, the FP point estimates are -0.5 and -0.7, and
statistically significant, at the regional and cantonal aggregation level, respectively. These
are obviously large estimates of the negative happiness externality caused by one additional
Ferrari or Porsche per 1000 population.
Columns 4 to 6 of Table 1 repeat the analysis with a number of individual control variables
included. For instance, the effect of being married is close to +0.5 – equivalent to a more than
doubling of income. Moving down and across the columns shows the usual stable patterns
that are rather insensitive to specification: the happiness - age profile is convex; men are
less happy than women; the effect of unemployment is negative and very large. Perhaps a
new finding, worth further explorations, the French speaking population within Switzerland
is, ceteris paribus, considerably less happy than the German speaking population. Including
these variables attenuates the FP effects somewhat, e.g. to -0.4 in the regional equation and
to -0.6 in the cantonal equation, but they remain large and statistically significant.
12
In the models discussed so far, there is a plausible concern that the FP effect may in
truth capture regional differences other than that of conspicuous consumption. It is for
example to be expected that the incidence of conspicuous consumption increases with the
prosperity of a region which, in turn, means that living expenditures, mainly housing, will
tend to be higher. In order to alleviate the concern of misinterpreting the FP effect, the
three equations were re-estimated with regional indicators of living cost and logarithmic
population size included. The results are shown in the left panel of Table 2. Essentially,
the FP effects remain statistically significant, although their size has once more decreased.
At the municipal level, the estimated “zero-FP” effect is now +0.14; each unit increase in
regional (cantonal) FP is predicted to lower the happiness index by 0.23 (0.31). Of course,
there could be other local differences not captured by these two controls. As long as they are
time invariant and common within a canton, the right panel of Table 2 with canton dummies
can account for those. The point estimates (for zero-FP and FP, repectively) remain in the
same range, and they are, with the exception of the third equation, statistically significant.
− − − − −− Table 2 about here − − − − − − −
Overall, the results are strongly suggestive of negative effects of conspicuous consumption
on happiness. While the estimates are associated with a considerable amount of uncertainty,
and their magnitudes depend on specification, the signs are always negative, and the implied
compensating income variations are large by any standard.
It would be attractive to report additional estimates from a full panel model with individual specific effects. This is not done here, however, for two reasons. First, two thirds of
all individuals are included in one year only. A panel analysis would thus lose most of the
observations, making the estimates from the remaining sample so imprecise as to factually
being useless. Second, in a world with potentially diverging short-term and long-run effects,
13
the fixed effects estimator identifies the former, whereas we are more concerned with the
latter.
Some alternative robustness checks are reported in Table 3. Specifically, interactions
between FP and three variables were considered: being male, having a left political orientation, and being poor (here: a household equivalent income in the first quartile of the income
distribution). One might expect that men are more susceptible to conspicuous car consumption, as are individuals who see themselves on the right of the political spectrum. However,
these conjectures are not supported by the evidence. Only the third interaction, that between being poor and FP, is found to be statistically significant: the negative consumption
externality appears to work almost exclusively among the poor.
− − − − −− Table 3 about here − − − − − − −
It is perhaps premature to draw specific policy conclusions from these results. The empirical investigation in this section suggests, however, that consumption externalities caused
by envy are an empirically relevant phenomenon that can usefully be studied with happiness
equations.
5
Conclusions
The traditional tools of welfare analysis rely exclusively on behavioral responses. This limits
the scope of questions one can answer. It is for example not possible to quantify the welfare
loss resulting from envy-related overconsumption, or to quantify the value of a public good.
To value such intangibles, the happiness approach offers a promising alternative. Related
examples are van Praag and Baarsma (2005) who use the happiness approach to value the
cost of airport noise, Luechinger (2009) for air quality and Frey, Luechinger and Stutzer
14
(2009) for terrorism.
... to be done: no citations in conclusions
Kuhn et al. lottery winnings do not make neighboring households less happy (six month
after win)
The primary objective of this paper was to explore the happiness approach as a conceptual
framework for thinking about conspicuous consumption and consumption externalities. A
secondary objective was to provide an illustrative application for Switzerland, in order to
determine the welfare costs of luxury cars. The results indicated that people living in an
area with a higher prevalence of luxury cars are indeed less happy, ceteris paribus.
Interestingly, luxury cars have become a recent political issue in Switzerland. The socalled “off-roader initiative” (that mandates for example a sales ban on cars with a weight
exceeding 2.2 tons and puts limits on CO2 emissions) has obtained more than 100’000 signatures of support. A date for the country-wide referendum has not yet been set. Once the
referendum is completed, it will be interesting to compare voting outcomes with luxury car
prevelance measure used in this paper. If people “scratch where it itches”, one should find
a positive association.
15
6
References
Arrow, K.J. and P.S. Dasgupta (2009) Conspicuous Consumption, Inconspicuous Leisure,
forthcoming in: Economic Journal.
Charles, K.K., E. Hurst and N. Roussanov (2009) Conspicuous Consumption and Race,
Quarterly Journal of Economics, 124, 425-467.
Clark, A.E. and A.J. Andrew J. (1996) Satisfaction and comparison income, Journal of
Public Economics, 61, 359-381.
DeLeire, T. and A. Kalil (2009) Does Consumption Buy Happiness? Evidence from the
United States. mimeo.
Di Tella, R., J. Haisken-De New and R. MacCulloch (2007) Happiness Adaptation to Income
and to Status in an Individual Panel, NBER Working Papers 13159.
Dupor, B., and W.-F. Liu (2003) Jealousy and Equilibrium Overconsumption, American
Economic Review, 93, 423428.
Dynan, K.E., and E. Ravina (2007) Increasing Income Inequality, External Habits, and
Self-Reported Happiness, American Economic Review, 97, 226231.
Hsee, C.K., F. Xu and N. Tang (2008) Two Recommendations on the Pursuit of Happiness,
Journal of Legal Studies, 37, S115- S132.
Frank, R.H. (1991) Positional externalities, In: Zeckhauser, R.J. (Ed.), Strategy and Choice,
MIT Press, Cambridge, MA, 2547.
Frank, R. (1999) Luxury fever: Why money fails to satisfy in an era of excess. New York:
Free Press.
16
Frank, R. (2004) How not to buy happiness, Deadalus, 133(2), 69-79.
Frey, B.S. Luechinger and A. Stutzer (2009) The life satisfaction approach to valuing public
goods: The case of terrorism, Public Choice, 138, 317-345.
Johansson-Stenman, O. and P. Martinsson (2006) Honestly, why are you driving a BMW?
Journal of Economic Behavior & Organization, 60, 129-146.
Kahneman, D. and R.H. Thaler (1991) Economic Analysis and the Psychology of Utility:
Applications to Compensation Policy, American Economic Review 81, 34146.
Krueger, A.B. and D.A. Schkade (2008) The Reliability of Subjective Well-Being Measures,
Journal of Public Economics, 92, 1833-1845.
Kuhn, P., P. Kooreman, A.R. Soetevent and A. Kapteyn (2008) The Own and Social
Effects of an Unexpected Income Shock Evidence from the Dutch Postcode Lottery,
Rand Working Paper WR-574.
Layard, R. (1980) Human Satisfactions and Public Policy, Economic Journal, 90, 737-750.
Layard, R. (2005) Happiness Lessons from a New Science, Penguin Press, New York.
Luechinger, S. (2009) Valuing Air Quality Using the Life Satisfaction Approach, Economic
Journal, 119, 482-515.
Luttmer, E.F.P. (2005) Neighbors as negatives: relative earnings and well being, Quarterly
Journal of Economics, 120, 963-1002.
Marx, K. (1987) Wage, Labour and Capital.
Neumark, D. and A. Postelwaite (1998) Relative income concerns and the rise in married
women’s employment, Journal of Public Economics, 70, 157193.
17
Oswald, A.J. and M.D. Rablen (2008) Mortality and Immortality: The Nobel Prize as an
Experiment into the Effect of Status upon Longevity, Journal of Health Economics,
27, 1462-1471.
SFSO (2005) Eidgenössische Volkszählung 2000: Die Raumgliederung der Schweiz, Neuenburg: Swiss Federal Statistical Office.
Solnick, S. and D. Hemenway (1998) Is more always better?: a survey on positional concerns.
Journal of Economic Behavior and Organization, 37, 373383.
Stutzer A. and B.S. Frey (2008) Stress that Doesn’t Pay: The Commuting Paradox, Scandinavian Journal of Economics, 110, 339-366.
van Praag, B.M.S. and B.E. Baarsma (2005) Using Happiness Surveys to Value Intangibles:
The Case of Airport Noise, Economic Journal, 115, 224-246.
Veblen, T. (1925) The Theory of the Leisure Class: An Economic Study of Institutions
(1899, [1925]).
Veenhoven, R. (2008) Healthy happiness: effects of happiness on physical health and the
consequences for preventive health care, Journal of Happiness Studies, 9, 449-469.
Wilkinson, R. (2000) Mind the Gap: Hierarchies, Health and Human Evolution, Weidenfeld
& Nicolson, London.
18
Table 1. Happiness and Conspicuous Consumption (Ordered Logit)
Dep. var.: Happiness
Logarithmic income
Ferrari&Porsche/1000 pop.(FP)
Zero Ferrari&Porsche
munic.
0.465
(0.045)
0.044
(0.033)
0.216
(0.045)
Aggregate variables at level of...
region canton munic. region
0.483
0.472
0.572
0.586
(0.044) (0.044) (0.047) (0.047)
-0.467 -0.680
0.042
-0.355
(0.078) (0.114) (0.033) (0.081)
0.158
(0.046)
Logarithmic household size
0.257
0.268
0.263
(0.054) (0.054) (0.054)
0.482
0.490
0.500
(0.062) (0.062) (0.062)
-0.109 -0.107 -0.109
(0.011) (0.011) (0.011)
0.001
0.001
0.001
(0.000) (0.000) (0.000)
-0.359 -0.341 -0.343
(0.050) (0.049) (0.049)
-1.082 -1.068 -1.062
(0.172) (0.175) (0.174)
-0.183 -0.170 -0.180
(0.043) (0.043) (0.043)
-0.152 -0.125 -0.114
(0.036) (0.036) (0.037)
Married (Yes=1)
Age
Age2 /100
Romandie (Yes=1)
Unemployed (Yes=1)
Male (Yes=1)
Year 2007
Observations
canton
0.576
(0.047)
-0.468
(0.120)
8745
8779
19
8827
8745
8779
8827
Table 2. Happiness and Conspicuous Consumption (Ordered Logit)
Dep. var.: Happiness
Logarithmic income
Ferrari&Porsche/1000 pop.(FP)
Zero Ferrari&Porsche
Cost of living
Log population
Canton fixed effects
munic.
0.556
(0.048)
0.039
(0.048)
0.142
(0.059)
-0.103
(0.078)
-0.006
(0.017)
Aggregate variables at level of...
region canton munic. region
0.608
0.590
0.579
0.617
(0.048) (0.047) (0.043) (0.041)
-0.225 -0.312
0.063
-0.248
(0.097) (0.148) (0.053) (0.118)
0.155
(0.059)
-0.406 -0.488
0.097
-0.376
(0.217) (0.270) (0.088) (0.296)
-0.075 -0.037
0.003
-0.022
(0.031) (0.030) (0.018) (0.035)
No
No
No
Yes
Yes
canton
0.602
(0.041)
-0.505
(0.344)
0.474
(1.310)
-0.109
(0.296)
Yes
Table 3. Effect heterogeneity by gender, political orientation, and poverty
Ratio Ferrari&Porsche(FP)
Male
Male×ratio
Politically left
Left×ratio
Poor
Poor×ratio
region
-0.230
(0.114)
-0.201
(0.059)
0.101
(0.149)
Aggregate variables at level of...
canton region canton region canton
-0.287 -0.213 -0.269 -0.043 -0.067
(0.169) (0.105) (0.157) (0.098) (0.148)
-0.209
(0.075)
0.095
(0.219)
-0.178 -0.172
(0.061) (0.077)
0.067
0.046
(0.156) (0.227)
0.006
0.047
(0.082) (0.099)
-0.595 -0.759
(0.183) (0.268)
20
Fig 1. Cantonal ranking in luxury car registrations per 1000 pop., 2001
.4
.2
be lu ur ar ju bs so fr ow gr gl ne sg vs sh tg ag ai bl nw vd zh ge ti sz zg
Source: Federal Roads Office, own calculations.
21
0
Newly registered Ferrari & Porsche
.6