25
Empirical tests of status consumption: Evidence from women’s cosmetics Angela Chao a , Juliet B. Schor b,* a Foremost Maritime Group, 310 Madison Ref. Avenue, New York, NY 10017, USA b Women’s Studies, Harvard University, 12 Quincy Street, Barker Center, Cambridge MA 02138, USA Received 5 February 1996 Abstract Nearly all applied work in consumer demand assumes that the preference functions of in- dividuals are independent, thereby ruling out status consumption and other inter-personal comparison motivations. Surprisingly, the validity of the ‘‘independence’’ assumption has not been tested. However, it is feasible to conduct tests which dierentiate between status-mo- tivated demand and consumer demand in which no status motivations are present. This paper provides such a test, which is based on the fact status consumption ordinarily occurs only with publically visible products. We investigate brand buying patterns among four cosmetics prod- uct, and find, as hypothesized, that visible status goods have a lower price-quality correlation (i.e., a higher status premium) and that the pattern of brand buying favors higher-priced (i.e., status) brands. We also find, as expected, that income and occupational status are positively associated with the propensity to engage in status-purchasing, as are urban and suburban res- idence, and being a Caucasian. Ó 1998 Elsevier Science B.V. All rights reserved. PsycINFO classification: 3900 Keywords: Status consumption; Cosmetics; Conspicuous consumption; Consumer preferences Journal of Economic Psychology 19 (1998) 107–131 * Corresponding author. Tel.: 617 495 9022, x 9199; fax: 617 496 9855; e-mail: [email protected]. 0167-4870/97/$17.00 Ó 1998 Elsevier Science B.V. All rights reserved. PII S0167-4870(97)00038-X

Empirical tests of status consumption: Evidence from women's cosmetics

Embed Size (px)

Citation preview

Page 1: Empirical tests of status consumption: Evidence from women's cosmetics

Empirical tests of status consumption: Evidence fromwomen's cosmetics

Angela Chao a, Juliet B. Schor b,*

a Foremost Maritime Group, 310 Madison Ref. Avenue, New York, NY 10017, USAb Women's Studies, Harvard University, 12 Quincy Street, Barker Center, Cambridge MA 02138, USA

Received 5 February 1996

Abstract

Nearly all applied work in consumer demand assumes that the preference functions of in-

dividuals are independent, thereby ruling out status consumption and other inter-personal

comparison motivations. Surprisingly, the validity of the ``independence'' assumption has

not been tested. However, it is feasible to conduct tests which di�erentiate between status-mo-

tivated demand and consumer demand in which no status motivations are present. This paper

provides such a test, which is based on the fact status consumption ordinarily occurs only with

publically visible products. We investigate brand buying patterns among four cosmetics prod-

uct, and ®nd, as hypothesized, that visible status goods have a lower price-quality correlation

(i.e., a higher status premium) and that the pattern of brand buying favors higher-priced (i.e.,

status) brands. We also ®nd, as expected, that income and occupational status are positively

associated with the propensity to engage in status-purchasing, as are urban and suburban res-

idence, and being a Caucasian. Ó 1998 Elsevier Science B.V. All rights reserved.

PsycINFO classi®cation: 3900

Keywords: Status consumption; Cosmetics; Conspicuous consumption; Consumer preferences

Journal of Economic Psychology 19 (1998) 107±131

* Corresponding author. Tel.: 617 495 9022, x 9199; fax: 617 496 9855; e-mail: [email protected].

0167-4870/97/$17.00 Ó 1998 Elsevier Science B.V. All rights reserved.

PII S 0 1 6 7 - 4 8 7 0 ( 9 7 ) 0 0 0 3 8 - X

Page 2: Empirical tests of status consumption: Evidence from women's cosmetics

1. Introduction

Can consumer demand be understood primarily as a relation between in-dividuals and goods, or are social, i.e., inter-individual, determinants of con-sumption crucial? Outside the ®eld of economics, the social dimensions ofconsumption are simply assumed. And even within economics, a pantheonof ``greats'', from Smith to Veblen to Keynes, have noted the importanceof social comparison in consumer behavior. 1 But despite this impressive ped-igree, the concept of social comparison, or ``relative consumption'', as it isknown within the ®eld, has been widely neglected. The hegemony of perma-nent income and life cycle approaches to consumption have coincided withthe near-universal assumption, particularly in applied research, that individ-ual preferences are independent. Thus, positional or ``status'' motives of con-sumers are ruled out of court. 2

In this paper, we present a new approach which we hope will stimulate in-terest in the ``relative'' or ``status'' approach to consumption. We have de-vised a simple method of di�erentiating between situations in whichconsumers make purchases in order to achieve relative position (or whatwe will call ``status'') and situations in which they are purchasing intrinsicproduct attributes (such as quality) or non-social extrinsic attributes. Usingdata on purchases of women's cosmetics across di�erently priced brands,we ®nd considerable evidence of status-buying. Standard theory, in whichthere is no relative standing or status component of the purchase decision,is not supported by our evidence. Having established the existence of statusconsumption, we then proceed to investigate a number of hypotheses aboutthe prevalence of status-consumption among population subgroups. We ®nd,as expected, that income and occupational status are positively associatedwith the propensity to engage in status-purchasing, as are urban and subur-ban residence, and being a Caucasian.

1 Smith notes the importance of relative position in The Wealth of Nations. Following his lead, we can

®nd similar references in the work of J.S. Mill, Marx, Marshall, Pigou, Keynes, Veblen, and Milton

Friedman, among others. For speci®c quotations, see the survey in McAdams (1992), pp. 10±14.2 Indeed, even the longstanding textbook nod to James Duesenberry's relative income hypothesis as an

alternative to life cycle and permanent income theories has disappeared. It has become almost ritualistic in

papers on relative preferences to note the discipline's neglect of this approach. See McAdams (1992), p. 10

and Frank (1985a), p. 101.

108 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 3: Empirical tests of status consumption: Evidence from women's cosmetics

2. Relative preferences and status consumption

Modern consumer theory is quite general, and as such is compatible with avariety of speci®c models of consumer motivation. However, as noted above,applied work on consumer demand nearly always assumes that individualpreferences are independent. For example, a recent survey by Deaton(1992) does not mention interdependent preferences.

However, the importance of status-seeking and comparative processesacross individuals was once an important topic within economics. The classicstatement is of course that of Veblen (1967, orig. 1899), whose Theory of theLeisure Class introduced the concept of ``conspicuous consumption''. Veblenbelieved that individuals become subject to ``invidious distinctions'' (pp. 16±21), which form the ``conventional basis of esteem'' (pp. 28,29), and ``self-res-pect'' (p. 31), thus prompting ``conspicuous consumption,'' i.e., publicallyvisible consumption which yields status for the individual. Thus, spending be-comes an essential element of establishing one's social position, and the in-trinsic attributes of goods are less important than their social meaning.

Following the tradition of Veblen, Harvard economist James Duesenberryargued in a well-known book published in 1949, that an individual's utilityfrom any given level of consumption depends not only on the absolute levelof spending, but also how that spending compares to that of others. This wasdubbed the ``relative income'' approach, and is also known in popular parlanceas the process of ``keeping up with the Joneses''. In Duesenberry's formulation,households are exposed (through ``demonstration e�ects'') to the consumptionpatterns of those in their reference groups, and seek to replicate those patterns.

The legacies of Veblen and Duesenberry have been ``relatively'' neglected,however, there are some exceptions, most of them theoretical. These includeEasterlin (1995), Hirsch (1976), Frank (1985a, b), Congleton (1989) andRauscher (1993). See also James (1987) for the developing country context.Status games concerning investment in capital goods are treated by Abel(1990) and Galli (1994). Among the few empirical treatments in this areaare Neumark and Postlewaite (1995), and Clark and Oswald (1996), bothof whom ®nd evidence which supports the importance of ``relativity'', i.e.,preference inter-dependence. 3 See also Alessie and Kapteyn (1991), Kapteyn

3 Kosicki's approach is not decisive. He tests the proposition that savings levels should be equal across

income classes, a prediction of life cycle/permanent income approaches, and does not ®nd equality. See

Frank (1985a), pp. 109±111 for a review of the evidence and Kosicki (1987, 1990). However, this criterion

is indecisive because if not all consumption is status consumption, then lower income groups could have

lower marginal propensities to save, but also be engaging in less status consumption.

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 109

Page 4: Empirical tests of status consumption: Evidence from women's cosmetics

and Herwaarden (1980), Basmann et al. (1988) and Brown's (1994) interest-ing measures of ``status'' consumption in middle class American familiesthroughout the twentieth century.

Let us begin by de®ning theories of consumption based on relative stan-ding. In a standard formulation, an individual's utility depends negativelyon the consumption of others.

Ui � f Ci

Xn

j�1

,aCj

!f 0 > 0; �1�

where U is utility, C is consumption and a is the weight applied by individuali to j's consumption. Thus, individual i's utility depends on the ratio of his orher own consumption to a weighted sum of others' consumption. Of course,other formulations are possible. Utility may depend on not having less thanan average level of consumption, for example. The common view is that forVeblen utility depends on exceeding the consumption level of others (con-suming to achieve a superior status position), while for Duesenberry it de-pends on not having lower consumption than others (``keeping up'' tosome standard). For our purposes, we ignore these possibilities in favor ofthe symmetrical formulation of Eq. (1). 4 In so doing, we want to stress thatwe do not attach any negative connotations to the term ``status''. 5

4 The issue of ``reference groups'' (i.e., the identity of the j's) is not well understood. With whom are

people comparing themselves? It has been often noted that comparisons are made locally, rather than

globally. In related research we are gathering empirical evidence on the constitution and dynamics of

reference groups. See Schor (1998). See also Tefertiller (1994) on reference groups among middle-income

American teens, and the impact of a national mass media on a formation of reference groups. See also

Festinger (1954), Park and Lessing (1977), Bearden and Etzel (1982) and Bearden et al. (1989) among

others.5 This point is not uncontroversial. For example, Lovell (1994) has argued that Veblen and what he calls

the Duesenberry±Sen approach, are opposed. In Veblen's conspicuous consumption, one enjoys

consuming more than others of one's reference group, whereas with Dueseberry±Sen, one su�ers a loss

of utility if one consumes less than others. We do not believe such an interpretation is supported by the

relevant texts. Duesenberry, like Veblen, cites the ``maintenance of self-esteem'' as the motive for acquiring

goods, ``the di�erentiated social status'' which characterizes our society (p. 28,29), and the idea that ``high

standards of consumption become established as criteria for high status.'' (p. 30). Veblen's own caveats

(see note 7, and his discussion of ``waste'', pp. 97±101) also suggest an interpretation similar to that of

Duesenberry. Furthermore, because we cannot distinguish between them in our empirical work, we have

chosen not to make a distinction between consuming to gain status and consuming to avoid a loss of

utility.

110 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 5: Empirical tests of status consumption: Evidence from women's cosmetics

Where consumption is for the purpose of gaining status or relative position(hereafter called ``status consumption''), at least two preconditions musthold. 6 First, individuals must share some degree of commonality in theirranking of the relative desirability (or status) of products and brands. Sec-ond, consumption of the products must be socially, or publically, visible. 7

(Hence the terms ``conspicuous consumption'' and ``demonstration e�ects''.)Social visibility is important because of the moral hazard problem associatedwith non-visible goods. Moral hazard arises because individuals have an in-centive to exaggerate their consumption in order to gain social position. Ifconsumption is not externally veri®able, self-reported levels of consumptionare not credible. A second issue is that self-reporting about consumption pat-terns (i.e., boasting) reveals the individual's concern with status, which insome contexts undermines status. (Appearing not to care too much aboutstatus is often necessary to attaining it.)

For these reasons, social visibility is a key dimension of status consump-tion. Casual inspection of consumption patterns suggests as much, andmay well explain why households spend much more money furnishing publicrooms (living and dining rooms) than private ones (bedrooms); why expen-sive ``under-the-hood'' automotive options (four-wheel drive, anti-lockbrakes) are now advertised on the outside of cars; or why the expansion ofdesigner fashion apparel to a large and informationally less savvy populationled to the widespread use of logos. (When designer apparel was worn only bya small group of the wealthy, the identity of designers was recognizable to allparticipants in the status competition.) By contrast, if utility does not dependon others' consumption, and the standard approach to consumption is correct,the level of social visibility of a product should be irrelevant in the pattern ofconsumer purchases.

We are able to exploit this di�erence in theoretical perspectives to designsimple tests which allow us to di�erentiate between consumption in which sta-tus-buying is present and consumption in which status-buying is absent (e.g.,the standard formulation). First, we chose a group of products with similar

6 In light of the perjorative meanings Veblen's terms have acquired in contemporary usage, it is worth

noting that he explicitly argued against such an interpretation. Discussing the term ``invidious'', Veblen

(1967), p. 34 said: ``There is no intention to extol or depreciate, or to commend or deplore any of the

phenomenon which the word is used to characterize. The term is used in a technical sense as describing a

comparison of persons with a view to rating and grading them in respect of relative worth or value...''.7 We do not intend to imply that all consumption is driven by relative concerns. The degree to which

``relativity'' matters varies across commodities. Furthermore, multiple consumption motivations for a

single commodity are of course possible, indeed are probably the norm.

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 111

Page 6: Empirical tests of status consumption: Evidence from women's cosmetics

use and function, but which vary on the dimension of social visibility. (Ideally,this would be the only dimension on which they vary.) Then we ranked theproducts by the social visibility dimension through an informal survey. Hy-pothesizing that the extent of status consumption varies positively with the ex-tent of social visibility, we measured status consumption in two ways.

The ®rst is through a ``status-premium'', which we de®ne as an incrementto the price which exceeds all quality-based price di�erentials. Productswhich are only used privately and are not exposed to public view shouldcarry a zero or small status premium, in comparison to products whichare publically displayed and utilized. The second form of measurement isto analyze patterns of brand-buying. Where status consumption is occur-ring, patterns of brand buying should be skewed to higher priced brands,for any given level of quality. This is because in the usual case, higherpriced brands carry more status. These two considerations (the existence/size of a status-premium, and the pattern of brand buying) raise the possi-bility of two testable hypotheses.

Hypothesis 1. The correlation between price and intrinsic quality is lowerwith products where status consumption is occurring than with productswhere no status consumption is occurring.

Hypothesis 2. There is more purchasing of higher priced brands with productswhere status consumption is occurring than with products where no statusconsumption is occurring.

In both cases, we use social visibility as a proxy for the extent of statusconsumption. Thus, the hypotheses can be re-stated as follows:

Hypothesis 1(a). The correlation between price and intrinsic quality variesnegatively with social visibility.

Hypothesis 2(a). The likelihood of purchasing expensive brands variespositively with social visibility.

If these two hypotheses are supported by the evidence, there is then astrong case that status consumption is occurring. By contrast, neithershould be supported if the standard theory, without social comparison ef-fects, is correct. (For a discussion of alternative explanations, see the pen-ultimate section.)

112 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 7: Empirical tests of status consumption: Evidence from women's cosmetics

3. Who buys status?: Income, education, residence and race

So far we have merely addressed the presence of status-motives in purchasebehaviors. Uncovering evidence for the existence of such motives is of greatrelevance to the discipline of economics, where they are typically assumed notto exist. Outside the ®eld of economics, however, status-seeking is taken forgranted. Nevertheless, despite what is now a longstanding presumption, thereis relatively little empirical evidence about the nature and distribution of sta-tus-seeking behavior. Are some groups of individuals more likely to seek sta-tus? What socio-economic variables a�ect the propensity to seek status? Arethere predictable dynamics to positional competitions?

The literature on status consumption does not give us extensive guidance inthese areas, but does contain some hints about factors which in¯uence the``demand for status''. Here we consider four: education, income, race, and ur-ban/rural residence. Following Veblen (1967), we predict that more highly ed-ucated people are more likely to engage in status buying. At ®rst blush such anassumption may seem counter-intuitive. This is because highly educated peo-ple may be assumed to be more highly educated consumers, and thus more at-tuned to intrinsic product attributes. However, this perspective neglects thevery real social and economic bene®ts achieved through status consumption.In addition to obvious social bene®ts (e.g., membership in desirable, but ex-clusive social groups), status consumption can be associated positively witheconomic returns (i.e., earning power), through signalling e�ects. Status-seek-ers may spend money to ``signal'' their success, and by implication, their owncompetencies. Owning an expensive car or wearing designer suits can enhancean individual's professional or career position, and thereby raise income.Thus, engaging in status consumption is not necessarily devoid of ``utilitari-an'' dimensions for the individual, but may be a deliberate strategy. (In thedata we use, we cannot distinguish between those individuals who desire sta-tus for its own sake and those who are merely trying to send positive ``sig-nals''.) Higher-educated individuals are better positioned to take advantageof these bene®ts. Because the importance of status signalling is much greaterin high prestige occupations, and these occupations require higher levels of ed-ucation, we expect a positive correlation between education and status-seek-ing. A second possible connection is that the education process itselfinduces people to care about status, through the inculcation of middle classvalues, and that this is increasingly true as one ascends the educational ladder.

A second variable we consider is income. Duesenberry (1949) argued thatincome and status-seeking are correlated. This may be because of the corre-

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 113

Page 8: Empirical tests of status consumption: Evidence from women's cosmetics

lations between income, education, and occupation. But income itself mayalso a�ect status purchasing, a view which is suggested by the theoreticalmodels of Congleton (1989) and Rauscher (1993). In these models there existsboth an ``ampli®cation'' e�ect (higher status consumption by others increasesthe marginal return to additional status consumption for an individual andtherefore encourages status consumption) and a ``discouragement'' e�ect(in which people drop out of the game on account of the extra status seekingactivities of others). We hypothesize that people with lower income will dis-play larger discouragement e�ects, and are thus, on average, less likely to en-gage in status consumption.

For similar reasons, (and following Duesenberry) we also predict that non-Hispanic whites do more status buying than Hispanics and Blacks. The argu-ment about racial groups stems from the economically and socially dominantposition of Caucasians in American society. Because Blacks and Hispanicsoccupy, on average, a lower position on the social ladder, it is more di�cultfor them to achieve a given status position. Therefore, discouragement e�ectsare more prevalent. Duesenberry found evidence that, controlling for income,Blacks save more than Whites, thereby supporting the view that less ``keepingup with the Joneses'' was occurring.

Finally, we expect that an urban setting leads to a greater need to use visibleconsumption to gain status, following Veblen. This expectation is based on theobservation that the social culture of urban areas is more ¯uid, thereby makingnon-consumption dimensions of status less salient. In a rural area, status-char-acteristics such as parental class and occupation, and own class and occupa-tion, are widely known. Therefore it is more di�cult for individuals to gainsocial status by spending money. By contrast, in an urban setting, higher levelsof anonymity make consumption-based status more possible. Furthermore,lower levels of social familiarity and knowledge also make it more imperativeto use products to establish identity and status. We expect that suburban loca-tion would be intermediate between rural and urban in this regard.

Hypothesis 3. The variables income, education, being a non-HispanicCaucasian, and living in an urban or suburban area are all positivelyassociated with the propensity to engage in status-consumption.

4. Testing for the existence of status consumption

In this section we test Hypotheses 1(a), 2(a), and 3. We have chosen wom-en's cosmetics as the broad commodity category for our inquiry. This is for

114 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 9: Empirical tests of status consumption: Evidence from women's cosmetics

two reasons. First, in the popular literature, descriptions of the industrystress the lack of intrinsic product quality and the importance of marketingand advertising. (See Allen, 1981; McKnight, 1989; Wolf, 1991.) Second,and more importantly, this was the only commodity category for which wewere able to readily ®nd su�cient detail on brand-buying to carry out ourtest. In the following, we distinguish between ``products'' (particular typesof cosmetics); ``brands'' (the name or marketer of the product) and ``lines''(speci®c varieties of product within a brand). We will be looking at four prod-ucts (lipstick, eyeshadow, mascara, and facial cleansers); numerous brands(e.g., L'Oreal, Maybelline, Estee Lauder, Revlon); and some speci®c lineswithin brands (waterproof mascara, long-lasting lipstick).

We have used two sources of data. The ®rst is the Mediamark ResearchIncorporated Adult Personal Care Annual Report, 1989±1991, from whichwe used brand-buying patterns and socio-economic/demographic informa-tion on consumers. Mediamark is a market research ®rm which collects largequantities of proprietary data for sale to businesses. It is one of the few mar-ket research ®rms which also makes this data publically available. Originallywe had hoped to obtain the company's data tapes for this paper, but the costwas prohibitive. Working from the published tabulations entailed certainlimitations on the kinds of models we could estimate.

Our second source of data is Consumer Reports, from which we used priceand product quality data by brand. We matched the two data sources byproduct and year. The Consumer Reports data tend to be somewhat more de-tailed by lines than the Mediamark data, although in cases where one lineholds a large market share it is listed individually in the Mediamark data.Where the two data sets did not match exactly, we averaged the prices ofthe various lines to arrive at an average brand price. Consumer Reportsranked 39 lipsticks, 34 mascaras, 31 eyeshadows and 36 facial cleansers.Matching these against the Mediamark data, we were left with 22 lipsticks,16 mascaras, 18 eyeshadows, and 5 facial cleansers.

4.1. Visibility rankings

Those who are unfamiliar with the habits of cosmetics users may be sur-prised to ®nd marked di�erences in the level of social visibility in the usageof our four products. However, we expected that actual consumers would id-entify clear di�erences. Our a priori view was that lipsticks would have thehighest level of visibility, because women take lipsticks out in public, at res-taurants, in meetings, on subways, airplanes, etc., as well as in ``powder

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 115

Page 10: Empirical tests of status consumption: Evidence from women's cosmetics

rooms''. Facial cleansers, by contrast, are ordinarily left at home and have nopublic visibility. Eyeshadows and mascaras are in an intermediate category.They are mainly taken out in semi-public areas such as powder rooms.Our a priori ranking was con®rmed in an informal survey of 20 female stu-dents at Harvard University (Table 1). Lipstick was ranked most visible byall 20 respondents; facial cleansers were ranked least visible by 18 of 20. 8

Mascara was ranked second, and eyeshadow third.

4.2. Quality rankings

Consumer Reports publishes the results of so-called product ``taste-tests'',using testers who are not given the identity of the brands. The method ofevaluation varied by product. For lipsticks and eyeshadows, professional

Table 1

Visibility rankings by product a survey results

Respondent Lipstick Mascara Eyeshadow Facial cleanser

1 1 2 3 4

2 1 2 3 4

3 1 2 3 4

4 1 2 4 3

5 1 2 3 4

6 1 2 3 4

7 1 2 3 4

8 1 2 3 4

9 1 2 4 3

10 1 2 3 4

11 1 2 3 4

12 1 2 3 4

13 1 2 3 4

14 1 2 3 4

15 1 2 3 4

16 1 2 3 4

17 1 2 3 4

18 1 2 3 4

19 1 2 3 4

20 1 2 3 4

a Respondents were asked to rank the four products by level of social visibility. All respondents were

female Harvard college students. 1�most visible, 4� least visible.

8 The ``public'' nature of status goods has been noted. See James (1987), Bearden and Etzel (1982) and

Holman (1981).

116 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 11: Empirical tests of status consumption: Evidence from women's cosmetics

make-up artists evaluated the products. For mascaras, 22 women whoordinarily wear mascara used and evaluated the products. For facial cleans-ers, a panel of 90 women were recruited who used the products over a 10week period. Laboratory tests were also conducted on mascaras. In all cases,testers were asked to rank the products on a variety of dimensions (e.g., easeof application, various appearance categories, ease of removal, durability,smell, e�ect on skin).

The results of these rankings are presented in Fig. 1: The raw data forFig. 1 is given in Appendix A. Strikingly, lipstick brands, ranging in pricefrom roughly $1.00±10.00, were ranked as equivalent, that is, with no qualitydi�erentials. 9 Consumer Reports also conducted a one-on-one test betweenFlame Glow (a low-end brand with little name recognition) and Chanel (per-haps the most prestigious of major brands), using the brand containers, butswitching half the lipsticks inside. About as many women preferred FlameGlow as preferred Chanel. At the time, Flame Glow cost $2.26 for 0.12 oz($18.83/oz) and Chanel cost $14.50 for 0.08 oz ($181.24/oz). In the eyeshad-ow tests, identical quality rankings also emerged for all brands. With mascar-as, a 0.424 correlation between price and quality was found; for facialcleansers the correlation was 0.408.

These results provide support for Hypothesis 1. The most highly visiblegood (lipstick) has a zero price/quality correlation, whereas the least visibleproduct (facial cleansers) has a positive correlation between price and qual-ity. It seems that women demand more quality with the non-visible product.The results for mascara are somewhat anomalous, in that they reveal a higherprice/quality correlation than we would have expected. This may be due tothe fact that mascaras were the only one of the four products to undergo lab-oratory testing, so that the evaluation procedures were not identical. In themascara rankings, we can also see evidence of heteroskedasticity, with a hor-izontal relationship among the low-priced brands and a slight downwardsloping relation in the higher-priced ones.

4.3. Patterns of brand buying

To assess di�erences in patterns of brand buying (Hypothesis 2), we usedthe Mediamark survey. For lipsticks, the data is from 1989, and covers

9 The two respondents who ranked eyeshadow last said they did so because they did not know anyone

who used it.

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 117

Page 12: Empirical tests of status consumption: Evidence from women's cosmetics

Fig. 1. Price vs. quality.

118 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 13: Empirical tests of status consumption: Evidence from women's cosmetics

12,573 respondents. For mascaras and eyeshadows the data is from 1990 andthere were 12,724 respondents. And for facial cleansers (1991), there were12,146 respondents. The published data is only available in cross-tabulatedformat. Thus, for each brand we know the percentage of women (amongthe 12,000-plus) who bought that brand. For each of the demographic or so-cio-economic categories (e.g., by education or income), we have the fractionof women in that group who bought each brand.

In Fig. 2, we present the brand buying patterns for the four productsamong the entire sample of women. The row data for Fig. 2 is given in Ap-pendix B. We used two statistical measures to determine the existence of sta-tus consumption. First, we tested whether or not the percentage of womenbuying expensive brands (de®ned as the top three brands) increases withthe visibility of the product. We ®nd this is the case. The percentage of wom-en buying the top three brands of facial cleansers, eyeshadows, mascaras, andlipsticks, were 10.88%, 13.19%, 16.19%, and 17.94%, respectively. The chi-square statistic indicates signi®cant di�erences by product visibility in thepercentage of women buying the top three brands, with a p-value less than0.0005. Second, we tested whether or not women within each income levelbought more of the most expensive brands (de®ned as the top ®fth of a prod-uct's brands, ranked by price). Again, the chi-square statistic is highly signi-®cant, with a p-value of less than 0.0005. Within each income level, thefraction of women buying an expensive visible product is signi®cantly morethan that of the less visible product.

In Table 2, we present simple ordinary least squares regressions for each ofthe four products, using the entire sample of respondents. The dependentvariable is constructed by taking one observation for each brand and is mea-sured as the weighted percentage of women buying that particular brand (ofeach of the four products). The independent variable is the price of thatbrand. So, for example, in the lipstick regression, there are 22 observations,in which the dependent variable is the fraction of women who purchased eachbrand of lipstick and the independent variable is the price of that brand. 10

To test for the presence of status consumption we compare the coe�cientsof the price variable across equations. The null hypothesis is that no status-consumption is occurring. If it is correct, the coe�cients on the price vari-

10 The price ranges for the four products were as follows: lipstick $0.81±10.80; mascaras $2.88±12.33;

eyeshadows $1.97±15.00 and facial cleansers $2.72±17.00. The brands are only those for which we had

both Mediamark and Consumer Reports data.

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 119

Page 14: Empirical tests of status consumption: Evidence from women's cosmetics

Fig. 2. Product distribution graphs.

120 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 15: Empirical tests of status consumption: Evidence from women's cosmetics

ables should be equal, or should at least not decline as the visibility of theproduct decreases. If status consumption is occurring, the coe�cient on priceshould decrease monotonically as visibility falls.

This is exactly what we ®nd. The coe�cient on price (b1) clearly varies byvisibility (i.e., from lipsticks to facial cleansers), getting both more negative(from )0.157 to )1.57) and more signi®cant (t-statistic goes from )0.407to )3.79). Thus, price is a signi®cantly more important determinant of de-mand for less visible products. Indeed, for lipsticks, price is not even a signi-®cant negative determinant of quantity demanded. By contrast, it does have asigni®cant negative impact for the less socially visible products. Notice alsothat for the two intermediate goods, the price coe�cient is measured at anintermediate level. Indeed, the linearity of estimates for the price coe�cientis rather striking, and correlates strongly with the a priori rankings of visibil-ity. These regressions provide strong support for the existence of status buy-ing in women's cosmetics products.

5. Testing for variations in the propensity to purchase status: Education,income, residence and race

As discussed above, we expect that the prevalence of status consuming var-ies among groups within the population. In particular, we hypothesized thatthe prevalence of status consumption is positively correlated with income andeducation, that non-Hispanic Whites are more likely to engage in status con-sumption, and that persons living in urban and suburban locales do more sta-tus buying. These hypotheses are tested in the regressions presented inTables 3±6. (Because our data are in cross-tabulated form, we cannot doone regression with all these independent variables.) And because the dataare only available as the fraction of women buying a particular brand, we

Table 2

Regressions of brand-buying

Lipstick Mascara Eyeshadow Facial cleanser

Constant 6.20 9.27 9.53 23.85

(2.89) (3.84) (7.09) (6.78)

b1 (brand price) )0.157 )0.464 )0.420 )1.57

()0.407) ()1.28) ()2.62) ()3.79)

t-statistics are in parentheses.

Dependent variable is weighted log of the percentage of women buying each brand. Independent variable

is the price of the brand. X1� brand price.

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 121

Page 16: Empirical tests of status consumption: Evidence from women's cosmetics

used a weighted logistic regression. The data we use are identical to thoseused above (Table 2), but in this case we add dummy variables for the vari-ous socio-economic and demographic groups.

The dependent variable is the logic transformation on the percentage ofwomen who buy a particular brand. The independent variables, in Ta-bles 3±6, are education, income, race, and urban/suburban dwelling, inthe form of dummy variables. The ®rst equation, for education, is the fol-lowing (de®nitions of the other dummy variables are given in the respectivetables):

Y � b0 � b1X1 � b2X2 � b3X3 � b4X4 � l; �2�where

Y � log np�1ÿ p�;where, p is the fraction of women buying a speci®c brand within a given ed-ucational level; n is the weighting by the number of women buying the brand;X1 is the price of brand; X2� 1 if graduated high school, 0 otherwise; X3� 1 ifattended college, 0 otherwise; X4� 1 if graduated college, 0 otherwise, and lis the error term. (The omitted category is attended high school.) 11

In all cases, we also added interaction terms between the price and the oth-er variables in order to determine whether the coe�cient on the price termvaries with the X's. If the coe�cient on price becomes less negative as educa-tion rises, for example, that implies a higher level of status buying by educa-tion. In each case we performed a generalized F-test to see if the interactionterms added explanatory power to the equation.

We begin with the education results in Table 3. We ®nd that the level ofeducational attainment is a signi®cant predictor of lipstick purchasing pat-terns. Considering the regression including the interaction terms, we ®ndthat these terms are signi®cant, and indicate that having more educationleads women to buy higher priced lipsticks. That is, the coe�cient on pricevaries with education level. Evaluating the full coe�cient at a lipstick priceof $1.00, we ®nd that the full price coe�cient equals )0.19 for women witha high school diploma. Among college-educated women, the coe�cient onthe price term is no longer negative, indicating the presence of what has

11 The weight we use is np(1 ) p) where n equals the number of women buying the particular brand of

the product, and p equals the proportion of women buying the brand. This weighting allows us to account

for the fact that the cheaper brands have larger market shares, by normalizing the percentages and

ensuring constant variance in the errors.

122 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 17: Empirical tests of status consumption: Evidence from women's cosmetics

been called in the literature ``snob'' demand, i.e., cases where a higher priceinduces higher demand (Leibenstein, 1950). For women who have attendedcollege, the coe�cient is 0.044, and for women with college degrees, theprice coe�cient is 0.117. Once interaction e�ects are properly accountedfor, the price coe�cient rises steadily as education increases, from )0.118for women who have not graduated high school, to 0.117 to those with col-lege diplomas. Thus, as price rises there is a greater ``positive'' contributionof price to demand.

By contrast, the facial cleansers equation produced very di�erent resultsthan for the other three products. In the regressions without interaction terms,the coe�cient on the price term is more positive for lipsticks (0.006) than forfacial cleansers (0.001). Furthermore, the additional explanatory power of theinteraction terms is lower for mascara and eyeshadows, than for lipsticks, andthere is no additional explanatory power from the interaction terms in the fa-cial cleansers regressions. (The F-test is )0.644, which is not signi®cant at

Table 3

Brand-buying by education level

Lipstick Mascara Eyeshadow Facial cleanser

Constant )1.21 )0.715 )0.168 )1.31 )1.60 )1.19 )0.887 )0.831

()12.38) ()6.37) ()18.79) ()10.08) ()15.17) ()9.52) ()8.66) ()4.7)

b1 0.006 )0.118 0.004 )0.077 0.007 )0.098 0.001 )0.011

(0.43) ()4.96) (0.37) ()3.12) (0.60) ()3.88) (0.06) ()0.31)

b2 0.897 0.514 1.35 1.07 1.21 0.942 0.578 0.357

(8.59) (0.146) (14.72) (6.74) (10.05) (6.15) (7.87) (1.49)

b3 )0.244 )0.926 0.432 0.001 0.330 )0.192 )0.741 )0.703

()2.1) ()5.66) (4.41) (0.01) (2.56) ()1.17) ()8.66) ()2.51)

b4 )0.414 )1.50 0.263 )0.403 0.165 )0.580 )1.12 )1.02

()3.38) ()8.48) (2.62) ()2.35) (1.24) ()3.43) ()12.0) ()3.3)

b5 0.099 0.062 )0.074 0.047

(3.36) (2.13) (2.57) (0.97)

b6 0.162 0.092 0.125 )0.083

(5.09) (3.04) (4.19) ()0.15)

b7 0.235 0.136 0.165 )0.022

(0.038) (4.45) (5.48) ()0.35)

Generalized F-

test

18.91 7.27 12.24 )0.644

t-statistics in parentheses.

Dependent variable is weighted log of the percentage of women in each educational category who buy the

brand, X1� brand price, X2 � 1 if graduated high school, 0 otherwise, X3� 1 if attended college, 0 other-

wise, X4 � 1 if graduated college, 0 otherwise, X5�X1 ´ X2, X6�X1 ´ X3, X7�X1 ´ X4. Omitted catego-

ry is attended high school.

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 123

Page 18: Empirical tests of status consumption: Evidence from women's cosmetics

standard con®dence levels.) The failure of the interaction terms to add explan-atory power indicates that education level does not a�ect the price elasticity.By contrast, with mascaras and eyeshadows, we also ®nd evidence of statuspurchasing, but less than with lipsticks. The full price coe�cient (evaluatedat a product price of $1.00), varies from )0.077 for the lowest educationalgroup to 0.059 for the highest. Similarly, for eyeshadows, the variation is from)0.098 to 0.067. ``Snob'' demand consistently becomes more prevalent as ed-ucational level rises. Thus, considering the four products, we see that the ex-tent of status buying varies with the visibility of the product.

In Table 4, the income regressions, the ®ndings are similar. Women withhigher incomes do more status buying. In the regressions with interactions,the price coe�cient (evaluated at a lipstick price of $1.00) declines as incomedeclines, from 0.003 (in the highest income category) to 0.014, to )0.001, to)0.011, to )0.16 (for the lowest income category). This is strong evidence forthe variation of status consumption by income group. Furthermore, the ex-tent of status buying does not vary across categories for facial cleansers (un-like for the other products), where we see that the F-test indicates theinteraction terms add no explanatory power to the equation. For the two in-termediate products, we do ®nd evidence of status-buying varying by income,as expected. In the mascara equation (again, evaluating the product at a$1.00 price), we ®nd that the price coe�cient falls from 0.017 in the highestincome group to )0.118 in the lowest group. Similarly, for eyeshadows, thecoe�cient falls from 0.057 to )0.08.

Race is also a signi®cant predictor of status buying (Table 5). Once again,the coe�cient on the price term falls as the visibility of the good declines,from 0.006 for lipstick to )221.3 for facial cleansers. Among White womenthe fraction buying each brand of lipstick rises as the brand price increases.Among Black and Hispanic women the reverse (standard) relation holds. Inthe equations with interaction terms, we see that status buying di�ers by racewith lipstick, but not with facial cleansers. (Interestingly, the semi-visibleproducts did not show race-status e�ects.)

Finally, our results show that women who live in urban and suburban ar-eas are more status-conscious (Table 6). The fraction of women living inthese areas who buy expensive lipsticks is higher than among women in rurallocales. Again, the F-test for the interaction terms with lipsticks is highly sig-ni®cant, while it is not signi®cant for facial cleansers. The results for the semi-visible products are slightly di�erent, with eyeshadow showing status e�ectsthrough the interaction terms, and mascara not. Unfortunately, we cannotsay what is causing this di�erence.

124 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 19: Empirical tests of status consumption: Evidence from women's cosmetics

Table 4

Brand-buying by income level

Lipstick Mascara Eyeshadow Facial cleanser

Constant 0.170 0.072 )1.83 )2.21 )1.82 )2.19 )2.82 )3.13

(1.56) (4.05) ()21.78) ()18.13) ()20.2) ()19.3) ()23.5) ()9.89)

b1 0.003 0.018 0.006 0.073 0.008 0.068 0.004 0.066

(0.18) (6.43) (0.77) (3.98) (1.17) (4.64) (0.25) (1.09)

b2 )0.722 0.010 )0.481 )0.155 )4.37 )0.373 0.091 0.431

()3.91) (0.37) ()4.5) ()0.82) ()3.58) ()2.15) (0.68) (0.98)

b3 0.044 0.123 )0.288 )0.110 )0.376 )0.158 0.085 0.436

(2.85) (5.40) ()2.8) ()0.6) ()3.15) ()0.93) (0.63) (0.95)

b4 )0.004 0.095 )0.137 0.294 )0.155 0.23 0.781 1.12

()0.22) (4.17) ()0.138) (1.67) ()1.36) (1.44) (6.56) (2.94)

b5 )0.029 0.171 0.107 0.578 0.061 0.590 1.16 1.52

(1.81) (7.48) (1.13) (3.38) (0.56) (3.84) (10.2) (4.09)

b6 0.035 0.198 0.143 0.634 0.228 0.741 1.15 1.49

(2.16) (8.58) (1.51) (3.73) (2.14) (4.96) (10.03) (3.99)

b7 )0.015 0.027 0.567 )0.925 0.529 1.46 1.71

()3.21) (0.28) (3.23) ()0.81) (3.29) (13.1) (4.68)

b8 )0.014 )0.510 0.429 )0.38 0.3601 )0.51 1.64

()3.64) (0.28) (2.07) ()3.07) (2.07) (0.28) (4.45)

b9 )0.019 )0.056 )0.011 )0.070

()4.65) ()1.9) ()0.49) ()0.8)

b10 )0.029 )0.030 )0.034 )0.072

()7.04) ()1.07) ()1.49) ()0.79)

b11 )0.034 )0.076 )0.062 )0.069

()8.14) ()2.73) ()2.87) ()0.94)

b12 )0.084 )0.091 )0.072

()3.08) ()4.19) ()1.01)

b13 )0.088 )0.087 )0.069

()3.24) ()4.22) ()0.96)

b14 )0.098 )0.113 )0.050

()3.43) ()4.68) ()0.72)

b15 )0.191 )0.148 )0.073

()5.02) ()5.1) ()1.03)

Generalized

F-test

16.82 34.33 7.87 0.20

t-statistics in parentheses.

Dependent variable is weighted log of percentage of women in each income category who buy the brand,

X1�brand price. Note: In the lipstick equations, the income categories are $50,000±$59,000; $35,000±

$49,999; $25,000±$34,999; $15,000±$24,999; and less than $15,000. Omitted category is greater than

$60,000. X7�X1 ´ X2, X8�X1 ´ X3, X9�X1 ´ X4, X10�X1 ´ X5, X11�X1 ´ X6. For mascaras, eyeshad-

ows and facial cleansers, X2� 1 if $60,000 < income < $74,999, 0 otherwise, X3� 1 if

$50,000 < income < $59,999, 0 otherwise, X4� 1 if $40,000 < income < $49,999, otherwise, X5� 1 if

$30,000 < income < $39,999, 0 otherwise, X6� 1 if $20,000 < income < $29,999, 0 otherwise, X7� 1 if

$10,000 < income < $19,999, 0 otherwise, X8� 1 if income < $10,000, 0 otherwise, X9�X1 ´ X2,

X10�X1 ´ X3, X11�X1 ´ X4, X12�X1 ´ X5,X13 �X1 ´ X6, X14�X1 ´ X7, X15 �X1 ´ X8. The omitted

category is income greater than $75,000.

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 125

Page 20: Empirical tests of status consumption: Evidence from women's cosmetics

Table 5

Brand-buying by race

Lipstick Mascara Eyeshadow Facial cleanser

Constant 1.931 1.44 1.80 1.66 1.99 1.99 2326.2 2.70

(12.26) (9.13) (4.26) (2.82) (10.65) (8.41) (9.94) (8.83)

b1 0.006 0.131 )0.008 0.019 0.017 0.017 )221.3 )302.1

(0.21) (3.91) ()0.14) (0.2) (0.75) (0.47) ()5.19) ()4.9)

b2 )4.19 )3.03 )4.16 )4.15 )4.22 )4.35 )181.9 )356.1

()24.2) ()12.6) ()9.78) ()4.55) ()19.2) ()12.2) ()1.07) ()0.83)

b3 )4.62 )4.10 )3.62 )3.20 )4.53 )4.38 )736.1 )1793.9

()24.8) ()15.4) ()9.0) ()3.74) ()20.1) ()12.0) ()4.27) ()4.02)

b4 0.309 )0.000 0.026 37.71

()5.77) (0.0) (0.47) (0.44)

b5 )0.133 )0.080 )0.029 217.0

()2.58) ()0.58) ()0.52) (2.44)

Generalized

F-test

16.76 0.201 0.437 0.371

t-statistics in parentheses.

Dependent variable is weighted log of the percentage of women in each racial group who buy the brand,

X1�brand price, X2� 1 if black, 0 otherwise, X3 � 1 if Spanish-speaking, 0 otherwise, X4 �X1 ´ X2,

X5�X1 ´ X3. Omitted category is white.

Table 6

Brand-buying by location of residence

Lipstick Mascara Eyeshadow Facial cleanser

Constant 0.359 )0.569 )0.666 )0.602 )0.4862 )0.605 )0.645 )0.298

(19.48) ()4.83) ()4.6) ()3.01) ()5.12) ()5.67) ()3.98) ()1.32)

b1 0.002 0.006 0.026 )0.014 0.008 0.031 )0.004 )0.080

(0.81) (0.28) (1.28) (0.42) (0.74) (1.95) ()0.13) ()1.7)

b2 0.064 0.087 0.202 0.094 0.096 0.152 0.193 )0.196

(3.56) (0.53) (1.47) (0.33) (0.92) (1.01) (1.64) ()0.62)

b3 )0.146 )0.375 )0.659 )0.758 )0.876 )0.432 )0.299 )0.976

()7.15) ()2.02) ()4.29) ()2.32) ()7.45) ()2.43) ()2.42) ()2.95)

b4 0.040 0.020 )0.010 0.083

(1.31) (0.45) ()0.47) (1.31)

b5 )0.093 )0.019 )0.100 0.143

()2.55) (0.33) ()3.54) (2.18)

Generalized

F-test

)21.46 0.11 6.75 2.40

t-statistics in parentheses.

Dependent variable is weighted log of the percentage of women by location who buy the brand,

X1�Brand Price, X2� 1 if suburban dweller, 0 otherwise, X3� 1 if neither city nor suburban dweller,

X4�X1 ´ X2, X5�X1 ´ X3. Omitted category is urban dweller.

126 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 21: Empirical tests of status consumption: Evidence from women's cosmetics

6. Alternative explanations

In the foregoing, we have interpreted social visibility as a proxy for status,and reasoned that the apparent e�ect of social visibility on price/quality cor-relations, brand-buying patterns, and the elasticity of price in the demandequation are all support our interpretation. Of course, because we have notdirectly measured status-motives or visibility, this methodology is open to al-ternative interpretations. The strength of our case rests on the relative plau-sibility of alternative explanations and our own.

To date we have been confronted with only two alternative explanations,both o�ered by an anonymous reviewer. The ®rst is that consumers ®nd iteasier to judge the quality of facial cleansers than lipsticks, and they thereforebuy higher priced lipsticks in order to avoid the risk of low quality. The sec-ond explanation is similar, positing that the risk of poor quality is higher forfacial cleansers. Essentially, both arguments say consumers have imperfectinformation about product quality, and that this accounts for the patternsof buying. We ®nd these arguments ultimately unconvincing. First, the Con-sumer Reports data clearly show that there are not measurable quality di�er-ences among lipsticks. This means that it should be easier for consumers todiscern quality di�erences among lipsticks than facial cleansers. Second, toexplain our pattern of results, there must not only be a di�erence betweenthe two extreme products (lipsticks and facial cleansers) but also with thetwo intermediate products (eyeshadows and mascaras). The alternative ex-planations must posit a continuum of ``di�culty of judging quality'' or ``risk-iness of failure'' among all four products. While this is logically possible, itseems less plausible to us than our interpretation. Furthermore, it is not ob-vious that the risk of failure is greater for facial cleansers. One could arguethat lipstick failures such as messy lips, lack of lipstick, etc. are socially em-barrassing in a way that mediocre skin cream results are not.

But we also believe there is a larger problem with using imperfect informa-tion approaches to explain these results. In order to be convincing, any ``im-perfect information'' approach must be able to explain why consumerscannot obtain perfect information about the products. Where products areused or purchased rarely, or are very expensive, informational imperfectionsmay be important. But cosmetics are used daily, they are purchased repeat-edly, and they are cheap, so that trial-and-error purchasing can occur. Fur-thermore, consumers can and do share word of mouth information, andConsumer Reports also publishes ratings. It is unlikely that through decadesof cosmetic use, women would continue to be uninformed about the true

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 127

Page 22: Empirical tests of status consumption: Evidence from women's cosmetics

qualities of the products they are purchasing. Indeed, because of their lowprices and frequent purchase and repeat purchase, cosmetics are preciselythe kinds of products about which quality information is easy to obtain.The imperfect information assumption needs to explain why, year in and yearout, women will pay a $12 premium on every tube of lipstick they buy. In ourview, the nature of this product makes it likely that status, rather than infor-mational imperfections explain our ®ndings.

7. Conclusion

We believe that the foregoing evidence provides strong support for the ex-istence of status motives in purchases of women's cosmetics. In contrast tothe predictions of standard consumer theory, which would not expect anysystematic di�erences to emerge across products by the level of social visibil-ity, we ®nd substantial evidence of such di�erences. Without invoking socialfactors it is di�cult to explain why women will pay $181 an ounce for a lip-stick which is intrinsically no di�erent than one they can purchase for $18 anounce. It is unlikely that women are systematically mislead by marketers. Amore plausible explanation, we believe, is that socially visible goods are asso-ciated with social standing. We feel the evidence we have presented here pro-vides a good ®rst test of one set of products, and hope this paper stimulatesfurther research on a wider range of products. If additional products yieldsimilar ®ndings, this should prompt researchers to re-think basic assumptionsin the ®eld of consumer demand.

Acknowledgements

The authors would like to thank Linda Pedelty and Mieke Lustenhouwerfor assistance.

Appendix A

The data for Fig. 1 is given below.Price vs. quality graphs data

128 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 23: Empirical tests of status consumption: Evidence from women's cosmetics

Appendix B

The data for Fig. 2 is given below.Product distribution graphs data:

Mascaras Facial cleansers Lipsticks Eyeshadows

Quality

rank

Price ($) Quality

rank

Price ($) Quality

rank

Price ($) Quality

rank

Price ($)

1 4.67 1 5.98 1 0.81 1 4.87

2 3.95 2 17.00 1 1.99 1 1.99

3 4.50 3 8.25 1 1.99 1 2.99

4 12.00 4 16.50 1 2.09 1 3.01

5 12.33 5 3.19 1 2.16 1 4.50

6 4.14 6 13.00 1 2.26 1 2.14

7 3.18 7 2.72 1 2.29 1 9.50

8 11.00 8 4.79 1 2.50 1 2.69

9 4.00 9 6.50 1 2.80 1 1.81

10 7.50 1 2.82 1 2.98

11 3.35 1 2.83 1 3.61

12 9.63 1 2.86 1 15.00

13 3.82 1 2.88 1 11.50

14 9.50 1 3.21 1 14.00

15 4.46 1 3.25 1 4.17

16 3.35 1 3.72 1 2.22

17 7.50 1 3.84 1 4.75

18 4.14 1 4.07 1 1.97

19 3.20 1 4.07 1 3.64

20 4.69 1 4.19 1 3.49

21 3.15 1 4.25 1 3.51

22 3.15 1 4.32 1 8.00

23 3.32 1 4.46 1 2.57

24 3.62 1 4.94 1 5.33

25 2.88 1 5.00 1 14.67

26 3.29 1 8.50

27 3.91 1 8.50

1 9.00

1 9.00

1 10.00

1 10.00

1 10.50

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 129

Page 24: Empirical tests of status consumption: Evidence from women's cosmetics

References

Abel, A.B., 1990. Asset prices under habit formation and catching up with the Joneses. Amer. Econom.

Rev. 80, 43±47.

Adult Personal Care Annual Report, 1989. MediaMark Survey, Mediamark Research, New York.

Adult Personal Care Annual Report, 1990. MediaMark Survey, Mediamark Research, New York.

Adult Personal Care Annual Report, 1991. MediaMark Survey, Mediamark Research, New York.

Alessie, R., Kapteyn, A., 1991. Habit formation, interdependent preferences and demographic e�ects in

the almost ideal demand system. The Economic Journal 101, 404±419.

Allen, M., 1981. Selling Dreams: Inside the Beauty Business. Simon & Schuster, New York.

Basmann, R. et al., 1988. A Note on Measuring Veblen's Theory of Conspicuous Consumption, Review of

Economics and Statistics, 531±535.

Bearden, W.O., Etzel, M.J., 1982. Reference group in¯uence and product and brand purchase decisions. J.

of Consumer Research 9, 183±194.

Percentage of women buying particular brand of Lipsticks, Facial Cleanser, Mascara and

Eyeshadow

Lipsticks Facial cleansers (FC) Mascara Eyeshadow

Average

price ($) of

brand

Percentage

buying

Average

price ($) of

brand

Percentage

buying

Average

price ($) of

brand

Percentage

buying

Average

price ($) of

brand

Percentage

buying

0.81 7.3 2.96 22.9 3.17 7.59 1.94 12.5

2.13 2 4.79 18.1 3.18 19.85 2.14 6.44

2.26 1.8 5.98 9.8 3.32 3.03 2.49 13.4

2.29 7.8 7.38 9.7 3.35 11.75 2.49 12

2.81 2.1 15.50 1.1 3.35 6.31 3.49 5.42

2.83 5.6 3.72 3.24 3.55 9.11

2.86 4 4.05 4.58 3.61 8.99

2.88 4.5 4.14 4.02 3.76 5.39

2.99 24.1 4.25 12.72 3.95 8.77

3.23 5.8 4.29 7.63 4.17 3.62

3.47 7.6 4.75 2.42 4.87 3.07

3.84 1.8 7.5 5.46 8.00 2.74

3.96 1.5 9.75 1.75 9.50 9.85

4.14 3.8 10.25 6.08 11.00 2.69

4.32 5.2 12 6.7 14.00 3.51

4.7 5.7 12.33 3.41 14.50 1.92

8.5 8.7 15.00 7.97

8.5 1.5 15.00 1.86

9 1.5

9.5 12.3

9.75 3.1

10 2.5

130 A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131

Page 25: Empirical tests of status consumption: Evidence from women's cosmetics

Bearden, W.O., Netemeyer, R.G., Teel, J.E., 1989. Measurement of consumer susceptibility to

interpersonal in¯uence. J. of Consumer Research 15, 473±481.

Brown, C., 1994. The Standard of Living. Blackwell, Cambridge, MA.

Congleton, R., 1989. E�cient status seeking: Externalities, and the evolution of status games. J. of

Economic Behavior and Organization 11, 175±190.

Consumer Reports, February 1988, June 1989, February 1991.

Deaton, A., 1992. Understanding Consumption. Clarendon Press, Oxford.

Duesenberry, J.S., 1949. Income, Saving and the Theory of Consumer Behavior. Harvard University

Press, Cambridge.

Easterlin, R., 1995. Will raising the incomes of all increase the happiness of all? J. of Economic Behavior

and Organisation 27, 35±47.

Festinger, L., 1954. A theory of social comparison processes. Human Relations 7 (2), 117±140.

Frank, R.H., 1985a. The demand for unobservable and other nonpositional goods. Amer. Econom. Rev.

75 (1) 101±116.

Frank, R.H., 1985b. Choosing the Right Pond: Human Behavior and the Quest for Status. Oxford

University Press, New York .

Galli, J., 1994. Keeping up with the Joneses: Consumption externalities, portfolio choice, and asset prices.

J. of Money, Credit, and Banking 26 (1), 1±8.

Hirsch, F., 1976. The Social Limits to Growth. Harvard University Press, Cambridge, MA.

Holman, R., 1981. Product use as communication: A fresh appraisal of venerable topic. In: Enis, B.M.,

Roering, K.J. (Eds.), Review of Marketing. American Marketing Association, Chicago, Illinois, pp.

106±119.

James, J., 1987. Positional goods, conspicuous consumption, and the international demonstration e�ect

reconsidered. World Development 15 (4), 449±462.

Kapteyn, A., Herwaarden, F.G.v., 1980. Interdependent welfare functions and optimal income

distribution. J. of Public Economics 14, 375±397.

Kosicki, G., 1987. A test of the relative income hypothesis. Southern Economic Journal 54, 422±433.

Kosicki, G., 1990. Income redistribution and aggregate consumption: Implications for the relative income

model. American Economist 34, 40.

Leibenstein, H., 1950. Bandwagon, snob and Veblen e�ects in the theory of consumers' demand. Quart. J.

Econom. 64, 183±207.

Lovell, M., 1994. Private communication.

McAdams, R.H., 1992. Relative Preferences. The Yale Law Journal. The Yale Law Journal, New Haven,

pp. 1±104.

McKnight, G., 1989. The Skin Game: The International Beauty Business Brutally Exposed. Sedgewick

and Jackson, London.

Neumark, D., Postlewaite, A., 1995. Relative Income Concerns and the Rise in Married Women's

Employment, NBER Working Paper #5044. NBER, Cambridge, MA.

Park, C.W., Lessing, V.P., 1977. Students and housewives: Di�erences in susceptibility to reference group

in¯uences. J. of Consumer Research 4, 102±110.

Rauscher, M., 1993. Demand for social status and the dynamics of consumer behavior. The Journal of

Socio-Economics 22, 105±113.

Schor, J.B., 1998. The Overspent American. Basic Books, New York.

Tefertiller, T., 1994. Children of Abundance: Exploring the Consumption Patterns of Teenage Girls.

Undergraduate Thesis, Harvard University.

Veblen, T., 1967. The Theory of the Leisure Class. Penguin Books, New York (Originally published 1899).

Wolf, N., 1991. The Beauty Myth: How Images of Beauty Are Used Against Women. Doubleday, New

York.

A. Chao, J.B. Schor / Journal of Economic Psychology 19 (1998) 107±131 131