Conspicuous consumption and political regimes: Evidence from East and West Germany

Contents lists available at ScienceDirect

European Economic Review

European Economic Review 67 (2014) 62–81

0014-29http://d

n CorrE-m1 Su2 W

journal homepage: www.elsevier.com/locate/eer

Conspicuous consumption and political regimes: Evidencefrom East and West Germany

Tim Friehe a,b,n, Mario Mechtel c,d

a University of Bonn, Center for Advanced Studies in Law and Economics, Adenauerallee 24-42, 53113 Bonn, Germanyb CESifo, Munich, Germanyc IAAEU Trier, Behringstr., 54286 Trier, Germanyd University of Trier, Germany

a r t i c l e i n f o

Article history:Received 6 February 2013Accepted 6 January 2014Available online 28 January 2014

JEL classification:D12D62P36

Keywords:Conspicuous consumptionStatus-seekingPolitical regimeBehavioral economics

21/$ - see front matter & 2014 Elsevier B.V.x.doi.org/10.1016/j.euroecorev.2014.01.005

esponding author. Tel.: +49 228 73 1737; faail addresses: [email protected] (T. Friech an influence is discussed by Cooter (199ithin this paper, we refer to the political en

a b s t r a c t

This paper investigates the influence of political regimes on the relative importanceof conspicuous consumption. We use the division of Germany into the communist GDRand the democratic FRG and its reunification in 1990 as a natural experiment. Relyingon household data that are representative for Germany, our empirical results stronglyindicate that conspicuous consumption is relatively more important in East Germany.Significantly, although we find some convergence, a considerable gap in conspicuousconsumption expenditures remains even 18 years after the German reunification.

& 2014 Elsevier B.V. All rights reserved.

1. Introduction

A large segment of economics research is devoted to designing institutions in order to induce socially desirable choicesfor given preferences of decentralized agents. However, the institutions themselves may also affect preferences.1 After morethan four decades of separation, the reunification of the communist German Democratic Republic (GDR) and the democraticFederal Republic of Germany (FRG) in 1990 constitutes a unique situation that allows us to explore the potential influence ofpolitical institutions on preferences in a natural setting.2

The present paper explores whether the political regime influences preferences by adopting the consumption patternsof households in East and West Germany as the object of study, using a representative dataset covering the periodfrom 1993 to 2008. In particular, we are interested in the importance of so-called conspicuous consumption. Conspicuousconsumption, a concept that can be ascribed to Veblen (1899), refers to consumption that is intended to communicateone's economic status to others. The category of conspicuous consumption reflects the observation that peoplecompare themselves to others in a multitude of ways, with relative performance being important for subjective

All rights reserved.

x: +49 228 73 9111.he), [email protected] (M. Mechtel).8), Fehr and Hoff (2011), and Hwang and Bowles (2011), among others.vironment of the GDR using the term “communism”.

www.sciencedirect.com/science/journal/00142921

www.elsevier.com/locate/eer

http://dx.doi.org/10.1016/j.euroecorev.2014.01.005



http://crossmark.crossref.org/dialog/?doi=10.1016/j.euroecorev.2014.01.005&domain=pdf



mailto:[email protected]

mailto:[email protected]


T. Friehe, M. Mechtel / European Economic Review 67 (2014) 62–81 63

well-being.3 In the attempt to establish one's economic status relative to others, consumption is often used as a signal. Thisfinding has been derived theoretically by Corneo and Jeanne (1998) and supported empirically by Heffetz (2011), amongothers. The importance of this type of consumption is due to the fact that many consumption choices (such as which car todrive) are easily observable by others, whereas aspects such as financial wealth cannot be readily observed. Goods that areparticularly suited for conspicuous consumption given their impact on social rank are commonly referred to as positionalgoods.4 The fact that positional concerns are important and that goods differ with regard to their positionality (i.e., certaingoods have a higher relevance for relative standing in society) has been confirmed in several empirical studies.5 Our analysiscomplements these studies by investigating non-experimental consumption patterns from East and West Germany.

The different political regimes experienced by East and West Germans may have shaped their preferences with regard toconspicuous consumption for several reasons. First, in contrast to the experience of West Germans, the communist regime severelylimited people's choice sets, thereby restricting the possibilities to signal status through the selection of the upscale product variety(see, e.g., Fulbrook, 2009). Moreover, the consumption of conspicuous goods in East Germany was often seen as an indicator ofcollaboration with the state security service (Staatssicherheit), as access to such goods was generally restricted to the politicallyprivileged (see, e.g., Fulbrook, 2009; Schäfgen, 1998); this created the potential for social isolation and stigmatization. In addition,the ideology of the GDR (which portrayed itself as a nation of workers and peasants) included the idealization of the working class,presenting an additional contrast to West Germany with probable repercussions for the relative importance of conspicuousconsumption (see, e.g., Fulbrook, 2009). Moreover, the unobserved heterogeneity in economic success was significantly smaller inEast Germany, given the system's emphasis on egalitarianismwithin and across professions. Consequently, there were substantiallysmaller differences in individual economic achievements in comparison to West Germany. For instance, in the GDR in 1988, theaverage net income of individuals with a university degree was 15% higher than that of blue-collar workers, while inWest Germanythis difference amounted to 70% (Schäfgen, 1998). As a result, people from East and West Germany were habituated in verydifferent conditions with respect to the use of consumption to reveal one's economic status to others.

Taking for granted the validity of our identifying assumption that observable differences in behavior are driven by thepopulations' experiences under the different political regimes for a moment (we will provide a detailed discussion in Section 2),we now consider possible hypotheses. We would not observe any systematic differences between East and West Germany afterreunification when controlling for factors such as household income, socio-economic household characteristics, education, andthe distribution of income in the state (Bundesland) of residence if the political regime had no effect on individual preferences.Alternatively, if the regime had an effect, in principle two directions are possible. One possibility is that people in East Germanywould emphasize conspicuous consumption more than individuals in West Germany, perhaps in an attempt to make up for therestricted choices they experienced before reunification. Another reason to expect conspicuous consumption to be moreimportant for East Germans after reunification is that East Germans seem less likely to attribute economic outcomes to luck thanpeople from West Germany (see Alesina and Fuchs-Schündeln, 2007). This perception of the determinants of economic successalso corresponds with the finding that more East Germans believe in the importance of hard work (Corneo, 2001). Whendifferences in economic outcomes are perceived to be driven by differences in ability and/or effort, this could support theimpulse to signal one's economic success to others. The opposite hypothesis is that people might have internalized the emphasison egalitarianism of the political regime in the GDR (at least to some extent) and would thus be less inclined to distinguishthemselves from their peers by engaging in conspicuous consumption. Models of conspicuous consumption usually consider asetting in which everybody knows that a given individual belongs to a group for which the distribution of income levels iscommonly known (see, e.g., Bilancini and Boncinelli, 2012; Charles et al., 2009; Glazer and Konrad, 1996). With regard to thedistribution of income in divided Germany, it is clear that inequality was much lower in the GDR than in the FRG; this continuedto be the case for a brief period after reunification as well (Fuchs-Schündeln et al., 2010). However, inequality in real householdmarket income in East Germany overtook that in West Germany some years after reunification (see, e.g., Grabka et al., 2012).However, such differences in inequality are not easily translatable into hypotheses about conspicuous consumption in East andWest Germany, since theoretical predictions concerning the impact of low levels of inequality are conditional on the modeling.6

One could argue that conspicuous consumption should be a relatively more important phenomenon inWest Germany, given thataverage earnings are higher in the West (e.g., Corazzini et al., 2012; Hopkins and Kornienko, 2004).

This paper uses a dataset covering the period from 1993 to 2008 that is representative for Germany and includeshousehold characteristics, a detailed breakdown of household expenditures, and information about household income.Our empirical results indicate that there are indeed significant differences between consumption patterns in East and WestGermany, and that these differences support the hypothesis that conspicuous consumption is relatively more important in

3 For instance, Dohmen et al. (2011) provide evidence for the importance of relative income for subjective well-being using functional magneticresonance imaging (fMRI). Further empirical evidence of the importance of relative income positions for individual happiness and actions can be found inStutzer (2004) and Frey et al. (2008).

4 The expression “positional good”was first introduced by Hirsch (1976) and has been taken up by other scholars, including Frank (1985a, 1985b, 2008)and Solnick and Hemenway (1998, 2005). Cars are usually considered a prime example of a positional good. Along these lines, Winkelmann (2012)establishes for Switzerland that the prevalence of luxury cars in an individual's own municipality decreases income satisfaction, and Kuhn et al. (2011) findthat the neighbors of people who have won a car in the lottery have significantly higher levels of car consumption than the general population.

5 See, for example, Alpizar et al. (2005), Carlsson et al. (2007), Carlsson and Qin (2010), Caporale et al. (2009), Clark et al. (2008), Clark and Senik (2010),Hillesheim and Mechtel (2013), Johansson-Stenman et al. (2002), Solnick and Hemenway (1998, 2005), and Solnick et al. (2007).

6 Bilancini and Boncinelli (2012) show that whether the status function is ordinal or cardinal decisively influences the impact of greater equality onconsumption expenditure. In Glazer and Konrad (1996), the result hinges on the curvature of conspicuous consumption as a function of income.

T. Friehe, M. Mechtel / European Economic Review 67 (2014) 62–8164

East Germany. Importantly, although we find evidence of convergence, significant differences with regard to conspicuousconsumption in East and West Germany remain, even 18 years after reunification. Based on our analysis, full convergencemay be expected around 2018. In testing for the relative importance in East and West Germany, we seek to control for othersocio-economic determinants of conspicuous consumption and thereby also consider the influence of unemployment,urbanization, education, and gender, yielding a host of findings secondary to our result on the effect of political regimes onpreferences. For example, by establishing that – based on our regression results – women spend more on visible goods thanmen, we contribute to the evolving literature on gender effects (see, e.g., Andreoni and Vesterlund, 2001; Croson andGneezy, 2009).

Our paper is most closely related to Charles et al. (2009) who analyze differences in spending on conspicuousconsumption between races (i.e., African-Americans, Hispanics, and Whites) using US data from the Consumer ExpenditureSurvey from 1986 to 2002. The authors find that spending patterns differ significantly between races. Accounting fordifferences in income, African-Americans and Hispanics spend about 25% more on visible goods than Whites. Charles et al.(2009) further include a control for mean race/state reference group income, establishing that this explains most of theracial gap in visible spending. This finding is compatible with the interpretation that consumption is used as a signal forunobservable income. Our study likewise includes a proxy for the mean income of the reference group, which consists of thepopulation of the state (Bundesland) in which the household is located.7 This measure identifies the reference group usingonly one household attribute but has explanatory power with regard to the level of expenditures on conspicuous goods.8

Kaus (2013) reproduces the analysis by Charles et al. (2009) using data from South Africa.We study potential differences in the preferences of East and West Germans as revealed by real economic behavior. There

is already some evidence that hints at possible differences between East and West Germans; however, most of these studiesrely on surveys. Torgler (2003), for example, uses World Value Survey data to determine that East Germans' tax morale washigher at the beginning of the nineties, later converging with that of West Germans. Rainer and Siedler (2009) study theextent to which East and West Germans trust institutions and other people, finding that East Germans persistently showless trust than West Germans. Similarly, Bauernschuster and Rainer (2012) find no evidence of convergence in attitudestowards gender roles in East and West Germany. Another contribution relying on survey data is that of Alesina and Fuchs-Schündeln (2007), who show that the different political regimes significantly influenced preferences for redistribution andstate intervention, with East Germans supporting such policies more than West Germans. Their data analysis leads theauthors to conclude that the preferences of the two populations do converge, but that it takes one to two generations for fullconvergence to take place. Corneo (2001) and Corneo and Grüner (2002) also establish that East Germans are relativelymore supportive of redistribution than West Germans.9 Interestingly, there is also evidence that East Germans redistributeless in practice when there are no mandates from the state. Building on the experimental findings of Ockenfels andWeimann (1999), Brosig et al. (2011) study experimental data from the solidarity game, focusing on the influence of thepolitical regime on fairness preferences; the authors determine that East Germans continue to show much less solidaritythan West Germans, even 20 years after reunification. Somewhat closer in focus to the present paper in terms of interest ininterdependent preferences, Ferrer-i-Carbonell (2005) conducts an empirical analysis of the importance of comparisonincome for individual happiness. The author finds that East and West Germans differ in that the impact of relative income onsubjective well-being is asymmetric for the latter but symmetric for the former (where an asymmetric effect means that thewell-being of poorer individuals is negatively affected by having income below the reference income, but richer individualsdo not benefit from having a higher income than the average).

This paper contributes to the literature in several ways. First, it demonstrates the influence of the political regime in acountry on preferences based on real economic decisions rather than making use of survey or experimental data. Second, weprovide a detailed analysis of the relative importance of conspicuous consumption for East and West Germany, adding to thefindings on differences between races from the USA and South Africa. Third, the data analysis yields findings regarding theinfluence of gender on the relative importance of conspicuous consumption expenditures, an issue that has not yet beenaddressed in the literature. The remainder of our paper is organized as follows: In Section 2, we discuss our research designin more detail. In Section 3, we describe the data used in our analysis. In Section 4, we present our empirical analysis. Thefinal section concludes.

2. Research design

The present paper treats the German separation into the Federal Republic of Germany and the German DemocraticRepublic as a natural experiment and proposes that differences in consumption behavior measured after the reunificationare related to having treated people living in East and West Germany with two different political regimes during the timeof separation. In order to isolate the causal impact of the communist regime on preferences regarding conspicuous

7 Similarly, Persky and Tam (1990) assume that the reference group includes all inhabitants of the same region, while Easterlin (1995) implicitlyassumes that individuals compare themselves to all others in the country. For a recent discussion, see Clark et al. (2013).

8 In contrast, when the reference group is identified by state and education, the control variables lose explanatory power; this may be due to therelatively imprecise information on education in the data.

9 Corneo and Grüner (2002) actually argue that support for redistribution in East Germany is stronger than in Bulgaria, Czechoslovakia, Hungary,Poland, and Russia.


consumption, a key identifying assumption is that East and West Germans did not differ from each other in terms ofpreferences prior to the German separation. Lacking precise data on consumption patterns, we propose that similaritybetween East and West Germans with respect to other variables may be sufficiently indicative that our assumption isreasonable.10

First, when the different political regimes were imposed, the inhabitants in the treatment and control regions must havebeen comparable in important dimensions. This also necessitates that the imposition of the political regime was not uponrequest of the inhabitants. In fact, which political regime was imposed was a result of how the United Kingdom, the UnitedStates, and the Soviet Union agreed on a protocol for the partition of prewar Germany in 1944, which happened so as toallow for a division of the territory into three sectors of roughly equal population size (e.g., Burchardi and Hassan, 2013).Accordingly, the actual imposition of the regimes was unrelated to the preferences of the inhabitants. In this vein, Reddingand Sturm (2008) highlight that the decisions determining the partition of East and West Germany are unlikely to becorrelated with prewar characteristics of respective regions. Moreover, as argued by Alesina and Fuchs-Schündeln (2007),for example, the regions that became the FRG and the GDR were similar in terms of pre-World War II average per capitaincome levels (in the years 1928, 1932, and 1936) and in terms of the amount of destruction during World War II. Theavailable data confirms similarity of the two regions in other regards as well. This applies, for instance, to the split of theworking population across industries (Schäfgen, 1998) or the political orientation of voters at the turn of the century(Alesina and Fuchs-Schündeln, 2007), and to the population density (Hubert, 1998). Moreover, in conclusion of hisconsideration of trade costs, Wolf (2009) states that Germany was by the end of the Weimar Republic in 1933 aneconomically well-integrated area, such that the separation into East and West Germany that existed between about 1946and 1989 was hardly predictable in 1939. Based on such evidence, Rainer and Siedler (2009) conclude that the two parts ofGermany were indistinguishable prior to the separation. Similarly, given the arguments just discussed, we are confident thatEast and West Germany were decidedly comparable before the separation.

However, it must be conceded that the actual imposition of the political regime in East Germany prompted around threemillion people to emigrate to West Germany before the Berlin Wall was built in August 1961 (e.g., Heidemeyer, 1994;Heineck and Süssmuth, 2013, Hubert, 1998), whereas there was little migration after 1961 or from West to East Germany.Importantly, the sample of migrants was not a random draw of the population; instead, intellectuals and entrepreneurswere overrepresented among the migrants (e.g., Heidemeyer, 1994).11 This migration during our treatment period ispotentially problematic for our identification of a treatment effect when the migrating population differed consistentlyregarding their preferences for conspicuous consumption – a possibility that is not testable due to the lack of data.

Next, it is relevant how comparable East and West Germans were when the EVS (Einkommens- und Verbrauchsstichp-robe) data we rely on in this study was actually collected. The present paper explores consumption choices by households inEast and West Germany. Consumption is a function of preferences and a multitude of individual characteristics such asincome, job prospects, and savings. The East German households in our representative sample differed from those in WestGermany in some characteristics. For example, households in East Germany had lower savings when the Berlin Wall fell (thesame result is also reported by Fuchs-Schündeln, 2008) and the share of households with a university degree was higher inEast than in West Germany.12 However, East and West German households were very similar in a large number ofcharacteristics, such as the age of the head of household, the number of children and the number of adults per household,the share of households living in a city, the share of households with German citizenship, and the share of householdswithout school leaving certificate.13 In our empirical estimation, we make use of the richness of our dataset and seek tocontrol for such effects when trying to identify the influence of the political regime on preferences. In addition, we allow forthe possibility that the levels of these variables have an effect on conspicuous consumption that is specific to householdsliving in East and West Germany, respectively. Fortunately, the dataset we utilize is very detailed when it comes to thepattern of consumption at a particular point in time and the control variables required to account for differences insocioeconomic characteristics of the households in our empirical analysis. However, the dataset contains no informationabout where the household resided when the Berlin Wall came down in 1989. The questionnaire of the EVS in fact only asksabout the current location of residence. In other terms, migration after the treatment period is a potential challenge to ouridentification of an influence of the regime on preferences. Generally speaking, it may be the case that some households thatwe categorize as East German because they lived in East Germany at the time of the survey had not actually experienced thecommunist regime, and vice versa.14 With regard to the mobility of the German population during and after the year 1989,migration from East to West Germany occurred in two waves, with particularly high levels of migration after the fall of the

10 The specific characteristics of the post-war period would make any interpretation of consumption data difficult, even if it were available. Suchparticularities include the scarcity of many goods and services and non-market allocation of resources (e.g., Merkel, 1999 pp. 310–312).

11 Nevertheless, Schäfgen (1998, p. 58) asserts that the structure of the society regarding educational achievement and implied differences betweensocial groups remained relatively stable and comparable in East and West Germany.

12 Average savings in the EVS data amounted to 4811 (7,483) Euros in East (West) Germany in the year 1993. The share of heads of households with auniversity degree was 24.7% (13.5%) in East (West) Germany. With respect to income, the mean values amounted to 36,047 Euros for East Germanhouseholds and 54,872 Euros for West German households in the first EVS wave after reunification.

13 The means of the corresponding variables are: age of the head of household (E: 47.4 years in 1993, W: 49.2), number of children per household (E:0.83, W: 0.86), number of adults in the household (E: 1.76, W: 1.71), the share of households living in a city (E: 0.39, W: 0.36), the share of households withGerman citizenship (E: 0.998, W: 0.98), and the share of households without school leaving certificate (E: 0.004, W: 0.014).

14 Fuchs-Schündeln and Izem (2012) face the same difficulty in their study on labor productivity.


Berlin Wall (in fact, the flows in 1989 and 1990 each represented 2.5% of the East German population) and during a secondwave that started in the late 1990s (Fuchs-Schündeln and Schündeln, 2009; Hunt, 2006). The population of the migrants ispredominantly young and more skilled than the population of those who stayed (Hunt, 2006). In other terms, manyindividuals who emigrated to West Germany did not have many years of exposure to the political regime of the GDR. Alongthe same lines, older migrants with more experience in the communist regime were more likely to return to the East (i.e., tobe a temporary migrant) when compared to younger migrants (Fuchs-Schündeln and Schündeln, 2009). These facts makethe extensive migration somewhat less problematic for our analysis, given that there is other evidence that East Germanswho lived under the communist regime for a longer time show bigger differences from West Germans (Bauernschuster andRainer, 2012; Alesina and Fuchs-Schündeln, 2007). Migration from the West to the East remained relatively stable (at lessthan 100,000 people per year) throughout the time period we consider. Fuchs-Schündeln and Schündeln (2009) report forthe period 1991–2006 that 2.45 million individuals migrated from the former GDR to the former FRG, whereas 1.45 millionindividuals moved in the opposite direction (though including a significant number of people moving back and forth).Among the various waves of the dataset we use, it seems likely that the 1993 data would be the least affected by migration.However, we believe that potential miscategorization of households will in general be negligible, since a large share of themigration to East Germany represents returning East Germans, and because there is presumably some adaptation towardsthe local norm, as reported by Brosig et al. (2011).15

Based on the aforementioned arguments and in line with other scholars (Alesina and Fuchs-Schündeln, 2007;Bauernschuster and Rainer, 2012; Heineck and Süssmuth, 2013; Rainer and Siedler, 2009), we believe that our identifyingassumption that observable differences in behavior are driven by the population's experiences under different politicalregimes seems justified. However, we cannot rule out the possibility that migration during and after the treatment periodconfounds our results.

3. Data

The present study relies on the income and expenditure sample (Einkommens- und Verbrauchsstichprobe (EVS))conducted by the German statistical office (Statistisches Bundesamt), which is the largest sample of its kind in the EuropeanUnion. This quinquennial cross-section dataset is representative for Germany. Households voluntarily participate in thesurvey, providing information on socio-demographic household characteristics and supplying data on household incomeand expenditures, savings, durable consumer goods, and the housing situation.16 The scope of the EVS is thus similar to thatof the US Consumer Expenditure Survey. The EVS has been used in other research on topics such as household savingsbehavior and inequality (see, e.g., Börsch-Supan et al., 2001; Fuchs-Schündeln et al., 2010; Kopetsch and Rauscher, 2006;Scheicher, 2010 ).17 We seek to identify the effect of living under the political regime of the GDR on preferences by focusingon the relative importance of conspicuous consumption; presumably, this can be best undertaken by examination of thesample collected shortly after reunification (i.e., EVS 1993). However, in addition to studying the difference between Eastand West Germans shortly after the abolition of the communist regime, we seek to determine whether any potentialdiscrepancy persists when both populations are subject to the same institutions over a longer period of time. This wouldallow us to more clearly attribute differences to the influence of the regime (since effects that are relevant shortly afterreunification, such as higher levels of purchasing when previously unobtainable durable goods become available, will nolonger be of importance in later years). Thus, in order to check for possible persistence, we also include data from 1998,2003, and 2008 in our analysis.

Conspicuous consumption is understood as the use of money or other resources to display one's high social status inrelation to others (Veblen, 1899). Goods that are particularly suited to this objective should (i) be readily observable, (ii) givethe impression that individuals who consume more of them are, on average, better off than those who consume less ofthem, and (iii) be portable across a variety of interactions. On the basis of an online survey and common-sense assumptions,Charles et al. (2009) consider visible consumption to consist of expenditures on apparel (including jewelry), personal care,and vehicles. We use a wider definition of conspicuous consumption expenditures (see Table 1). Our definition incorporatesevidence that the respect and admiration one earns in face-to-face interactions with groups such as colleagues and friendsare a major determinant of status concerns (see Anderson et al., 2012; Clark and Senik, 2010; Senik, 2009). Accordingly, ourdefinition of “visible goods” also includes items that can be regarded as positional only vis-a-vis colleagues, friends, andfamily. This wider definition reflects the idea that items such as expensive TVs, golf clubs, furniture, and pianos arepresumably also chosen with an eye to how they will be perceived by others.18. Furthermore, our definition assignsadditional items to the category personal care used by Charles et al. (2009), such as dental treatments. We will use the

15 Since 1991, about 750,000 East Germans have returned to East Germany from the West (see Lang and Nadler, 2012). In the sample of Fuchs-Schündeln and Schündeln (2009), for example, 20% of East-West migrants return to the East later.

16 For further information on the EVS, see, e.g., Statistisches Bundesamt (2005a, 2005b).17 Bursztyn and Cantoni (2012) use EVS data to study the influence of Western television on consumption patterns in East German households, finding

that exposure to Western TV ads before reunification had a (vanishing) influence on the composition of consumption baskets when Western productsbecame available after reunification.

18 This inclusion is in accordance with the visibility survey conducted by Heffetz (2011), the results of which similarly assign a relatively high visibilityscore to furniture, for example.

Table 1Definitions of visible goods.

Category Baseline Intermediate Charles et al. (2009) Heffetz (2011)

Motor vehicles (new/used), motorbikes, bikes, water vehicles, aircraft(including commodities for motor vehicles)

X X X X

Shoes X X X XApparel (women, men, children, babies), purses, shoulder bags X X X XJewelry, watches X X X XSkin and body care: commodities and services X X XHosiery goods/headpieces X XDental treatments & protheses X XFurniture X X XValuable electronic household appliances

(other than washing machine, tumble dryer, fridge, freezer, or heater)X X X

Phones, TVs, radio sets, cameras X X XOptical instruments, collections, art objects, music

instruments, sporting and other leisure goods (e.g., games, toys)X X

Food and drinks in restaurants X XHolidays X


definitions proposed by Charles et al. (2009), Heffetz (2011),19 and an intermediate basket of visible consumption goods(situated between our baseline definition and that used by Charles et al.) as robustness checks for our baseline analysis.Conspicuous consumption does not include items with little or no visibility and/or limited status effects, such as insurancepremiums, books, food consumed at home, utilities, tobacco, education and training, and pharmaceutical products. Wefollow Charles et al. (2009) in excluding expenditures on rent, albeit for a different reason. Whereas the decision of Charleset al. (2009) is based on their focus on racial discrimination in the housing market, we opt for this course of action becauserents are systematically and markedly lower in East Germany than in the West. For example, according to the Germanmicro-census, the average rent in East Germany in 2010 amounted to only 75.9% of the average rent in West Germany (seeStatistisches Bundesamt, 2012). Furthermore, there is an additional reason to exclude rents from our analysis: theproportion of households owning a house or a flat is significantly larger in West Germany, which can primarily beattributed to expropriations of East German households after the Second World War.

Having defined our terms, we now present the descriptive statistics for the variables of interest for the pooled dataset.This data provides information on 176,751 households (see Table 2), 21.5% of which were residing in East Germany at thetime of the survey.20 The variable visible spending is key to our analysis, as it captures a household's expenditures on goodsthat are classified as being easily observable by others and status-relevant. Mean spending on visible goods is higher in WestGermany, irrespective of the definition applied.21 The likelihood of the head of the household being a woman is higher inEast Germany than in West Germany. While the number of adults per household does not greatly differ, there are on averagefewer children in East German households. Table 2 clearly shows that our dataset includes relatively few students and manyretirees in both East and West Germany. Our dataset allows us to distinguish between survey respondents with a schoolcertificate from those without one, and between those with a university degree from those without one.22 As can be seen inTable 2, the dataset consists almost exclusively of German respondents, especially in East Germany.

19 Heffetz conducted a survey with the main survey question being “Imagine that you meet a new person who lives in a household similar to yours.Imagine that their household is not different from other similar households, except that they like to, and do, spend more than average on […]. Would younotice this about them, and if so, for how long would you have to have known them, to notice it? Would you notice it almost immediately upon meetingthem for the first time, a short while after, a while after, only a long while after, or never?” for 31 different expenditure category titles. Based on the resultsof the survey, he develops a measure of expenditure visibility that ranks different purchases in terms of how readily they are noticeable to other people.The most visible expenditure category is cigarettes, followed by cars and clothing. In our robustness checks, we use the positions 2–11 from Heffetz (2011)(including all EVS items most similar to the goods from the Heffetz scale), that is, we omit cigarettes. This omission is based on our belief that cigarettes donot fulfill the second requirement of our definition of conspicuous consumption – namely, that the good should give the impression that those whoconsume more of it are, on average, better off than individuals who consume less. The empirical analysis by Heffetz strongly suggests that expenditures oncigarettes decrease for income levels above 40,000 US-Dollars (Heffetz, 2011, Figure 2), that is, cigarettes seem to be inferior goods for a large interval of theincome scale. While our judgment regarding the exclusion of cigarettes accords with that of Charles et al. (2009), for example, it is not decisive for therobustness of our results when using the Heffetz definition. Results for the case with cigarettes included are part of the Online Appendix.

20 The dataset allows us to distinguish between East and West Berlin.21 All monetary figures reported have been converted into Euros and deflated to the year 1993. The deflators were constructed using the consumer

price indices provided by the German Federal Statistical Office (Destatis).22 Data on whether or not the head of household received an Abitur, the high school diploma enabling recipients to progress to university, is

unfortunately not included in all waves.

Table 2Descriptive statistics. Pooled over all waves (i.e., 1993, 1998, 2003, 2008).

Variable Obs Mean Std. Dev. Min Max

Eastern Germany 176,751 0.2152916 0.4110256 0 1Income 176,751 52,677.82 33,682.85 0 893,848.4

West 138,698 55,930.05 35,078.02 0 893,848.4East 38,053 40,823.87 24,613.45 0 260,558.1

Female 176,751 0.3129204 0.4636836 0 1West 138,698 0.2812658 0.4496185 0 1East 38,053 0.4282974 0.4948385 0 1

Age 176,751 49.95225 14.8522 20 85West 138,698 50.022 15.0301 20 85East 38,053 49.69803 14.18219 20 85

No. of adults 172,348 1.7002 0.4581717 1 2West 135,114 1.694813 0.4604882 1 2East 37,234 1.719745 0.4491296 1 2

No. of children 171,876 0.739027 1.010776 0 5West 134,773 0.7554184 1.037355 0 5East 37,103 0.6794868 0.9052232 0 5

City 176,751 0.3183858 0.4658514 0 1West 138,698 0.3321533 0.4709874 0 1East 38,053 0.2682049 0.4430307 0 1

Savings 176,751 8140.303 22,084.45 0 2,354,567West 138,698 8815.399 23,314.54 0 2,354,567East 38,053 5679.667 16,627.36 0 2,354,567

Vis. spending 176,751 8989.581 12,483.07 0 430,765.8West 138,698 9383.842 12,938.37 0 430,765.8East 38,053 7552.554 10,536.18 0 211,029.9

Vis. sp. (intermediate) 176,751 6269.867 11,359.98 0 425,642.9West 138,698 6509.57 11,749.94 0 425,642.9East 38,053 5396.182 9758.706 0 209,267.3

Vis. sp. (Charles et al.) 176,751 4593.582 10,489.6 0 422,914.3West 138,698 4787.122 10,864.44 0 422,914.3East 38,053 3888.153 8956.835 0 207,574.8

Vis. sp. (Heffetz) 176,751 7515.761 11,760.71 0 429,350.2West 138,698 7842.805 12,193.66 0 429350.2East 38,053 6323.731 9935.054 0 208,776.1

Total cons. expenditures 176,751 31,625.37 24,757.53 677.2227 454,509.8West 138,698 33,099.43 26,007.2 677.2227 454,509.8East 38,053 26,252.63 18,572.35 1703.181 223,347.8

German 176,751 0.9844923 0.1235609 0 1West 138,698 0.9810163 0.1364677 0 1East 38,053 0.9971619 0.0531993 0 1

No school certificate 176,751 0.0157566 0.1245329 0 1West 138,698 0.0185655 0.1349851 0 1East 38,053 0.0055186 0.0740831 0 1

University degree 176,751 0.173668 0.3788248 0 1West 138,698 0.1562315 0.363076 0 1East 38,053 0.2372218 0.425385 0 1

Unemployed 176,751 0.0416744 0.1998447 0 1West 138,698 0.0320985 0.1762624 0 1East 38,053 0.0765774 0.2659232 0 1

Student 176,751 0.0164242 0.1271006 0 1West 138,698 0.0179743 0.1328584 0 1East 38,053 0.0107744 0.1032407 0 1

Self-employed 176,751 0.0541836 0.2263802 0 1West 138,698 0.0604983 0.2384087 0 1East 38,053 0.0311671 0.1737713 0 1

Non-working 176,751 0.0185628 0.1349754 0 1West 138,698 0.0173975 0.1307477 0 1East 38,053 0.0228103 0.1493003 0 1

Retiree 176,751 0.260949 0.4391534 0 1West 138,698 0.2643297 0.4409772 0 1East 38,053 0.2486269 0.4322227 0 1

Year of birth 176,751 1950.612 15.39899 1908 1988West 138,698 1950.473 15.58062 1908 1988East 38,053 1951.118 14.7066 1908 1988

Unemployment rate 176,751 10.78472 4.27835 4.3 24.6West 138,698 8.996211 2.459645 4.3 20.8East 38,053 17.30357 2.969076 11.1 24.6



4. Empirical analysis

4.1. Baseline regressions

4.1.1. Econometric modelGiven our research question, household spending on visible consumption is the variable of central interest to our

econometric analysis. The dependent variable lnðvisible spendingiÞ captures household i's expenditures on goods that areclassified as easily observable by others and status-relevant. As indicated in Section 3, we rely on a baseline definition ofwhat constitutes expenditures on “visible goods”; we check the robustness of our findings by reference to our intermediatedefinition and the definitions proposed by Charles et al. (2009) and Heffetz (2011) in Section 4.2.1.

Using the four waves of the EVS data, our first aim is to determine whether expenditures on conspicuous consumptiondiffer between East and West Germany. To identify whether or not there is a significant difference in the relative importanceof conspicuous consumption for East and West Germany, we introduce the dummy variable East Germanyi. This variabletakes the value of 1 whenever household i's residence is located within the borders of the former GDR. Our analysis theninvestigates how any differences evolve over time. For this purpose, we utilize dummy variables for the 1998, 2003, and2008 waves of the EVS and interact them with East Germanyi. The coefficients of these interaction terms show the wave-specific East Germany effects in comparison to the 1993 wave. This allows us to test whether potential differences betweenEast and West Germany are persistent over time even though East and West Germans have been subject to the sameinstitutions since reunification.

Our set of control variables includes household characteristics that might have an effect on expenditures on visiblegoods. We first control for household income, which is the variable most likely to be a key predictor for consumptionexpenditures.23 In order to reflect the concept of wealth, we utilize the household's level of savings as a measure of actualpermanent income. We also consider total consumption expenditures as an explanatory variable.24 Acknowledging the factthat households with different characteristics have different consumption demands, we use the number of adults andchildren in the household as well as the age, gender, and employment status (with unemployed, student, retiree, self-employed, and non-working as alternatives for the default “employed”) of the head of household as control variables. Withregard to education, our estimations include two dummy variables indicating whether or not the head of household holds auniversity degree or has obtained any school-leaving certificate. The dummy variable Femalei (Germani) takes the value of 1whenever the head of household i is female (a German citizen). For the location of residence, we include a dummy variableCityi that takes the value of 1 whenever the number of inhabitants in the household's location of residence exceeds100,000.25 This inclusion follows from Veblen's idea that signaling one's wealth via conspicuous consumption tends to bemore important when social cohesion is low (as otherwise it is futile to resort to such a costly instrument of informationtransmission). To account for the fact that conspicuous consumption occurs in the context of the distribution of income for ahousehold's reference group (as argued in Section 1), we use the EVS data to construct proxies for the income distribution ofthe reference group at time t, t¼1993, 1998, 2003, 2008, by creating lnðmean incomejtÞ and lnðstddv: incomejtÞ, where thepopulation of the household's state of residence j is employed as a reference group.

Our full econometric model can be stated as follows:

lnðvisible spendingiÞ ¼ α0þα1 lnðincomeiÞþα2ðlnðincomeiÞÞ2þα3lnðtotal cons: expendituresiÞþα4 East Germanyiþα5 East Germanyi � 1998þα6 East Germanyi � 2003

þα7 East Germanyi � 2008þα8 Femaleiþα9 Ageiþα10 Age2iþα11 #of adultsiþα12 #of childreniþα13 Cityiþα14 Germani

þα15 No school certificateiþα16 University degreeiþα17 Unemployedi

þα18 Studentiþα19 Self�employediþα20 Non�workingiþα21 Retireeiþα22 lnðsavingsiÞþα23 lnðmean incomejtÞþα24 lnðstddv: incomejtÞþα25 1998þα26 2003þα27 2008þεi;

for household i¼ 1;…;176751 living in state j¼ 1;…;17.

4.1.2. Estimation resultsWe begin including our explanatory variables stepwise into the regressions. All models are estimated using OLS with

heteroskedasticity-robust Huber–White standard errors. The results are presented in Table 3. As income is likely to be animportant determinant of spending behavior, we start by considering income and its square as the only explanatoryvariables (column (1)). In column (2), we also include a household's level of overall consumption spending. We find that

23 Squared income also enters our econometric model to account for potential nonlinearities.24 Similar to previous studies (e.g., Charles et al., 2009), expenditures on visible and total consumption, income, and savings will enter our empirical

model in log form; this allows us to interpret the respective coefficients as elasticities.25 This definition follows the German specification for a city (“Großstadt”) and allows us to optimally exploit the information on household location

provided in the EVS.

Table 3Determinants of conspicuous consumption expenditures.

(1) (2) (3) (4) (5) (6)Baseline definition Baseline definition Baseline definition Baseline definition Baseline definition Baseline definition

ln(income) 1.34nnn 0.88nnn 0.808nnn 0.8nnn 0.708nnn 0.712nnn

(0.398) (0.217) (0.147) (0.146) (0.153) (0.171)(ln(income))2 �0.019 �0.025nn �0.04nnn �0.04nnn �0.033nnn �0.034nnn

(0.019) (0.01) (0.007) (0.007) (0.007) (0.008)ln(total cons. exp.) 0.945nnn 1.532nnn 1.533nnn 1.59nnn 1.591nnn

(0.003) (0.004) (0.004) (0.004) (0.016)East Germany 0.071nnn 0.165nnn 0.217nnn 0.292nnn

(0.003) (0.005) (0.005) (0.024)East Germany � 1998 �0.093nnn �0.107nnn �0.121nnn

(0.008) (0.008) (0.018)East Germany � 2003 �0.119nnn �0.126nnn �0.147nnn


(0.008) (0.008) (0.019)Female 0.012nnn 0.012nnn

(0.004) (0.004)Age �0.012nnn �0.012nnn

(0.001) (0.001)Age2 7�10�5 nnn 7�10�5 nnn

(9�10�6) (1�10�5)# of children �0.085nnn �0.085nnn

(0.001) (0.002)# of adults �0.124nnn �0.123nnn

(0.005) (0.008)City 0.012nnn 0.018nn

(0.003) (0.009)German 0.05nnn 0.052nnn

(0.012) (0.015)No school certificate 0.05nnn 0.049nnn

(0.014) (0.014)University degree �0.059nnn �0.059nnn

(0.004) (0.009)Unemployed �0.156nnn �0.156nnn

(0.011) (0.012)Student 0.09nnn 0.089nnn

(0.015) (0.027)Self-employed �0.021nnn �0.023nnn

(0.006) (0.007)Non-working �0.045nnn �0.045nn

(0.014) (0.021)Retiree 0.007 0.007

(0.007) (0.013)lnðSavingsÞ 0.013nnn 0.013nnn

(0.0004) (0.001)lnðmean incomejt Þ 0.208nn

(0.095)lnðstddv: incomejt Þ �0.02

(0.063)Constant �3.545n �7.5nnn �11.848nnn �11.841nnn �11.642nnn �13.728nnn

(2.095) (1.138) (0.767) (0.762) (0.811) (1.018)Wave dummies Yes Yes Yes Yes

N 176,751 176,751 176,751 176,751 170,780 170,780R2 0.35 0.6 0.72 0.72 0.74 0.74F (East Germany, 1998) 139.06 325.85 68.62p-value F-test 0.0000 0.0000 0.0000F (East Germany, 2003) 45.06 176.96 24.03p-value F-test 0.0000 0.0000 0.0000F (East Germany, 2008) 3.99 38.16 30.53p-value F-test 0.0459 0.0000 0.0000

Notes: Coefficient estimates from ordinary least squares regressions. The dependent variable is ln(conspicuous consumption expenditures), whereconspicuous consumption follows our baseline definition (see Table 1). East Germany is the explanatory variable of central interest. This dummy variabletakes the value of 1 whenever a household is located in East Germany. 1998, 2003, and 2008 are dummy variables that indicate the three last waves in ourdataset, such that the year 1993 is our baseline. The dummy variable City takes the value of 1 whenever a household resides in a city with more than100,000 inhabitants. The control variables Female, German, No school certificate, University degree, Unemployed, Student, Self-employed, Non-working, andRetiree are dummy variables. All other explanatory variables are non-binary (see Table 2 for descriptive statistics). lnðmean incomejt Þ and lnðstddv: incomejt Þrepresent moments of the income distribution in state j and wave t and characterize a household's reference group. The number of children and/or adults isunclear for some households in the EVS data. We therefore end up using 170;780 observations in all estimations including these two explanatory variables(columns 5 and 6). Heteroskedasticity-robust Huber-White standard errors are in brackets. Standard errors are clustered at the reference group level jt incolumn 6. F (East Germany, wave) shows F-statistic for East GermanyþEast Germany�wave¼0.

n Significant at the 10% level.nn Significant at the 5% level.nnn Significant at the 1% level.



conspicuous consumption expenditures are linearly increasing in the log of total consumption expenditures,26 while theeffect of income on conspicuous consumption is strictly concave.

Column (3) presents the first estimation that includes the East Germany dummy variable. Its coefficient is positive and highlysignificant, indicating that households in East Germany spend 7.1% more on average on conspicuous consumption. Havingestablished this significant East Germany effect, column (4) addresses how the difference between East and West Germanyevolves over time. Including the three wave dummy variables and interacting them with the East Germany dummy variable, wefind that the level of conspicuous consumption spending was 16.5% larger in East Germany than in West Germany in 1993. Thecoefficients of all three interaction terms are negative and highly significant; this shows that the gap between East and WestGermany has decreased but has not vanished, even 18 years after reunification. On the basis of F-tests, we find that the hypothesisthat East Germany þEast Germany�wave¼0 can be rejected for all waves (po0:0001 in 1998 and 2003, p¼0.0459 in 2008).

As explained in Section 4.1.1, we consider other household-specific control variables such as the head of household's ageand education and information on the household structure (number of children, number of adults) as importantdeterminants of spending on conspicuous consumption. The regression results based on the full set of household-specificcontrols are presented in column (5).27 The coefficients of the variables of central interest are again highly significant andindicate some but not complete convergence.28

Our preferred model additionally includes information on the income distribution of the household's reference group.Column (6) depicts the estimation results for the full model. In this estimation, standard errors are clustered at the referencegroup level (state-year) as these distribution variables are fixed at the state-year level. The absolute size of the East Germanycoefficient and the interaction terms increases due to the inclusion of the income distribution control variables. Thecoefficient of the East Germany dummy variable shows that the level of conspicuous consumption expenditures was 29.2%larger in East Germany immediately after reunification. This gap became smaller in the 1998 wave of the EVS, but remainedstatistically highly significant (po0:0001 according to the F-test for East Germany þ East Germany �1998¼0). We find astepwise reduction of the gap in the 2003 and 2008 waves. However, the combination of the East Germany and EastGermany�2008 coefficients indicates a remaining gap of more than 8% (significant at the 1%-level) in 2008 – that is, 18years after the reunification of Germany. In view of these estimation results, we conclude that the data strongly suggeststhat there was indeed a highly significant influence of the political regime on individual preferences, leading people in EastGermany to spend more on conspicuous consumption. The gap between East and West German spending on conspicuousgoods decreased during the 18 years after reunification, but had not yet disappeared altogether.

The coefficients for most of our control variables are also highly significant. Columns (5) and (6) in Table 3 reveal that womenspend about 1.2% more on visible goods than men. This is interesting, given that Charles et al. (2009) do not report any gendereffect for their dataset; the literature on the importance of relative standing presents mixed conclusions in this regard.29 We alsoobserve that expenditures on visible goods decline with the head of household's age;30 this corresponds with the findings ofCharles et al. (2009). Location in a city with more than 100,000 inhabitants seems to result in greater conspicuous consumption,which would be in line with Veblen's (1899) ideas about the influence of social cohesion. Furthermore, the coefficient for theGerman citizenship of the head of household is positive and significant. Our findings regarding the influence of the level ofeducation suggest that households with higher education tend to spend less on conspicuous consumption. If the head of thehousehold did not finish school, conspicuous consumption expenditures are significantly higher. If the head of the householdobtained a university degree, conspicuous consumption expenditures are significantly lower. Due to the structure of our data, wecannot control for educational achievements in greater detail as information on the type of high school diploma enablinguniversity admission (Abitur) is missing from two of the four waves. With respect to the influence of savings, we obtain theintuitive finding that households with more savings (which may be understood as those households with higher permanentincome when current income is held constant) choose to spend more on visible goods. Higher mean income in the referencegroup increases spending on visible goods, whereas the variance has no discernible significant effect.31 Finally, according tocolumns (5) and (6), the presence of more adults or children diminishes the tendency to consume conspicuously.32

26 In unreported regressions, we obtained an insignificant coefficient when including a squared term for total consumption expenditures.27 Due to the structure of the EVS data, the number of children and/or adults is unclear for some households. We therefore end up using 170,780

observations in all estimations including these two control variables.28 F-tests on East Germany þ East Germany�wave¼0 can be rejected at the 1%-level for each of the three waves.29 For example, a survey of Costa Rican students conducted by Alpizar et al. (2005) also indicates that women care more about relative income and

consumption than men. A similar finding is obtained by Corazzini et al. (2012). However, based on survey results, Pingle and Mitchell (2002) argue thatgender does not influence the probability that an individual will show a positional concern for income. Likewise, Dohmen et al. (2011) report that theimportance of relative standing is comparable for the two sexes.

30 Although the coefficient of Age2 is positive, it is dominated by the negative linear age effect in the relevant range.31 When we test an alternative definition of the reference group, exploiting the information about whether or not the head of household holds a

university degree in combination with the state of residence, we obtain insignificant coefficients for both the mean and the standard deviation. However, itremains unclear whether this is due to the lack of more detailed educational information or whether households’ reference groups do not depend heavilyon educational levels. It is important to note that the coefficients of our variables of central interest are very similar to those presented in our preferredmodel, leading us to conclude that the East Germany effect is not driven by the definition of the reference group (see the Online Appendix for detailedregression results).

32 When thinking about potentially relevant additional control variables, the level of unemployment comes to mind, because East and West Germanyshow persistent differences with regard to the rate of unemployment (see Section 2). Including this variable as an additional control, we find that


4.2. Robustness checks

In this section, we test the robustness of our findings in several dimensions. First, we show that our findings are robust tovarying the definition of our dependent variable in our empirical model. Next, we explore finer distinctions in severalregards to isolate the potential importance of unobserved heterogeneity in our baseline regressions. As a next step, weconsider cohort effects. We conclude by discussing an artificial division of German states (North-South) unrelated to thepolitical division as a kind of placebo test.

4.2.1. Varying the definition of the dependent variableIn this section, we consider whether or not the identified East Germany effect is robust to different definitions of what

constitutes conspicuous consumption expenditures and to using relative conspicuous consumption expenditures instead oftheir level as our dependent variable.33

As a first robustness check, we use the intermediate definition of conspicuous consumption described in Table 1. As isevident from column (1) in Table 4, the magnitude and significance level of the East Germany effect barely changescompared to our baseline regression. Importantly, we again find that spending patterns seem to have converged, but thatthere still is a significant difference between East and West Germany almost twenty years after reunification. In a secondvariation, column (2) in Table 4 applies the definition of visible goods proposed by Charles et al. (2009). Again, we findsupport for the hypothesis that the relative importance of conspicuous consumption is higher in East Germany. Theestimation results reveal that the difference between East and West Germany is still statistically significant in the year 2008.The same picture emerges whenwe apply the definition specified by Heffetz (2011), which is used as the dependent variablein column (3) of Table 4. These regression results support the findings based on our baseline definition of conspicuousconsumption expenditures and demonstrate that our findings do not depend on the definition of the dependent variable.

Another variation of our baseline model uses a relative rather than an absolute measure as the dependent variable (i.e.,the ratio of expenditures on conspicuous consumption over total consumption expenditures instead of an absolute value forconspicuous consumption expenditures). In column (4) of Table 4, we report the results, which are very similar to thosefrom our preferred model, in that they show a highly significant East Germany effect that declines over time but is stillsignificant in the year 2008.

4.2.2. Dealing with (unobserved) heterogeneityIn this section, we explore the repercussions of introducing finer distinctions into our empirical model. We first introduce

dummies for each East German state, consider a model which allows for an impact specific to East Germany in alldimensions next, and, finally, consider a model based on different types of households (with a set of binary variablesrepresenting different combinations of socioeconomic characteristics).

The results documented in Section 4.1.2 show evidence of an East Germany effect that we attributed to the treatment bythe different political regime. We now analyze whether this positive and persistent effect of the GDR's political regime onconspicuous consumption expenditures exists for each individual East German state. This focus on the state level allows usto determine whether our findings are driven by certain states and whether there is heterogeneity between states. Similareffects for all East German states would clearly support our conjecture that the overall East Germany effect was indeeddriven by the political regime. To analyze these issues, we replace the East Germany dummy variable with six dummyvariables, one for each East German state (Brandenburg, Mecklenburg-West Pomerania, Saxony, Saxony-Anhalt, Thuringia,and East Berlin), and estimate a model that includes all of the explanatory variables from our preferred model. The results ofthis estimation are presented in column (1) of Table 5.34 The results mirror the findings from our baseline model: there is apositive and highly significant effect for 1993 and a subsequent decrease in the positive state-specific conspicuousconsumption effects over time. When we conduct F-tests for all states and waves, we find that the hypothesis State þState�wave¼0 can be rejected at the 1%-level for all waves and states. We can conclude that our results using the EastGermany dummy variable are not driven by any individual state, but similarly arise for each single East German state.

Next, we consider a model that accounts for potential differences in the effects of the explanatory variables in East andWest Germany. This model includes interaction terms for all independent variables with the East Germany dummy variable

(footnote continued)conspicuous consumption expenditures decrease modestly with the unemployment rate, and that the inclusion of this variable does not influence theidentified East Germany effect (see the Online Appendix for detailed regression results).

33 Since a potential objection to our findings is that East Germans may simply have higher overall consumption levels, we also estimated a model usinghousehold i's total consumption expenditures as the dependent variable and the same explanatory variables as in our preferred model (omitting totalconsumption expenditures) in order to test whether total expenditures differ between East and West Germany. We find a statistically significant negativecoefficient of the East Germany dummy variable. The coefficient of the East Germany�2008 interaction term is positive and significant (but smaller inabsolute size than the coefficient of the East Germany dummy), while the coefficients of the other two interaction terms are insignificant. F-tests show thatEast Germany þEast Germany�wave¼0 can be rejected for 1998 and 2003, but not for 2008. We therefore conclude that our findings presented above arenot driven by effects resulting from total consumption spending; this supports the validity of our conspicuous consumption interpretation. Detailed resultsare presented in the Online Appendix.

34 In order to save space, we focus on the state coefficients in this table. The coefficients of all of the control variables are similar to those from Table 3.Complete tables are available upon request.

Table 4Determinants of conspicuous consumption expenditures for alternative definitions of conspicuous consumption.

(1) (2) (3) (4)Intermediate definition Charles et al. definition Heffetz definition Share (baseline def.)

lnðincomeÞ 0.357nnn 0.501nnn 0.751nnn 0.043n

(0.109) (0.18) (0.205) (0.025)

ðlnðincomeÞÞ2 �0.021nnn �0.026nnn �0.034nnn 0.001

(0.005) (0.008) (0.009) (0.001)ln(total cons. exp.) 1.609nnn 1.424nnn 1.506nnn

(0.018) (0.022) (0.018)East Germany 0.277nnn 0.22nnn 0.31nnn 0.047nnn

(0.028) (0.031) (0.027) (0.005)East Germany � 1998 �0.098nnn �0.095nnn �0.181nnn �0.019nnn



(0.023) (0.024) (0.023) (0.004)Female 0.145nnn 0.237nnn �0.3nnn 0.003nnn

(0.008) (0.013) (0.004) (0.001)Age �0.01nnn �0.01nnn �0.006nnn �0.001nnn

(0.001) (0.001) (0.001) (0.0002)Age2 4�10�5 nn 6�10�5 nnn 4�10�5 nn 3�10�6

(1�10�5) (1�10�5) (2�10�5) (2�10�6)# of children �0.027nnn 0.01nn �0.075nnn �0.015nnn

(0.003) (0.004) (0.003) (0.001)# of adults 0.017 0.122nnn �0.115nnn �0.003nnn

(0.011) (0.016) (0.009) (0.001)City �0.028nnn �0.031nnn 0.001 0.009nnn

(0.01) (0.01) (0.008) (0.002)German 0.019 0.022 0.086nnn 0.012nnn

(0.02) (0.019) (0.014) (0.003)No school certificate 0.087nnn 0.072nnn 0.018 �0.006n

(0.02) (0.019) (0.022) (0.003)University degree �0.117nnn �0.139nnn �0.057nnn �0.006nnn

(0.008) (0.011) (0.009) (0.002)Unemployed �0.15nnn �0.169nnn �0.109nnn �0.013nnn

(0.011) (0.014) (0.014) (0.002)Student �0.064n �0.076nnn 0.139nnn 0.027nnn

(0.034) (0.032) (0.029) (0.004)Self-employed �0.046nnn �0.003 �0.017nn �0.002

(0.01) (0.009) (0.007) (0.001)Non-working �0.053nn �0.075nnn �0.032 0.011nnn

(0.024) (0.025) (0.02) (0.003)Retiree �0.061nnn �0.074nnn 0.005 0.023nnn

(0.01) (0.01) (0.012) (0.002)ln(savings) 0.007nnn 0.008nnn 0.01nnn 0.003nnn

(0.001) (0.001) (0.001) (0.0001)lnðmean incomejt Þ 0.298nnn 0.34nnn 0.108 0.02

(0.107) (0.116) (0.092) (0.016)lnðstddv: incomejt Þ �0.147nn �0.148n 0.018 0.012

(0.066) (0.077) (0.062) (0.013)Constant �11.829nnn �11.82nnn �12.648nnn �0.771nnn

(0.885) (1.335) (1.185) (0.161)Wave dummies Yes Yes Yes Yes

N 170,780 170,780 170,780 170,780R2 0.64 0.55 0.69 0.26F (East Germany, 1998) 58.71 20.54 38.37 42.7p-value F-test 0.0000 0.0000 0.0000 0.0000F (East Germany, 2003) 26.89 15.6 22.95 13.23p-value F-test 0.0000 0.0002 0.0000 0.0005F (East Germany, 2008) 8.32 3.38 11.93 17.11p-value F-test 0.0053 0.0704 0.001 0.0001

Notes: Coefficient estimates from ordinary least squares regressions. The dependent variable is ln(conspicuous consumption expenditures) in Columns (1)–(3) with alternative definitions following the exposition in Table 1. The dependent variable in Column (4) is the share of conspicuous consumptionexpenditures according to our baseline definition over a household's total consumption expenditures. East Germany is the explanatory variable of centralinterest. This dummy variable takes the value of 1 whenever a household is located in East Germany. 1998, 2003, and 2008 are dummy variables thatindicate the three last waves in our dataset, such that the year 1993 is our baseline. The dummy variable City takes the value of 1 whenever a householdresides in a city with more than 100,000 inhabitants. The control variables Female, German, No school certificate, University degree, Unemployed, Student, Self-employed, Non-working, and Retiree are dummy variables. All other explanatory variables are non-binary (see Table 2 for descriptive statistics).lnðmean incomejt Þ and lnðstddv: incomejt Þ represent moments of the income distribution in state j and wave t and characterize a household's referencegroup. The number of children and/or adults is unclear for some households in the EVS data. All estimations rely on the sample of 170,780 households forwhich the number of children and/or adults is unambiguously given in the dataset. Heteroskedasticity-robust Huber–White standard errors clustered atthe reference group level jt are in brackets. F (East Germany, wave) shows F-statistic for East GermanyþEast Germany�wave¼0.



Table 5Determinants of conspicuous consumption expenditures with East German state dummies included.

(1) (2)Baseline definition Baseline definition

Brandenburg 0.315nnn

(0.019)Brandenburg � 1998 �0.168nnn



(0.013)Mecklenburg-West Pomerania 0.286nnn

(0.022)Mecklenburg-West Pomerania � 1998 �0.134nnn



(0.011)Saxony 0.33nnn

(0.022)Saxony � 1998 �0.129nnn

(0.012)Saxony � 2003 �0.125nnn

(0.014)Saxony � 2008 �0.228nnn

(0.015)Saxony-Anhalt 0.29nnn

(0.023)Saxony-Anhalt � 1998 �0.098nnn

(0.01)Saxony- Anhalt � 2003 �0.137nnn

(0.014)Saxony-Anhalt � 2008 �0.193nnn

(0.013)Thuringia 0.314nnn

(0.024)Thuringia � 1998 �0.142nnn

(0.011)Thuringia � 2003 �0.209nnn

(0.015)Thuringia � 2008 �0.198nnn

(0.013)East Berlin 0.225nnn

(0.025)East Berlin � 1998 �0.055nnn

(0.011)East Berlin � 2003 0.053nnn

(0.012)East Berlin � 2008 �0.165nnn

(0.017)South Germany 0.04n

(0.02)South Germany � 1998 �0.004

(0.027)South Germany � 2003 �0.016

(0.028)South Germany � 2008 0.008

(0.029)Constant �13.908nnn �8.871nnn

(1.034) (1.023)Wave dummies yes yes

N 170780 170780R2 0.74 0.73

Notes: Coefficient estimates from ordinary least squares regressions. The dependent variable is ln(conspicuous consumption expenditures), whereconspicuous consumption follows our baseline definition (see Table 1). In Column (1), we include state dummy variables for each East German stateand their interaction with the wave dummy variables 1998, 2003, and 2008, such that the year 1993 is our baseline. In Column (2), we alternatively includea dummy for South Germany and its interaction with the wave dummy variables. The definition we use here includes Bavaria, Baden-Wuerttemberg,Hesse, Rhineland-Palatinate, Saarland, Saxony, and Thuringia in South Germany. The same additional explanatory variables as in column 6 of Table 3 are


included: ln(income), (ln(income))2, ln(total consumption expenditures), Female, Age, Age2, # of children, # of adults, City, German, No school certificate,University degree, Unemployed, Student, Self-employed, Non-working, Retiree, ln(savings), lnðmean incomejt Þ, and lnðstddv: incomejt Þ. The dummy variable Citytakes the value of 1 whenever a household resides in a city with more than 100,000 inhabitants. The control variables Female, German, No school certificate,University degree, Unemployed, Student, Self-employed, Non-working, and Retiree are dummy variables. All other explanatory variables are non-binary (seeTable 2 for descriptive statistics). lnðmean incomejt Þ and lnðstddv: incomejt Þ represent moments of the income distribution in state j and wave t andcharacterize a household's reference group. All estimations rely on the sample of 170,780 households for which the number of children and/or adults isunambiguously given in the dataset. Heteroskedasticity-robust Huber-White standard errors clustered at the reference group level jt are in brackets. F-testsfor StateþState�wave¼0 can be rejected at the 1%-level for all East German countries. F-test for South GermanyþSouth Germany�wave¼0 cannot berejected for the 2003 wave.n Significant at the 10% level.nn Significant at the 5% level.nnn Significant at the 1% level.


and thereby allows for a specific (parameterized) form of heterogeneity of the effects. The results of the correspondingordinary least squares estimation are presented in Table 6. We find that the effects of some regressors indeed differ betweenEast and West Germany. For example, the positive effect of savings on conspicuous consumption expenditures issignificantly smaller in East Germany. Additionally, the absolute size of the negative effect of the number of children andadults is larger for East German households. Overall, however, we still find a significant East Germany effect on conspicuousconsumption, although we do not observe any convergence. It is not straightforward to interpret the coefficient of the EastGermany dummy variable as one has to take into account all explanatory variables interacted with this particular dummy.We evaluate all continuous variables at their East-West-specific means. All dummy variables are set in a way thatcharacterizes their majority of observations (that is, it takes the value 1 if the corresponding dummy variable equals 1 formore than 50% of the households). This yields an East Germany effect in 1993 of 35.5% which is similar to the effectestimated in the baseline model. As can be seen from the East Germany�wave interaction terms, the difference inconspicuous consumption expenditures between East and West Germany does not decrease over the subsequent waves.These results support our finding of a general positive East Germany effect; however, they suggest that the convergence maybe driven by the distribution of some of the underlying control variables. Because it is not clear whether we can capturedifferences in the distributions of these control variables in the econometric models presented thus far, we now turn to adifferent approach.

In an attempt to capture unobserved differences, we choose an approach in which we first create a variety of binaryvariables to identify households with similar characteristics (i.e., values of explanatory variables) and then estimate theeffect of the political regime of the GDR within these types of households. More specifically, we first create a set of dummyvariables for different categories of explanatory variables (such as income and age); we then consider all possiblecombinations of these dummy variables. For example, one type of household might be characterized by an income below15,000 Euros, a male head of household between 31 and 45 years old, located in a city, etc. We include income (5 groups),age (4), gender (2), citizenship (2), location in a city (2), presence of children in the household (2), single-adult or largerhousehold (2), possession of a university degree (2), level of savings (2), and total consumption expenditures (4) asexplanatory factors.35 Next, the full set of combinations of these characteristics is included in a regression, together with theEast Germany dummy variable and the East Germany �wave interactions. Estimating the East Germany effect within thesetypes of households minimizes the potential for unobserved effects. The results of the corresponding OLS estimation arepresented in Table 7. We find a highly significant positive coefficient of the East Germany dummy variable, supporting thefindings from our preferred model. This coefficient indicates that the level of conspicuous consumption expenditures was17.4% higher among East German households with similar characteristics in 1993. As in our preferred model, we findnegative and highly significant coefficients for the East Germanyi�wave interaction terms for all three waves. The absolutevalue of these coefficients increases over time. Most importantly, F-tests for East Germany þ East Germany�wave¼0 can berejected at the 1%-level for all three waves.

4.2.3. Cohort effectsWe now consider the effects of the number of years under communism on individual preferences. Column (1) of Table 8

shows ordinary least square estimation results including the head of household's year of birth and its interaction with theEast Germany dummy variable. The interaction term's coefficient has a negative sign and is statistically significant at the1%-level, indicating that a longer treatment in the GDR yields an increased level of conspicuous consumption expenditures.In column (2), we re-estimate our baseline regressions by distinguishing between five different cohort groups: born in orbefore 1930 (cohort 1), born between 1931 and 1945 (cohort 2), born between 1946 and 1965 (cohort 3), born between 1966and 1975 (cohort 4), and born after 1975 (cohort 5). Members of the most recent cohort spent only their childhood undercommunism; this is the omitted group in the regressions. This cohorts-based estimation supports the results of ourpreferred model. It is also interesting to focus on estimations that include not only the cohort dummy variables but also

35 We rely on the following categories: income (o15;000; 15,000–34,999; 35,000–59,999; 60,000–129,999; Z130;000), age (o31; 31–45; 46–65;Z66), gender (male; female), citizenship (German; foreign), location in a city (yes; no), presence of children in the household (yes; no), single-adult orlarger household (single; larger), possession of a university degree (yes; no), level of savings (o5;000; Z5;000), and total consumption expenditures(o15;000; 15,000–29,999; 30,000–59,999; Z60;000).

Table 6Determinants of conspicuous consumption expenditures with interactions of all explanatory variables withEast Germany dummy variable included.

(1)Baseline definition

East Germany 3.831n

(2.15)East Germany � 1998 0.119nnn

(0.038)East Germany � 2003 0.135nn

(0.053)East Germany � 2008 0.046

(0.051)lnðincomeÞ 0.756nnn

(0.225)lnðincomeÞ � East Germanyi �0.28

(0.259)

ðlnðincomeÞÞ2 �0.035nnn

(0.01)

ðlnðincomeÞÞ2 � East Germanyi 0.013

(0.012)ln(total cons. exp.) 1.561nnn

(0.017)lnðtotal cons: exp:Þ � East Germanyi 0.158nnn

(0.025)Female 0.014nnn

(0.005)Female� East Germanyi 0.007

(0.008)Age �0.011nnn

(0.001)Age� East Germanyi �0.007nnn

(0.002)Age2 5�10�5 nnn

(1�10�5)

Age2 � East Germanyi 1�10�4nnn

(3�10�5)# of children �0.081nnn

(0.002)# of children� East Germanyi �0.023nnn

(0.005)# of adults �0.105nnn

(0.007)# of adults� East Germanyi �0.106nnn

(0.015)City 0.015

(0.01)City� East Germanyi 0.012

(0.021)German 0.055nnn

(0.016)German� East Germanyi 0.017

(0.065)No school certif. 0.057nnn

(0.014)No school certificate� East Germanyi �0.095

(0.06)University degree �0.083nnn

(0.008)University degree� East Germanyi 0.072nnn

(0.011)Unemployed �0.171nnn

(0.015)Unemployed� East Germanyi 0.08nnn

(0.02)Student 0.08nn

(0.034)Student� East Germanyi 0.067

(0.055)Self-employed �0.026nnn

(0.007)


Table 6 (continued )

(1)Baseline definition

Self�employed� East Germanyi 0.027(0.018)

Non-working �0.063nn

(0.028)Non�working� East Germanyi 0.08nn

(0.034)Retiree �0.0001

(0.014)Retiree�EastGermanyi 0.035

(0.024)ln(savings) 0.013nnn

(0.001)lnðsavingsÞ � East Germanyi �0.004nnn

(0.001)lnðmean incomejt Þ 0.153

(0.103)lnðmean incomejt Þ � East Germanyi 0.282

(0.224)lnðstddv: incomejt Þ 0.042

(0.063)lnðstddv: incomejt Þ � East Germanyi �0.646nnn

(0.19)Constant �13.77nnn

(1.268)Wave dummies Yes

N 170,780R2 0.74F (East Germany, 1998) 3.33p-value F-test 0.0724F (East Germany, 2003) 3.34p-value F-test 0.072F (East Germany, 2008) 3.19p-value F-test 0.0784

Notes: Coefficient estimates from ordinary least squares regressions. The dependent variable is ln(conspicuous consumption expenditures), where conspicuous consumption follows our baseline definitionin Table 1. East Germany is the explanatory variable of central interest. This dummy variable takes the valueof 1 whenever a household is located in East Germany. 1998, 2003, and 2008 are dummy variables thatindicate the three last waves in our dataset, such that the year 1993 is our baseline. The dummy variableCity takes the value of 1 whenever a household resides in a city with more than 100,000 inhabitants. Thecontrol variables Female, German, No school certificate, University degree, Unemployed, Student, Self-employed,Non-working, and Retiree are dummy variables. All other explanatory variables are non-binary (see Table 2for descriptive statistics). lnðmean incomejt Þ and lnðstddv: incomejt Þ represent moments of the incomedistribution in state j and wave t and characterize a household's reference group. All explanatory variablesare interacted with the East Germany dummy. All estimations rely on the sample of 170,780 households forwhich the number of children and/or adults is unambiguously given in the dataset. Heteroskedasticity-robust Huber-White standard errors clustered at the reference group level jt are in brackets. F (EastGermany, wave) shows F-statistic for East GermanyþEast Germany�wave¼0.



their interactions with the East Germany dummy. The regression results are presented in column (3) of Table 8. As in theprevious estimations, we find a highly significant positive East Germany effect that decreases over time. The results revealthat this positive effect of the GDR treatment is larger for older households with a longer treatment duration: thecoefficients of the Cohort1� East Germany and Cohort2� East Germany interaction terms are positive and significant(po0:0001 and p¼0.063, respectively). However, the difference in the size of the treatment effect between cohorts 3/4 andthe reference cohort is not statistically significant. Overall, the most important result we obtain from focusing on cohorts isthe very intuitive finding that longer exposure to the political regime of the GDR makes the treatment effect morepronounced. Similar results were obtained by Alesina and Fuchs-Schündeln (2007), for instance.

4.2.4. North-South: Placebo testIn this section, we evaluate the reliability of our findings by means of a placebo approach, testing a different division of

German states by employing the dummy variable South Germany. While the East-West split is unambiguous, classificationwith respect to North and South is somewhat arbitrary. The definition we use includes Bavaria, Baden-Wuerttemberg, Hesse,

Table 7Determinants of conspicuous consumption expenditures when different types of households are represented by a large set of binary variables.

(1)Baseline definitionBin categories estimation

East Germany 0.174nnn

(0.007)East Germany � 1998 �0.089nnn

(0.011)East Germany � 2003 �0.109nnn

(0.011)East Germany � 2008 �0.122nnn

(0.011)Constant 8.787nnn

(0.199)Wave dummies Yes

N 170,780R2 0.64F (East Germany, 1998) 116.76p-value F-test 0.0000F (East Germany, 2003) 60.61p-value F-test 0.0000F (East Germany, 2008) 43.53p-value F-test 0.0000

Notes: Coefficient estimates from ordinary least squares regressions. The dependent variable is ln(conspicuous consumption expenditures), whereconspicuous consumption follows our baseline definition in Table 1. East Germany is the explanatory variable of central interest. This dummy variabletakes the value of 1 whenever a household is located in East Germany. 1998, 2003, and 2008 are dummy variables that indicate the three last waves in ourdataset, such that the year 1993 is our baseline. The explanatory variables are dummies that capture different types of households: We first create a set ofdummy variables for different categories of explanatory variables and then consider all possible combinations of these dummy variables. Income, age,gender, citizenship, location in a city, presence of children in the household, single-adult or larger household, possession of a university degree, level ofsavings, and total consumption expenditures are included as explanatory factors. We rely on the following categories of these explanatory factors: income(o15;000; 15,000–34,999; 35,000–59,999; 60,000–129,999; Z130;000), age (o31; 31–45; 46–65; Z66), gender (male; female), citizenship (German;foreign), location in a city (yes; no), presence of children in the household (yes; no), single-adult or larger household (single; larger), possession of auniversity degree (yes; no), level of savings (o5;000; Z5;000), and total consumption expenditures (o15;000; 15,000–29,999; 30,000–59,999;Z60;000). The full set of combinations of these categories is included as explanatory variables in our regression, together with the East Germany dummyvariable and the East Germany�wave interactions. All estimations rely on the sample of 170,780 households for which the number of children and/oradults is unambiguously given in the dataset. Heteroskedasticity-robust Huber-White standard errors are in brackets. F (East Germany, wave) shows F-statistic for East GermanyþEast Germany�wave¼0.n Significant at the 10% level.nn Significant at the 5% level.nnn Significant at the 1% level.


Rhineland-Palatinate, Saarland, Saxony, and Thuringia in South Germany because they geographically represent thesouthern half of the country. Our analysis yields a small and weakly significant positive South Germany effect (10%-level),as can be seen in column (2) of Table 5. However, this effect is notably smaller in size than the East Germany effect from ourprevious regressions. Note that the coefficient of the South Germany dummy variable is positive, like the coefficient for theEast Germany dummy, such that the East Germany effect reported above cannot be an artifact of a North-South differencewhereby conspicuous consumption is less important in South Germany. Most importantly, when we estimate a modelincluding both the South Germany and the East Germany dummy variables and all their interactions with the wave dummies,we find that the inclusion of South Germany (and the corresponding wave interactions) has barely any influence on ourinferences regarding the East Germany effect.36

5. Conclusion

This paper takes advantage of the natural experiment created by the division and reunification of Germany to analyzewhether political regimes influence preferences. In particular, we are interested in whether or not conspicuous consumption(as a means of signaling status to others) is relatively more important in East Germany. Our empirical analysis makes use of adataset that provides detailed information about consumption patterns and household income and is representative forGermany. The influences of political regimes (here, a communist regime that promotes egalitarianism and idealizes theworking class versus a democratic market economy) on how people seek to signal their status to others is an interesting

36 The coefficients are very similar to those reported in column (6) of Table 3: 0.276 instead of 0.292 for East Germany, �0.117 instead of �0.121 for EastGermany�1998, �0.142 instead of �0.147 for East Germany�2003, and �0.197 instead of �0.208 for East Germany�2008. All F-tests for East Germany þEast Germany�wave¼0 can be rejected at the 1%-level. Moreover, the F-tests for South Germany þ South Germany�wave¼0 cannot be rejected for the1998 and 2003 waves.

Table 8Determinants of conspicuous consumption expenditures with possible cohort effects.

(1) (2) (3)Baseline definition Baseline definition Baseline definition

East Germany 6.432nnn 0.29nnn 0.24nnn




(0.021) (0.019) (0.021)Year of birth 0.006nnn

(0.0003)Year of birth � East Germany �0.003nnn

(0.001)Cohort 1 (born 1930 or earlier) �0.288nnn �0.313nnn

(0.016) (0.018)Cohort1 (born 1930 or earlier) � East Germany 0.134nnn

(0.034)Cohort 2 (born 1931–1945) �0.219nnn �0.236nnn

(0.015) (0.019)Cohort 2 (born 1931–1945) � East Germany 0.073n


(0.015) (0.019)Cohort 3 (born 1946–1960) � East Germany 0.023


(0.015) (0.019)Cohort 4 (born 1961–1975) � East Germany �0.029

(0.038)Wave dummies Yes Yes Yes

N 170,780 170,780 170,780R2 0.74 0.74 0.74F (East Germany, 1998) 40.34 65.81 9.91p-value F-test 0.0000 0.0000 0.0025F (East Germany, 2003) 39.79 22.83 6.11p-value F-test 0.0000 0.0000 0.016F (East Germany, 2008) 39.32 28.51 3.30p-value F-test 0.0000 0.0000 0.0736F (East Germany, 1998, cohort 1) 61.00p-value F-test 0.0000F (East Germany, 1998, cohort 2) 80.87p-value F-test 0.0000F (East Germany, 1998, cohort 3) 44.19p-value F-test 0.0000F (East Germany, 1998, cohort 4) 19.44p-value F-test 0.0000F (East Germany, 2003, cohort 1) 35.12p-value F-test 0.0000F (East Germany, 2003, cohort 2) 34.18p-value F-test 0.0000F (East Germany, 2003, cohort 3) 20.85p-value F-test 0.0000F (East Germany, 2003, cohort 4) 8.43p-value F-test 0.005F (East Germany, 2008, cohort 1) 42.44p-value F-test 0.0000F (East Germany, 2008, cohort 2) 48.83p-value F-test 0.0000F (East Germany, 2008, cohort 3) 21.75p-value F-test 0.0000F (East Germany, 2008, cohort 4) 3.92p-value F-test 0.0518

Notes: Coefficient estimates from ordinary least squares regressions. The dependent variable is ln(conspicuous consumption expenditures), whereconspicuous consumption follows our baseline definition (see Table 1). East Germany is the explanatory variable of central interest. This dummy variabletakes the value of 1 whenever a household is located in East Germany. 1998, 2003, and 2008 are dummy variables that indicate the three last waves in ourdataset, such that the year 1993 is our baseline. The estimations account for treatment duration and therefore focus on households' years of birth. Thevariable Year of birth in column (1) varies between 1908 and 1988. Cohort1–4 are dummy variables that take the value of 1 whenever the head of the


respective household was born before 1931 (cohort 1), between 1931 and 1945 (cohort 2), between 1946 and 1960 (cohort 3), and between 1961 and 1975(cohort 4). Those born after 1975 form the reference category. The following additional explanatory variables are included: ln(income), (ln(income))2, ln(total consumption expenditures), Female, # of children, # of adults, City, German, No school certificate, University degree, Unemployed, Student, Self-employed,Non-working, Retiree, ln(savings), lnðmean incomejt Þ, and lnðstddv: incomejt Þ. The dummy variable City takes the value of 1 whenever a household resides in acity with more than 100,000 inhabitants. The control variables Female, German, No school certificate, University degree, Unemployed, Student, Self-employed,Non-working, and Retiree are dummy variables. All other explanatory variables are non-binary (see Table 2 for descriptive statistics). lnðmean incomejt) andlnðstddv: incomejt Þ represent moments of the income distribution in state j and wave t and characterize a household's reference group. The number ofchildren and/or adults is unclear for some households in the EVS data. All estimations rely on the sample of 170,780 households for which the number ofchildren and/or adults is unambiguously given in the dataset. Heteroskedasticity-robust Huber-White standard errors clustered at the reference group leveljt are in brackets. F (East Germany, wave) shows F-statistic for East GermanyþEast Germany�wave¼0. F (East Germany, wave, cohort t) shows F-statisticfor East GermanyþEast Germany�waveþ East Germany� cohort¼0.n Significant at the 10% level.nn Significant at the 5% level.nnn Significant at the 1% level.


issue on many levels. We show that there are significant differences in consumption patterns between East and WestGermany, with East Germans spending more on visible goods. While our analysis suggests that there has been convergencein consumption decisions between East and West Germany in this regard, a significant gap remains even 18 years afterreunification.

The results of the present paper suggest that institutions – in our case, political regimes – influence preferences. Thisfinding may be partially attributable to social norms that were shaped by the respective political regimes and passed on tosubsequent generations. Cohort-specific estimations undertaken as robustness checks for our main empirical model supportthis conjecture. This path dependence could be an important factor for policy makers to take into consideration. Our resultsmay also be of policy importance because conspicuous consumption has been shown to have far-reaching socialconsequences.37 The incentives for excessive conspicuous consumption could be reduced by means of policy instrumentssuch as region-specific taxation, a strategy already in use in the US.

Acknowledgments

We thank Florian Hett, Laszlo Goerke, Markus Pannenberg, Katrin Schmelz, and the participants of both the EconomicsWorkshop 2012 in Tübingen and the Annual Meeting of the German Economic Association (Verein für Socialpolitik) 2012 inGöttingen for all their helpful suggestions. We gratefully acknowledge the comments received from two anonymousreviewers and the editor in charge, Jörg Oechssler.

Appendix A. Supplementary data

Supplementary data associated with this article can be found in the online version at http://dx.doi.org/10.1016/j.euroecorev.2014.01.005.

References

Alesina, A., Fuchs-Schündeln, N., 2007. Good-bye Lenin (or not?): the effect of communism on people's preferences. Am. Econ. Rev. 97, 1507–1528.Alpizar, F., Carlsson, F., Johansson-Stenman, O., 2005. How much do we care about absolute versus relative income and consumption? J. Econ. Behav. Organ.

56, 405–421.Anderson, C., Kraus, M.W., Galinsky, A.D., Keltner, D., 2012. The local-ladder effect: social status and subjective well-being. Psychol. Sci. 23, 764–771.Andreoni, J., Vesterlund, L., 2001. Which is the fair sex? Gender differences in altruism. Q. J. Econ. 116, 293–312.Bauernschuster, S., Rainer, H., 2012. Political regimes and the family: how sex-role attitudes continue to differ in reunified Germany. J. Popul. Econ. 25, 5–27

.Bilancini, E., Boncinelli, L., 2012. Redistribution and the notion of social status. J. Publ. Econ. 96, 651–657.Börsch-Supan, A., Reil-Held, A., Rodepeter, R., Schnabel, R., Winter, J., 2001. The German savings puzzle. Res. Econ. 55, 15–38.Brosig, J., Helbach, C., Ockenfels, A., Weimann, J., 2011. Still different after all these years: solidarity behavior in East and West Germany. J. Publ. Econ. 95,

1373–1376.Burchardi, K.B., Hassan, T.A., 2013. The economic impact of social ties: evidence from German reunification. Q. J. Econ. 128, 1219–1271.Bursztyn, L., Cantoni, D., 2012. A tear in the Iron curtain: the impact of Western television on consumption behavior. CEPR Discussion Paper 9101.Caporale, G.M., Georgellis, Y., Tsitsianis, N., Yin, Y.P., 2009. Income and happiness across Europe: do reference values matter?. J. Econ. Psychol. 30, 42–51.Carlsson, F., Johansson-Stenman, O., Martisson, P., 2007. Do you enjoy having more than others? Survey evidence of positional goods. Economica 74,

586–598.Carlsson, F., Qin, P., 2010. It is better to be the head of a chicken than the tail of a phoenix: concern for relative standing in rural China. J. Socio-Economics

39, 180–186.Charles, K.K., Hurst, E., Roussanov, N., 2009. Conspicuous consumption and race. Q. J. Econ. 124, 425–467.

37 For example, Moav and Neeman (2010, 2012) find that conspicuous consumption may influence the likelihood of poverty traps. Similarly, Frank(2000) argues that conspicuous consumption may deplete resources that could otherwise be spent on healthcare, for example.



http://refhub.elsevier.com/S0014-2921(14)00007-5/sbref1



















Clark, A.E., Frijters, P., Shields, M.A., 2008. Relative income, happiness, and utility: an explanation for the Easterlin paradox and other puzzles. J. Econ. Lit. 46,95–144.

Clark, A.E., Senik, C., 2010. Who compares to whom? The anatomy of income comparisons in Europe. Econ. J. 120, 573–594.Clark, A.E., Senik, C., Yamada, K., 2013. The Joneses in Japan: income comparisons and financial satisfaction. Mimeo.Cooter, R., 1998. Expressive law and economics. J. Legal Stud. 27, 585–608.Corazzini, L., Esposito, L., Majorano, F., 2012. Reign in hell or serve in heaven? A cross-country journey into the relative vs absolute perceptions of wellbeing.

J. Econ. Behav. Organ. 81, 715–730.Corneo, G., 2001. Inequality and the state: comparing U.S. and German preferences. Annales d'Economie et de Statistique 63–64, 283–296.Corneo, G., Grüner, H.P., 2002. Individual preferences for political redistribution. J. Publ. Econ. 83, 83–107.Corneo, G., Jeanne, O., 1998. Social organization, status, and savings behavior. J. Publ. Econ. 70, 37–51.Croson, R., Gneezy, U., 2009. Gender differences in preferences. J. Econ. Lit. 47, 1–27.Dohmen, T., Falk, A., Fliessbach, K., Sunde, U., Weber, B., 2011. Relative versus absolute income, joy of winning, and gender: brain imaging evidence. J. Publ.

Econ. 95, 279–285.Easterlin, R.A., 1995. Will raising the incomes of all increase the happiness of all? J. Econ. Behav. Organ. 27, 35–47.Fehr, E., Hoff, K., 2011. Tastes, castes, and culture: the influence of society on preferences. Econ. J. 121, 396–412.Ferrer-i-Carbonell, A., 2005. Income and well-being: an empirical analysis of the comparison income effect. J. Publ. Econ. 89, 997–1019.Frank, R.H., 1985a. Choosing the Right Pond. Oxford University Press, Oxford.Frank, R.H., 1985b. The demand for unobservable and other nonpositional goods. Am. Econ. Rev. 75, 101–116.Frank, R.H., 2000. Luxury Fever: Money and Happiness in an Era of Excess. Princeton University Press, Princeton.Frank, R.H., 2008. Should public policy respond to positional externalities? J. Publ. Econ. 92, 1777–1786.Frey, B.S., Schmidt, S.L., Torgler, B., 2008. Relative income position, inequality and performance: an empirical panel analysis. In: Andersson, P., Ayton, P.,

Schmidt, C. (Eds.), Myths and Facts About Football: The Economics and Psychology of the World's Greatest Sport, Cambridge Scholars Publishing,Newcastle upon Tyne, pp. 349–369.

Fuchs-Schündeln, N., 2008. The response of household saving to the large shock of German Reunification. Am. Econ. Rev. 98, 1798–1828.Fuchs-Schündeln, N., Izem, R., 2012. Explaining the low labor productivity in East Germany – a spatial analysis. J. Comp. Econ. 40, 1–21.Fuchs-Schündeln, N., Krueger, D., Sommer, M., 2010. Inequality trends for Germany in the last two decades: a tale of two countries. Rev. Econ. Dyn. 13,

103–132.Fuchs-Schündeln, N., Schündeln, M., 2009. Who stays, who goes, who returns? Econ. Trans. 17, 703–738.Fulbrook, M., 2009. History of Germany 1918–2008, 3rd edition Wiley-Blackwell, Oxford.Glazer, A., Konrad, K., 1996. A signaling explanation for charity. Am. Econ. Rev. 86, 1019–1028.Grabka, M.M., Goebel, J., Schupp, J., 2012. Has income inequality spiked in Germany? DIW Econ. Bull. 12, 3–14.Heffetz, O., 2011. A test of conspicuous consumption: visibility and income elasticity. Rev. Econ. Stat. 93, 1101–1117.Heidemeyer, H., 1994. Flucht und Zuwanderung aus der SBZ/DDR 1945/1949-1961. Droste Verlag, Düsseldorf.Heineck, G., Süssmuth, B., 2013. A different look at Lenin's legacy: social capital and risk taking in the Two Germanies. J. Comp. Econ. 41, 789–803.Hillesheim, I., Mechtel, M., 2013. How much do others matter? Explaining positional concerns for different goods and personal characteristics. J. Econ.

Psychol. 34, 61–77.Hirsch, F., 1976. Social Limits to Growth. Harvard University Press, Cambridge, MA.Hopkins, E., Kornienko, T., 2004. Running to keep in the same place: consumer choice as a game of status. Am. Econ. Rev. 94, 1085–1107.Hubert, M., 1998. Deutschland im Wandel. Geschichte der deutschen Bevölkerung seit 1815. Franz Steiner Verlag, Stuttgart.Hunt, J., 2006. Staunching emigration from East Germany: age and the determinants of migration. J. Eur. Econ. Assoc. 4, 1014–1037.Hwang, S.H., Bowles, S., 2011. The sophisticated planner's dilemma: Mechanism design with endogenous preferences. Mimeo.Johansson-Stenman, O., Carlsson, F., Daruvala, D., 2002. Measuring future grandparents' preferences for equality and relative standing. Econ. J. 112, 362–383.Kaus, W., 2013. Conspicuous consumption and race: evidence from South Africa. J. Dev. Econ. 100, 63–73.Kopetsch, T., Rauscher, M., 2006. Zur Einkommenselastizität der Nachfrage nach Gesundheitsleistungen - Eine Analyse von Querschnittsdaten. Schmollers

Jahrbuch: J. Appl. Soc. Sci. Stud. 126, 59–81.Kuhn, P., Kooreman, P., Soetevent, A., Kapteyn, A., 2011. The effects of lottery prizes on winners and their neighbors: evidence from the Dutch postcode

lottery. Am. Econ. Rev. 101, 2226–2247.Lang, T., Nadler, R., 2012. Re-Turn: Regionalentwicklung durch Rückwanderung. Leibniz-Institut für Länderkunde.Merkel, I., 1999. Utopie und Bedürfnis. Die Geschichte der Konsumkultur in der DDR. Böhlau Verlag, Köln.Moav, O., Neeman, Z., 2010. Status and poverty. J. Eur. Econ. Assoc. 8, 413–420.Moav, O., Neeman, Z., 2012. Savings rates and poverty: the role of conspicuous consumption and human capital. Econ. J. 122, 933–956.Persky, J., Tam, M.Y., 1990. Local status and national social welfare. J. Reg. Sci. 30, 229–238.Pingle, M., Mitchell, M., 2002. What motivates positional concerns for income? J. Econ. Psychol. 23, 127–148.Ockenfels, A., Weimann, J., 1999. Types and patterns: an experimental East-West comparison of cooperation and solidarity. J. Publ. Econ. 71, 275–287.Rainer, H., Siedler, T., 2009. Does democracy foster trust? Evidence from the German reunification. J. Comp. Econ. 37, 251–269.Redding, S.J., Sturm, D.M., 2008. The costs of remoteness: evidence from German division and reunification. Am. Econ. Rev. 98, 1766–1797.Schäfgen, K., 1998. Die Verdoppelung der Ungleichheit. Sozialstruktur und Geschlechterverhältnisse in der Bundesrepublik und in der DDR. Humboldt-

University Berlin, Germany.Scheicher, C., 2010. Does work always pay in Germany? Ger. Econ. Rev. 11, 266–277.Senik, C., 2009. Direct evidence on income comparisons and their welfare effects. J. Econ. Behav. Organ. 72, 408–424.Solnick, S., Hemenway, D., 1998. Is more always better?: A survey on positional concerns J. Econ. Behav. Organ. 37, 373–383.Solnick, S., Hemenway, D., 2005. Are positional concerns stronger in some domains than in others? Am. Econ. Rev. 95, 147–151.Solnick, S., Hong, L., Hemenway, D., 2007. Positional goods in the United States and China. J. Socio-Economics 36, 537–545.Stutzer, A., 2004. The role of income aspirations in individual happiness. J. Econ. Behav. Organ. 54, 89–109.Statistisches Bundesamt, 2005a. Einkommens- und Verbrauchsstichprobe Aufgabe, Methode und Durchführung der EVS. Fachserie 15, Wirtschaftsrech-

nungen, Heft 7, Statistisches Bundesamt, Wiesbaden.Statistisches Bundesamt, 2005b. Einkommens- und Verbrauchsstichprobe Einnahmen und Ausgaben privater Haushalte. Fachserie 15, Wirtschaftsrech-

nungen, Heft 4, Statistisches Bundesamt, Wiesbaden.Statistisches Bundesamt, 2012. Bauen und Wohnen Mikrozensus-Zusatzerhebung 2010, Bestand und Struktur der Wohneinheiten, Wohnsituation der

Haushalte. Fachserie 5, Heft 1, Statistisches Bundesamt, Wiesbaden.Torgler, B., 2003. Does culture matter? Tax morale in an East-West-German comparison. FinanzArchiv 59, 504–528.Veblen, T., 1899. The Theory of the Leisure Class: Economic Study of Institutions. Random House, New York.Winkelmann, R., 2012. Conspicuous consumption and satisfaction. J. Econ. Psychol. 33, 183–191.Wolf, N., 2009. Was Germany ever united? Evidence from intra- and international trade, 1885–1933. J. Econ. Hist. 69, 846–881.
































































Documents

Conspicuous consumption and political regimes: Evidence from East and West Germany