KEEPING UP WITH THE SCHMIDTS
Do Richer Neighbours Make People Unhappier?
Gundi Knies
DIW Berlin and Centre for Market and Public Organisation, University of Bristol
< incomplete draft. Do not quote without author’s permission!>
Abstract
We test empirically whether people’s life satisfaction depends on their relative income position in the neighbourhood, drawing on a unique dataset, the German Socio-economic Panel Study (SOEP) matched with micro-marketing indicators of population characteristics.Relative deprivation theory suggests that individuals are happier the better their relative income position in the neighbourhood is. To test this theory we estimate micro-economic happiness models for the years 1994 and 1999 with controls for own income and for neighbourhood income at the zip-code level (roughly 9,000people). There exist no negative and no statistically significant associations between neighbourhood income and life satisfaction, which refutes relative deprivation theory. We find positiveassociations between neighbourhood income and happiness in all cross-sectional models and this is robust to a number of robustness tests, including adding in more controls for neighbourhood quality, changing the outcome variable, and interacting neighbourhood income with indicators that proxy the extent to which individuals may be assumed to interact with their neighbours. In particular, there are no effects of neighbourhood income on happiness once we control for differences between East and West Germany. Such regional differences are not usually being controlled for in the international happiness research.
1.Introduction
Research in the field of happiness has shown that people are happier the more income they have, and that, more importantly, they care about how this income compares to that of others (Clark and Oswald 1996; Frey and Stutzer 2002; Layard 2005). On the other hand, social scientists placed growing emphasis on geography as an explanatory factor for social inequalities between individuals (Jencks and Mayer 1990; Buck 2001). In this paper we integrate these two strands of research by an empirical investigation of whether levels of and changes in happiness depend on one’s financial position within one’s neighbourhood.
1.1Happiness Research and the Neighbourhood
Subjective well-being has been a heavily researched area in psychology, sociology and more recently in economics (see Diener, Suh et al. 1999 for a review). While an impact of neighbourhood and community contexts has been suggested in the happiness research (e.g., Layard 2005), few empirical studies have actually considered the neighbourhood context as a relevant variable in the prediction of happiness.
The role of the neighbourhood context in happiness research is twofold. One strand of this research looks at how variations in life circumstances can explain the variation in how happy individuals are with their life overall. Typically, domain-specific satisfaction reports are regressed on overall life satisfaction. An example is the study by Sirgy and Cornwell (2002), who examine the interplay between satisfaction with the neighbourhood, the community, the home and housing to explain overall life satisfaction. They show that overall life satisfaction depends on satisfaction with the neighbourhood through the mediator of community satisfaction, while housing satisfaction affects life satisfaction through the satisfaction with the home. When neighbourhood satisfaction is further differentiated into satisfaction with social features on the one hand and economic features on the other, the former is shown to affect life satisfaction via satisfaction with the community and the latter via satisfaction with the house and home (ibid.).
A problem with this kind of research is that the explanatory variables are likely to be highly correlated with life satisfaction and that the direction of causation is not clear. How happy individuals are with their life overall is certainly linked to how happy they are with different aspects of their life circumstances. Thus, these models contain endogeneity problems. A way around this is offered by Shields and Wooden (2003). The authors regress objective social and economic neighbourhood characteristics on life satisfaction, and identify sizeable neighbourhood effects. In particular, neighbourhoods perceived by respondents as places of high interaction between neighbours exert positive effects on people’s happiness. A theoretical explanation as to why it makes people happy to be around neighbours who interact frequently with each other is not provided in this research. It may be that people in neighbourhoods with good contact among neighbours also interact frequently with their neighbours and that being integrated into a community makes people happy. However, we could also think of occasions where living in close-knit neighbourhoods makes people feel miserable: for instance, when they are the subject of gossip or feel excluded from neighbourhood activities.
The second strand of happiness research that considers the neighbourhood context as a relevant variable looks at just this. It asks how individuals react to objectively existing or perceived differences in their own life circumstances compared to those of others. Michalos (1985), for instance, analyses the impact of a number of perceived discrepancies between what individuals have and what relevant others have on overall life satisfaction. Among other things survey respondents were asked how their life as a whole measures up to the average for most people their own age in their area. In most models no relationship between self-rated happiness and the indicator relating to neighbours was found significant.
More popular examples of comparison groups drawn on in the happiness literature are the society as a whole and people from the same profession. This type of research is mostly under the umbrella of testing the relative income hypothesis which states, in brief, that individual utility[1] is derived not so much from one’s absolute income position but from one’s income position within a relevant group. This theoretical framework explains, for instance, the apparent paradox that the happiness of a nation is only weakly associated with growing total incomes of its population, despite the fact that, at the level of individuals, gains in own income are associated with greater happiness (see, for instance, the empirical work by Easterlin 1974). In another study, Clark and Oswald (1996) find that reported levels of happiness with own income are at best weakly correlated with absolute levels of income, however, negatively correlated with relative income deprivation levels. The comparison group here is other people in the same profession, and it is argued that the incomes of other people in the same profession present a reference level against which individuals assess their own achievements. If incomes do not keep up with what is a normal income in the field, this leads to unhappiness.
Though neighbours have not typically been chosen as a comparison group in happiness research that addresses the relative income hypothesis, there are good reasons to assume that it also holds for this reference group. In the local housing market, for instance, it may not be some absolute amount of money that will ensure that the richest individual gets the best-quality land and property, but rather that income position of all people with a demand for land and property
in the city or town will determine who gets what and how much it costs to get the best spot.[2] The implications for happiness are that if we observe two individuals that are statistically identical apart from living in different neighbourhoods where the incomes of the neighbours in one neighbourhood are higher and lower in the other, the individuals in the richer neighbourhood will be unhappier with their lives (assuming that satisfaction with housing and the home affects life satisfaction, see Sirgy and Cornwell 2002), because their income will not have allowed them to find as nice a place as would have been possible in the neighbourhood with less affluent competitors.[3] Luttmer (2005) discusses the role which the neighbours’ income position plays for overall life satisfaction in this argument. The author employs income as a proxy for consumption and argues that “if utility depends on relative consumption, one person’s increase in consumption has a negative externality on others because it lowers the relative consumption of others” (ibid., p. 964). Using socio-economic panel data linked with Census data for the US, Luttmer finds a strong negative association between neighbour’s earnings and self-reported levels of happiness that is robust to a wide range of model specifications (including controlling for individual fixed effects, individual relocations, and interaction effects).[4]
1.2Neighbourhood Research and Happiness
From the perspective of neighbourhood effects studies, this is an attempt of testing the relative deprivation theory. Among the theories that have been put forward to explain the mechanisms through which neighbourhood context impacts on people’s life chances, the theory of relative deprivation is distinct in that it is the only one which suggests negative outcomes of living in a better-off neighbourhood (see Jencks and Mayer 1990; Buck 2001; Dietz 2002).
Relative deprivation concepts have been employed, for instance, to explain why some individuals have a higher propensity to get involved in riots (e.g., Gurr 1970; Canache 1996), but also to elucidate why students of a low socio-economic background are more successful in schools where students have an on-average low socio-economic status than in schools where the average student is from a family of high socio-economic status (e.g., Davis 1966; Meyer 1970). More recently, Lopez-Turley (2002) looked at children’s self-esteem and behaviour using PSID family and child development files 1995 linked with Census data 1987 and finds positive effects of relative deprivation on the test score results, and on the emotional and behavioural measures.
1.2.1Relative Deprivation and Comparison Theory
The term relative deprivation has been coined, yet only loosely defined, by Stouffer (1949). In order to explain why career prospects that objectively looked good were not accordingly appreciated, Stouffer proposes that people compare their own career prospects and achievements with that of others: when there are many opportunities for a promotion but a person fails to get promoted she feels worse than would be the case if the chances for a promotion had been lower. Similarly, individuals that get promoted appreciate this more when they were among few who did get promoted.
While comparison of self to others is assumed to build the basis for feelings of relative deprivation, it remains a question of both high theoretical and high empirical value under which conditions individuals compare themselves to others, and to whom. Suggestions for this can be derived from social comparison theory. Merton and Rossi, for instance, point out that “some similarity in status attributes between the individual and the reference group must be perceived or imagined for the comparison to occur at all” (1968, p. 296). This, in principle, gives rise to an indefinite number of possible reference groups. And this is a major obstacle for empirical tests of the theory of relative deprivation: unless surveys directly ask respondents against which others’ lot they assess theirs, and unless objective measures of the respective others’ lot are available at the same time, the social scientist is very likely to always be able to suggest more or less randomly a frame of reference the respondents to the survey might have taken so as to make sense of the empirical findings. Only those reference groups, however, are relevant for the sociologist that “are patterned by the social structure” (ibid., p. 298) and which thus cannot be assumed to vary randomly across individuals.
We know from studies in the field of social psychology that the groups that individuals compare themselves to are less random when comparison is undertaken along characteristics that are socially structured (like social status or income class) than when comparison is along more individualised characteristics (like intelligence or physical strength).[5] Social comparisons of the latter type have a stark tendency to be undertaken within the same, or with the adjacent strata. Yet, mere comparison within one stratum or across neighbouring strata is a weak condition for engendering feelings of deprivation. Runciman defines relative deprivation as “a psychological effect deriving from comparison with others who have achieved something that would be feasible to achieve for oneself, that one desires but does not have himself” (Runciman 1966, p. 9).
2.Methodology
Reference group theory suggests that three requirements need to be met when we want to subject to an empirical test relative deprivation theory in the neighbourhood context. At first it is necessary to show that people living in the same neighbourhood can indeed be understood as a reference group in the sociological sense (see, e.g., Suttles 1972). This requirement is met when we can demonstrate that membership in the neighbourhood provides a relevant and significant structure along which individuals organise their activities, behaviour or thoughts. Generally, the array of arguments with which this can be confirmed is broad. Galster and Killen (1995), for instance, provide an illustration of how local circumstances influence neighbours through numerous ways (e.g., through the quality and quantity of schools, the enforcement of law and order, and the housing market). In our study, we make the assumption that the particular neighbourhood units we employ (i.e., German zip-code areas, see description in Section 3.3.3 below) provide a frame of reference for the subjects of our study. We use geographical units as neighbourhoods that are commonly known by the residents, and often even by others. Some of these units refer to entire villages or towns and thereby have a name of their own. Due to the rather large spatial scale, we may think of these areas as providing a wide range of local institutions and other amenities that structure the neighbours’ lives.
Secondly, the characteristics of the neighbours that we employ to assess the effects of relative deprivation within the neighbourhood should be a relevant category for comparison. Relevant categories for comparison are those that are socially structured. Our comparison category is (own and others’) income. A great number of sociological studies on social inequality have shown that the income position that individuals (and, via socially structured partner selection and family formation, households) occupy is a function of parental income, education, and employment. The educational system and the job system are highly organised social systems, and income, as a highly socially structured unit, can thus be assumed to bear a high potential for representing a basis for social comparisons.
Last but not least, we want to be able to discriminate between deprivations that we can objectively identify and those deprivations that cause emotional distress. We undertake an empirical test of the relative deprivation theory in the neighbourhood context applying it to the outcome of self-reported levels of happiness.[6] Veenhoven (1984, p.10) defines life satisfaction “as the degree to which an individual judges the overall quality of his life-as-a-whole favourably”. A judgement of the overall quality of one’s life is the result of a process that involves an assessment of one’s objective living conditions but also of interpersonal and intertemporal comparisons. If feelings of relative deprivation are present we expect this to show in lower happiness scores.
However, not all lower happiness scores should be regarded as deriving from not being as well off as people think they deserve to be. In the absence of a dataset that observes objective and subjective deprivations at the same time, we need to operationalise relative deprivation on the basis of some measure of how people match up to others. In our study, this is the relative income position that individuals occupy in their neighbourhood. We define that people who have less income than their average neighbour are relatively deprived and assume that relatively deprived people will be unhappier.
Our empirical analyses unfold as follows. In a first step, we provide descriptive tables and graphs that show the mean life satisfaction by classes of household and neighbourhood income at two points in time, 1994 and 1999. We proceed to a multivariate model where we predict life satisfaction controlling for other aspects that are regarded to influence how satisfied people are with their lives.
2.1Multivariate Predictions
We include in our models controls for five domains of life that have been shown to impact on subjective well-being. These spheres are family context, financial situation, work, community and friends (which we proxy with neighbourhood characteristics), and health, respectively.[7] In addition, we include basic characteristics such as age and gender. A source for heterogeneity that is not usually considered in the international research on happiness is regional differences.
In the German context, it has been shown that people in East Germany are unhappier with their lives than people living in West Germany (e.g., Ferrer-i-Carbonell 2005). In addition, average personal incomes in East Germany are lower than in West Germany. This does not necessarily mean that all neighbourhoods in the East are poorer than neighbourhoods in West Germany. In fact, research by Knies and Krause (2006) has shown that some regions within West Germany are on average poorer than regions in East Germany. Also, it has been shown that the levels of happiness in the two regions converged over time and that most of the increase in East German’s happiness cannot be attributed to increases in personal income (Frijters, Haisken-DeNew et al. 2004). We therefore control for regional heterogeneity.