Fabian Stephany – Who are Your Joneses?

Trust is a good approach to explain the functioning of markets, institutions or society as a whole. It is a key element in almost every commercial transaction over time and might be one of the main explanations of economic success and development. Explaining the roots and patterns of trust across time and space, researchers have identified economic inequality to be one of the main threads to trust. Trust diminishes the more we perceive others to have economically different living realities. In most of the relevant contributions, scholars have taken a macro perspective on the inequality-trust linkage, with an aggregation of both trust and inequality on a country level. However, patterns of within-country inequality and possibly influential determinants, such as perception and socioeconomic reference, remained undetected. This paper offers the opportunity to look at the interplay between inequality and trust at a more refined level. With the use of 2009 EU-SILC data, measurements of income inequality are developed for socio-specific groups of society in 21 countries. A sizable variation in inequality measures can be noticed, even within known low inequality societies. Identifying the full spectrum of socio-specific inequality improves the estimation of trust in supplementation to actual income inequality. Findings show that the negative effects of inequality are amplified for individuals with high levels of trust.


The essential argument why inequality reduces generalized trust is that differences between individuals are too large to trust each other. Two major lines of reasoning are relevant in this respect. On the one hand, scholars like (Tumin 1953) have shown that generalized trust diminishes the more economically stratified a society is. Distinct spheres of everyday life emerge (segregated housing, public versus private schooling etc.) within which understanding and trust for “outsiders” fade away. This association is described as the stratification effect. On the other hand, claims have been made (Brockner 1996 that perceptions and believes about the distribution of resources and how they should ideally be allocated determine our trust in others by the mechanism of the so-called perception effect (Uslaner 2005). Previous studies have claimed to capture the effects of both stratification and perception with the use of one inequality indicator. Though, if one believes that the comparison with others is key to understand how inequality influences trust, the concept of economic reference groups should be described accordingly. Instead to comparing one’s own economic well-being to the rest of the society, it is much more likely, that we chose a reference group to which a meaningful comparison is possible, in the first place. For an evaluation of the \emph{perception effect} of inequality, “our family Jones”, to use the term of social class benchmarking, still needs to be defined.

However, the inequality-trust linkage has so far been examined only by looking at stratification and the society as a whole. Though, it can be assumed that personal characteristics, as age and education, play an important role in the deeper understanding of how income distributions are associated with generalized trust, for mainly two reasons. First, our understanding of justice strongly determines to what extent inequality might affect our trust. At the same time, our views on equity and justice alter with experience and education during the life course. In addition, it can be questioned that the extent to which we are exposed to an assumed stratification effect is the same in all stages of our life. The degree of interaction with individuals from other social groups might be different for a student compared to a senior employee or a pensioner. It is therefore necessary to claim that the association between inequality and trust changes as we age and gain knowledge. Second, scholars advocating the perception effect have rarely specified the corresponding reference group. If personal frustration about unjust allocation and distribution of economic resources influences our level of trust (Pickett 2009), then the question arises to whom we actually compare ourselves. A twenty-year old student would perceive the economic success of a forty-year old manager rather as something aspiring than something unjust. One probable assumption is that the perception of inequality within our age-education-specific socio-economic reference groups strongly determines our level of generalized trust.


In a first step, the EU-SILC micro data provides us with the information of the monthly gross household income on 32.377 households in 30 European Countries. Additionally, the age and educational attainment of the person providing the households accommodation is reported. With the use of the Gini formula derived by (Milanovic 1997), age- and age-education specific Gini coefficients are calculated for each country. In stage two, these socio-specific Gini coefficients will be added to the ESS 5 data set. Here, a measure of generalized trust emerges from the survey question phrased “…generally speaking, would you say that most people can be trusted, or that you can’t be too careful in dealing with people?”, with a response scale from 0 (“You can’t be too careful”) to 10 (“Most people can be trusted”).

Lastly, the ESS 5 supplies us with the individual’s age, other personal characteristics (sex, marital and employment status, level of education etc.), and a large set of macro-economic indicators (GDP per Capita, percentage of foreigners in a country etc.) as additional control variables.Two dimensions are taken into consideration for a specification of the calculation of a socio-specific Gini coefficient, age and education. The first measure assembled is an age-specific Gini coefficient. The coefficient is calculated for each age, taking into consideration individuals within an eleven-year age-span (individuals five years younger and five years older than the reference age). In this context several smaller age-spans (3, 5, 7, 9 years) have been tested – with no significant change of results. Results lose significance once a one-year-span is chosen. For the second measure, education in addition to age is taken into account. For the calculation of the socio-specific Gini, only individuals within the same age group and with a similar level of educational attainment (ISCED 0-1, ISCED 2, ISCED 3, and ISCED 4-5) are considered.



We consider the descriptive results of figure 1. The graphical display of 21 European countries is structured as followed from top to bottom. First, the “standard” country level Gini is shown. Secondly, in the middle of the graph, distributions of age-specific Gini coefficients (age span of 11 years) is display. In the lowest part of the graph, the distribution of age- and education-specific Ginis is depicted. It is important to notice that the third part does not just separate the information of the middle section by educational attainment. The measures of inequality, in the lowest part, are calculated from different populations. Here, only individuals within one age and educational group are considered. In general, it can be noticed that countries with a low level of national trust, like Bulgaria, Portugal and Cyprus, have a high level of general and socio-specific inequality. While high trust countries like Sweden, Norway, Denmark or the Netherlands, show low levels of reference-specific inequality on the other side.

As disregard by the “standard” inequality measure in the first row, income inequality varies significantly within countries. While Scandinavians are – as expected – societies with low income imbalances, countries like Greece, Bulgaria and Portugal appear to have relatively skewed income distributions. With the introduction of age- and education-specific Gini coefficients, additional patterns of income inequality emerge. At first sight, sizable difference in the age-specific imbalances the within countries and across education levels are present. As an example, in the Netherlands, a country with egalitarian roots and known for it’s low level of average inequality, age-specific Ginis decrease with level of education. In Portugal, the country with the sample’s highest level of general inequality, age-specific imbalances seem to increase with educational class. Societies like Germany or Poland, middle-ranked in Europe’s score of income inequalities, do not show any distinct differences in age-specific inequality across education. Another interesting case is the example of Lithuania (LT). Looking at figure 1, the Baltic society seems to stick out of the average pattern with a relatively high level of general inequality. Though, the split-up by age and education shows that high levels of country inequality are mainly driven by income imbalances in the lowest educational groups.

The results of this study underline that the perception of income imbalances in Europe play an important role in explaining generalized trust. Economic comparison with other members of society impacts on how much we trust each other. The question to whom we compare our standards of living is relevant and the concept of an adequate socio-economic reference group needs to be considered. Age and education, among many other possible characteristics, are use to build examples of socio-economic groups, within which measure of income inequality are calculated. The introduction of these sub-categories of inequality has uncovered that income imbalances within European societies vary to a sizable extent across age and educational groups.

As one, of many, illustrative examples; in Sweden, a society with low levels of national inequality, the variation of age-specific inequality is strikingly different comparing low and high levels of education. Differences range from low levels of group-specific inequality around 0.2 up to values certainly above 0.35. The fact that socio-economic reference determines our mutual confidence underlines the importance to consider this spread.

In the full version of this paper, we explain the importance of each inequality measure with the use of a multi-level ordered logit model and show that marginal effects vary significantly across different levels of trust.


Brockner, Joel (1996): „Understanding the interaction between procedural and distributive justice: The role of trust.

Milanovic, Branko (1997): „A simple way to calculate the Gini coefficient, and some implications.“ Economics Letters 56, Nr. 1, 45–49.

Pickett, Kate / Wilkinson, Richard (2009): „The Spirit Level: Why more equal societies almost always do better.“ London: Allen Lane .

Tumin, Melvin M. (1953): „Some principles of stratification: A critical analysis.“ American Sociological Review, 387–394.

Uslaner, Eric M. / Brown, Mitchell (2005): „Inequality, trust, and civic engagement.“ American politics research 33, Nr. 6, 868–894.



Fabian Stephany – PhD student – Vienna University of Economics and Business

Leave a Reply