Abel Brodeur and Sarah Flèche, Review of Income and Wealth, 2019, 65(2): 217-238.
Sarah Flèche is an Assistant Professor at AMSE. She holds a PhD from Paris School of Economics and was a Research Economist at the London School of Economics. Her research interests lie in labour, education and behavioural economics. She was a consultant for the OECD between 2011-2016 and co-wrote a book on The Origins of Happiness published by Princeton University Press in 2018.
Classical economists agree that utility is relative, that individuals compare themselves with others around them. For instance, the impact of comparisons on consumption and saving behaviours were thoroughly analysed in the seminal work of Veblen and Duesenberry. More recently, a growing literature shows that when people assess their own life satisfaction, they tend to compare their situation to that of their neighbours. However, these recent studies yield contradictory findings. Some provide empirical support for a negative relationship between neighbours’ income and individual self-reported well-being, while others find that the richer their neighbours are, the happier individuals are. Such a divide in the literature is puzzling.
In this paper, we argue that the conflicting evidence arises because estimates of neighbours’ effects are sensitive to the neighbourhood definition that is used. As the scale of the neighbourhood changes (from block/zip code, metropolitan area (MSA), county to state-level aggregates), the relationship between respondent and referent neighbours’ income is likely to differ. Actually, individuals are generally unaware of the income of neighbours who live farther away. Moreover, individuals tend to compare with others who are like them, and close neighbours may have more in common than more distant neighbours. While feelings of envy or jealousy towards these close neighbours may arise, their income can also provide clues as to an individual’s own future prospects. Finally, neighbours’ income over a large area may capture confounded effects other than income comparisons, such as local amenities or local labour-market conditions.
In this paper, we use rich U.S. local data and take a multi-scale approach, looking at neighbours’ incomes at the zip code, MSA, county and state levels simultaneously. This allows us to empirically test whether the association between neighbours’ income and well-being varies with the different levels of aggregation.
Our results provide evidence that conditional on own income, the association between neighbours’ income and life satisfaction is positive at the zip code level. Zip codes correspond roughly to 24,000 inhabitants and proxy for neighbours within the same local community. By contrast, we find a negative and significant relationship between neighbours’ income and life satisfaction at more aggregated levels (i.e., the MSA, county and state levels), consistent with previous studies. We also distinguish between individuals below and above neighbours’ median income. We provide evidence that the positive association between neighbours’ income and well-being at the zip code level is greater for poorer individuals.
To better understand what drives these associations, we then test explicitly whether the inclusion of local variables such as neighbours’ socio-economic characteristics, number of schools and health establishments, economic environment or criminal records in the neighbourhood affects our coefficients of interest. Including these variables in the model makes the relationship between zip code neighbours’ income and life satisfaction statistically insignificant. This suggests that at the most disaggregated level (zip code), social and economic features of the environment fully explain the positive association between neighbours’ income and life satisfaction. At more aggregated levels, other mechanisms are likely to play a role.
This analysis raises interesting perspectives. However, further research is needed in at least two directions. From our results, we cannot conclude on whether feelings of envy or jealousy are likely to play a role in the relative income effects. Also, we cannot tell whether the degrees of connectivity and similarity between respondent and referent neighbours are key factors in the observed relationships. To answer these questions, we plan to use data from the Somerville Happiness Survey, which will allow us to observe neighbours’ income effects within one city at the street level. In addition, further research is needed to identify random income shocks at the neighbourhood level. This would enable us to evaluate the causal effects of neighbours’ income on individual well-being.