Geographic reference groups and the determinants of life satisfaction

Christopher Paul Barrington-Leigh
Department of Economics
University of British Columbia

PhD Dissertation
Supervisor: Thomas Lemieux
January 2009


This dissertation combines three contributions to the literature on the determinants of well-being and the social nature of preferences. Departures from self-centred, consumption-oriented decision making are increasingly common in economic theory and are empirically well motivated by a wide range of behavioural data from experiments, surveys, and econometric inference. The first two contributions are focused on the idea that reference levels set by others' consumption may figure prominently in both experienced well-being and in decision making. In the first paper, the well-being question is addressed empirically through the use of self-reported life satisfaction and high-resolution census and survey data in Canada. Strong negative income externalities are found at multiple spatial scales after controlling for various confounding factors.

The second paper explores the general equilibrium consequences of a utility function having an explicit comparison with neighbours' consumption. The question is investigated in a model in which decision makers knowingly choose their neighbours --- and hence their consumption reference level --- as well as their own consumption expenditure, thereby helping to set the reference level for nearby others. For both discrete and continuous distributions of types in an economy with a heterogeneous population undergoing such endogenous formation of consumption reference groups, there exist general equilibria in which differentiation of neighbourhoods occurs endogenously. The novel welfare implications of growth in such economies are described.

The final paper addresses econometric reservations about the use of subjective reports as dependent variables. The date and location of survey interviews are combined with weather and climate records to construct the random component of weather conditions experienced by respondents on the day of their interview. Standard inferences about the determinants of life satisfaction remain robust after taking into account this significant source of affective bias.

The full dissertation is available in PDF form.