Zhou XUN co-authored with Michel LUBRANO
In this paper, we study the relationship between individual’s subjective well-being (SWB) and individual income. The Easterlin paradox questions this relationship by noting that at the individual level, income and well-being are positively correlated while at the country level, GNP per capita has followed an upward trend when mean satisfaction remained constant during the same period. This paradox questions the way income can enter a utility function. A common explanation is that individuals are sensitive not to the level of their income, but to their relative income as measured by the ratio between their income and the mean income of their reference group.
After reviewing the different ways of defining a reference group, in particular by basing it on education (a human capital characteristics), we state the several assumptions that are necessary to be able to relate well-being data which are observed on a Cantril scale to a list of three income variables when using panel data : short-term variations, long term value, reference group value. We also state the assumption necessary to identify the two visions of income inequality (inequality as an opportunity or as a risk). The rest of the paper is devoted to an empirical application using the six last waves of the BHPS. When introducing the reference income, we are confronted to an empirical puzzle as for keeping the same utility level, a small increase in the reference income must be compensated by a much larger increase in the individual income. Considering non-linearities does not solve the problem which can also be detected in many other empirical studies. We show that the reference group has to be characterised not only by a central measure (the reference income) but also by its dispersion characteristics. We thus introduce a measure of within-group inequality using a Gini index. Individuals consider within-group inequality as an opportunity, a reward of their effort and talent as it appears with a positive sign. And individuals with a low level of education have a tendency to over-estimate their opportunities. When considering this type of asymmetry, we manage to solve the empirical puzzle as now an increase in the reference income can be compensated by an exact equivalent increase in personal income. We finally consider the possibility of measuring the impact of inequality as a risk, implying risk aversion. For that, we introduce a second reference group, defined this time independently of education, using only location. We manage to identify a slight aversion to inequality, but reduced to lower educated people.