Personality, well-being and the marginal utility of income: What can we learn from random coefficient models?

Abstract

Fixed effects models are the gold standard in empirical well-being research, however, their applicability is limited to controlling for intercept heterogeneity and identifying effects of time-varying variables. This paper investigates the usefulness of random coefficient models in controlling for heterogeneity in well-being and the marginal utility of income, and explores whether these forms of heterogeneity depend on the Big-Five personality traits. Using unique Australian longitudinal data that have personality measures available in two time periods we show that a Mundlak-adjusted random coefficient model yields almost identical results as the fixed effects model, making it a powerful modelling alternative when interest lies in multiple forms of heterogeneity. Big-Five personality explains 10 percent of the variation in intercept heterogeneity and 6-7 percent of the variation in the marginal utility of income. For women, we suggest that the marginal utility of income is significantly linked to personality, implying important gender-differences in the expected effectiveness of financial incentives to influence behaviour.