Subjective wellbeing, mean reversion and risk

Abstract

This thesis examines the possibility of directly measuring utility via reported subjective wellbeing and considers the structure of utility functions such measurements imply. Chapter one surveys the practical and theoretical arguments in favour of such an approach, and outlines the key technical difficulties associated with obtaining usable utility data from self-reported subjective wellbeing. Chapter two links subjective wellbeing to the model presented by Robson (2001) whereby restrictions on utility are predicted by evolutionary models. A specific model of mean reversion is then suggested, demonstrating the relationship between mean reversion and time preference, and data from the British Household Panel Survey (BHPS) is used to estimate the speed at which habituation occurs. Links between evolutionary models and the problem of mapping from reported subjective wellbeing to underlying utility are then used to generate a method for recovering cardinal utility data from ordinal subjective wellbeing using observed response frequency. This approach is then demonstrated using the BHPS data. Chapter three considers the aggregation of subjective wellbeing through time and introduces the concept of `peak aversion', a preference for smoothness in utility across states and through time, and relates it to traditional risk aversion measures and general state-dependent utility. The application of peak aversion to subjective wellbeing, Quality Adjusted Life Years (`QALYs'), tortious compensation and social choice theory are then considered. An empirical estimate of the strength of the preference is obtained using data on differences in Standard Gamble and Time Tradeoff QALYs and is used to calculate the curvature of the inequality averse social welfare function and appropriate QALY weightings based on severity of illness.