Inequality Aversion and Self-Interest: An Experimental Approach

Preferences relating to inequality aversion, the trade-off between equality and efficiency, and self-interest, the degree to which the ‘self’ is weighted in relation to ‘others’, are incorporated within structural models to explain prosocial behaviour, the act of giving to others. To observe such behaviour, incentivised laboratory and lab-in-the-field experiments are run. Structural preferences parameters are then estimated, at the individual, cluster and sample level, within the utility functions proposed. Noise in decision making is formally modelled with the Beta and Dirichlet distributions, which are formulated as random behavioural models. In the first chapter, distributional decision problems amongst groups of three are presented to participants within a laboratory experiment. Using multiple experimental designs and alternative perspectives, within-subject treatment effects are tested. The second chapter incorporates oneness, the closeness of connection to others, within a structural model to better explain the differential effects that social distance can have on distributional decision making. In a lab-in-the-field experiment in Mbale, Uganda, modified three-person dictator games are presented to participants to enable the observation of such behaviour, alongside an extensive survey. Finally, the third chapter focuses on N-person giving. Five alternative utility functions are formulated, which incorporate differing behavioural preference parameters; accounting for the distinction of self-other and between-other inequality aversion, congestion and minimum threshold levels. Both the goodness-of-fit and predictive accuracy of each model are compared, to identify the ‘best’ model for each individual. Within each of the three chapters, results show extensive heterogeneity in prosocial behaviour, which is accounted for through the estimated preference parameters. On average, participants have a substantial regard for others and a preference for reducing inequality, rather than increasing efficiency. The experimental design, perspective, oneness levels and number of recipients are shown to have significant, but differential, effects on prosocial behaviour.