Reference Points in the Wild

Abstract

The aim of this thesis is to investigate the effect of reference points on households’ decisions and outcomes using field data. The thesis looks at several different applications on multiple UK household datasets, all of which relate to the role of reference points in life cycle decisions about consumption, savings and income. The first chapter studies the influence of reference points and loss aversion on consumption-savings behaviour. It has long been theorised that past income acts as a reference point, affecting individuals’ consumption-smoothing. A simple two-period model shows that individuals below their reference point will save less, driven by a lower marginal propensity to save owing to loss aversion. The chapter proposes to measure reference points through subjective reports of financial changes in the British Household Panel Survey (BHPS), allowing individuals’ own subjective perceptions to determine their reference points. Results show that ceteris paribus individuals below their reference point save less on average, which is driven by a marginal propensity to save about a third lower than those at or above their reference points. Further analysis investigates the effects of the SFS variables in more detail, including decomposing them using factor analysis and testing the influence of partner’s SFS in couple households. The second chapter tests the same class of models from another empirical perspective. Life cycle models featuring reference-dependent preferences and loss aversion imply that individuals who are below their reference point have lower overall utility but a higher marginal utility of consumption, which has important consequences for welfare and policy analysis. This chapter is the first to use ‘experienced utility’ in the form of subjective well-being (SWB) to corroborate this hypothesis. Two potential determinants of reference points are proposed: last year’s consumption (habit formation) and average regional consumption (relative income). Parameter estimates using the combined BHPS & Understanding Society dataset are consistent with the null hypothesis of loss aversion and indicate that it is characterised by a smooth utility function rather than a discontinuous one. The analysis favours a reference point based on habit formation as opposed to relative income, though the estimated parameters suggest model mispecification. Additional analysis decomposes the SWB measure into positive and negative affect to test the hypothesis that loss aversion is largely driven by negative affect as in psychological theories where losses are qualitatively distinct from gains, but this hypothesis is not supported. The third chapter investigates intergenerational mobility in contemporary England and builds a model where parents’ earnings act as an aspiration, modelled as a reference point, to explain the empirical results. Estimates of intergenerational mobility – Intergenerational Earnings Elasticities (IEEs) – have been shown to vary systematically across the joint earnings distribution of children and parents, but there is no agreed upon explanation for the observed pattern. A contemporary cohort dataset, the Longitudinal Study of Young People in England (LSYPE) is used to provide the most up to date estimates of IEEs in England, which are low on average though likely downward biased. Quantile regression is used to estimate IEEs across the distribution and results show they are lower for richer children but higher for richer parents. In order to explain this pattern a human capital model is constructed where parents’ earnings act as an aspirational reference point for children. Children whose earnings are below their aspiration will have a higher marginal utility from investing in human capital and will compensate by increasing their future earnings faster than children who have reached their aspirational level.

Date of Award	31 Dec 2019
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Horst Zank (Supervisor) & James Banks (Supervisor)