On the empirical economics of income persistence and class identity in the UK

Abstract

Socioeconomic gradients in life outcomes are important political issues and a key area of research for many applied economists. This thesis – comprising three empirical chapters – represents the culmination of a process of research and investigation into the measurement of income persistence and the role of class identity in shaping behaviour and health in the UK. The first empirical chapter explores measures of intergenerational income persistence and their sensitivity to researcher decisions. Using data from the British Cohort Study, 810 estimates of intergenerational persistence are produced. The chapter is the first to estimate and report the entire profile of persistence estimates with precedent in the literature for a given dataset. Through a novel application of the techniques from meta-regression, the effects of each researcher decision are estimated. The results confirm the existence of errors-in-variables and lifecycle bias and show that seemingly innocuous changes to variable definitions can lead to large fluctuations in persistence estimates. Rank coefficients are found to be the least sensitive to researcher decisions and are thus preferred over elasticities or correlations for comparative purposes. The thesis progresses from measuring the intergenerational transmission of advantage, to the process of how socioeconomic gradients occur. Specifically, from a health economics perspective, how can we better understand differences in behaviours between socioeconomic groups? The thesis applies a range of econometric methods to provide the first evidence on the economic role of subjective class identity on smoking behaviour and the relationship between relative deprivation and health. The second empirical chapter focuses on subjective class identity and risky behaviour, as measured by smoking. Data from the British Household Panel Survey (BHPS) suggests around 1 in 3 people in the UK state that they belong to a social class. A range of Two Part Models is used to demonstrate behavioural differences between individuals with a class identity and those without class identities. Working class identity individuals are more likely to smoke and, those that do, consume more cigarettes than individuals with no class identity. These are the first empirical findings to show that associations exist between subjective class identity and health behaviours even after controlling for a host of socioeconomic characteristics. These findings are consistent with predictions from the identity economics framework, given the literature on class, smoking and stigma. iii The third empirical chapter again uses data from the BHPS, to examine whether class identity has differential effects on the relationship between relative deprivation and measures of health (self-assessed health and psychological well-being). Heterogeneity analyses allow the associations to differ across subjective class identity groups. In measuring relative deprivation, the reference groups - with which individuals compare themselves – are allowed to vary with class identity. For each sample, outcome, and reference group, OLS is used alongside models designed to control for unobservable heterogeneity. Once time constant unobserved differences are controlled for there is no evidence of self-assessed health or psychological health costs from relative deprivation for males. For females, there is a significant positive relationship between relative deprivation and psychological well-being. This effect is driven by working class women and hypothesised to be an aspiration effect, i.e. as incomes in the reference group increase, the subject anticipates that their income too will increase, improving well-being. Each empirical chapter focuses on an aspect of the central theme, socioeconomic gradients in life outcomes. The key contributions are on the measurement of income persistence and the role of subjective class identity on health behaviours and outcomes. The central implications are the importance of transparency in the reporting of income persistence estimates and the need for accurate comparisons. As well as the need to consider self-assessed class identity in studies of risky behaviour and health inequalities, as these may provide more insight into the effects of health interventions and the existence of hard-to-reach individuals. Throughout, results also differ across gender, highlighting the need to consider gender when targeting policy messages.