There is increasing interest in the “economics of happiness”, reflected by the number of articles that are appearing in mainstream economics journals that consider subjective well-being (SWB) and its determinants. This paper provides a detailed review of this literature. It focuses on papers that have been published in economics journals since 1990, as well as some key reviews in psychology and important unpublished working papers. The evidence suggests that poor health, separation, unemployment and lack of social contact are all strongly negatively associated with SWB. However, the review highlights a range of problems in drawing firm conclusions about the causes of SWB; these include some contradictory evidence, concerns over the impact on the findings of potentially unobserved variables and the lack of certainty on the direction of causality. We should be able to address some of these problems as more panel data become available.
For the last one hundred years, neoclassical economists have inferred the utility that an individual derives from goods and services from the decisions that she makes – the preferences that she reveals – in her market behaviour. This is based on the premise that individual utility or well-being is the extent to which the individual’s preferences are satisfied. If it is assumed that individuals are rational, fully informed and seek to maximise utility, then the choices they make are those that, by definition, maximise expected utility.
However, economists and psychologists have become increasingly concerned that preferences are often not a very good guide of the well-being associated with the consequences of choices, and are turning to alternative ways of thinking about and measuring utility. Self-reported measures of utility are more familiar within psychology. Subjective well-being (SWB) is often used by psychologists as an umbrella term for how we think and feel about our lives (see Diener, Suh, Lucas, & Smith, 1999). Despite earlier concerns, these appear to be relatively robust indicators of a person’s SWB (Dolan & White, 2007). Rather than the ‘decision utility’ approach of revealed preferences (as reflected in market behaviour) or stated preference studies (e.g. using the contingent valuation method), SWB takes an individual’s well-being to be their overall assessment of their life (Sumner, 1996).
Through the analysis of large datasets, economists and psychologists have gained important insights into the determinants of SWB, such as the effect of income and relative income (Clark & Lelkes, 2005) and the possible effects of the trade-offs between inflation and unemployment (Di Tella, MacCulloch, & Oswald, 2001). Studies on the determinants of well-being adopt the general form:
SWBreport=r(h)SWBreport=r(h)
Turn MathJax on
where the self-reported SWB, often a response to a single life satisfaction or overall happiness question, is some reporting function (r) of true SWB (h), and true SWB is determined by a range of social, economic and environmental factors (X’s). This is usually modelled empirically as an additive function:
SWBit=α+β1X1it+β2X2it+⋯+εitSWBit=α+β1X1it+β2X2it+⋯+εit
Turn MathJax on
where individual differences in reporting are captured within the error term.
How SWB responses are treated differs across studies: some empirical work treats SWB responses as cardinal whilst others respect the strict ordinality of the data and treat true SWB as a latent variable (analysed by ordered logit or probit). However they are estimated, interpreting the coefficients from empirical work relies upon the assumptions within the model; critically, that causality runs from explanatory to dependent variable, and that no unobserved variables are correlated with the included explanatory variables.
This paper reviews the evidence relating to how a range of personal, economic and social factors are associated with SWB. We make no substantive claims about the superiority of SWB over preferences as a representation of individual utility but rather seek to provide fresh insights into the determinants of SWB so that others can more fully consider their relevance to policy, etc. We focus our review on analyses of large datasets and, as such, we do not consider the results from studies conducted by Kahneman, Krueger, Schkade, Schwarz, and Stone (2004), for example, which measure well-being as the aggregation of mood over the course of day. The measures of SWB we review here are ‘experienced’ in the sense that individuals assess how well life is going but these assessments do not have to be duration-weighted aggregations of well-being over time in the way that Kahneman’s conception of ‘experienced utility’ does (Kahneman, Wakker, & Sarin, 1997).
A significant review of the economic literature was conducted by Frey and Stutzer (2002) but, since then, the number of studies exploring SWB has burgeoned. In particular, there are now many more papers using panel data, which allow us to shed more light on the vexing issue of causality than was possible five years ago and control for time-invariant individual effects, such as personality. Our aim is to provide economists, psychologists and other researchers interested in SWB the opportunity to learn more about the state-of the-art research being carried out in the economics literature, including the measures and analytical techniques used as well as emerging results. The degree to which such evidence is robust and provides any suggestions about causality is given particular emphasis.
In Section 2, we present the review strategy and in Section 3, we present the results of the review. In Section 4, we consider some implications of the findings. One firm conclusion that can be drawn is that the existing evidence base is not quite as strong as some people may have suggested and there are some important avenues for future research that could be explored with the existing panel datasets. This, in addition to the lack of clear evidence on causality, makes it difficult to make clear policy recommendations at this stage. Nevertheless, our findings suggest researchers – and perhaps policy makers too – should be aware of the impact of income, relative income, health, personal and community relationships and employment status in their analyses.
