Andrew Clark

Subjective wellbeing data are increasingly used across the social sciences. Yet, despite the widespread use of such data, the predictive power of approaches commonly used to model wellbeing is only limited. In response, we here use tree-based Machine Learning (ML) algorithms to provide a better understanding of respondents’ self-reported wellbeing. We analyse representative samples of more than one million respondents from Germany, the UK, and the United States, using data from 2010 to 2018. We make three contributions. First, we show that ML algorithms can indeed yield better predictive performance than standard approaches, and establish an upper bound on the predictability of wellbeing scores with survey data. Second, we use ML to identify the key drivers of evaluative wellbeing. We show that the variables emphasised in the earlier intuition- and theory-based literature also appear in ML analyses. Third, we illustrate how ML can be used to make a judgement about functional forms, including the existence of satiation points in the effects of income and the U-shaped relationship between age and wellbeing.

We use quarterly panel data from the COME-HERE survey covering five European countries to analyse three facets of the experience of loneliness during the COVID-19 pandemic. First, in terms of prevalence, loneliness peaked in April 2020, followed by a U-shape pattern in the rest of 2020, and then remained relatively stable throughout 2021 and 2022. We then establish the individual determinants of loneliness and compare them to those found in the literature predating the COVID-19 pandemic. As in previous work, women are lonelier, and partnership, education, income, and employment protect against loneliness. However, the pandemic substantially shifted the age profile: it is now the youngest who are the loneliest. We last show that pandemic policies affected loneliness, which rose with containment policies but fell with government economic support. Conversely, the intensity of the pandemic itself, via the number of recent COVID-19 deaths, had only a minor impact. The experience of the pandemic has thus shown that public policy can influence societal loneliness trends.

The distribution of job quality across workers and the change in job quality over time can be reflected in various measures of job outcomes, or single-item job-satisfaction scores. This paper takes both approaches to establish the evolution of job quality over a period from the mid-1990s to the mid-2010s in 13 OECD countries, using data from the three latest ISSP Work Orientation modules. The rise in job satisfaction from 1997 to 2005 has continued through 2015, despite the 2008 Great Recession. This improvement is also found in most of the joboutcome domains, despite some evidence of harder and more stressful work. Workers reported that job security was the most-important job aspect every year, and the percentage of workers with secure jobs rose over time. There has been a small rise in the dispersion of job satisfaction, but the good news regarding better job quality over a 20-year period does not seem to be dampened by large changes in its inequality.

It is well-known that unemployment leaves scars after re-employment, but does this scarring effect persist even after retirement? We analyse European data on retirees from the SHARE panel, and show that the well-being of the retired continues to reflect the unemployment that they experienced over their working life. These scarring effects are somewhat smaller for older retirees, but larger for those who arguably had higher expectations regarding the labour market when they were active. The lower well-being from lifetime unemployment does not reflect lower retirement income. This long-run scarring for those who have left the labour market underlines that contemporaneous correlations significantly under-estimate the well-being cost of unemployment.

We here evaluate the link between job insecurity and one of the most-important decisions that individuals take: homeownership. The 1999 rise in the French Delalande tax on firms that laid off older workers produced an unexpected exogenous rise in job insecurity for younger workers. A difference-in-differences analysis of panel data from the European Community Household Panel shows that this greater job insecurity significantly reduced the probability of becoming a homeowner. This drop seems more attributable to individual preferences rather than greater capital constraints, consistent with individuals reducing their exposure to long-term financial commitments in more-uncertain environments.

We use Australian panel data to examine the impact of retirement on individual locus of control, a socio-emotional skill with substantial explanatory power for a broad range of life outcomes. Exploiting the eligibility age for the Australian Age Pension, we find that retirement leads to increased internal locus of control. This greater internal control explains around one-third and one-fifth of the positive effects of retirement on health and subjective well-being, respectively. We also show that locus of control is much more malleable at retirement than the other socio-emotional skills of the Big-Five personality traits, risk and time preferences, and trust.

There is much still to learn about the relationship between income and well-being, and in particular how this may depend on the economic and social context. We use Russian data to estimate individual Welfare Functions of Income, and examine two potentially context-dependent concepts: self-assessed income needs and welfare sensitivity to income (how well-being changes with income). The considerable geographical diversity in Russia provides within-country variation in GDP, inequality, population density, and unemployment. We first show that income needs exceed actual income on average in Russia, and that these needs are less sensitive to changes in income than in other countries. Second, income needs vary by individual characteristics, while welfare sensitivity does not. Welfare sensitivity is however related to the regional context. Last, our estimated contextual results help us to understand why the existing literature has produced such a wide range of results.

Information on both wages and job quality is needed in order to understand the occupational dispersion of wellbeing. We analyse subjective wellbeing in a large UK sample to construct a measure of ‘overall reward’, the sum of wages and the value of job quality, in 90 different occupations. If only wages are included, then labour market inequality is underestimated: the dispersion of overall rewards is one‐third larger than the dispersion of wages. Our findings are similar, and stronger, in data on US workers. We find a positive correlation between job quality and wages in all specifications, both between individuals in the cross‐section and within individuals in panel data. The gender and ethnic gaps in the labour market are larger than those in wages alone, and the overall rewards to education on the labour market are underestimated by earnings differentials alone.

It is well-known that the wealthier are more likely to have Right-leaning political preferences. We here in addition consider the role of the individual’s starting position, and in particular their upward social mobility relative to their parents. In 18 waves of UK panel data, both own and parental social status are independently positively associated with Rightleaning voting and political preferences: given their own social status, the upwardly-mobile are therefore more Left-wing.We investigate a number of potential mediators: these results do not reflect the relationship between well-being and own and parents’ social status, but are partly linked to the individual’s beliefs about how fair society is. We replicate these findings using US data and show that, in both countries, the choice of specification when controlling for the respondent’s own status is crucial and may help explain some of the mixed findings in the literature.