Work, health, and health inequality

This project aims to consider how extended working lives affects health and health inequalities. The correlation between work and health is strongly positive in observational data - a recent review for DWP can be found at https://www.gov.uk/government/publications/is-work-good-for-your-health-and-well-being

However, remarkably there is almost no evidence that provides an understanding of the extent to which this relationship shows that work causes health. Indeed, while this official review noted the issue, it nonetheless came to the strong conclusion that work was good for health. What evidence as does exist mostly relies on fixed effect estimation to track the effect of unemployment on health – where the findings largely point to strong adverse mental health effects.

Recent changes in legislation and economic circumstances have dramatically affected the incentives to retire in the UK. These events now allow us to better understand how changes in work for the elderly that are mediated through the structure of incentives alone affect their health. In particular, there have been recent changes in the age at which individuals become eligible for their state pension (upon which the majority of the population depend almost exclusively for their post-retirement income). While these changes have been flagged well in advance (which might allow a degree of adaptation) we would, nonetheless, expect that they would impact on working behaviour past the old pension age and even beyond the new pension age. Moreover, the financial crisis in 2008+ have had important impacts on pension wealth for those with private DB pension provision that will not affect those with DC provision.
Note that the pension age changes affect everyone but (because pension income is taxable) they affect high-income people less than low income, while the financial crisis affects those with wealth and not those without. Because these changes affect different groups of people differently we would expect them to change health inequalities - but it is unclear quite how and how much, without doing the research.

These events give rise to a number of identification strategies beyond simple matching methods that attempts to compare like with like on the basis of observables alone. Nonetheless, there are a number of threats to the analysis that will take good data and careful statistical analysis to steer around. For example, these changes affect incomes as well as working behaviour. Moreover, there are other reforms/events taking place over time that affect work (disability testing in the welfare system), or income (changes in the real value of state pensions are imminent that will mostly benefit those with private/occupational pension income), or health itself. Good data and clear statistical methodology will be required. The supervisors have extensive experience on estimating causal effects in other contexts and the work will rely mostly on the English ELSA cohort study (http://www.elsa-project.ac.uk/)