Facilitating Personal Recovery in the Community: a Qualitative Evaluation of a Peer Support Group for Adults with Severe Mental Health Challenges in Rural Australia.

Estimating the effects of education on socioeconomic outcomes has been an important part of applied microeconometrics over the past three decades. While most of the literature has focused on labor market outcomes, health effects have also been investigated. In Table in the Appendix we list 24 studies that estimate health effects of education and use instrumental variable estimation methods for identification. More than half of these studies find no statistically significant effects overall or in relevant subgroups. All of these studies have in common that they aggregate effects across age groups, often over several decades. However, this may miss relevant patterns. Kaestner et al. [] extend the classic [] model of demand for health and conclude that “it is unlikely that the relationship between education and health is constant over the life cycle and that education is likely to have little effect on health at younger ages when there is little depreciation of the stock of health” []. Thus, an estimated small and insignificant effect averaged over younger and older individuals does not necessarily imply that health is not causally affected by education. It may well be that the effect occurs late in life and that the aggregate average is obscured by a zero effect for younger individuals. It is well known that the socioeconomic status-health gradient increases over the life cycle [, ]. This descriptive pattern has also been shown more specifically for the education-health gradient. For example, Kaestner et al. [] find no differences in mortality by education until the age of 60, after which hazard rates diverge by education. In contrast, they find an education-morbidity gradient only for the 45-60 age group, but explain this with possible selective mortality. Bijwaard et al. [] find an increasing difference in mortality between those with primary education and those with more than primary education, mostly after the age of 60. They find that the differences are mainly due to selection effects (based on cognitive ability) at early ages, while the role of education increases after age 60. Leopold and Leopold [] find differences in self-rated health between higher and lower educated individuals from ages 30 to 80, which increase from age 50 (for men). Ross and Mirowsky [] find a gap in physical impairment between the well-educated and the poorly educated over the life cycle, which is more pronounced for women. These studies provide a descriptive picture of the education-health gradient over the life cycle, but do not claim causality. We are aware of only two studies that explicitly look at causal health effects of education by age-group, thus allowing to take on a life-cycle perspective of health effects. Clark and Royer [] find that the increase in minimum school leaving age from 14 to 15 in 1947 and from 15 to 16 in 1972 in Britain did not affect mortality overall, but also not within separate 5-year age groups between 20-24 and 65-69. Gehrsitz and Williams Jr [] study the effects of the same 1972 reform but restrict the analysis to Scotland and report results by age for 30-55 year olds. They find no effect on self-reported health, but a reduction in hospital admissions for selected conditions. This is mainly true for men and starts after the age of 40.