Börsch-Supan and Schuth 2013 — Early Retirement, Mental Health and Social Networks

retirementagingcognitive-functionsocial-networkseuropesharemental-healthearly-retirementwell-beingols

Summary

This chapter uses Survey of Health, Ageing and Retirement in Europe (SHARE) Waves 1–4 (25,000\approx 25{,}000 Europeans aged 50+) to examine how retirement affects mental health, cognitive functioning, and social networks. Ordinary least squares (OLS) regressions with country and time fixed effects show that early retirees experience gradual cognitive decline (0.019-0.019 words/year on delayed recall) and social network contraction (0.014-0.014 persons/year); controlling for network size attenuates the retirement-cognition coefficient, indicating partial mediation. Disability retirees follow a distinct trajectory: a large immediate CASP-12 well-being loss (0.510-0.510) driven by status loss rather than elapsed time, and increased reliance on formal helpers.

Key Claims

Concepts Introduced or Extended

Entities Mentioned

Quotes

My Take

This is a descriptive OLS study with no causal identification. Country and time fixed effects control for aggregate confounders but cannot address individual selection: workers who retire early differ systematically from those who remain employed in ways that also predict cognitive decline and social withdrawal. The mediation finding is susceptible to reverse causation (declining cognition causing social withdrawal) and omitted variable bias (poor health causing both retirement and network shrinkage). The disability retirement finding — a sharp immediate status-based CASP-12 penalty that does not decay with duration — is intuitively plausible but equally observational. The paper's main contribution is a descriptive quantification of diverging trajectories for early vs. disability retirees across multiple SHARE waves, providing a useful baseline. It contrasts sharply with Coe and Zamarro (2011), who find null causal effects on cognition using IV; the discrepancy likely reflects unaddressed negative selection in OLS.