Social Capital and Health - Exploring the Causality Issue at the Individual Level

Presenter: Nicolas Sirven, IRDES (Institute for Research and Development in Health Economics)

Abstract

The influence of social capital on health has been studied for many years now. At the individual level, it is still not completely established whether good health is the result of social capital or whether social capital is the result of good health and/or other unmeasured personal characteristics that determine both health status and patterns of social engagement. The causal influence of social capital on health has not yet been addressed in a coherent micro framework because individual instruments are difficult to find. Although general attitudes towards “religion” are almost never suggested as a valid instrument at the micro scale (because it is a good predictor of health), the distinction between religious beliefs and religious rituals (e.g. attending churches, meeting people) or other forms of social participation, helps supporting the assumption that social participation is a potential mediator through which religious beliefs may benefit health. “Religious beliefs” could thus be a good instrument of “membership in associations” (or social capital). Using individual cross-section and panel data of Europeans aged 50 years old and over, in eleven countries from the two waves of the Survey on Health, Ageing, and Retirement in Europe (SHARE) in 2004 and 2006, a binary index of social participation is derived from self-reported questions on membership in associations, and help provided to family, friends, and neighbours. The influence of this variable on health is estimated with and without use of “religious beliefs” as the sole instrument. The stability of the causal relationship between social capital and health is investigated using five different binary health outcomes (self-reported health, depression, cognitive impairments, CVD, and difficulties in ADL), and through a time dimension (to account for a potential lagged effect of social capital). In addition to the parametric approach, the influence of social capital on health is estimated using the method of Propensity score matching (PSM). This semi-parametric approach is a good alternative to test the validity of the social capital coefficient estimated in the IV model since the usual tests (designed for 2SLS estimates) are not relevant for ML estimates.