The Causal Effect of Social Activities on Cognition: Evidence from 20 European Countries

Abstract

Using harmonized data from 20 European countries, we examine the causal effect of being socially active on old age cognition. To address the endogeneity of social participation, we employ nonparametric partial identification methods that bound the average treatment effect for the population under fairly plausible, and thus credible, assumptions. We find strong evidence that social activities have a positive impact on all cognitive dimensions we analyze. At their upper bound these effects are quite large, while at their lower bound they are more modest but still salient. Additionally, we show that ignoring the endogeneity of social activities severely underestimates the uncertainty about their causal effect on cognition.

Keywords: Cognition, Social Activities, Aging, Partial Identification, SHARE

JEL Classification: I10, J14, C14