Proceed with caution: Social capital measurement

Abstract

Researchers often use the number of memberships to voluntary associations to proxy social capital. We investigate the proxy’s suitability in the context of income in the United States. While the aggregated social capital variable correlates with higher incomes, a disaggregated measurement shows inconclusive evidence. We find the commonplace inclusion of union membership in “social capital” overestimates the income effect, the average treatment effect of an additional membership varies across occupational groups, and not all voluntary associations are correlated to higher income levels. Finally, we find evidence the aggregate measurement is not endogenous to income, though we caution against causal claims