Abstract

In this paper, we show that the positive estimated coefficient of average social security expenditure, often detected in cross-country growt regression, can not be imputed to reverse causation, that is on economic growth pulling social security expenditure, nor to omitted variables or other misspecification problems. Morover, we show that the positive effect of social security expenditure on growth is much stronger in poor countries than in rich countries. As for the channel through which the positive effect of social security expenditure on growth takes place, our results point out that the social security influences human capital formation. On the other hand, we do not find support for theories claiming that generous social security benefits should enhance investment productivity and growth by inducing retirement of unproductive workers or by improving political stability and social cohesion.