هزینه های تامین اجتماعی و رشد اقتصادی: یک ارزیابی تجربی

In this paper, we analyse the empirical relationship between social security expenditure and economic growth, using cross-country data for a sample of 61 countries and panel data for a sample of 20 industrialized countries. We find that, whenever a statistically significant association between social security expenditure and growth exists, it has a positive sign. The positive estimated coefficient of social security expenditure seems robust to various forms of misspecification and appears to be larger in poor countries with relatively underdeveloped social security systems. As for the channels through which the positive effect of social security expenditure on growth should take place, our results seem to indicate that social security has a positive influence on human capital formation.

One widely accepted prediction of most economic growth models is that purely redistributive policies have a depressive effect on physical capital accumulation and growth, since they imply a reduction We are indebted to Roberto Perotti, Jos´e Victor Rios-Rull, two anonymous referees and seminar participants at University of Bologna for useful comments. We would also like to thank Francisco Rodriguez for kindly providing data on social security for OECD countries. All errors are ours alone.

This paper contributes to the empirical literature on the association between social security expenditure and growth in two main ways. First, our work pays a great deal of attention to assessing the sensitivity of empirical results to various estimation problems that are usually present in growth regressions. After correcting for these problems, we find evidence of a positive relationship between social security and growth in a cross-country analysis for a large sample of 61 countries and in a panel data analysis for 20 OECD countries. Second, we perform an original investigation of the channels through which the positive effect of social security expenditure on growth may take place. We find no evidence of a statistically significant association between social security expenditure and investment in physical capital. This may be due to the fact that crowding-out and crowding-in effects of social security on saving cancel out. Instead, our results seem to support the view that social security expenditure stimulates investment in human capital and therefore growth. Explanations based on the idea that generous social security enhance efficiency by improving political stability and social cohesion are not equally supported by our estimation results.Other arguments according towhich social security induces early retirement of old unproductive workers and fosters growth are harder to test with available data. Yet, a tentative test of this theory does not provide encouraging evidence in its favour. In our opinion, an issue that deserves further work is the joint treatment of the probable endogeneity of some explanatory variables, including social security, and of unobservable countryspecific effects. In a recent paper, Caselli et al. (1996) deal with this problem by exploiting dynamic panel data methods. However, the application of these methods to our context is severely limited by the SOCIAL SECURITY EXPENDITURE AND ECONOMIC GROWTH 267 availability and reliability of social security expenditure data for long term spans in developing countries. Alternatively, potentially fruitful insights into the relationship between social security and growth may come from the specification of a complete model, where social security expenditure explicitly appears as an endogenous variable, along with growth. However, few valuable indications on how such system should be specified can be found in recent theoretical work on the issue. As shown by Perotti (1996), the ‘‘political economy approach’’, which suggests that income inequality should be used as an instrument for social security expenditure, is not empirically successful. In Bellettini and Berti Ceroni (1999), a model is developed where the size and composition of government expenditure are jointly determined together with growth as the outcome of an intergenerational game. Unfortunately, this framework does not indicate a set of instruments sufficiently large to identify a system where fiscal policy variables and growth can be jointly estimated.