Determinants of Social Spending in Latin America. A Dynamic Panel Data Error-correction Model Analysis

The aim of our research is the study of economic, demographic, and political determinants of social spending in Latin America between 1990 and 2010 We apply system General Method of Moments (GMM) estimation for panel data error-correction models. The empirical results suggest a long-term relationship between social expenditure and its determinants, in all cases indicating diminishing returns; this confirms the convergence hypothesis toward an upper-bounded steady state. Globalisation and growth of interest payments on debt crowd out total social expenditure in the first case and health expenditure in the second, while higher unemployment rates increase total social spending. For demographic factors, Latin American governments seem to prioritise social security for the working-age and elderly populations at the expense of the young population. On the other hand, the movement to less-democratic governments increases total social expenditure, as well as spending on education and health; the latter is also negatively affected by left-wing governments. Finally, countries' specific structural characteristics not included in the model absorb an important proportion of the explanatory power of our model. There are positive effects in the case of health and education, and negative effects for social security expenditure.