In many countries social security is a large fraction of the government budget. Why should this be so, given that at any moment in time the number of recipients of social security benefits is smaller than the number of contributors? More generally, what determines the size of social security? To answer these questions, this paper studies an overlapping generations model in which all individuals currently alive vote on social security. There is no commitment to preserve the legislation inherited from the past. Voters are weakly altruistic, and there is heterogeneity within each generation. The paper shows that in equilibrium the size of social security is larger the greater is the proportion of elderly people in the population, and the greater is the inequality of pre-tax income. Both predictions of the theory are supported by the empirical evidence in cross-country data.
