I study an overlapping generations model where physical and human capitals are used in production and can be accumulated by withholding resources from current consumption. Human capital is accumulated through a schooling system which can be finance either by private expenditures or by taxes on current income or by a combination of both. IN a political equilibrium with majority voting, the median voter may approve of public school financing as an instrument to solve a "free rider problem". It improves the skills of next period's workers which in turn increases the expected return on capital, something which could not be achieved by means of private school financing. Public schools, moreover, turns out to be an instrument for intergenerational income distribution so that they may be preferred to private schools for this motive as well. The model is shown to display a poverty trap (poor societies vote to invest too little in education) as well as persistent growth. I am able to fully characterize the global dynamics of the model, which delivers a number of interesting and potentially testable hypotheses on the relation between income growth , capital accumulation and the development of public education I also endogenize the dynamic behavior of school attendance rates, as well as the choice between public and private financing of schools in the presence of parental altruism. A particular attention is paid to the different performances of a publicly provided school system vis-a-vis a publicly financed school system. It is shown that, under very general conditions, the latter tends to create a better environment for the accumulation of human capital as it fosters support for public education trough the voting mechanism. All through the paper I concentrate on specific functional forms that allow for a closed form solution of the equilibrium dynamics, but all the important results can be shown to apply for a general class of utility and produciton functions.