High inequality goes hand in hand with low intergenerational earnings mobility across countries. Little is known about why the US is characterized by high inequality and low mobility, while the opposite tends to hold for Scandinavian countries. In an overlapping generations model, calibrated to the US, education policies are endogenized via probabilistic voting. By exploiting cross-country variation in the bias in voter turnout towards the educated and elderly, the model replicates the negative relation between inequality and public education expenditures and accounts for more than a quarter of the variation in inequality and mobility. For the US, I find that compulsory voting could foster mobility, whereas inequality would be hardly affected.