In this paper we conduct a quantitative analysis of a number of stylized educational loan systems. We develop a stochastic general equilibrium model of a closed economy with a competitive firm sector and a government that levies taxes and administers educational loans. Individuals are heterogeneous in their talent for education and ability to learn on the job and face uninsurable idiosyncratic labour productivity risk during their working career. We calibrate the model to the US mortgage loan system and subsequently consider two possible reforms. The first is a Graduate Labour Tax (GLT) system whereby grants to students are financed by means of a tax on the labour income of educated individuals. We find that in the long run the proportion of uneducated workers stays roughly constant but the average educational attainment of students increases.As there exists a considerable amount of transitional dynamics in the model the welfare effects of the reform differ by generation. Cohorts alive at the time of the shock are worse off while ex-ante welfare of future cohorts increases. The gains to the latter are large enough to – at least in principle – compensate the losers from thepolicy reform and generate an overall welfare gain. The second possible reform we study is a Comprehensive Labour Tax (CLT). It is very similar to the GLT except for the fact that the educational tax is levied on all workers, including those who are uneducated. In contrast to the GLT reformthe proportion of uneducated workers dropssubstantially. Generations that become economically active soon after the policy reform are worse off and the aggregate ex-ante welfare effect is negative.