This paper studies the importance of dynamic earnings modeling for the design of income contingent student loans (ICLs). ICLs have been shown to be theoretically optimal in terms of efficiency in the presence of risk aversion, adverse selection and moral hazard, and have attractive equity properties. Recognition of their benefits has led to their adoption for tertiary education tuition fees in countries including Australia, New Zealand, and the UK. Since the design of ICLs relies on the prediction of the underlying costs, we explore the extent to which the complexity of earnings modeling affects the estimation of loan subsidies. The use of Australian data allows us to compare our simulated debt repayments to actual repayments under the Australian Higher Education Contribution Scheme (HECS). Our findings reveal that the complexity of earnings modeling has considerable implications for the calculation of loan subsidies.
educational finance dynamic stochastic modeling panel data income contingent loans