Forecasting pell program applications using structural aggregate models

Demand for postsecondary financial aid offered under the Pell Grant program has proved notoriously difficult to predict in recent years. The U.S. Department of Education maintains a microsimulation model to generate forecasts of applications, but the data requirements of this model are fairly severe and the results subject to considerable error, at least in some years. This paper proposes an alternative model for predicting Pell program applications that uses aggregate data, yet avoids some of the limitations of simple linear models. Using the moments of the aggregate income distribution and other macrovariables, this model empirically performs no worse on average than the microsimulation model, and better in some periods. [ JEL I21, I28, C53]