Predictive Modeling: Linking Enrollment and Budgeting.
Dale Trusheim, Carol Rylee · Planning for higher education · 2011
Introduction The hard choices that must be made to balance budgets at higher education institutions can be painful and have dramatic consequences that may linger for years. If enrollment projections and therefore tuition income/budgeting projections for future years are inaccurate, then the result may be unnecessary or insufficient budget reductions, both of which can be problematic and contribute to financial instability. This article addresses the need to link budgeting and planning by describing an innovative solution for developing enrollment projections created through a partnership between the Office of Institutional Research and the Budget Office at the University of Delaware. Planning for an uncertain future in the face of challenging economic conditions continues to be a priority for higher education administration. The purpose of this article is to document an accurate, simple, and effective enrollment forecast model. We also illustrate how the Budget Office uses these data in budget formation. The number of publications on the topic of enrollment projections for individual institutions is rather limited. The authors are only aware of a few chapters or publications in the literature (e.g., Brinkman and McIntyre 1997; Kroc and Hanson 2001). Recently, the Association of Institutional Research published a paper by Chau-Kuang Chen (2008) that utilizes autoregressive integrated moving average (ARIMA) methodology and linear regression to forecast enrollment for Oklahoma State University. Other quantitative prediction techniques that have been the topics of discussion and papers are the cohort survival method, exponential smoothing, Bayesian inference, and trend analysis. This suggests that there is not a widely accepted, best-of-breed methodology for the forecasting of individual school enrollment. Furthermore, many of these methods require the knowledge and use of advanced statistical techniques. Not every institution has administrative staff with the expertise to conduct these kinds of quantitative analyses. The first part of this article presents a very simple method of estimating future enrollment. Of course, simplicity by itself is not valuable unless the prediction has an acceptable degree of accuracy. We have used this enrollment methodology for over 10 years and have been well satisfied with the results. We will present evidence of prediction accuracy at the conclusion of the next section. Enrollment Projection Method Colleges and universities do not operate in a vacuum. Institutional enrollments can clearly be affected by larger economic and market forces. However, the prediction of future enrollment can be thought of as a fairly simple equation. The conceptual framework for our model is that enrollment in any subsequent semester is a function of two variables: new students and continuing students. For example, the prediction of fall semester enrollment is based on how many new students enroll (e.g., readmitted students) and how many of the students who attended in the previous spring semester continue to enroll. The researcher must be able to identify on a term-by-term basis the entry-action status of enrolled students. As an example, there must be a way to determine that, in a given semester, 1,000 students are classified as first-time freshmen, 200 as transfers, 50 as readmits, and 5,000 as continuers. Note that there may well be more codes available in the records system to classify students; for the purpose of the prediction model, however, only these four codes are necessary. Students can be temporarily reassigned to the correct category. Also, it is important to note at the outset that for the purposes of this article, we are interested only in the prediction of total undergraduate enrollment. Technically, the methodology of this prediction model could be used at the graduate level or even down to the specific undergraduate college or department level. …