Jeff Pidcock
Director of Budgeting and Business Transformation, Miami University
The views and opinions expressed are those of the author and do not necessarily represent the views or opinions of EAB.
Topic and problem
It would hardly be an understatement to say that undergraduate enrollment is important to Miami University. For the fiscal year 2023, the institution’s unrestricted education and general (E&G) revenue budget was made up of 78% net tuition, 18% from the state share of instruction (SSI) in Ohio, and 4% from other sources. Additionally, Miami faces strong headwinds similar to those faced by many other institutions of higher education.
The COVID-19 pandemic hindered international and domestic recruitment, resulting in a nearly 20% decline in undergraduate net instructional revenue from just five years ago. As we pushed to ‘make’ the class, discount rates pushed up to the point that per student net instructional revenue decreased by 9%. As the budget tightened we needed to learn what student characteristics should we incorporate into the long-range budget plan to improve the forecasting precision of student enrollment.
Current state
Miami’s budget used a relatively simple model (using cohort and residency status) to forecast student enrollment. We would calculate how many students from old cohorts retained to their second term or third term (and so on), and apply that percentage to each term for each enrolled cohort.
This worked well for Miami, but as we approached December 2019 it became apparent that Miami’s students weren’t retaining the way they used to. Feeding enrollments through this model resulted in forecasting errors of $1.9M over actual results for one year and $1M under actual results for another.
Interested in the Rising Higher Education Leaders Fellowship?
Learn more about the program and future cohorts.
Future-state solution
I will be shadowing a retention model in our long-range budget plan that includes student residency status and ethnicity. Several resources converged on student ethnicity as being a key predictor in retention models. Hanover Research noted “race/ethnicity” as one of the most consistent predictors of retention. Other researchers have also found ethnicity to significantly predict student retention (Lopez-Wagner et al.). Including this variable in Miami’s long-range enrollment forecast reduced the forecasting variability to a range of just over $1.2M ($620K in either direction).
I also considered whether to include some form of measure of student quality. Hanover Research’s (2011) report mentions that several high school achievement factors (GPA, SAT/ACT test scores, high school rank) are among the most consistent predictors of retention. Unfortunately, Miami became test-optional in the fall of 2020 and our collection of data such as high school GPA and rank was inconsistent. Including these variables resulted in a forecasting model that performed no better (and arguably worse) than the old model when compared to the 2020 through 2022 fiscal years.
Resources
Hanover Research (May 2011). Predicting College Student Retention.
Lopez-Wagner, M.C., Carollo, T., Shindledecker, E. Predictors of Retention: Identification of Students At-Risk and Implementation of Continued Intervention Strategies.
See the fellows’ blogs from the capstone projects
Jeff Pidcock and others participated in EAB’s Rising Higher Education Leaders Fellowship in fall 2022
