Forecasting performance of grey prediction for education expenditure and school enrollment

GM(1,1) and GM(1,1) rolling models derived from grey system theory were estimated using time-series data from projection studies by National Center for Education Statistics (NCES). An out-of-sample forecasting competition between the two grey prediction models and exponential smoothing used by NCES was conducted for education expenditure and school enrollment under the assumption that grey prediction was as promising as NCES's forecasting technique in dealing with univariate time-series data while some other determinants of the variables under examination were excluded. The purpose of this study, therefore, was to verify that the GM(1,1), and GM(1,1) rolling models would provide forecasts that were at least as accurate as the NCES's approach to extrapolating education expenditure and school enrollment. The findings revealed that the forecasting efficiency of GM(1,1) rolling model was superior to exponential smoothing and GM(1,1) model. The results can offer valuable insights and provide a basis for further research in model building for short-term estimation on educational statistics.