Investigating the Use of Holt-Winters Time Series Model for Forecasting Population at the State and Sub-State Levels

Amanda Walters, University of Virginia
Qian Cai, University of Virginia

Much of the research concerning the evaluation of time series models for population forecasting has been focused on using ARIMA models, but few have evaluated the accuracy of the Holt-Winters model for population forecasting. In this study, we evaluate the accuracy of the Holt-Winters model at the state as well as the sub-state level for Virginia. Our research shows that time series forecasts for larger areas are not necessarily more accurate than forecasts for smaller areas. In addition, we found that the overall accuracy of the Holt-Winters model is comparable to trend extrapolation techniques and the cohort component method for short-term projections. Therefore, based on the results of this study, we believe that the Holt-Winters model demonstrates desirable accuracy for state and sub-state projections 5 to 10 years into the future.

  See paper

Presented in Session 106: Sub-National Population Estimates and Projections