Model Building and Forecasting with Multicollinear Time Series

Abstract

An explanatory regression model from time series data allows us to identify performance drivers and forecast performance given specific driver values, just as regression models from cross sectional data do. When decision makers want to forecast future performance in the shorter term, a time series of past performance is used to identify drivers and fit a model. A time series model can be used to identify drivers whose variation over time is associated with later variation in performance over time.