login

Collinearity and forecasting

Journal of ForecastingPublished 1 April 1984
David A. Belsley
Citations47
SJR quartileQ1
SJR score0.70
SNIP1.09

Abstract

Abstract Old saw: Collinearity will not harm forecasts as long as it continues into the forecast period. If, however, collinearity is unlikely to continue but has already harmed estimation, then corrective action (introduction of prior information) to improve estimates should improve forecasts. This paper marshals the diagnostic information needed to assess collinearity's continuance and, when required, to direct meaningful corrective action. The process is illustrated by an example involving forecasts using energy prices.

Keywords

MathematicsDecision Sciences