Serial Correlation and the Combination of Forecasts

Francis X. Diebold

Journal of Business and Economic Statistics · 1988 · 74 citations · 10 references

Concepts

Abstract

It is shown that regression-based methods of forecast combination lead to serially correlated combined prediction errors. The form of the serial correlation is characterized, and specification, estimation, and prediction are treated. A fully optimal combined predictor, which exploits the serial correlation, is developed and compared with existing regression-based methods in a numerical example, leading to decreases in mean squared prediction error.

References

10