Page 358 - Special Topic Session (STS) - Volume 3

STS550 Angelia L. Grant et al.
approach quickly adapts to changes in model performance. A forecast
combination framework is particularly useful when it is necessary to forecast
over a three-year forecasting period.

5. Concluding Remarks
This paper outlines a methodology for forecasting the components of
household final consumption expenditure. It uses a forecast combination
approach with autoregressive models, regressions on relative prices and the
almost ideal demand system developed by Deaton and Muellbauer (1980).
The forecast combination that weights the forecasts based on forecasting
performance according to rolling squared forecast errors generally performs
better than the currently-used almost ideal demand system. The forecast
combination takes advantage of the forecasting performance across the
different individual models for the different components of consumption
expenditure and across different forecasting horizons. The forecast
combination is particularly useful when it is necessary to forecast over a three-
year forecasting period, given significant differences in forecasting
performance of models across different forecasting horizons.

References
1. A. Deaton and J. Muellbauer. An almost ideal demand system.
American Economic Review,70(3):312-326, 1980.
2. H. Gerard and K. Nimark. Combining multivariate density forecasts using
predictive criteria. RBA Discussion Paper 2008-02, 2008.
3. M. Guidolin and A. Timmermann. Forecasts of US short-term interest
rates: A flexible forecast combination approach. Journal of Econometrics,
150(2):297-311, 2009.
4. D.E. Rapach, J.K. Strauss, and G. Zhou. Out-of-sample equity premium
prediction: Combination forecasts and links to the real economy. Review
of Financial Studies, 23(2):821-862, 2010.
5. A. Timmermann. Forecast combination. Handbook of Economic
Forecasting, edited by: G. Elliott, C.W.J. Granger and A. Timmermann,
North-Holland, 2006.

347 | I S I W S C 2 0 1 9