Theory and practice of multivariate arma forecasting
- 1 July 1984
- journal article
- research article
- Published by Wiley in Journal of Forecasting
- Vol. 3 (3) , 309-317
- https://doi.org/10.1002/for.3980030308
Abstract
We compare univariate and multivariate forecasts based on ARMA models. In theory we cannot do worse by using a multivariate model instead of a univariate one, but we can risk getting no improvement. Conditions for no improvements are discussed as well as cases where large improvements occur. The effect of estimated parameters is examined and found to be small granted that a good method of estimation is used. However, multivariate models could be very sensitive to structural changes. This is illustrated via an example involving monetary data, where the multivariate forecasts perform considerably worse than the univariate ones. This seems to put a limitation on the use of multivariate ARMA forecasting models.Keywords
This publication has 20 references indexed in Scilit:
- Bias of some commonly-used time series estimatesBiometrika, 1983
- EMPIRICAL IDENTIFICATION OF MULTIPLE TIME SERIESJournal of Time Series Analysis, 1982
- The accuracy of extrapolation (time series) methods: Results of a forecasting competitionJournal of Forecasting, 1982
- Testing of Hypotheses for Distributions in Accelerated Life TestsJournal of the American Statistical Association, 1982
- Predictions of multivariate autoregressive-moving average modelsBiometrika, 1981
- A new autoregressive spectrum analysis algorithmIEEE Transactions on Acoustics, Speech, and Signal Processing, 1980
- Asymptotic prediction mean squared error for vector autoregressive modelsBiometrika, 1979
- Covariance Characterization by Partial Autocorrelation MatricesThe Annals of Statistics, 1978
- Estimating the Dimension of a ModelThe Annals of Statistics, 1978
- Investigating Causal Relations by Econometric Models and Cross-spectral MethodsEconometrica, 1969