|
| << previous | next >> |
Journal description
All volumes of the corresponding journal
All issues of the corresponding volume
All articles of the corresponding issues
Details for article 24 of 26 found articles
| Title: |
The efficiency of dynamic linear model estimators applied to a linearly aggregated time series |
| Author: |
Bolstad, William M. |
| Appeared in: |
Communications in statistics |
| Paging: | Volume 19 (1990) nr. 1 pages 83-90 |
| Year: |
1990 |
| Contents: |
Many time series are weighted sums of component time series. If the same dynamic linear model generates all of the component time series, it also generates the linearly aggregated time series. This paper establishes the result that if the prior distribution of the dynamic parameter estimators is the same for the dynamic linear model applied to the linearly aggregated time series as for the weighted sum of the component dynamic linear model estimators, the posterior covariance matrix for the linearly aggregated model estimators equals the posterior covariance matrix for the weighted sum of the component model estimators plus the weighted sum of the covariance matrices of the posterior mean vectors averaged over all possible component observations such that the weighted sum is the linearly aggregated observation. This paper determines the relative efficiency of the two estimators and shows that the generalized variance of the posterior estimators for the linearly aggregated model is always greater than or equal to the generalized variance of the weighted sum of component posterior estimators. |
| Publisher: |
Taylor & Francis |
| Source file: |
Elektronische Wetenschappelijke Tijdschriften |
Details for article 24 of 26 found articles
| << previous | next >> |