Robust joint modelling of labour force series of small areas - ARCHIVED

Articles and reports: 12-001-X199300214458

Description:

In this article we report the results of fitting a state-space model to Canadian unemployment rates. The model assumes an additive decomposition of the population values into a trend, seasonal and irregular component and separate autoregressive relationships for the six survey error series corresponding to the six monthly panel estimators. The model includes rotation group effects and permits the design variances of the survey errors to change over time. The model is fitted at the small area level but it accounts for correlations between the component series of different areas. The robustness of estimators obtained under the model is achieved by imposing the constraint that the monthly aggregate model based estimators in a group of small areas for which the total sample size is sufficiently large coincide with the corresponding direct survey estimators. The performance of the model when fitted to the Atlantic provinces is assessed by a variety of diagnostic statistics and residual plots and by comparisons with estimators in current use.

Issue Number: 1993002
Author(s): Pfeffermann, Danny; Rubin-Bleuer, Susana

Main Product: Survey Methodology

FormatRelease dateMore information
PDFDecember 15, 1993

Related information

Subjects and keywords

Subjects

Date modified: