|
Activity Number:
|
386
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Thursday, August 15, 2002 : 8:30 AM to 10:20 AM
|
|
Sponsor:
|
Section on Survey Research Methods*
|
| Abstract - #300646 |
|
Title:
|
Nonlinear Mixed Effects Cross-Sectional and Time Series Models for Unemployment Rate Estimation
|
|
Author(s):
|
Yong You*+ and Edward Chen and Jack Gambino
|
|
Affiliation(s):
|
Statistics Canada and Statistics Canada and Statistics Canada
|
|
Address:
|
, Ottawa, Ontario, K1A 0T6, Canada
|
|
Keywords:
|
Benchmarking ; Hierarchical Bayes ; LFS ; Nonlinear mixed model ; Small area ; Unemployment rate
|
|
Abstract:
|
You, Rao and Gambino (2000) proposed a linear mixed effects cross-sectional and time-series model for small area estimation. In particular, they applied the proposed model to the Canadian Labour Force Survey (LFS) and produced efficient model-based unemployment rate estimates for sub-provincial areas such as Census Metropolitan Areas (CMAs) and Census Agglomerations (CAs) across Canada. In this paper, we extend the model of You, Rao and Gambino (2000) by considering a nonlinear linking model for the parameters of interest. In particular, we propose a log linear unmatched cross-sectional and time series model for the LFS unemployment rate estimation by extending You and Rao (2002) unmatched sampling and linking models to cross-sectional and time series data. We apply the proposed model to produce model-based HB unemployment rate estimates for the CMA/CAs under a hierarchical Bayes (HB) framework. Bayesian model evaluation based on the posterior predictive distribution is also studied. The HB estimates can also be benchmarked so that the benchmarked HB (BHB) estimates add up to the direct LFS estimates for large areas.
|
