Abstract
The Brazilian Labour Force Survey publishes monthly national indicators based on 3-month rolling data. This paper presents state-space models to produce state-level single-month unemployment rate estimates. The models account for sampling errors and the increased dynamics in the labour force series due to the unforeseen SARS-COV-2 pandemic. Bivariate time series models with claimant count auxiliary data and multivariate models combining survey data of several states are investigated. The results demonstrated the benefits of the univariate state-space approach to produce unemployment official statistics for Brazil. Additionally, the regional multivariate model shows promising results but requires further investigation.
Original language | English |
---|---|
Pages (from-to) | 1707-1732 |
Number of pages | 26 |
Journal | Journal of the Royal Statistical Society Series A-Statistics in Society |
Volume | 185 |
Issue number | 4 |
DOIs | |
Publication status | Published - Oct 2022 |
Keywords
- official statistics
- sampling error
- state-space model
- time series
- unemployment rate
- SMALL-AREA ESTIMATION
- TIME-SERIES
- DISCONTINUITIES
- LEVEL