Affiliations: [a] João Pinheiro Foundation (FJP): Division of Statistics and Information, Minas Gerais, Brazil | [b] Brazilian Institute of Geography and Statistics (IBGE), Rio de Janeiro, Brazil
Correspondence:
[*]
Corresponding author: Caio César Soares Gonçalves, João Pinheiro Foundation (FJP): Division of Statistics and Information, Minas Gerais, Brazil. E-mail: [email protected].
Note: [1] Note: Here is presented only the abstract of the 3rd place winner paper of the IAOS YSP 2021. After the submission of this article and before the results being announced, the authors extended the paper in collaboration with two other researchers. The extended version was accepted in another journal, and it is currently under review.
Abstract: The Brazilian Labour Force Survey (BLFS) is a quarterly rotating panel survey with 80% sample overlap between two successive quarters. Monthly unemployment rate estimates are regularly produced based on a three-month average of direct estimates. Due to the unforeseen situation of COVID19 pandemic and its effects in the economy and labour market, there was a need to investigate model-based estimation procedures to obtain unemployment rate single-month estimates. We present structural time series models developed to produce model-based single month estimates at national level as well as small area (state-level) estimates at a higher frequency than those currently being published. Using the state-space framework, the models account for the autocorrelation due to sample overlap and the increased dynamics in the labour force series in 2020. In addition, bivariate models that combine claimant count and survey data are investigated. The models not only yield estimates with better precision than direct estimates, since the latter were affected by a rise in non-response, but they can deliver reliable state-level official statistics at a monthly frequency that are presently required. The new improved model-based estimates were proposed as experimental statistics for the Brazilian national statistical office (IBGE).
Keywords: Official statistics, unemployment rate, state-space model, sampling error, small area estimation.
