191 – Variable Selection for Regression
State Space Model for the UK Labour Force Survey
Duncan Elliott
Office for National Statistics
Ping Zong
Office for National Statistics
The UK Labour Force Survey (LFS) publishes a single estimate for each rolling quarter based on a rotating wave design. The purpose of this research is to propose and empirically assess a State Space Model for LFS wave-specific rolling quarterly data with two distinct features. First, the state vector for UK LFS unemployment will not only consider the rolling quarter but will take all relevant monthly periods into account for all state variables in order to capture the characteristics of the rolling quarterly data. Second, the survey error in wave i at time t is correlated with the survey error in wave (i+1) at time t+3 for all waves due to the rotating panel survey design. With the development of such a model we aim to deal with potential discontinuities as a result of changes to the survey design and improve estimates of LFS unemployment. Additional improvements include the use of administrative data on unemployment benefits, and the potential for a monthly unemployment estimate.