Modelling the UK Labour Force Survey using a Structural Time Series Model

The UK Labour Force Survey (LFS) is a quarterly survey based on a rotating panel design, which generates about 80% sample overlap between two successive quarters and can induce correlated sampling errors and rotation group bias (RGB). The standard time series analysis packages fail to account for the effect of the sampling error autocorrelation (SEA) and the use of these procedures may produce spurious trends. This paper represents the first attempt to set up a structural time series model for the UK Labour Force Survey series that accounts for the correlation between panel estimates. The proposed model is a first step towards the development of a state-space model to estimate the rotation group effects and discontinuities that may arise due to the transferring of the LFS to the Integrated Household Survey.