The measurement of low pay in the UK labour force survey

Consideration of the National Minimum Wage requires estimates of the distribution of hourly pay. The UK Labour Force Survey (LFS) is a key source of such estimates. The approach most frequently adopted by researchers has been to measure hourly earnings from several questions on pay and hours. The Office for National Statistics is now applying a new approach, based on an alternative more direct measurement introduced in March 1999. These two measures do not produce identical values and this paper investigates sources of discrepancies and concludes that the new variable is more accurate. The difficulty with using the new variable is that it is only available on a subset of respondents. An approach is developed in which missing values of the new variable are replaced by imputed values. The assumptions underlying this imputation approach and results of applying it to LFS data are presented. The relation to weighting approaches is also discussed.