British Household Panel Survey Employment Histories: Waves 1-11, 1991-2002

Abstract copyright UK Data Service and data collection copyright owner.This study was compiled using data from Waves 1-11 of the British Household Panel Survey (BHPS - held at the UK Data Archive under GN 33196). The main objective of the research project was to investigate the impact of recall bias on measured labour market behaviour and to highlight how and to what degree the biases in the reported data may affect the estimation of models of labour market dynamics. The labour force histories in the BHPS provided a unique opportunity to analyse this potential bias by collecting information for the same time period from two different interviews (the 'wave overlaps'). Employment histories from the original data were constructed in two different ways, mimicking that which might be collected by repeated interview, and also that collected by a single retrospective interview. This allowed comparison of the measured labour dynamics from each approach. A 'reconciled' data set was also constructed as a base comparison, attempting to provide the most accurate picture of behaviour. For the second edition (December 2010), an updated version of the user guide was deposited. Changes have been made to the introductory text and the details given for the program bhpshist3.Main Topics:The dataset consists of eight data files derived from data from the first 11 waves of the BHPS. Each dataset contains a record for each full interview provided by each individual. The data contain a series of variables defining the individual's labour force spells since leaving full-time education until the time of the interview. The spells are defined in terms of spell state, start date and end date. This series of spells has been used to construct summary measures of the number of months in employment since leaving full-time education and employer tenure in months. Each file contains the same variables and the same sample, but the employment history variables in each file have been constructed under different methods. Two types of data source and four construction methods define the eight data files. The two types of data sources consist of 'main activity' and 'all-work' sources. The 'main activity' information derives from questions about the individual's self-defined main status, while the 'all-work' sources aim to record all employment spells. At each interview in the BHPS, information on labour force spells is collected for a period extending prior to the previous interview. This can generate inconsistencies in the spells reported during this 'overlap' period. The four methods of construction are based on which source of information is used for this overlap period. The first method, called the 'traditional panel', ignores any information prior to previous interview. The second method, named 'latest interview rules' only uses information from previous interviews to complete gaps. The 'reconciled' method uses case-by-case correction of inconsistencies, while the final method, 'selected no problems', uses only those observations with no inconsistencies.