Was the Mid-2000s Drop in the British Job Change Rate Genuine or a Survey Design Effect?

Abstract

The year-on-year job change rate fell sharply, from 18% in 2005 to around 13% in 2006, according to British Household Panel Survey (BHPS) estimates. This fall coincides with the introduction of dependent interviewing to the BHPS, intended to reduce measurement error and improve consistency. Estimates from models of job change misclassification rates (Hausman et al., Journal of Econometrics, 1998) show that reduced measurement error cannot account for the fall in the job change rate. This suggests that the fall was genuine.

Keywords: job change, dependent interviewing, misclassification error, feed forward

JEL Classification: J62, C25, C81