Longitudinal Analysis of Labour Force Data

Abstract

The Household Labour Force Survey (HLFS) is a quarterly panel survey that is used to provide a snapshot of the New Zealand labour force at a point in time. Although originally intended for cross-sectional purposes, the fact that the occupants of the same households are interviewed for up to eight quarters makes it possible to extract longitudinal information, such as labour force dynamics. This paper will discuss some of the longitudinal uses of the data, and some potential problems and issues.

One issue discussed is data cleaning. The HLFS imputes some variables, such as age and sex, when data has not been provided by respondents. However, as the main objective is to produce cross-sectional estimates, there is no attempt to achieve longitudinal consistency, so apparently people may change sex or be rejuvenated. I will discuss some methods for cleaning the data and investigate whether this has any significant effects on longitudinal estimates, such as hazard rates.

The results suggest that it is feasible to obtain longitudinal information about transitions in labour force status from the HLFS data, but it is necessary to concatenate panels to obtain reasonable accuracy. Editing seemed to make little difference to the conclusions.