Measuring Job Tenure Using Linked Employer-Employee Data

Abstract

The publication of job tenure statistics from Statistics New Zealand's Linked Employer-Employee Dataset (LEED) in 2006 provided the first comprehensive source of information in this area for New Zealand. The most noteworthy aspect of the new statistics was the high number of jobs with short tenure. LEED job tenure statistics are constructed from administrative data that is collected for tax, not statistical, purposes. For this reason, Statistics NZ has had to address a number of issues before determining the appropriate methodology to measure job tenure. Two key issues were: (1) How breaks in job tenure are best identified from monthly data. (2) How to best correct for administrative chum in the dataset. This paper looks at the impact on the level o f short-tenured jobs when varying the treatment o f these two issues and the appropriateness of doing so. Consequently, if also explores the challenges of deriving longitudinal statistics from administrative data. This work is part of a review of LEED methods and outputs that Statistics NZ has been undertaking over the past year.