The main objective of this paper is to investigate the use of earnings spells in LEED as a measure of job tenure. The paper was motivated by whether LEED could be used to identify and describe seasonal employment patterns, which typically only last for a few months. The paper explores the extent to which employment relationships in LEED contain multiple job (earnings) spells and the impact on the tenure distribution if individual job spells, between an employer and employee, are joined together.
The study found that one in five jobs (21.1 percent) in LEED, as at March 2006, were repeat spells with the same employer and nearly half (44.4 percent) of repeat-job spells started following a single month of non-employment and only 16.2 percent of repeat spells occurred after a non-employment period of over 12 months. Imputing all non-employment periods as employment had a measurable, but not a particularly dramatic effect on the job tenure distribution. For example, the share of job spells with elapsed tenure of 12 months or less falls by only 10 percentage points from 48.1 percent to 38.0 percent, a decline of around 20 percent. A distinctive pattern among repeat-job spells was for an earnings spell to end in December and for a new spell to begin in February. Around a quarter of all repeat spells, separated by a single month, start in February, in particular, 63.1 percent of job spells in the education industry fall into this category.
It is likely that LEED-based measures of job tenure that define job spells as a contiguous set of months with non-zero earnings will bias downwards mean job tenure and raise the share of jobs that have only been going for a few months. There is some evidence that many repeat job spells are in fact continuing relationships between an employer and employee and not the beginning of a new job.