Volume 49 - Article 33 | Pages 865–904
Background: The distribution of job tenure plays an important role in demography, economics, and sociology. Job tenure in a labor market is analogous to age in a population. Demographers have used indirect methods based on variable-r methods to estimate parameters for life table models. The variable-r method can also be employed to estimate the parameters of a job tenure table model that yields the expected length of job tenure and related measures.
Methods: Only two retrospective surveys of current employee tenure lengths and a count of between-survey hires are required to estimate the parameters of a period tenure table using the variable-r method. Tenure-specific sources of decrement allow an analyst to estimate the parameters of multiple-decrement tenure tables and associated single-decrement tenure tables that isolate the proximate contribution of a specific decrement to the job separation process. I illustrate and evaluate the method using publicly available US data.
Results: Variable-r methods generated reasonable parameter estimates: The expected job tenure was 2.48 years at 2002–2004 decrement rates. Multiple-decrement methods can estimate the fraction of employment relationships that end via job displacement. Cause-deleted tenure tables can capture the static effect of eliminating a particular risk to the population of employment relationships.
Contribution: Arthur and Vaupel (1984) provide a framework for studying nonstable populations that subsume the variable-r relations that I utilize in this work. Vaupel had an interest in formal demography throughout his life but started his academic career in business statistics. This paper combines those interests.
Most recent similar articles in Demographic Research
Cited References: 64
Download to Citation Manager