DISTRIBUTIONS OF JOB TENURE FOR US WORKERS IN SELECTED INDUSTRIES ANDOCCUPATIONS

Identifying data to support a distribution or value to describe the le ngth of time that a person may be exposed to contaminants at a hazardo us waste site is, in our experience, one of the most challenging tasks in calculating risk estimates (or developing risk-based cleanup targe ts). To evaluate potential risks for workers when no site-specific dat a on exposure duration are available, current job tenure is often used as a surrogate for projected job tenure. Surveys conducted by the U.S . Bureau of the Census provide data on the length of employment at a c urrent job. By modeling the behavior of U.S. workers based on these su rvey data, projected job tenure can be inferred. In this paper, we est imate distributions of projected job tenure for U.S. workers in select ed industries and occupations using January 1987 survey data on the le ngth of employment at a current job. Our analysis indicates that avera ge projected job tenure is shorter than average current job tenure. Th e whole distribution for projected job tenure also exhibits this well- known but seemingly counter-intuitive result. Based on the results of our analysis, we believe that use of current job tenure as a surrogate for projected job tenure overestimates potential exposures for all gr oups of workers (and for developing remedial goals, produces risk-base d cleanup targets which are too stringent). Since numerical estimates of risk and risk-based cleanup targets impact management decisions and costs associated with carrying a site through the remedial process, w e recommend that state and federal regulatory agencies adopt a fully p robabilistic approach in lieu of current deterministic risk assessment techniques, which tend to produce overly conservative risk estimates and cleanup targets. Specifically, rather than using a point estimate of current job tenure to characterize the length of time that workers may be exposed to contaminated media, we urge these agencies to apply the methodology presented in this paper to develop distributions of pr ojected job tenure.