Why Outsourced Workers Count as Neither Layoffs nor New Hires

The Job Openings and Labor Turnover Survey, or JOLTS, published by the US Bureau of Labor Statistics, is an important indicator for tracking the US labour market.

But there is a detail in these statistics that markets can easily overlook. Suppose an employee continues working in the same position, with unchanged pay and duties. Only the employer named in the contract changes.

How is that worker recorded within the JOLTS statistical framework?

Outsourcing changes the reporting employer, even when the worker stays in the same job

Low layoff readings in JOLTS can raise a question: why do the figures show relatively few layoffs when job stability appears to be weakening?

The explanation is that JOLTS tracks changes in jobs at establishments, rather than the employment paths of individual workers. When a company shifts employees to an outsourcing arrangement, those workers disappear from the original employer's records, affecting its employment, hiring, vacancy and separation measures. The workers have not become unemployed; their employment is now attributed to the outsourcing provider.

BLS is avoiding double counting, rather than overlooking workers

BLS has an explicit rule: employees of staffing agencies, employee-leasing firms and outsourcing contractors are reported by the employer with which they are actually contracted.

The original company stops including those employees because the employer responsible for reporting them has changed. BLS also provides a separate data-collection channel for staffing and employee-leasing businesses. This arrangement is designed to prevent the same worker from being counted under two employers.

Why can employment data diverge from people's experience?

US employment statistics use two different survey frameworks, which answer different questions. The establishment survey focuses on jobs; the household survey focuses on employed people.

If a person holds two part-time jobs, the household survey counts one employed person, while the establishment survey counts two jobs. A divergence between the figures does not necessarily mean the data are wrong. It reflects a difference in what the surveys measure.

Employment counts show whether people have work, rather than the quality of that work

Employment statistics can tell us how many people are employed. They cannot directly show changes in job stability, benefits or workers' bargaining power.

Even if companies switch to outsourcing or short-term contracts, total employment may not fall. The data themselves are not necessarily incorrect. They simply answer a different question from the one the market wants to understand.

Moving employees to an outsourcing arrangement therefore does not mean their employment has disappeared. It means the employer to which that employment is attributed has changed.

Employment data tell us about changes in jobs and the number of employed people. Changes in job stability, benefits and bargaining power do not appear directly in those numbers. Macroeconomic data are not wrong simply because they fail to capture every aspect of economic experience: each series answers the question it was designed to answer.