Why are measurements of the gig economy so out of tune?

The gig economy—where workers are independent contractors, moving from job to job and setting their own schedules and workloads—is often described as a growing trend. However, measurements of the size and growth of the gig economy by various organizations and government agencies deliver wildly varying numbers.

The Gallup organization estimates that 36 percent of American workers are in the gig economy. But not so fast—a recent report from the Bureau of Labor Statistics found that approximately 10 percent of American workers were involved in “alternative work arrangements”—which was a one percent decline from 2005! And 2018 data from the Federal Reserve said that less than one third of adults in this country do some form of gig work—either as their primary job or to supplement other income.

Related: Are your employees ready to go gig?

As the New York Times said recently, “You can see the gig economy everywhere but in the statistics.” The conflicting numbers don’t leave economists or employers with a clear picture of where this trend is going.

A new report from the National Bureau of Economic Research looks at the problem of why there is such a wide range of conclusions about the gig economy. “Measuring the Gig Economy: Current Knowledge and Open Issues,” acknowledges that the current systems of measuring employment simply aren’t doing a very good job of getting a handle on the gig economy.

“The recent resurgence of interest in non-traditional work arrangements reflects the perception that new technology, along with the restructuring of business enterprises made possible by this technology, is producing an accelerated pace of change in the organization of work that is having important effects on both workers and firms,” the study notes.

However, the researchers add, the traditional means of measuring labor statistics, which include household surveys and administrative data, were simply not built to measure independent contracting and freelance work. “The current system of economic measurement is designed for a world in which workers have a traditional employment relationship or operate a formal business,” the report notes. “Non-employee work may not be fully captured in existing data sources.”

The study also notes that disagreements between household survey data and administrative measurements can be due to the complexity of self-employed work.

“There are many different types of self-employment work,” the report said. “Neither the household survey data nor the administrative data may be ideally suited to pick up all of that activity.” The researchers note that income being reported by individuals for tax purposes is often not reflected in administrative data—suggesting the need for new formulas in regards to that information.

To start with, the researchers call for better definitions. “Identifying the key attributes that characterize different forms of non-employee work may help us close in on the traits of jobs that are most appropriately characterized as gig work,” they say.

One question the report raises is where gig-type works fits in to the career paths of workers. The researchers say that much gig work is supplemental, and may or may not be related the participants’ career goals.

Linked data sets would also be very helpful in measuring this type of work, the report suggests. “Linking tax data with household survey data gives us not only the worker’s demographic characteristics, but also the worker’s family characteristics—something that is crucially important for understanding how gig employment is related to family income and health insurance coverage,” the report says.

A true understanding of the gig economy will depend on better data—and linked data sets are the most likely way forward, the report concludes. “Recognizing the limitations of each of the individual available sources of data, efforts to develop linked data sets that combine household survey data, tax data, employer survey data and, potentially, naturally occurring private sector data are likely to have a high payoff, permitting greater insight in to the changing nature of work than is possible using any single data set.”