The Gig Economy and BLS Surveys

This is a cautionary column. The staff at the Bureau of Labor Statistics have an impossible task. They claim the “Employment Cost Index or ECI measures changes in the cost of employees to employers over time.”  But increasingly the so-called gig economy opens to question the accuracy of their statistics. 

To be fair, in the strictest understanding of the terms “employees” and “employers,” the ECI sentence may be accurate. However, it’s not stated in many BLS announcements that their surveys rely heavily on administrative data from state unemployment insurance programs. Two core points flow from that linkage:  

1. State unemployment laws differ on a number of somewhat obscure points, which means the data are not strictly comparable from state to state, and 
2. All individuals paid for their services are not covered for unemployment insurance—think gig workers—and therefore their work information is not included in BLS data.  

That undermines our confidence in BLS statements such as, “The Quarterly Census of Employment and Wages (QCEW) . . . covers more than 95% of U.S. jobs available at the county, Metropolitan Statistical Area (MSA), state, and national level . . .”  That is simply not accurate.

BLS data could include sample data from virtually all “traditional” jobs but experts who have written about the gig economy argue the government’s data miss a rapidly growing number of workers. With each year that passes, the voids in the data are expanding; the fact is government does not understand the gig economy and currently does not have the capability to document relevant data.

Why does this matter? BLS survey data are the statutory basis for the deliberations by the Federal Salary Council to recommend adjustments to the General Schedule. The Federal Employee Pay Comparability Act mandates the use of the data in question. It’s never been explained but in the three decades since passage of FEPCA each President, Republicans and Democrats, has chosen not to base their decision on the BLS data. This further erodes the value of the data.

Gig Economy Estimates

At this stage, experts are quick to acknowledge there is no agreement on the facts. Perhaps the most credible source is the research reported on the National Bureau of Economic Research website. An August 2018 article, “Measuring the Gig Economy: Current Knowledge and Open Issues,” appears to be the most recent. Two of the four authors work for the Census Bureau. The article focuses on the measurement problems and reaches a key conclusion: Government simply does not understand this growing segment of the labor force.  

When FEPCA was enacted it was assumed that BLS pay data would be limited to the then standard jobs and employees who 1) are paid a wage or salary; 2) have an expectation of job security; 3) may be full time or part time; 4) have hours and earnings that are reasonably predictable; and 5) are supervised by the same firm that pays their wage or salary. That includes on-call workers as well as direct hire temporary workers. The test is whether a worker appears on a payroll and in 1990 that was close to universal (except for part time jobs held by youth, personal service workers and illegal workers).

It’s the workers and jobs that do not satisfy those criteria that are the basis for the gig economy. One estimate of the size of this workforce, based on a 2019 “Freelancing in America” survey, claims it includes 57 million workers over the age 18. BLS surveys show the US labor force has roughly 155 million “wage and salary” workers in all sectors. That means the gig workforce is 36% of the size of the labor force in more traditional jobs.  

Significantly the freelancing survey also shows:

• The full time freelancers increased from 17% in 2014 to 28%.
• Skilled services are the most common type of freelance work, with 45% providing skills such as programming, marketing and business consulting.
• The top reason for choosing freelance work is flexibility. A high percentage work from home.
• Freelancing opens the door to opportunities for those who otherwise might not be able to work.
• Freelancers were found in all age groups—29% of Baby Boomers, 31% of Gen X workers, 40% of Millennials, and 53% of Gen Z workers.

As I was writing this, I noticed the Wall Street Journal website was running a sponsored-content campaign that succinctly summarizes what’s unfolding: 

The Future Is Here—And It’s Flexible: In the past, large companies used independent professionals sparingly; now, they increasingly see them as integral to their success.

As confirmation, the Federal Reserve’s May 2018 report on the economic well-being of U.S. households, found that 31% of adults are engaged in independent work—up 3 percentage points from the 2016 report. Other studies by McKinsey and Upwork show that 36% of the American working age population engages in independent work. There is no accepted definition of the gig economy but it is large and growing.

Those workers are not included in BLS pay estimates and that is a serious omission. Everyone is aware of the gig workers making home deliveries (e.g., DoorDash) and providing rides (e.g., Lyft) but no one to date has developed a methodology to document the numbers of workers, their occupations, hours of work, locations or, most importantly, the pay levels. They clearly are competing with the more traditional employees reflected in BLS surveys.

Gig workers are seen as a cost saving alternative to adding full time employees. There have been reports that gig workers making deliveries are dissatisfied with how much they earn. That may be true in other gig roles as well. There is also scattered data showing the average hourly pay in some gig jobs is higher than the pay of full-time employees. Gig employees, however, are rarely eligible for benefits or paid time off.  Plus, since they are not employees, they can be terminated at any time with no recourse. But the facts have never been documented; we simply don’t know how the compensation of gig workers compares.

What is the Value of BLS Data?

Each year the Salary Council and the Pay Agent fulfill the roles set forth in FEPCA and submit recommendations for GS salary increases based on BLS survey data. FEPCA also allows the President to reject the recommendations and rely on the alternative plan authority in proposing increases. Each president, starting with Clinton, has annually rejected the recommendations.  

That track record raises questions about the value or utility of the BLS data, more specifically the value of the National Compensation Survey. The cost is buried in the BLS budget but it has to be in the millions. As a far less costly alternative, government could purchase or gain access at no cost by participating in the surveys used by other employers. There are hundreds covering every sector and occupation.

Those surveys are not statistically valid but the participating employers are known, the reliance on the data is solidly established, and the data are integral to talent management. The use of survey data is routine and universally accepted in other sectors.

BLS claims their data are used for several purposes. The first topping the list: “Business owners and human resources professionals make decisions about pay and benefits to stay competitive in the labor market.” But there is no evidence of this in the HR literature.

It's not clear if or when the BLS has ever asked data users listed on the BLS website—the business owners and HR professionals, budget and contract specialists, etc.— how they use the data or what they think of the data. Instead, BLS has focused internally on developing an incomprehensible statistical methodology.

The most obvious shortcoming that undermines the value of BLS data for employers is the inability to report what jobs are paid. Employers also need to understand program management practices like the role of cash incentives. BLS fails to report that information. That’s what determines the value of pay surveys.  

A simple step to assess BLS data would be to develop a comparative database. Government, as with every employer, needs to be competitive for essential talent. BLS data cannot address that core question.

For years BLS has functioned in its ivory tower while the country’s labor markets and people management practices have undergone significant change. It’s no longer 1990. Now the rapid expansion of the gig economy means BLS cannot claim their data are representative of workers’ pay and benefits across the United States. That’s a problem.