Politicians extol the virtues of small businesses as an engine of the U.S. economy, but often leave out a very important flip side. They also have a shorter life span than larger firms, and thus toss more people into joblessness, making them disproportionately the generators of unemployment. .
Yet there’s no good way to know what’s happening in those small businesses fast enough. That’s particularly true in times of crisis, such as 2008, when employers were shedding jobs by the hundreds of thousands every month and the lack of real-time data about small business was troubling.
“A key issue for policymakers following the financial crisis of 2008 was whether small businesses were disproportionately affected because of credit constraints,” economist Alan Krueger wrote in 2011. “The lack of timely, comprehensive data on the performance of small businesses was a hindrance to policy development.”
The most frequently turned to data on what’s happening to businesses of different size is the Labor Department’s Business Employment Dynamics data, which gives a read once quarterly on a period that ended six months earlier. That’s useful for economic researchers wanting to know what happened over several years, not as useful for anyone needing to know what happened last month.
Given that small businesses are both created and destroyed at a pace that differs from larger establishments, the lack of real-time data remains a problem.
“Once you account for the age of a business, then small business does not contribute disproportionately to overall job growth,” said Heidi Shierholz, a labor economist for the Economic Policy Institute, a liberal economic think tank, who supports efforts to find better measures. “It helps us know whether we really should give preferential treatment in support of them as disproportionately strong job creators.”
EPI researchers found that less than 20 percent of the employed in 2011 were at companies with 20 or fewer workers, a share that’s been stable for a decade. But from 2000 to 2007, small businesses also accounted for about 40 percent of all job gains, underscoring the importance of understanding what’s happening in real time.
At the time of his writing, Krueger had recently left the Treasury Department. He would later go on to head the White House Council of Economic Advisers, a job he left in June. His concerns translated into efforts at the Labor Department to experiment with ways to measure changes in companies by size and in real time.
At Treasury, Krueger received from the Bureau of Labor Statistics unofficial real-time tabulations from the Job Openings and Labor Turnover Survey, known by economists as JOLTS data. The BLS provided information on hiring and firing by companies of six different sizes, which was then further massaged into reporting on companies of three size groupings: fewer than 50 employees, 50 to 249 workers, and those with at least 250 employees.
Initial efforts faced an immediate problem, said Keith Hall, who headed the Bureau of Labor Statistics from 2008 to 2012.
“The challenge is the difference between establishments and firms,” said Hall, now a senior researcher at George Mason University’s Mercatus Center in Virginia. “A Starbucks is a small establishment, not many people are working there.”
In reality, Starbucks is a massive corporate chain. Statisticians had to tease out establishments from the data collected and then, using identifying information from employer unemployment insurance notifications, reclassify the company into a category based on firm size. The process was repeated for all sorts of companies.
“They weren’t really set up to do this link between establishments and firms very easily,” Hall said.
The Labor Department eventually launched experimental monthly estimates on job openings and losses by business size, created a database based on a survey of 16,000 non-farm business establishments, and now offers information that begins in December 2000 and ends in May 2013. The effort is not yet ready for prime time, however, remaining in the experimental stage. The sobering initial conclusion is that it’s really hard to measure small business in real time.
“Additional research performed on these experimental estimates since their initial release suggests that hires and separations in the lowest size class may be somewhat understated,” the BLS said on a webpage for the experimental data.
Another branch of the BLS that’s responsible for the Current Employment Survey has also experimented with publishing monthly data about employment, earnings and hours worked by size of business. That effort has been on hold since February 2012 as statisticians try to determine why there is variance with the Business Employment Dynamics that’s published currently on a quarterly basis.
“We want to be confident in what we’re putting out there,” said Kirk Mueller, a BLS branch chief who oversaw the effort. “There’s a lot of interest in size-class data.”