WASHINGTON, Nov 29 (Reuters) - As a Harvard University graduate student in the 1980s, Katharine Abraham tried to fill a hole in U.S. economic data by piecing together an estimate of labor demand, essentially a best guess of the number of open jobs in the country.
She drew from a patchwork of data, including surveys some states had conducted, a bit of information from the manufacturing sector, and even a pinch of data from Canada.
It was the best available information at the time, she said in a recent interview. But later, in the 1990s as the Clinton administration's Bureau of Labor Statistics commissioner, she converted that project into one of the most important, if still somewhat obscure, sets of labor market data beyond the monthly U.S. jobs report itself.
The Job Openings and Labor Market Turnover Survey (JOLTS), first issued to the public in 2002 and with information dating back to December 2000, is young compared with unemployment statistics that reach back to the 1940s, and was even scuttled by budget cutters the first time Abraham proposed it.
But it is having a moment, with a track record now long enough that Federal Reserve Chair Jerome Powell has made the JOLTS job vacancy estimate a touchstone in how he views the labor market and, as a result, the possible course of interest rates.
Analysts have taken note.
When the survey for October is released on Wednesday, "all eyes will be on the job openings data" and whether an expected decline in openings reaffirms the Fed's hope that the super tight hiring conditions seen through much of the COVID-19 pandemic are easing, Tim Duy, chief U.S. economist at SGH Macro Advisors, wrote in a note.
'OUT OF LINE'
A big surprise in either direction could tilt estimates of the Fed's expected path of interest rates.
The JOLTS vacancy estimate "has been unusually important in this cycle because it has been so out of line," with roughly two jobs open for each unemployed person, Powell said in a news conference after the end of the central bank's Nov. 1-2 policy meeting. "We keep looking for signs that ... the beginning of a gradual softening is happening ... I don't see the case for a real softening just yet."
The monthly unemployment rate, the product of a much larger sample of about 60,000 people each month compared with the 21,000 firms polled by the Labor Department for JOLTS, still captures the bigger headlines as the bellwether number for the state of the job market.
But that is a broad and blunt statistic. In recent years, researchers have looked for supplements, in particular to data like JOLTS, to provide more nuance about job market dynamics.
A worker who quits a job and one who is laid off may seem both headed for unemployment, for example. But a "quitter" is more likely simply flowing into another job in a sign of a strong economy, while rising layoffs are a sign that things may be weakening.
JOLTS measures both, and former Fed Chair Janet Yellen, when still vice chair of the U.S. central bank, elevated the quits rate to key status in her monitoring of the job market. Quits estimates would later fuel debate about a "Great Resignation" when they spiked during the pandemic.
The JOLTS data has become prominent enough that the Biden administration wants to double the size of the survey, and boost the budget from around $5.5 million to $9.6 million, so that estimates can be produced without the current one-month lag, provide more detail by industry and state, and fill what are seen as longstanding gaps.
The current survey was designed to make it easy on businesses and encourage responses, with a one-page form that asks for six pieces of data.
One issue, however, is that the job openings recorded by businesses don't reflect how intensively a company is trying to fill its available jobs, and, consequently, whether executives are primed to bid up wages or are passively waiting for the right candidate, said Steven Davis, an economics professor at the University of Chicago Booth School of Business.
"When JOLTS came along it was stepping into a data void that it has done a good job of filling. It is just incomplete," Davis said. "What is missing is anything about the effort that the employer is putting forth - like advertising, how quickly you interview ... whether the standards are tight or easy. There is zero on that."
An expanded JOLTS survey may get directly at that and other issues in the future, said Paul R. Calhoun Jr., who was involved with developing the survey in the 1990s and is its current manager.
Among the substantive holes to potentially fill, he said, is the pay offered for open jobs - data that, if coupled with occupational estimates that JOLTS economists also want to develop, could help show how wages are expected to behave, another key point for the Fed in its inflation control efforts.
"You got all these job openings," Calhoun said. "How can I characterize them? Are these good jobs or not?"
Expansion is a far cry from where JOLTS began.
Abraham said after her original survey proposal was turned down, it was resurrected when the unemployment rate dove towards 4% during the technology-driven boom of the mid-to-late 1990s and President Bill Clinton's White House wanted more information about what was happening.
"We had the unemployment rate, so we knew how many people there are who are looking for work and don't have a job. We didn't have anything analogous on the employer side," she said. Similar to the current situation, "people were very concerned about labor shortages and difficulty employers were having in filling jobs."
Even then, JOLTS staffers said the program had a tenuous grip for the first few years.
The priority among staff at that time was to build a simple survey "to get this off the ground for a few years and hopefully survive because that was certainly, from my standpoint, not a guarantee," said Mark Crankshaw, the JOLTS lead statistician who also helped build the program.
But "it has been going upwards. Now ... I listen to podcasts and people start talking about JOLTS ... I didn't expect it at all," he said.
Our Standards: The Thomson Reuters Trust Principles.