Tavis Coburn
While the US workforce-participation rate languishes near historic lows, many employers complain they are having a tough time finding qualified workers to fill open positions, and not only in the “glamour” field of information technology.
The Solar Foundation reports that two-thirds of solar-energy installers can’t find qualified job applicants. Health-care providers of all kinds worry there won’t be enough trained people to serve an aging population.
Throughout the country, manufacturers are desperate to hire skilled welders. According to Monster.com, carpenters, electricians, mechanics, and plumbers are among the jobs employers have found most difficult to fill.
Meanwhile, in many areas across the US, people say they’re desperate to find good work. “I will be the greatest jobs producer that God ever created,” Donald Trump proclaimed in January, shortly before taking the presidential oath of office. But four months later, a worried factory worker from Racine, Wisconsin, told a CNN reporter that his job is moving to Canada and he might lose his house. Almost 15 million people are looking for full-time work.
Economists say this is a labor-market mismatch. Companies want workers, and workers want jobs, but they’re not in synch. What could help? Education and training is the primary answer given by many economists. But also something else: data.
From the numbers, the US looks close to or at full employment, according to many economists. The economy has added 15.8 million jobs since October 2010, posting a record 80 consecutive months of gains. The unemployment rate was 4.3 percent in May, its lowest in 16 years. Even wages are rising again: after years of stagnation, average hourly wages rose 2.5 percent over the 12 months ending in May.
“The job market has almost totally recovered from its devastating decline during the Great Recession,” Harry Holzer, former chief economist of the US Department of Labor, told the Atlantic this past January. “Significant wage growth, which was lacking early in the recovery, has returned.”
Federal Reserve Chair Janet Yellen struck a similar note in March. “The economy has essentially met the employment portion of our mandate, and inflation is moving closer to our 2 percent objective,” she told a meeting of the Executives’ Club of Chicago. Since then, the rate-setting Federal Open Market Committee has twice raised the federal funds rate by a quarter point. At a news conference following the FOMC meeting in March, Yellen declared, “The simple message is the economy is doing well.”
But the picture described by Holzer and Yellen, and seemingly supported by unemployment and wage data, is incomplete. There are millions of people who are underemployed, or who have dropped out of the workforce entirely.
A different view of the labor market
The official US unemployment rate excludes some notable segments of the working-age population.
A better picture of the labor market emerges when you consider another set of data from the Bureau of Labor Statistics, called U6, which includes the officially unemployed, people who are working part time but would prefer to work full time, and “marginally attached” workers who have looked for work over the year but grown discouraged. In May, the U6 stood near 8 percent, almost double the official unemployment rate.
Furthermore, some economists are alarmed by the labor-force-participation rate, which the BLS defines as the percentage of the total population aged 16 and above that is either employed or officially unemployed. That plummeted from a peak of about 68 percent in April 2000 to a low of 62.5 percent in November 2015, and hasn’t climbed much since. It indicates that an aging population aside, many able-bodied individuals have simply left the labor market.
According to the Organisation for Economic Co-operation and Development, the US ranked 39th in labor-force participation for prime-age workers in 2015, the latest year for which data were available. In this measure, the US trailed even the much smaller economies of Greece, Bulgaria, and Romania. And Maximiliano A. Dvorkin and Hannah Shell at the Federal Reserve Bank of St. Louis found in 2015 that the US was the “only country in our sample experiencing a recent decline in the aggregate labor force participation rate.”
The US is “doing much worse” than advanced European countries including France, Germany, and the United Kingdom, says University of Pennsylvania’s Ioana Marinescu. Even though France, for example, has a much higher unemployment rate than the US—10 percent, as of January 2017—“more people are working in France as a percentage of the prime-age population than in the US, and to me that shows there’s something really worrying about the US economy.”
For years, many believed they were seeing a cyclical problem, the result of a deep recession that trailed a financial crisis. Ben Bernanke, then chairman of the Federal Reserve, initially accepted the classical Keynesian explanation that high long-term unemployment and low labor-force participation were caused by a huge shortfall in aggregate demand.
Toward the end of his tenure, however, he worried that structural factors were playing a big role and that high unemployment and underemployment “damage the productive potential of the economy as a whole by eroding workers’ skills and . . . by preventing many young people from gaining workplace skills and experience in the first place.”
In retrospect, we can see that the housing boom masked the structural change. When manufacturing was shedding jobs, boomtowns such as Las Vegas, Phoenix, Miami and Tampa in Florida, and California’s Central Valley put people to work in construction. But when 2 million construction jobs disappeared from August 2006 to July 2010, it revealed the weakness in the labor market, especially among relatively unskilled workers who hadn’t gone beyond high school. “Employment rates for prime-age workers are well below historical levels, and for the men they’re at historically low levels,” says Chicago Booth’s Erik Hurst.
The statistics and data affirm the message voters collectively sent politicians last fall: many people want jobs, and better ones.
Meanwhile, on the labor-demand side of the economy, the situation looks entirely different. There were 5.5 million job openings nationwide this past December, according to the BLS. Even in manufacturing, the sector that has suffered the biggest job losses, employers complain of a dearth of prospective employees with the requisite skills. A 2015 report by the Manufacturing Institute projected that retirement among skilled workers and lack of training among younger ones will leave 2 million manufacturing jobs unfilled by 2025.
Companies are willing to pay well to fill open positions—if people have the requisite skills.
The wages of low-skilled or unskilled workers have stagnated or fallen, but “individuals near the top of the wage distribution enjoyed rapid and sustained wage growth,” write Chicago Booth’s Kevin M. Murphy and Robert H. Topel. In fact, “market fundamentals favoring more skilled workers are the driving force behind rising inequality.”
“Our current statistical infrastructure is not well suited to inform us about granular details of the labor market on the demand side.”
What you see is a skills mismatch. Jobs are available, but a sizable percentage of the population lacks the skills, education, and training needed to do them. When Chicago Booth Review surveyed economists about jobs, the skills gap came up repeatedly, and the economists offered a number of recommendations to address it, including offering vocational training, making college more accessible, and improving and investing in elementary and secondary education.
Yet data could also better help employers and workers find each other. Online personals have united couples, and online job searches have united workers with adoring, or at least approving, human-resources departments. Research indicates that this labor-market matchmaking could be far more widespread and effective, and researchers are developing tools and data sets that could help.
For starters, labor-market data as currently collected can give only a high-level picture of the situation and need more specificity to be useful to hiring managers. So some researchers are turning to online job postings.
The most-authoritative data on the labor market come from the BLS, which gathers and disseminates key statistics on pricing, productivity, wages, and employment. It tracks employment trends in industries going back decades and drills down to the state and metropolitan-area level. Its most widely watched and influential report is the monthly jobs report, which can move markets and shape monetary policy.
But while researchers consider the data gathered through it and other surveys and publications to be critical, these and other BLS data sets have limitations, especially if you’re looking for particular information that could perhaps guide corporate decision making. “Our current statistical infrastructure is not well suited to inform us about granular details of the labor market on the demand side,” says Chicago Booth’s Steven J. Davis. “It doesn’t tell us exactly what kinds of jobs employers find easy to fill, what kinds they find hard to fill, and why they find them easy or hard to fill.”
Hence Davis and University of Chicago PhD candidate Brenda Samaniego de la Parra are building a database of online vacancy postings. Thus far their database includes 9 million job openings posted since January 2012, along with 76 million applications for those jobs. The researchers worked with DHI Group, a New York–based company that owns several career and recruitment sites in industries such as technology, health care, finance, and energy.
The data, principally from DHI’s largest site, technology-oriented Dice.com, allowed Davis and Samaniego to drill down to the functional level, such as “software developer” or “project manager,” and even to specific skills, such as “JavaScript” or “Linux programming.” The database provides second-by-second tracking of job postings that let the researchers follow the volume and flow of job applications. (Davis is also a senior research advisor to DHI, and both he and Samaniego were compensated by the company to develop and analyze the database.)
The researchers also developed two new hiring indicators. First is the DHI-DFH Vacancy Duration Measure, which tracks the average number of working days it takes to fill a position. Based on methodology developed by Davis, the Federal Reserve Bank of Chicago’s R. Jason Faberman, and University of Maryland’s John Haltiwanger, the measure uses the BLS’s Job Openings and Labor Turnover Survey to estimate the average number of days it takes to fill open positions each month. The DHI-DFH Recruiting Intensity Index uses JOLTS data to quantify the intensity of employers’ efforts to fill vacant jobs.
Data behind job vacancies
Two indicators offer insight into the state of the hiring market, based on open positions and what companies are doing to fill them. The researchers maintain the data at DHIhiringindicators.com.
Davis et al., 2013; DHI Group
Analyzing the Dice.com data, Davis and Samaniego make some observations that illuminate the situation for companies and job seekers, and may help change the way employers view hiring. But Davis says the data sets he and Samaniego are building could also help illuminate issues that keep employees and employers apart—such as skills and location.
Imagine, for example, an employer looking for a Cisco programmer. The DHI data set and tools can help assess if there’s a surplus of these programmers in one part of the country and a shortage in another. “There are jobs and there are workers, but the workers may not be located in the same place as the jobs,” Davis says. “We’re building a machine to put out statistics every month so they can inform our understanding of labor markets not just two or three years after the fact but in near-real time.”
Other researchers are also using job postings to make observations. Marinescu and University of Toronto’s Ronald Wolthoff looked at all the vacancies posted on CareerBuilder.com for Chicago and Washington, DC, in early 2011, and find that job titles explained more than 90 percent of the variance in posted wages between jobs. “The words in the job title communicate important information about a position, and workers use them to direct their search,” they write, observing that titles say more than BLS occupational classifications do about wages and applications. As job titles appear to be the decisive factor in wage differences between posted positions that disclosed salary data, employers seeking workers should pay more attention to job titles.
Moreover, workers want jobs that are local. Marinescu and University of Warwick’s Roland Rathelot, also using CareerBuilder.com data, find that job seekers were 35 percent less likely to apply to a job 10 miles away from their zip code of residence because they “dislike applying to distant jobs.” While Marinescu and Rathelot call geographic mismatch “a minor driver of aggregate unemployment,” economists have become concerned about the macroeconomic impact of declining geographic mobility in the US. The percentage of the working-age population that moved to another state has halved since the 1980s, and fell below 1.5 percent in 2010.
But solving the greater mismatch problem involves more parties than job seekers and employers. Consider all the workers-in-training, students trying to decide what course of study to pursue. There’s a movement to give students more data to help them make more-informed education decisions by letting them know how much they can expect to make by graduating with a degree in, say, engineering rather than education.
What if students picked what to study based on not only outcomes but also specific, local workforce needs? “Maybe you have enough skills, but if you only added one additional skill, you’d be in a perfect position for a really in-demand job,” Marinescu says. “That’s something that, with the proper understanding of the landscape and algorithm, we can say, ‘Here’s the skill or couple of skills you’re missing to round out your résumé. And here’s a college where you can get a certificate.’”
How to hire, and get hired
Traditionally employers and economists have seen hiring as a sequential process, in which employers evaluate applications as they come in, and then decide whether to interview a candidate and make an offer. “The data we look at really don’t fit that picture,” says Chicago Booth’s Steven J. Davis.
What does the hiring process really look like? By mining data, researchers are learning things that may help change the way employers, economists, and applicants view hiring.
Don’t let a job ad languish. Job posters respond predominantly to recent job postings: some 39 percent of applications were for jobs posted in the past 48 hours, and 54 percent were for those posted in the past 96 hours. (Considering repostings—where postings are temporarily taken offline before being put back up—didn’t change the results.)
You have nine days to apply. The mean posting duration for single-position openings was nine days, which is only one-fifth of the mean vacancy duration in the JOLTS database. This implies that the search phase of the hiring process is much shorter than the selection and hiring phases.
Workers are looking on Mondays, and in January. Job seekers were more active early in the workweek. Application volume on Mondays and Tuesdays was more than 40 percent higher than on Fridays and more than four times greater than on Saturdays. Also, applications dropped 30 percent in December and rebounded in January, when volume was 60 percent heavier than in December.
Steven J. Davis and Brenda Samaniego de la Parra, “Application Flows,” Working paper, March 2017.
Moreover, what if the schools responsible for educating the present and future workforce had data to help them decide which courses they should offer? Community and four-year colleges do talk with local employers, but what if they shared detailed information about their hiring needs?
Marinescu is involved in a project to construct a powerful labor-market database through a partnership between the Labor Department, the National Economic Council, and the Center for Data Science and Public Policy at the University of Chicago. Supported by a grant from the Alfred P. Sloan Foundation, the Workforce Data Initiative will create a database that, according to the university, will “integrate data from national administrative sources . . . with privately held data from job websites, employment agencies, human resources management software,” and other sources. When it’s completed, the new data set—called the Skills Cooperative Research Database—may become a public resource akin to the Human Genome Project, containing the elemental information behind every job in the country.
“I think what’s really useful is to have a clever crosswalk between all the sources of data, so they can all be leveraged at the same time either for research or practical purposes,” says Marinescu. “That’s really the ambition. And the reason computer scientists can be so useful is there are lots of complexities involved in merging and matching all these disparate sources of data.”
This is the kind of process that could help fix the bifurcated US labor market, and could significantly scale up localized efforts to match employers with employees. BMW’s plant in Greer, South Carolina, set up a BMW Scholars Program with two technical colleges and a local community college, to give students specialized training in advanced manufacturing, as well as part-time employment and help with tuition. At the end of last year, 107 graduates of the program had been hired for jobs with BMW.
Data could help such efforts grow and be far more impactful. BMW’s 107 jobs may not change the state of the state or national labor markets, but 10,700 or 107,000 could start to do so. Such large-scale matchmaking would put people to work, grow business, and produce economic growth that everybody would love.
Conventional economic analysis of slavery overlooks the costs imposed upon the people who were enslaved.
Emancipation May Have Generated the Largest Economic Gains in US History
In the complex real economy, consumer price inflation can be a misleading indicator.
What’s Causing Inflation: Supply or Demand?
Quantitative easing may have played a part in the US financial sector’s current instability.
Did the Fed Contribute to SVB’s Collapse?Your Privacy
We want to demonstrate our commitment to your privacy. Please review Chicago Booth's privacy notice, which provides information explaining how and why we collect particular information when you visit our website.