Author
From the 1992 debut of the forerunner of Monster.com to the current ubiquity of LinkedIn and Indeed, hiring has moved heavily online.1 Researchers have been understandably tempted by the data: It is voluminous and high frequency, with linked information on firms, locations, occupations—and sometimes wages.
It turns out, however, that employers tend not to include much wage information in online job postings. When posts do include wage information, they tend to represent a biased subset of firms, according to findings from Minneapolis Fed economists Amanda Michaud and Simon Mongey (Staff Report 690, “Online Job Posts Contain Very Little Wage Information,” with Honey Batra of Cornerstone Research). They analyze the universe of online postings between 2012 and 2018; the economists determine that around this time, some job boards began imputing wage ranges on their own, causing a discontinuity in the data. In all, they study 140 million postings from 45,000 online sources, including third-party job boards and company websites.
These data are available through a contract between the Federal Reserve System and Lightcast (formerly Burning Glass Technologies). Many other researchers have used this same dataset to study the U.S. labor market, including wages. But none has asked the simple questions of how often wages are posted and how wage information is correlated with other aspects of the post, such as occupation, sector, local labor market conditions, and the firm’s position in the labor market.
The results from Batra, Michaud, and Mongey (forthcoming in the journal Labour Economics) suggest using the wage component of these data with caution: “Wage information in online job postings can be characterized as sparse, vague, and potentially selected,” they write, and their findings can be laid out along these lines.
“Sparse”
Only 13.5 percent of job posts in the sample contain any wage information at all. Behind this average, the distribution is heavily skewed across employers. The median firm provides wage information in less than 1 percent of its job posts. More than half of job postings are from firms that never post wage information (see figure).
Among the top 20 job-posting firms in the sample, only one included any wage information in more than 2 percent of its postings.
Most job posts that mention wages are by a small minority of employers. The right-tail cluster shows that about 10 percent of all posts in the sample are from firms that might be considered heavy wage-posters, with wage information in at least 25 percent of their posts.
“Vague”
Among the small share of all postings with any wage data, 57 percent feature a range. And these ranges are often wide; the average range is 28 percent of the midpoint. Examples close to this average would be $21–$28 an hour or $32,000–$42,000 a year.
Ranges are especially common in high-wage jobs, where 70 percent of all posts with wage information feature a range rather than a point. They are also more likely in the private sector.
The authors note that even policies to compel employers to include wage information could suffer from firms’ preference for imprecision. “Wide, variable ranges are a problem that remains salient even after the enaction of recent laws mandating wage information in job posting,” they write, based upon their preliminary analysis of more recent data from Colorado, one state that implemented such a law.
“Selected” (that is, biased)
If online wage data were sparse, vague, or otherwise noisy in a regular or random way, statistical techniques might compensate to some degree. But the economists find ample evidence that these issues are not random, with “systematic patterns in the availability and precision of wage information across occupations, firms, and labor markets.”
For example, postings for middle-wage occupations are about twice as likely to contain wage information as postings for lower- and higher-wage jobs. Posted wage information is more common for public sector positions, but also in certain private fields more likely to feature piece-rate or part-time work, such as trucking and security.
The type of wage information also varies in systematic ways. Postings for higher-paying jobs and by higher-paying firms are more likely to feature a range than a specific value. Among posts where ranges are provided, ranges are proportionately much wider for salaried versus hourly positions.
Online posts and the broader labor market
Bias across so many cross-cutting dimensions poses a steep challenge for researchers. This is layered atop evidence that online job posts imperfectly reflect the overall distribution of jobs in the U.S. economy. Batra, Michaud, and Mongey find that online job posts, regardless of wage information, are skewed toward higher-wage occupations. They cite research finding that online posts overrepresent computer-related roles and underrepresent jobs such as food preparation and construction.
When the economists compare wages in online job posts with administrative labor market data, they find a bias that might at first seem surprising. One might expect job posts to feature lower wages than the wages of average workers in an occupation (who are further along in their careers). But for lower-paid jobs, they find that posted wages are 30 to 40 percent higher than average wages reported by the U.S. Bureau of Labor Statistics. This suggests that in lower paid occupations, online posts with wage data overrepresent positions with higher-than-average pay. One plausible reason: Recruiters for such positions might have an incentive to signal that fact.
The economists also test whether patterns in the Lightcast wage data bear any relation to local labor market conditions. Intuitively, one might expect cities with lower unemployment rates to feature more—and perhaps more precise—wage information in online job posts. But the unemployment rate of a location appears uncorrelated with the share of posts with wage information, the share with pay ranges, and the width of posted ranges.
Batra, Michaud, and Mongey note that their findings do not mean most workers are unaware of compensation as they apply—only that they are not obtaining information about it from public job posts. And while they do document some intriguing patterns among the data, future researchers should not lose “the 1,000-foot view,” they write, “that hardly any of these postings contain wage information at all.”
Read the Minneapolis Fed staff report: “Online Job Posts Contain Very Little Wage Information”
Endnote
1 The first online job board is generally considered to be the Online Career Center, which launched in 1992 and later merged with Monster.com. By 2015, the Pew Research Center determined that applying for jobs online had become “nearly universal among recent job seekers,” although personal and professional connections were similarly important in the job hunt. In a 2025 iHire survey, 69 percent of employers said they did all or most of their hiring on job boards, not including LinkedIn; another 46 percent use LinkedIn, and 50 percent use their own websites.
Jeff Horwich is the senior economics writer for the Minneapolis Fed. He has been an economic journalist with public radio, commissioned examiner for the Consumer Financial Protection Bureau, and director of policy and communications for the Minneapolis Public Housing Authority. He received his master’s degree in applied economics from the University of Minnesota.
