February 7, 2012 by Joshua Wright
Editor’s note: EMSI has heard from many clients lately about real-time labor market information and its potential uses compared to standard labor market data to help colleges and regional workforce planners. The purpose of this article is to distinguish between real-time LMI and traditional employment data, highlight the strengths and weaknesses of job-posting and résumé data, and discuss how regional data analysis is bolstered when the short- and long-term perspectives are taken into consideration.
In some ways, it’s the new frontier. Policy groups, workforce agencies, and higher education professionals have recently started digging into “real-time” labor market information (essentially, current job postings and résumés in online labor exchanges). This has been facilitated by new technology that allows easier and more comprehensive access to the information. Already, some organizations have started using the data to help match the skills of the unemployed with available jobs, or to help training providers set up nimbler short-term educational programs to address immediate community needs.
The need for intelligence that will lead to better training programs is clear. An estimated 12.8 million Americans are unemployed, millions more are underemployed, and yet many firms complain that they cannot find the skilled talent they need. For workforce and education planners tasked with connecting the unemployed or underemployed with open positions, real-time information holds obvious value. It can provide quick insights, through keyword searches, into the job titles that employers are looking for and the requisite skills that they demand.
At EMSI, we see the value of this data too. Our labor market analysis tools – used by over 200 of the nation’s community colleges as well as many workforce investment boards and economic developers – aggregate current job postings from Indeed.com. Linked to our database, these postings supplement official labor market data from the Bureau of Labor Statistics, Census Bureau, state labor market offices, and many other government sources, which offer a more holistic and big-picture employment view.
After fielding questions from clients and seeing research about job-posting data, we feel it’s important to distinguish between real-time LMI and labor market data from the BLS or other sources. Real-time data gives a glimpse into current trends based primarily on recent want ads and a database of résumés. It’s very timely but sometimes incomplete. Traditional data, on the other hand, includes past trends and projections on hundreds of standardized industries and occupations, all of which have been collected for decades and are based on either mandatory tax records or mandatory employer participation in surveys. It’s generally more accurate and complete than postings data — but not as timely.
Although different, these types of data can complement each other. For colleges that set up short-term customized industry training, real-time data based on postings from the past few weeks or months is certainly informative and helpful. Yet it is also important to have the big picture in mind. Labor markets churn and fluctuate, the needs of employers are always changing, and understanding broad trends alongside short-term trends will help colleges make sharper, more effective decisions. This is why the long-range perspective supported by established labor market data is so valuable, and why the short-term nature of job-posting data — when looked at in isolation — can potentially be misleading for colleges deciding on which two-year programs to support or start.
Before we go further, it might be helpful to establish an actual definition for labor market data. One of the nation’s most respected labor market sources, Oregon’s Labor Market Information System (OLMIS), defines LMI this way:
Data available on a particular labor market, including geographic and industry employment and unemployment estimates, occupational employment projections and wage information, and industrial average hours and earnings data.
Online job postings, which sometimes include earnings ranges, would definitely be part of this definition. However, two difficulties often come with postings and résumés: there is no context, and there are no long-term trends or projections that can be used to validate or broaden the perspective for a particular job. On top of that, not every company actually releases online job ads when looking to fill openings, so real-time data (job postings, in this context) could give a limited view into a particular labor market. Further still, some companies put out postings from time to time simply to gather résumés for future use. If a particular company posts 500 ads and only has plans to hire a small percentage of those, it can be a bit misleading. Finally, in sectors such as retail or agriculture or sales, jobs can be seasonal. You might see a big bump in job postings before Christmas or in the summer, but this can be problematic if the seasonal variations in the data aren’t taken into account. Again, this is a complex issue, but knowledge of the potential pitfalls is key before you start to apply the data into your decision-making process.
Jobs for the Future did a great job of summarizing the limitations – as well as the advantages – of a job-posting data in a recent report:
As the building blocks for analyzing real-time labor market demand, online job ads have three substantial shortcomings:
- Not all job openings are posted online, distorting the employment picture.
- Deriving an accurate count of job openings posted online, comparable over time, is not yet possible because current technology cannot eliminate all duplicates and find all ads.
- Few online job ads include complete information about desired qualifications.
Traditional labor market data also has built-in weaknesses. The sample size for surveys can be small. Some sources are not updated regularly, which causes a lag in the numbers that are reported. (The most recent EMSI data is three months old.) Federal sites suppress data records when there are only a few establishments in a specific industry in an area.
EMSI does its best to work around each of these issues, but we still need nuance when we work with any kind of data.
To conclude, here’s a comparison of data metrics from a few sources for the “marketing managers” occupation in the New York City metro area. This is a fairly small occupation compared to, say, office clerks or registered nurses, but it’s large enough to give us a sense of the disparity in labor market statistics depending on the source.
EMSI estimates 516 yearly openings – a combination of new and replacement jobs – for marketing managers in the New York City area. Meanwhile, Indeed.com shows 3,840 job postings in New York City for this occupation.
Why are there eight times more postings than openings? The biggest reason is job churn. In a major metro like New York, the amount of movement from one job to a similar job in any given month can be staggering. Turnover tends to be particularly high in sales and marketing positions. But here’s the thing to remember: EMSI’s replacement job estimates (which are a key element to our annual opening figure) take into account only workers who leave the labor market (through retirement, death, etc.) or change occupations. The BLS – the source of EMSI’s replacement data – makes clear that it “does not count workers who change jobs but remain in the same occupation” when estimating replacement needs; there’s no way to make the same distinction with job postings.
Replacements are typically a much smaller number than those who leave one job and get another similar one (see this job churn example from Indiana). Yet, as the BLS mentions in the same section highlighted above, workers entering an occupation often need training; therefore, “these replacement needs, added to job openings due to growth, may be used to assess the minimum number of workers who will need to be trained for the occupation.” Implicit in this statement is the fact that workers who change jobs but stay in the same field do not usually need training. Because of this, churn data is not as worthwhile for training providers to analyze as new and replacement job statistics.
Nevertheless, there’s more to look at than just openings or postings. The Occupation Employment Statistics (OES) program, one of the BLS’ sources for occupation statistics, shows an employment estimate for marketing managers in the New York City area of 14,470 as of May 2010, the most recent data available. If we go to EMSI’s Complete employment dataset, which includes self-employed workers, the 2011 estimate is 18,438 – 25% higher than the BLS figure.
Past trends also show that marketing manager jobs have stabilized in recent years … after a big drop-off in the early 2000s (and during the recession). Furthermore, median wages are much higher in the NYC area ($60.55 per hour) than nationally ($45.78 per hour).
Taken together, these various data elements offer an enlightening perspective for workforce planners and colleges who want to nudge jobseekers in the right direction, or who are wondering if they should set up training for this occupation. But – and here is our point in showing this – if we take any of these sources in pure isolation (particularly the job postings) without further inspection of trends and current labor market conditions, the statistics can be misleading, and we can wind up making poor decisions.
Illustration by Mark Beauchamp.