Part 2: This is the second in a series of posts that provide insight on real-time job postings (see Part 1 here). In this post we discuss the findings of our side-by-side analysis of seven months of traditional labor market data and real-time information. Our research revealed real-time job postings can be a good leading indicator for future job growth (while by no means being indicative of actual growth), and that jobs postings are most concentrated in tech-intensive fields.
By Brian Points
EMSI Consulting Economist
EMSI gets a lot of questions about real-time job postings information — questions like, how does it work? How valuable is it? And increasingly, how does it compare to traditional labor market information? A few other organizations, such as Jobs for the Future and The Conference Board, have laid some initial groundwork for analyzing this topic, but no one has actually compared the two datasets side-by-side on a month-by-month basis. Which is exactly what we set out to do in this post.
For our analysis, we used EMSI’s proprietary employment data, which you can read more about here. The data were compared on a month-to-month basis from July 2012 to January 2013.[1. Monthly data was used for the purposes of this analysis. For users of our subscription-based tools, jobs numbers are available as annual averages.] EMSI’s data considers job creation due to economic growth (i.e., new jobs) and job creation due to attrition (i.e., replacement jobs).
Our initial research revealed the following facts:
1. Traditional LMI Covers Nearly All Workers; Real-Time Job Postings Measure a Small Slice of Labor Force
The underlying data sources for real-time job postings and traditional LMI are very different. Traditional LMI, such as that from EMSI’s Analyst tool, is based primarily on federal government sources that measure nearly all of the United States labor force. Real-time sources, on the other hand, only look at a relatively small slice of that labor force. For example, the Quarterly Census of Employment and Wages (QCEW), the backbone of EMSI’s dataset, considers 106 million wage-and-salary employees working for roughly nine million business establishments. Even the best of the real-time sources available capture no more than 60 million job postings.[2. QCEW Comparision Notes: http://www.bls.gov/cew/cewbultn10.htm#coverage]
2. Job Postings Far Outnumber Actual Created Jobs
When we considered all occupational types, we found jobs postings were five times higher than actual job creation over the seven months of data we analyzed, and were larger than actual job creation for every individual month (see Figure 1).
3. Job Postings Are Clustered in Tech and Professional Services Industries
Job postings are also much higher than actual job creation for technology-intensive jobs and business and professional services jobs. For example, the ratio of job postings to job creation for computer occupations is 38-to-1; for engineers, it is 22-to-1 (see Figure 2).
4. Production Fields and Others Have Far Fewer Postings
On the flip side, there are also many employment categories where real-time data are very small in comparison to job creation. The ratio of job postings to job creation for production occupations (i.e., manufacturing workers) is 1-to-2.4 (see Figure 3).
5. Occupations With Higher Wages (and Higher Ed Levels) Have More Postings Relative to Jobs Created
Job postings are not an unbiased reflection of job creation in other ways as well. For example, occupations with higher average wages, and occupations with higher average educational levels, tend to have a higher level of job postings relative to job creation.
We also noticed several points for data users to keep in mind:
- Given the seven months of data analyzed by EMSI, real-time data does have a statistically significant relationship with job creation. Therefore, as long as the data is interpreted correctly, real-time job postings can serve as a good leading indicator for future job growth.
- Real-time data primarily measures intentions, whereas traditional LMI measures actions. There are numerous reasons why a job posting may not be translated into an actual job. One of the more prevalent is a process called “bench-building,” in which companies keep postings up for months at a time just to continue collecting résumés. At some point they intend to hire but the number that they hire will likely fall far short of the number of postings that can be counted over that time series.
- Under no circumstances should real-time data from one period be compared with traditional LMI from a previous period to indicate “growth.” Doing so is like comparing the ranking of a university from two different evaluating agencies (such as Forbes and US News and World Report) from two different time periods.