Job Posting Analytics: Harnessing Multiple Sources for Maximum Clarity

By Isabella Minudri and Clare Coffey
January 22, 2020

Job posting analytics offers a wealth of insight into supply and demand in the labor market. But the way supply and demand interact in the market can be more complicated than we might expect. That’s why comparing real-time job postings with more structured BLS data can offer a composite picture richer than either source could offer on their own.

For instance, the US is facing an urgent shortage of truck drivers. And not just any truck drivers–specifically long-haul drivers, who transport goods such as Coca-Cola products, building materials, and Amazon packages across the country and deliver to your local vendor. 

According to the American Trucking Association, over 60,000 drivers were needed at the close of 2018, a number that is threatening to snowball in the coming years. But why is this shortage such a big deal? Are truckers really that important to our economy?

In short, yes. The backbone of success in our economy is born from the movement of goods from coast to coast, a service provided almost exclusively by heavy and tractor-trailer (long-haul) drivers. 

In some cases, we face a shortfall for some important service because the providing industry is shrinking due to financial problems–less money equals fewer jobs equals fewer drivers, and so on. But the demand for truckers hasn’t diminished in any significant capacity over the past few years, demonstrated by the sheer number of job postings advertised–in 2019, there were over 17M total postings for this industry. The problem actually lies in finding sufficient drivers. 

The map below can help you visualize how the number of postings for trucking occupations, shown in dark blue, far outnumbers postings for all other occupations, shown in green. 

Source: Emsi Burning Glass. Showing data for August-November 2019.

But it’s important to remember that this map only shows the mass of postings, and does not necessarily reflect an equal number of hirable positions. 

How job openings affect the data
The Bureau of Labor Statistics’ Job Openings and Layoff Turnover Survey (JOLTS) estimates the number of openings within each industry. 

For JOLTS, a position is considered open if it meets these three conditions: 

• a specific position exists and there is work available for that position. The position can be full-time or part-time, and it can be permanent, short-term, or seasonal, and
• the job could start within 30 days, whether or not the establishment finds a suitable candidate during that time, and
• there is active recruiting for workers from outside the establishment location that has the opening. 

When we cross-reference JOLTS with Emsi Burning Glass job posting analytics, we gain valuable insights into the character of the industry in question. There are three major ways (and of course, nearly infinite permutations) in which these two data sources interact.

More postings than openings
The chart below shows the number of job postings in the transportation and warehousing industry (which is largely populated by trucking occupations) in green, and the number of open positions according to JOLTS in black.

Source: Emsi Burning Glass. Showing data for January 2017-July 2019.

We can see that from January 2017 to July 2019, the average number of job postings in the transportation and warehousing industry was ~1.4M per month (green). But according to JOLTS, the number of hirable openings for the transportation industry during this two-year period never reached more than 360,000 per month (black). The difference between these postings and openings–almost a 75% margin–is staggering.

In an effort to fill a growing hole in the industry, trucking companies have aggressively distributed and cross posted job advertisements across many job boards. Even if only one or two positions are actually open for hire, a company might have, say, 100 postings for that position advertised in other cities.

These aggressive recruitment practices within the transportation industry and trucking occupations reflect the underlying market realities such as lack of labor supply, high turnover, and the difficulty in always being on the move. Job distribution and cross posting are common when talent or labor is hard to find, as job postings can be relatively effective.

Similar numbers of job postings and openings
One industry that exhibits consistent success in attracting talent through job postings is healthcare. In January of 2019, JOLTS recorded 1,158,000 openings for occupations such as registered nurses and general practitioners, and Emsi Burning Glass JPA data found 1,120,342 postings for similar positions. Though healthcare openings are slightly higher than the number of postings over the years, the disparity between the two is minimal, suggesting that the difficulty of finding talent is lower than what we saw with truckers (NOTE: finding talent is tough wherever you are – the data seems to suggest that the trucker problem is rather more extreme or pronounced as compared to something like nursing, where there seems to be more supply). It’s worth noting a spike in postings around July 2019, which may mean that this dynamic is beginning to shift as a generation of medical professionals retires. If the trend continues, it’s evidence that the labor market for healthcare is beginning to tighten up. 

Source: Emsi Burning Glass. Showing data for January 2017-July 2019.

Fewer postings than openings
Industries that can rely on a local talent pool and a strong word-of-mouth network may not need to post heavily, or even regularly. For example, the accommodation and food service industry, shown below, tends to be underrepresented in job boards on a regular basis. Occupations such as waiters and waitresses, cooks, hotel clerks, and housekeeping cleaners are often hired through means other than job posting sites, a pattern reflected in job posting data. A comparison of JPA Analyst and BLS JOLTS shows that, with some important fluctuations, the industry tends to have more openings than postings.

As with trucking, the relationship of job postings to actual openings offers insights into the characteristics of the industry in question. Here, it indicates both word-of-mouth hiring practices and a workforce that is, by necessity, local. Furthermore, fluctuations over the course of a year highlight the seasonal rhythms of the industry.

Source: Emsi Burning Glass. Showing data for January 2017-July 2019.

We can see here that there were just over 1 million openings for this industry around May of 2018, but less than 700K postings for those positions. Postings accelerate in number near July of each year, and then drop down again for the winter months. This reflects seasonal shifts in demand that hospitality businesses tend to experience–in most cases, the slower summer months prove a good time to hire and train, while the busier winter season sees less turnover.

Deduplication and beyond: Emsi Burning Glass’s approach to varying posting behavior
Disparities between postings and openings will always occur. At Emsi Burning Glass, our approach to these disparities is two-fold. First, we trim away repeated postings as much as possible. 

We home in on the number of positions that are actually available by churning job postings through a deduplication process: wherever multiple postings are advertised for a single open position, we reduce those postings down into a single, isolated or “unique” posting. 

Second, we cross-reference job postings data with JOLTS to show what incongruities still remain. Because our deduplication process removes much of the repetition standard to online postings, the multiple postings that remain represent valuable information about the state of demand. 

For instance, when companies begin to post in more than one city, it is difficult to discern if the postings represent multiple openings in multiple cities or one opening in a unique location–but the fact that companies are posting in multiple cities is itself worth noting. In the case of trucking, any attempt to pin down a single origin for any given job is difficult: long haul positions, by nature, do not reside in any particular city. Deduplicating these trucking postings would also remove the indication of real opportunity for employment in that region.

Conclusion
The incongruities in posting behavior revealed by comparing JPA with JOLTS offers valuable insights into both the nature of the industry and the challenges it may be facing.  When we see particular industries such as transportation & warehousing posting millions of job advertisements every month with relatively few open positions, we can infer that they’re having trouble meeting their talent needs. On the other hand, if we see enormous posting numbers complemented by an equal number of open positions, we might instead predict a high rate of turnover. A steady, moderate rate of both postings and openings suggests an industry at equilibrium. And a drop or spike in posting behavior across industries might indicate a broader economic trend, or a talent issue that affects the labor market as a whole.

We are constantly striving to provide accurate, organized data to our client base, and that includes transparency in our methodology. As job postings continue to be a driving force in the labor market, we encourage our customers to avail themselves of the multiple complementary data sources we offer, in order to create the most representative, nuanced, and insightful picture possible.