Measuring the Impact of Apple and the App Economy

We all know about the explosion of Apple tech products, the ever-expanding number of mobile applications in the App Store — and the near $100 billion in cash that Apple is hoarding. Yet one question that has gone mostly unanswered is how many jobs Apple has generated (and supported) with its mind-boggling growth.

Last week Apple released results of a study done by the Analysis Group of Boston — a firm commissioned by the tech giant — to measure its employment impact in the US. The results:

  • 47,000 current US jobs at Apple
  • 257,000 indirect (or support jobs) in manufacturing of components, transportation, professional services, etc.
  • 210,000 estimated iOS App Economy jobs, a figure derived using the same methodology as a study from Mike Mandel for TechNet.

The total estimate — 514,000 jobs created or supported by Apple — has already been criticized by at least notable one economist. But the methodology appears to be fairly conservative, if only because the authors did not include an estimated 187,000 additional jobs from induced effects (i.e., those that come increased spending at grocery stores, restaurants, etc.).

However, it’s the other component of Apple’s total job figure — the 210,000 for iOS developers — that warrants a further look.

The App Economy

Mandel’s in-depth study (PDF) on the App Economy was released in early February. Using want ads from The Conference Board Help Wanted OnLine database, Mandel and his colleagues estimated that 466,000 jobs in the US can be attributed to app firms and app-related jobs at companies such as AT&T, Electronic Arts … and of course Apple. Pretty remarkable considering that, as Mandel wrote, there were zero app jobs in the US prior to 2007. Apple looked at the study and:

Using the same keyword search methodology employed by the study’s authors at the time of its release, we found that 45 percent of job postings in the app economy are specifically tied to iPhone and iOS, indicating that at least 210,000 jobs are driven by the iOS app economy.

There’s a good reason why Mandel relied on want ads instead of conventional labor market data — app jobs only started cropping up in the last five years, making them hard to track down through the BLS and other sources. Concluded Mandel, “… the App Economy is far too new to show up in the government statistics.”

Job posting (or real-time labor market) data is constantly updated and allows for full keyword searches. This is valuable to those who want a very recent look at the most in-demand skills needed by employers in a particular industry or field. But, as with all data, there are inherent shortcomings to analyzing want ads. And more to the point, using job postings to estimate employment in a sector can be problematic. Consider Mandel’s approach:

Our procedure for estimating the number of App Economy jobs has several steps (see Table 1).

1. We identified a set of keywords that characterize want ads for App Economy computer and mathematical occupations, which for convenience we will call ‘tech jobs’;

2. We used historical relationships to estimate the ratio between the number of want ads for tech occupations and the actual level of tech employment [emphasis ours];

3. We examined a sample of third-party app developers to estimate the ratio of tech jobs to non-tech jobs in the App Economy;

4. We drew from the literature to derive a conservative estimate of the spillover effects to the broader economy;

5. We used the location data in The Conference Board database to estimate App Economy jobs by metro area and by state.

Mandel looked at four years of postings data, which suggested “that tech jobs and tech want ads tend to move together, except for anomalous periods such as 2009, at the bottom of the downturn.” So he took the roughly 3.5 million tech jobs (defined as computer and mathematical occupations) in fourth quarter 2011 and the 952,000 tech want ads over the same time to come up with a ratio of roughly 3.5 tech jobs for each non-duplicated tech want ad over a 90-day period.

During the 90 days’ worth of want ads that Mandel examined, he and his colleagues identified 44,400 non-duplicated postings for tech workers that had at least one of the keywords they chose to look for — a list that included “Android,” “iOs,” “iPhone,” “Facebook API” and others. Using the 3.5 ratio, those 44,400 want ads result in 155,000 tech jobs in the App Economy as of December 2011 (not including non-tech jobs in the App Economy).

Perhaps this is an accurate estimate, but a few things should be mentioned here. If Mandel had used a different 90-day period, he might have gotten a larger or smaller number of tech want ads. And a different — or longer — time period might also have shifted the employment-to-want ad ratio up or down.

Job posting data is volatile; it can change depending on the time of year, profession, hiring practices of employers, etc. Some employers post want ads to collect resumes for jobs they don’t intend to fill (or perhaps will fill internally). Also, the method of converting the text in want ads into generic job titles or standard occupation codes through keyword searches — a vital step for high-level analysis of real-time data — can cause discrepancies if not thoroughly vetted or fact-checked. A posting for a “team member” at a fast food joint can be misinterpreted as a “team assembler.” Things like this happen.

Pinning down existing app developer jobs is particularly tricky. A software developer might devote only 30% of the workday to apps or might spend a few months developing an app and move on.

More could be said here, but it’s clear that Mandel’s study provides insight into a burgeoning and heretofore unanalyzed part of the economy. Nonetheless, it’s helpful to have a lens with which to look through a study like this and real-time data in general.

See also: Making a Key Distinction: Real-Time LMI & Traditional Labor Market Data

Comments are closed.