How Many Programmers Does The US Economy Need?

October 9, 2023 Timothy Prickett Morgan

Throughout history, we strongly suspect that people always thought they lived in peculiar times. The difference is, we say with tongue firmly in cheek but also with a raised eyebrow and a pensive look, that we really do live in peculiar times. The others before us just thought they did.

Take the IT job market, for instance. Or leave it. Never before has information technology been so instrumental to the conduct of business as well as our lives outside of work. And never before has there been a force like generative AI that can put downward pressure on wages and make programmers question how they do their jobs and the value they bring to those jobs. How much automation is feasible or desirable? IT shops all over the world are trying to figure that out.

At the same time, inflationary pressures throughout the global economies are making employees ask for raises. And sometimes they are getting them. Workers at United Parcel Service, which is in a fight for its life against an Amazon that is building its own transportation network, in August got a 35.5 percent increase in pay for part-time workers, to $21 an hour, and full-time workers now average $49 an hour. At the end of this five year deal, UPS drivers will average $170,000 in pay and other compensation. This has emboldened the United Auto Workers union to negotiate very hard with Ford, General Motors, and Stellantis (the merger of PSA Group, Fiat, and Chrysler).

Speaking very generally across all jobs, the compensation and job placement experts at Robert Half have just released their 2024 Salary Guide and offer some insight. Some 63 percent of workers polled say that they plan to ask for a raise by the end of year, with inflation being cited as the main reason (39 percent of those polled), and taking on more responsibilities (26 percent) and feeling underpaid after checking the market (16 percent) being the two other big drivers of the ask. About the same percent of workers (62 percent) say they would stay in a job with lower salary if they have a more flexible work schedule. Let that be a note to business managers everywhere. Where we do the work, particularly for white collar jobs, doesn’t matter as much as that we do the work.

As for generative AI, four out of ten workers polled by Robert Half say that this new technology will have a positive impact on their careers. That does not necessarily mean that six out of ten think it will be bad, so be careful about jumping to that conclusion. Interestingly, 63 percent of the people in the IT department and 54 percent of the people in the HR department think that generative AI will create demand for their skills. Gen X and Baby Boomers – I am literally on the line between the two, either the first Xer or the last Boomer – are less optimistic about AI (in a general way, not just restricted to generative AI models), with only 30 percent of Xers and 21 percent of Boomers being sanguine. By contrast, of those in Gen Z that were polled by Robert Half, 63 percent were optimistic about the benefits of AI, and 57 percent of Millennials said they were optimistic, too.

Let’s talk a little more specifically about jobs. The Bureau of Labor Statistics, the number-crunching part of the US Department of Labor, just released its Employment Situation Summary report, which is does on the first Friday of every month covering the jobs gained and lost by industry and sub-segment across the US economy in the prior month just ended. You can see the October jobs report here, and when November rolls around, that link will always show the current report. I used to track this jobs report like a hawk during the Great Recession and do my own reporting on this, and I was never sure if the raw data or the seasonally adjusted data was more relevant to any current situation.

If you want to get a feel for the IT segment, you can build a proxy of sorts from the combination of job gains or losses in the telecommunication, data processing and hosting, content provider, computer systems design, and other information services sector. This is not a perfect analog because some IT jobs are at manufacturers even if the employees are not engaged in manufacturing. Companies are classified by industry, in a very general way, but they often participate across sectors and subsegments at a finer granularity. Which is why I have always said you need to track jobs by title, not by employer classification. I want to know how many programmers, system administrators, network engineers, system architects, and IT directors and CIOs there are right now, and how the count changes each month historically.

You would think that the Department of Labor would want to know the same thing. If I was President. . . . Let’s not even go there.

Surveys of businesses, what is called the establishment survey data, showed that the US economy created 336,000 jobs in September, which is a pretty strong showing considering how the Federal Reserve is raising interest rates to try to slow down inflation, and that usually weakens the job market. The average gain in jobs over the past twelve months was 267,000. So this is quite a bit better than average.

By industry, the leisure and hospitality sectors added 96,000 jobs, governments at all levels collectively added 73,000 jobs, and healthcare companies added 41,000 jobs. These accounted for 62.5 percent of all new jobs. Now, we don’t know how many of those jobs were actually IT jobs, and that is because the Department of Labor needs someone to bring it some coffee so it can wake up, as we pointed out above.

The good news is that the household survey data, the other dataset the BLS creates each month, showed the unemployment rate steady at 3.8 percent and the number of unemployed people – meaning those who are looking for work and who can’t find it but not including those who have stopped looking – stood at 6.4 million. The number of long-term unemployed people was 1.2 million as September came to an end, and there were an additional 5.5 million people who want a job but who are not in the workforce actually looking for a job at the moment.

To get that proxy summary of IT jobs, I am too lazy and I just got see what Victor Janulaitis, the owner of Janco who tracks IT employment, has to say about it. If you look at his summary reports, which go far beyond the analysis of the BLS report each month, the “IT market” as the BLS defines it had 3.22 million employees in 2000 and then bled down to 2.65 million employees in 2004, had some wiggling up and down around 3.65 million in 2015 through 2017, hit 3.83 million in 2019, dropped to 3.8 million in 2020 thanks to the pandemic, and as September 2023 came to a close was at 4.19 million.

We know that information technology companies have shed lots of jobs as we came out of the pandemic, but these are not necessarily IT jobs that were shed. In fact, the odds favor them being sales, marketing, and other back office jobs.

All of this data sounds like we know something, but these are only indicators of things and not the things themselves. That is now what the Department of Labor or the Department of Commerce are supposed to do. They are supposed to gather information sufficient to help better guide the economy or at least more accurately describe its state at any given moment so individual companies can try to reckon what they ought to do. If we tried to do server or network performance with this kind of proxy information in the datacenter, we would have very little idea of what is going on with the applications. Something bad would happen, and it would cause a resume generating event for sure.

This is, of course, unacceptable.

Here is what we need to do. We need an official list of job functions and when we employ someone, we have to pick one of these official job functions for each and every employee and track how their job functions change across companies. The data has to be anonymized by company and employee, of course, and self-employed people and those working at small companies cannot be exempt, either.

The next thing we need to do is compile all of the job openings that organizations have open by the same job function categories. We can’t infer this from the job sites because there would be double, triple, quadruple counting. We need to know how many of each they have and how many more they need of each job.

Now, assemble that once a month, just like the jobs report does. Now, we know how many programmers or plumbers or truck drivers we have and how many more we think we need. This is a useful set of information. And when the employee counts at companies go down, we will see it automagically. We will be able to see precisely where we lost jobs and where we did not. And this data can be helpful as all of us plot out courses through the global economy in our lives.

That is what data is supposed to be about: Being useful. What we have right now that passes for data is just simply not. I just logged into Claude 2, the generative AI program created by Anthropic, a company that just got a big influx of money from Amazon Web Services, and the data is really thin and speculative. As Claude 2 put it:

• According to Evans Data Corporation, there were 23 million software developers worldwide in 2019. This includes both professional developers and hobbyist/student developers.
• The IEEE estimates there were more than 20 million software developers worldwide as of 2016.
• On Stack Overflow’s 2021 survey of over 80,000 developers, the three countries with the highest representation were United States (21.7 percent), India (12.4 percent), and Germany (6.1 percent). This suggests India and the US likely have the highest developer populations currently.
• In the US alone, the Bureau of Labor Statistics estimates there were about 1.6 million software developers as of 2020. This doesn’t include web developers and other programmer roles.
• Estimates suggest China has around 4 million to 5 million software developers and India has around 5 million as well. However, the data is not very reliable.

So while there are no definitive counts, estimates generally put the global developer population in the range of 20 million to 30 million currently.

Thanks Claude 2. We were going for ironic, using generative AI to talk about generative AI and trying to answer the question – how many programmers does the US economy need? – that ought to be easier to answer than it is.

For the record: Whenever I use generative AI, it will be in red italics. I do my own work and my own thinking, and that will never change.

RELATED STORIES

IBM i Salaries: Underpaid, Yet Highly Valued And Hard To Replace

The Real IBM i Legacy Is The People

The IBM i Job Environment, It’s a Changin’

Innovating And Thriving with IBM i

So You Want To Do Containerized Microservices In the Cloud?

In Search Of Next Gen IBM i Apps

COVID-19: A Great Time for Application Modernization

Club Alan: Seiden Signs Liam, Corners Youth Market

RPG Programmer Shortage Blamed For CSC’s Earnings Miss

There Is No Lack Of RPG Programmers, IBM i Community Contends

IBM i Skills Shortage: Now You See It, Now You Don’t