I think it's possible that the term "STEM" obfuscates the problem. I don't believe believe there are any shortages in the maths and sciences. There's not a lot of science graduates in the school I graduated from to be sure, but there wasn't exactly a lot of demand for them either. Engineering is a wash -- it really depends on the discipline. Comp Sci guys though are in high demand -- basically every CS grad from my uni had job offers before they graduated, and it's not a prestigious school or hi-tech hotspot.

Maybe this is the money quote (from one of the links)

More than 40 percent of recent U.S. college graduates are underemployed or need more training to get on a career track, a poll released on Tuesday showed.

I'm not from the US, so I can't speak for the US education system, but in my country, there is a clear development towards the stance of "leaving nobody behind", especially in schools. While universities are still largely unaffected from that, getting a university degree means that you know the basics, and have the endurance to eat whatever bureaucratic crap you have to endure to get the degree, but not that you are fit for a job in your field. As far as I can tell, most employers see the university degree as a necessary, but not sufficient qualification. And for good reason. Just like every student, I have had to work multiple university projects in random teams, and the skillset ranges from people who could build the next big GPU arch to guys you wouldn't entrust with a sorting algorithm.

Those 40%, that are having trouble finding a job, can probably be split into two groups: Those that are capable, but have nothing to prove it (private projects, internships, ...), and those that are truly clueless. In both cases, them not getting hired is not an indication for the absence of a "STEM shortage", but for a failure of the education system.

Ok, to clarify: What I meant in my previous post is that there may or may not be a "STEM shortage" in the US, but that the number of college graduates, that are having troubles finding a job, is a bad metric for it. The number of open job positions would be a far better indicator.

As for the "catch 22" situation: Yes, it definitely is a problem. In some fields more than in others. While you can fill your portfolio with the small computer games that you wrote at home, there is no way to build an international airport in your backyard. But I believe, that employers in the latter field are also more likely to understand the lack of prior projects, then for example, an employer in the video games industry.

I'm still on my way out of university, so everything I say should be taken with a grain of salt, but my advice to everyone is to try and get a small *paid* part time job in your respective field for the last couple of years at university. IMHO getting paid is important, not for the money, but because it forces your employer to give you productive tasks. If possible, also do some private projects to fill your portfolio. In that respect I think that we (comp sci and comp engi) are really privileged, because most things, that are expected from us on the job, we can actually do at home.

You have to remember, that hiring the wrong guy can do significantly more damage than keeping the position open. As a part time or low wage intern, the amount of damage that you can cause is rather low, so it's a good way to get your foot into the door and build up a track record and connections, so that when you do graduate, you have more than a degree to show, that you are indeed the "right guy".

In the UK we have a supposed STEM graduate shortage as well. On the whole, STEM graduates can get higher paid jobs in unrelated fields. This feeds back into the subjects people choose to study. There are also frequent calls by companies that they need to import foreigners to do this work because of the supposed shortage (they'll generally work for less money). All sorts of education initiatives are being put in place to encourage more people to take up STEM subjects but the real problem we have is that companies are unwilling to pay STEM graduates salaries that are competitive with what they can achieve in other fields. Until that happens, I can't see the 'problem' going away.

It is both in my view.

There are too many people getting degrees in fields with limited usefulness. I knew some people getting degrees in medieval literature, and there isn't much of a market for that. University education is not job training, but some topics have better employment prospects than others.

On the other hand, companies have shifted from where they were years ago. Too many employers are looking for drop-in replacements and refuse to train anyone. Some fields like medicine and teaching have requirements for continuous training. In our field most employers refuse to provide continuous education, and then just fire the old team and hire a new team with updated skills. Perhaps a few people can get trained on newer technologies on the company's money, but most of us are expected to learn on our own time. Too many employers look for exactly-experienced workers who have been programming in old technology up until yesterday, but also have five years with the new technology.