The Fate of STEM PhDs (and More)

My post this evening will mainly involve a new report from the National Science Foundation (NSF) on how well STEM PhDs are faring in the job market, both short- and long-term.  But first I’ll bring in a couple of news items, and then connect them to the NSF, with the common theme of spinning almost any kind of news in a manner aimed at implanting in the American consciousness a positive view of expanded immigration policies.

So, first and foremost, congratulations to Kira Kazantsev, Miss America 2015.  She was my daughter’s classmate in high school and even junior high school, and though I’ve only seen her a few times, it has always been clear that she is headed for big things, and deservedly so.

But I was a little irritated to see the manner in which Ms. Kazantsev’s victory was announced by  “Did you know? The new Miss America 2015 is a first generation American and daughter of Russian immigrants.”  What possible relevance can that have?  It’s just PR indoctrination.  I’ve been critical of in this blog before for their manipulative ways, as many of you know.  By the way, is hiring interns for such work, which is troubling.

Maybe I need my own PR firm, which would make a proud announcement, “Did you know that H-1B critic Norm Matloff is the son of a Lithuanian immigrant?”

My second item before turning to the NSF report is this job-seekers advice column by Laszlo Bock, Senior VP of People Operations (read “HR”) at Google.  (I learned of this from an alert reader, thanks.)   The advice itself is good, but rather at odds with the Google/Bock stance on the labor market.  Bock writes

The toughest part is that for 15 years, I’ve continued to see the same mistakes [in re’sume’s] made again and again by candidates, any one of which can eliminate them from consideration for a job. What’s most depressing is that I can tell from the resumes that many of these are good, even great, people. But in a fiercely competitive labor market, hiring managers don’t need to compromise on quality. All it takes is one small mistake and a manager will reject an otherwise interesting candidate.

Huh?  Does Bock work for the same Google that insists there is a tech labor shortage, so severe that it must hire lots of H-1B visa workers?  Indeed, Bock testified to Congress in support of expanding the H-1B program.  In fact, in his testimony he talked about Google being in competition with other firms, here and abroad, for talented workers — a huge contrast to his statement above that even great workers are competing with each other for jobs, rather than employers competing to hire them.

Bock’s testimony, by the way, is laced with the term foreign-born in referring to the H-1Bs, rather than foreign.  I explained in an earlier post, titled “When Did Foreign Students Become ‘International’?”, that even this is calculated PR; nothing is left to chance.

A couple of days ago, a prominent journalist for whom I have enormous respect said to me, “The H-1B program has strong mainstream support.”  The above incidents show how such support is created by deft PR work and lots of money.

Which brings me, at last, to the new NSF report.  Keep in mind that the NSF has always been pro-H-1B, and it shows here:  The press release headline reads, “Unemployment for Doctoral Scientists and Engineers Below National Average in 2013.”  Wonderful!  But definitely misleading, for a number of reasons.  For example, those who bother to read the actual report will find that the NSF is counting someone as “employed” even if they are working just part-time.

Moreover, the comparison to the national average is inapropriate, since highly-educated people tend to be resourceful types, who will find some kind of job.  But WHAT kind?  The report actually answers that question, and there the picture is not so rosy.  Let’s look at computer science (CS), both because it’s my field and because it is the field with the largest number of H-1Bs.

Begin with Table 2, which shows that nearly 7% of CS PhDs are either working part-time or are unemployed (but seeking employment).  That’s a rather high rate, considering that the industry PR people say CS is such a red-hot field, and in light of the fact that the PR people often point to the high proportion of CS doctorates granted to foreign students by U.S. universities as a reason for hiring H-1Bs.

Things then get worse in the second half of the same table, where there is a breakdown by age.  I’ve emphasized many times that one of the major reasons employers like the H-1B program so much is that enables them to hire young H-1Bs instead of older (age 35+) Americans, and sadly, the table shows employment for PhDs declines markedly with age.

Table 3 compares, among other things, Americans to workers with temporary work visas (H-1B, L-1, F-1/OPT, J-1 etc.).  Look at the dramatic difference!  The percentage who are either working part-time or are unemployed is over 12% for U.S. citizens and permanent residents, but under 4% for the work visa people.  (Most of the unemployed in the latter group are presumably F-1s.)  A large part of this discrepancy is likely due to the age effect, but it is compelling in any case.

Finally, go to another NSF report, in data titled “Table 32-2, Involuntarily Out-of-Field Rate among Doctoral Scientists and Engineers, by Occupation: 2013,”  There we see the remarkable statistic that 11.3% of CS PhDs are involuntarily working outside the field.  Again, consider this in the context I described above — CS is supposed to be a hot seller’s market for employment, especially for PhDs.

The reality is that getting a doctorate in CS is overkill for most nonacademic jobs.  An HP manager told me, “The ‘sweet spot’ is the master’s degree,” and an Intel recruiter remarked, “Intel is not very interested in PhDs.”  Indeed, an HR person at Google told me that CS PhDs hired at Google typically get Software Engineer titles, just like bachelor’s-level workers.  And as I report in my Migration Letters paper, a study commissioned by the Computing Research Association (a consortium of university CS Departments across North America) found that the job market for CS PhDs is tough and will continue to be so.

I’ve often mentioned that a 1989 internal NSF report forecast (and spoke approvingly) that an influx of foreign doctoral students would keep PhD wages down, making doctoral study unattractive to Americans.  That is exactly what has occurred, as noted in the congressionally-commissioned NRC report in 2001, and put bluntly by Cisco Systems Vice President for Research Douglas Comer:  “…a Ph.D. in computer science is probably a financial loser in both the short and long terms, says Douglas Comer”  (Science Careers, April 11, 2008).

If having a PhD doesn’t add much value in CS jobs, then why does the industry hire them?   As noted, they actually don’t hire so many as they want the public to believe.  But the answer also lies in another point I often make about the attractiveness of the H-1B visa to employers — the visa (coupled with sponsorship for a green card, typical in firms like Cisco, Google etc.) renders the worker IMMOBILE, a status of tremendous value, as the employer doesn’t want an engineer leaving for another firm in the midst of an urgent project.  This is often pitched to employers by immigration attorneys as the advantage of hiring foreign students; the ad at the site should be required reading for anyone who wishes to understand the real dynamics of the H-1B visa and employer-sponsored green cards.

Once again, Senator Grassley’s comment about H-1B applies:  “No one should be fooled.”


23 thoughts on “The Fate of STEM PhDs (and More)

  1. @matloff: “a 1989 internal NSF report forecast (and spoke approvingly) that an influx of foreign doctoral students would keep PhD wages down, making doctoral study unattractive to Americans.”

    Does anyone have a URI for “the 1989 internal NSF report forecast,” aka House[1] 1989? Or, for that matter, its subsequent discussion by the Washington Post’s then science writer, Dan Greenberg[2]? For those wanting to “read the docs” online, see Weinstein[3] and Mervis 1991[4].

    @matloff: “That is exactly what has occurred,”

    For one analysis of the doctoral compensation clamp, see Borjas 2006[5]: “A 10 percent immigration-induced increase in the supply of doctorates lowers the wage of competing workers by about 3 to 4 percent. About half of this adverse wage effect can be attributed to the increased prevalence of low-pay postdoctoral appointments in fields that have softer labor market conditions because of large-scale immigration.” Interestingly, news of the “postdocalypse” has reached as far as NPR, who give it the corporate-funded-media treatment here[6].

    [1]: Peter House was then director of the NSF’s Policy Research and Analysis division (since disbanded).
    [2]: Daniel S. Greenberg, “A Shortage of Scientists and Engineers”, Washington Post, August 18, 1991, page C7. Cited by Weinstein[3], who also cites it as “(Wolpe Page 583)”: “Wolpe” apparently refers to Rep. Howard Wolpe’s Opening Statement as Chairman of the Hearing Before The Subcommittee on Investigations and Oversight of the Committee on Science, Space, and Technology, U.S. House of Representatives, One Hundred Second Congress, April 8, 1992. For the funny-but-sick story of the confrontation of House, Wolpe, Edward Tufte, et al, see Mervis 1991[4].
    [3]: (currently: the URI for this paper changes). No date cited, since this remains a “working draft.”
    [4]: : Jeffrey Mervis, October 28, 1991.
    [5]: : George J. Borjas, NBER Working Paper No. 12085, “Immigration in High-Skill Labor Markets: The Impact of Foreign Students on the Earnings of Doctorates”


    • Data showing increased foreign graduation from US graduate programs since 1989 or whenever would also be welcome.

      The claim of an American skills gap suffers from the lack of an account of how and exactly when it came to be—beyond the conservative think tank and Fox News story that it’s related to the establishment of a federal Department of Education affecting K-12—and proponents of that claim ought to be called on that.

      My personal experience and observation is that US hard science and engineering graduate schools `always’ (since the 1930s actually) had substantial fractions of foreign students, and that those fractions are large now doesn’t indicate Americans have dropped out in recent years, since, say, they put a man on the moon.

      What needs to be emphasized to the public at any rate is that the graduate department numbers are not at all representative of the much more numerous STEM baccalaureates, and that in limiting immigration the US is not failing to reap its higher educational investment—at least not before the very recent growth in popularity of US universities among Chinese bachelors’ candidates.


      • See for data on the growth of foreign students in computer science. For all fields, the government Educational Statistics Digest is very useful.

        The issue of K-12 schools is quite complex, far too much so for me to get into here. I’ll simply note a few points, as I have before. The U.S. states without a large underclass so quite well in terms of international test scores. Texas Instruments, one of the most strident firms urging Congress to allow the tech industry to hire more foreign engineers, testified that they have no shortage of qualified American applicants for engineering jobs at the Bachelor’s degree level. The Americans don’t go on for grad degrees because the salary differential isn’t high enough, in turn caused by the large numbers of foreign students — a phenomenon predicted by an internal NSF white paper back in 1989. In computer science, the average quality of the former foreign students now working in the U.S. is lower than that of their American peers (my EPI paper).


    • Many people think the K-12 situation is _very_ simple: there are public schools, teachers’ unions, and there’s federal pressure to lower standards and divert resources to accommodate minorities, so it’s no wonder, they say, we aren’t teaching high-tech skills, a competitive work-ethic, and engendering innovation and entrepreneurship. Even among segments of the public that don’t buy all that, the conviction that there’s a STEM shortage rooted in K-12 problems is unshakable.

      In chemistry, mathematics, physics, and electrical engineering, at least, there were numerous foreign students in US graduate schools in the 1960s, so in these fields that there are currently roughly comparable numbers to then, cannot reflect, and prove the existence of, a more recently developed indifference to, or inability to prepare for, graduate work among American students. (Computer science may be somewhat different in that it grew up more at the same time as guestworkers and immigrants became more prevalent.)

      Moreover, it s untrue that between 1/3 and 1/2 of all public expenditures (i.e., _at all levels_) on STEM higher education go to educating foreign students—as my congressperson among others seems to say she believes, and which much of the public seems to find a persuasive reason for increased guestworkers and immigrants.


      • Again, see that NSF report as to why fewer Americans go to graduate school: The extra salary is not worth the extra years of study and loss of industry-level income during that time. To my knowledge, no study has claimed that American students are not prepared for grad school.

        Foreign student enrollment has increased in all the STEM fields, I believe. As I mentioned, there are government data readily available that you can consult.


    • Foreign students aspiring to employment in the US are equally affected by the opportunity cost of time spent in graduate school.

      The data I’ve seen I think says that in the newer biomedical related fields, and in computer science, the fraction of foreign graduate students has grown as the fields have grown, but in the earlier established, already mentioned fields, where their fraction is still the largest—and which has been cited as illustrating the US’s investment in higher STEM education—it’s been roughly constant.

      It’s important to distinguish MSs from PhDs, and enrollment from degrees awarded; for one thing in some fields people seem to enroll to get a few strategic courses that do matter to their employment, but at least don’t make any haste toward getting even an MS to put behind their name.


      • No, the domestic and foreign students aren’t equally affected. The foreign students have the additional incentive of getting an employer-sponsored U.S. green card, which is of enormous value to many.

        Why distinguish MS from PhD?


      • Ed Coke September 20, 2014 at 11:25 pm: “Data showing increased foreign graduation from US graduate programs since 1989 or whenever would also be welcome.”

        See completion data below.

        Ed Coke September 23, 2014 at 10:32 pm: “The data I’ve seen I think says that in the newer biomedical related fields, and in computer science, the fraction of foreign graduate students has grown as the fields have grown, but in the earlier established, already mentioned fields, where their fraction is still the largest—and which has been cited as illustrating the US’s investment in higher STEM education—it’s been roughly constant.”

        My expertise in this field is strictly amateur (Dr Matloff presumably knows much more), but DuckDuckGo[1] is my friend 🙂 From what I can (cursorily) see, the problem is that most of the current relevant data stops in 2009-10. I do see the NSF’s “Science and Engineering Indicators 2012 Chapter 2. Higher Education in Science and Engineering”[2], which discusses total enrollments by year, not fractional change (which is what this discussion is about). However, that report *does* make datatables available, so someone could (and I suspect already has) combined them so as to generate the data wanted here (change in foreign graduate student proportion over time). It also says

        > Foreign students on temporary visas earned a larger proportion of doctoral degrees than master’s, bachelor’s, or associate’s degrees (appendix tables 2-17, 2-19, 2-26, and 2-28). The temporary residents’ share of S&E doctorates rose from 30% in 2000 to 33% in 2009. In some fields, foreign students earned sizeable shares of doctoral degrees. In 2009, foreign students on temporary visas earned half or more of doctoral degrees awarded in engineering, physics, computer sciences, and economics. They earned considerably lower proportions of doctoral degrees in other S&E fields, for example, 29% in biological sciences, 8% in medical/other life sciences, and 7% in psychology (appendix table 2-28).

        Ed Coke September 23, 2014 at 10:32 pm: “Foreign students aspiring to employment in the US are
        equally affected by the opportunity cost of time spent in graduate school.”

        But cost is only one part of decisionmaking: the other is benefit. I suspect the data above supports the hypothesis that foreign students see increasing return relative to cost for time spent in US graduate school than do US students, which I suspect is due to

        * for US students: the increasing cost of education relative to income, as assistance for graduate education declines, and median net graduate education cost increases faster than median income

        * for foreign students: increasingly justified belief that graduate study improves chance of acquiring US residency and citizenship



  2. Yah, I commented on this Google guy’s same post over on LinkedIn, that it sounds like they need better managers, if they throw away good resumes because of a typo. This is Google’s *real* attitude on hiring, hire the ones who most eagerly jump through hoops – and make managers out of the ones who make the peons jump through hoops.

    Google follows such strange business models, the myth is one thing the reality is another thing and neither makes much sense. Back in their heyday it was Xerox that had this reputation – they were a goofy and nonfunctional company but *something* about the product was so good they made money anyway. Google has done some amazingly good work, some in their “AI” of matching question keywords to answers, some in bigdata engineering, and some in gigascale server-farm engineering and networking, and they give away tons of stuff of immense value.

    But they hide almost all the details of how they actually get the good stuff done. For such a large and dominant company their internals are amazingly obscure. Then you get noise like this post from their HR guy carrying on about typos in resumes.


    • I’m glad to see you didn’t include self-driving cars as one of Google’s great accomplishments. 🙂

      In terms of getting better managers, the poor hiring policies are standard for the industry, not just at Google, as you know. The real issue is that they just don’t want to take the time to look carefully at the applicant pool, and as Bock himself says, there really is no need to so, given their huge number of qualified applicants.


      • I think there’s an anthropological, social, psychological – whatever term you prefer – aspect to it all, the industry *prefers* bad managers and mediocre workers. They would rather have a manager making $120k and ten workers making $80k (total $920k), then a manager making $200k and three workers making $150k (total $650k) – even if the smaller group would get the work done twice as fast, multiplying the difference. I’m just pulling these numbers out of nowhere, fiddle them as you like, we all know the situation.

        To be fair it is more difficult to build and run an elite staff. It is also more profitable. The risk/reward profile just does not appeal to top management these days, they view the actual R&D work as overhead, not productive asset. There may be some traditions along those lines, but they are pre-tech conventions.

        It does happen, you know, but usually only in the formative days of new startups, and the compensation is more usually contingent – founders stock and options. Few startups succeed *without* doing this. But once the prototype is running watch out, the next guys in are subject to the standard process.


  3. Senator Grassley is right.
    What is really the most troubling thing is that so many people ‘do’ believe it.
    Zombie facts with years of accompanying obfuscation of the truth by the press and the industries that control it, have created the situation that everyone knew it would 25 years ago.
    Our university system is a joke, our ability to innovate as a nation is gone and no American child wants anything to do with STEM.



    ‘The Ph.D Bust: America’s Awful Market for Young Scientists—in 7 Charts
    By Jordan Weissmann

    Politicians and businessmen are fond of talking about America’s scientist shortage — the dearth of engineering and lab talent that will inevitably leave us sputtering in the global economy.

    But perhaps it’s time they start talking about our scientist surplus instead.’

    ‘Most readers here know how brutal the employment situation is for chemists (especially those involved in drug discovery). Knowing that and seeing constant headlines about the crippling shortage of so-called STEM workers is always hard to take’

    ‘New Ph.D. recipients, welcome to the next stage of your career! It’s a time-honored tradition of academics, research, and stagnant purgatorial nonprogression. That’s right: It’s your postdoctoral appointment!

    As a postdoc, you’ll contribute vitally to the progress of science, simultaneously filling the roles of scientist, scholar, and sucker. But first, a few basic facts about postdocs:

    Postdoc positions were established to prolong the awesomeness of graduate school, which everyone loves, while simultaneously postponing the ability to make money, which everyone hates.

    …For every available opening for a tenure-track professorship in the United States, there are roughly seven postdocs. That’s not an exaggeration or a joke.

    …Unfortunately, many postdocs are treated like glorified lab techs … and it’s very sad that you felt a little good just now about the “glorified” part.’


    • A lot of the public hears all this and takes it as the whining of the losers: the insufficiently talented or earnest wannabe STEM workers who can’t cut it in the modern, technically advanced, admittedly demanding of near perfection, world.

      What do `we’ say about that?

      They try to shut you up as a spoiler: discouraging—even very dangerously `corrupting’—the youth, who rather need `positive’ role models.


      • I don’t hear that much at all. However, it could be that the industry lobbyists are saying this kind of thing in private on the Hill.

        Over the years, I’ve given a number of examples of top-quality people who have trouble getting work in tech.


    • You don’t hear it because you have a pretty good job, Norm.

      There have been discussions where some of the points in Kennedy’s post are raised, and academe and industry representatives present say modern science and high-tech is by its nature very demanding: there’s just no tolerance for error or space for less than excellence.

      As a practical matter it’s difficult for those affected to speak out and not end up making fools out of, or discrediting, themselves, and the point they are trying to make.

      You may say top quality people have trouble getting work, but others will say having trouble getting work in this context we have of a severe and desperate labor shortage, is prima facie proof positive of failure even to meet minimal levels of quality.


      • Ed Coke September 22, 2014 at 12:42 am: “others will say having trouble getting work in this context we have of a severe and desperate labor shortage, is prima facie proof positive of failure even to meet minimal levels of quality.”

        Which raises the more general question: how to deal with those who (claim to) assume the truth of a claim which one knows to be empirically false (or at least questionable)? The academic term for this is “affirming the consequent,”[1] so I’ll use “AC” ot refer to the behavior.

        The predominant cognitive bias among the “nice white progressives” with whom I regularly interact is what I call the “Platonic bias”: the belief that bad behavior (aka evil) is due to lack of knowledge[2], and that lack of knowledge can be cured by education (or some sort of magic or PR). Unfortunately for the Platonics, it seems empirically true not only that many people act on the basis of false beliefs, but that, for many people, confrontation with negating evidence fails to negate, or even *strenghens*, the false belief.[3] It is also true (though much harder to establish empirically) that some people will knowingly make false claims, or can convince themselves to believe falsely, for their own advantage.[4]

        Accordingly, I claim there are some combinations of {people, topics, time} for which rational discourse is futile, and that the best available strategy for discourse with opposing parties who AC resembles the following pseudocode:

        challenge OP to produce evidence *for* the claim under discussion[5]
        if (OP produces persuasive counterevidence) reconsider my position
        if (OP seems genuinely motivated to attempt to produce counterevidence) wait
        cease discourse on this topic with OP
        if (OP’s false belief is probably inconsequential[6]) ignore
        act to disable OP from acting on false belief[7]

        [1] See for a brief discussion. For some interesting examples of AC, see and .
        [2] While nuanced in several dialogues, Plato attributes to Socrates the belief that “true knowledge” produces virtue, esp in the Protagoras.
        [3] This has been called the “backfire effect.” For a survey, see
        [4] The classic modern study of this effect in science-related discourse is IMHO Oreskes and Conway 2010: see
        [5] Rather than just reasoning from it. E.g., what evidence supports the claim that most STEM PhDs are completely unqualified to work in the field?
        [6] E.g., if OP has no power, or is probably disinclined, to act on their false belief.
        [7] E.g., separate OP from their power to act in the domain affected by their false belief. Easily said 🙂 but must be done. To paraphrase Burke, all that is necessary for a bad policy to be made is for those who know better to allow it.


    • People have always claimed they are better qualified for the corner office than their boss: there’s a tendency to suspect some puffery and apply some deflator when someone says they are not employed at the level due their skills, even though it’s not strictly logical.

      I’ve seen discussions where essentially demoted American workers were given short shrift and humiliated, but where the self and bitterly proclaimed plight of employers who needed only a dozen or so tech workers, and who were disadvantaged in the H1b lottery relative to Microsoft and Wipro who apply for visas by the hundreds, was treated seriously and with sympathy.

      Krugman remarked yesterday in the NY Times that that almost everyone knows someone who’s had employment problems sometime during the last 6 to 8 years, apparently doesn’t discredit the assertion of John Boehner that laziness or `leisure preference,’ and choosing to freeride, has been responsible.


  5. I was told that a previous employer specifically required a Master’s degree because he knew it would weed out American workers. He had other ploys he would use, and would pass everything by his law firm to make sure that he got what he wanted. Last I heard he was pushing 100 developers over a 15 year period, and not a single American developer has been hired! That is mathematically impossible!

    I am not a “techie”, and it was common for our immigrant managers to try to hide or overplay how hard it would be to add or modify (change request) certain product features. So we hired a summer Intern to help us out from Harvard, something I pushed for. I had asked for some small modifications to our software application, and had been delayed for over 9 months. As an aside, I mentioned these items to our new Intern, who was working in a non-technical position. One change was simply adding a left-justify change to a certain web client (page) for text. A second was to be able to upload a file to our evolving web client. Our Intern had had one or two programming classes, that’s it. So he cracks open a book, and 2 days later he asks me, “Is this what you are after?” He had added both items! I was ecstatic, and the managers fumed!


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s