Thursday, April 25, 2013

Cowen on the EPI report

Well the report has made it to perhaps the most important node in the entire economics blogosphere: Tyler Cowen. He has good thoughts that in a general sense I fully agree with, but I'd make a few additional points (which regular readers should be familiar with by now). I'll reproduce in full:

Via Matt, Hal Salzman, Daniel Kuehn, and Lindsay Lowell suggest there is no shortage because we do not observe significant wage rises for STEM workers. That’s a fact worth knowing and I am happy to praise the study in that regard. But it’s painting the basic worry into too narrow a box with its use of the word “shortage,” interpreted so literally.

The core claim is that STEM sectors will be those which produce the future social increasing returns for the economies which house them. If true (I am not trying to prejudge this), that means we should invest in both more STEM workers and more complementary inputs, whether that be particle colliders, NIH funding, the right broadband infrastructure, legalizing driverless cars, better IP law, tougher schools, or whatever. With the new, additional STEM workers, and the complementary inputs, America will (supposedly) be much better off.

To point out that the current supply of STEM workers stands in proper proportion to the other inputs suggests only that we are at a local optimum, not a global optimum. Similarly, it could have been pointed out that, before the rise of Hyundai, South Korea had just the right number of auto workers (not many) for their factories (also not many). That could have been true enough, but still investing in more auto factories and more auto workers was for Korea a very good path forward.

On top of all that, the report shows a worrying lack of concern about the notion of an economic margin. Even without boosts in the complementary inputs, more STEM workers still can be put to good use, even if there is no “shortage” today.

Here is a related post by Adam Ozimek.
So my research interest is the STEM labor market. There's lots of good work on R&D investments and other elements that go into science and engineering, but my value-added is on the labor market side and I personally think it's fine to talk about shortage claims in that market.

What he's really referring to is similar to what Arrow and Capron called a "social demand shortage": we simply think there ought to be more science done, period.

I agree we have such a social demand shortage, and I think science policy should be decided accordingly.

But the question for me is - exactly where is there a problem in the market provision of this socially optimal level of science? Generally speaking, all of the market failures economists identify that pertain to science and engineering are on the demand side, not the supply-side. Most of what Tyler lists are inputs (supply-side factors). Some are examples of demand (NIH funding, for example).

It seems to me there's no great market failure to speak of in the STEM labor market. It operates fine to meet the level of demand. There's no great market failure in the broadband manufacturing and installation market. The market works fine - it's problems on the demand side that are the hold up. The Super Collider wasn't shut down for supply side reasons. We could build the thing. It was shut down because the demand for it fell through. Why aren't we on Mars today? Not because we don't have the technology - because of the lack of demand.

There are lots of demand-side market failures associated with science that most of you are probably aware of. That's where the emphasis needs to be, IMO (and that includes NIH funding and the public decision to build things like supercolliders that Tyler mentions).

So that's my big-picture policy statement on STEM. Aside from big-picture stuff, though, I'll generally be talking about labor markets. That shouldn't be construed as me thinking that the other stuff is less important.


  1. God dang it you nailed it buddy, just nailed it!

    Consider why demand is down for proprietary stuff that STEM labor market makes: the laws don't allow for demand to develop! Take stem cell research: banned on 'humanitarian' grounds. Mission to Mars to terraform Mars? Same--governments have a monopoly on space travel, by fiat (law). Higher value added products? Forbidden due to IP piracy--why spend more on a patented item when you can get nearly the same substitute from an infringer, at less cost? Innovation in the workplace? No need. Why invent something when employment contracts routinely ban employees from gaining any monetary compensation (these contracts should be struck down as against public policy, and in Germany and Japan indeed there is a limited ban on such contracts for pioneering inventions--Google "Blue Laser JP scientist", but in general such contracts are upheld). In short, we have a lack of demand due to our laws constraining us from going to the next level, the global maximum, and we're stuck, as Tyler Cowen says, at a "local maxima"--and a non-optimal solution. Thanks for reading!

  2. Congratulations on getting so much press about your new policy paper! I am not used to this sort of policy fare, and I'm trying to digest it. I have a couple of questions. I didn't know which post to put this under since you have a few, so I hope it fits here.

    1. First, your claim seems to be that wage stagnancy in IT indicates that there isn't really high demand for IT guys. I'm sympathetic, but IT is a broad class of people, going from a barely-there, do-as-little-as-possible sys admin all the way up to the guys that develop driverless cars at google. The trend in wages of H1B visa IT guys appears to be a better indicator, if you've got it. If those wages are stagnant, it would be more direct evidence that demand for the sort of things foreign IT guys do isn't growing. I can imagine a situation where there is a shortage of the best of the best, or at least certain types of IT. Can you see H1B wages in your data?

    2. I would be interested to see more quantiles of the IT wage distributions. My prior is that the average are stagnant, but that the upper quantiles are shooting up (and lower quantiles dropping). Again, best of the best type thing. You have microdata, do it!

    3. Why so much focus on computer programmers? There were other computer related CPS codes that go back before 2002 like software developer...Same question on the four categories included in the lazonick stuff. Are the trends robust to the other IT classifications?

    4. A little comment: I think you guys are reading a lot into noisy data in which there is a lot of stuff going on. I would be awesome if you could find some sort of natural experiment or even a plausible instrument to test a hypothesis such as: fixing industry wages, the IT sector will not hire more people if a foreign applicant pool becomes available (i.e. we can't find guys good enough to hire in the US for any price, and the visas prevent us from getting better people abroad).

    1. Thanks very much for these thoughts.

      I definitely agree on the broadness of the group. The Lazonick material was supposed to help with this since he broke it up more. The same goes for the high H1-B intensity metro areas (that narrows the group geographically).

      Your quantiles solution is a good one. Honestly we had tons of data we wanted to show, and we just never dug this deep into the averages. The wage piece was only one of many things we tackled in the report, as I'm sure you saw. We are talking about doing a short follow up in the next couple weeks (just to gin up more interest). I'll raise this point and try to get some quantiles in. There is no reason to think different segments aren't experiencing different trends. The question is, does that call for a whole class of guestworkers that by and large aren't the best and the brightest but instead are as garden variety as the domestic supply.

      Thinking more outside the box, there's also a possibility of a tournament-style labor market for the very high end which may be driving wages way up. Paula Stephan and Richard Freeman have talked about this sort of phenomenon in STEM more broadly. I can try to flesh that out in more detail later.

      I need to take a look at the CPS again - I thought programmers were the only ones that went back that far but I could be wrong. Ultimately, different categories were showing lots of the same things so we didn't worry about doing a lot more than that (and like I said above, this was one of many questions we tackled in the paper). I'll look into disaggregating more in the follow up too.

      Agree on the value of a natural experiment here. We have a great one for petroleum engineering that is coming out in an NBER volume. If you think of one for IT I'm all ears. I don't know the industry well enough myself (Hal is the one that has been studying it for years) to know what to look for.

    2. I really appreciate these thoughts - thanks again.

    3. And I appreciate the reply! I'll keep an eye out for your follow-up.

    4. I should say, the medium here is very different from an academic paper. It's a very careful balance between "just fight the bad narrative with a good narrative" and digging in to the finer points.

      A natural experiment paper, for example, would never come out as a research brief from EPI. It would go elsewhere and make a different sort of impact on different sorts of people.

  3. The most important node is not winterspeak?


  4. Would you consider inadequate demand for science to be a market failure or an unfortunate cultural value?

    1. :)

      I would say both, but I'd probably have to take my economist hat off before I gripe about the latter.

    2. Indeed you would (have to take your economist hat off).

  5. OK, I may get a lot wrong here, from what I read, I feel this is one of the larger pieces of bad info produced, by folks who live with data and do not live in the trenches. Also I do believe the saying: “Anyone can bend stats to suit their needs. “

    So my understanding from this is:

    1. You are focusing very tightly on IT/MIS arena in your definitions of STEM. Although you do bring healthcare and oil/gas in, when it suits your needs
    2. You are proposing the easy flow of H1's into and out of the USA in order to fill open positions. If your proposal is that people can flow back and forth, like water, to fill the shortages, please stop. This provides bad info and embarrassing conclusions for 3 authors from named institutions.
    Having recruited for 20 years, in many industries, I can tell you there IS a talent shortage. It is not going away soon, globally.

    1. Looking only at IT/MIS, it appears you may be ignoring, very large areas served by STEM. Engineering including EE/CE/SE/SE, who do no IT/MIS work, but use it in their jobs as tools. While you mention oil/gas and healthcare, I am not sure you used them in some of the other charts and info

    2. In IT/MIS there is a shortage. Take a look at the open jobs. Just off the top, look at Los Angeles, Boston and Seattle, all areas where IT, your focus, is begging for good talent. While some of your data suggests STEM graduates were having trouble finding jobs, I would suggest this has a number of non-statistical, reasons, any of which we could discuss. I know, I know, you only deal with statistics.
    3. Part of your argument appears to be, “bring in more guest-workers”. In my opinion, this faulty and shows a lack of real, down in the trenches, knowledge. Many companies can't, or won’t do this for a variety of reasons:
    A. They are not set up to sponsor H1's.
    B. They are, legitimately worried about IP property theft, (if we are talking about the IT/MIS world ). Idea theft can be more prevalent, and very hard as well as costly to fight/enforce, after the theft has occurred.
    C. If they hire a lower paid guest worker, how long will company A really retain that person until they have to go back to their home country, or, take another job in the US, for more money with company B? It is difficult enough for an employer concerned about non guest-workers. Guest-workers adds more risks to potentially lose a good worker, and there's enough to worry about without adding to the problem.
    D. Their experiences with guest-workers has not always been ideal. Just because a guest-worker can program in JAVA, does not mean they can work in, what may be described as, the corporate culture of a US company. And this is a big issue
    4. I am not sure if your study takes into account ramp-up time. New STEM grads, any new grad, needs time for their education to really implement it and be productive. Just looking at the grad #'s and saying they are sufficient, personally shows another lack of being in the trenches.
    5. Also would you have any opinion on the value of keeping expertise within the states. The space program, a good example of a program that spawned, I can’t count, how many products, technologies and companies. This provided a great deal of long-term economic growth.
    6. It appears a great deal of the reference material quoted from, is from the 3 authors themselves, and government sources. Did you attempt to cross reference data from any of the major commercial job information sources? Linkedin, CareerBuilder, Monster, Indeed, for starters.
    I am sure you and your co-authors are very educated, thoughtful, with the best of intentions. I would suggest you run a search on the phrase “talent crisis”. You will see the depth and breadth of this issue, from many of the people who have been, and are living with these real issues today.
    If your goal is to provide controversy, I applaud you, as I feel you will have done so. However, if your goal was to provide an objective, thoughtful, well-rounded commentary on the state of STEM, I believe you have not done a good job.

    1. Thanks for taking the time to comment -

      On #1, see my post on whether we conflate IT and STEM. I think the report is pretty clear that we don't. All fields are different. Life sciences is in much worse shape than engineering and both are in worse shape than health care or the energy sector. But I know of no good examples of persistent shortages (emphasis on the word "persistent" here) that aren't caused by government policy.

      On #2... no, not like water. But they do have much easier access points than low skill immigrants. Some channels (like F-1 to OPT) are extremely easy. H-1B is obviously more competitive. While some of my co-authors might disagree I actually would like to see more access to guestworker visas. What's problematic is the restriction to high skill workers.

      On your second #2 - we have data from the highest H-1B metro areas and some of the biggest tech corridors. The data seem to be suggesting something different. Openings are a problematic metric. Firms will often post multiple job openings or multiple H-1B applications in the hope of getting a much smaller number. It's interesting data to look at, but it's more reliable to look at the behavior of wages, unemployment, and underemployment figures.

      On #3... in a way that's my argument (not my co-authors necessarily). But in a way I'm just saying the problem is not what a lot of people assert it to be. Certainly some firms are going to be more interested in the guestworker route than others (I'm in the DC area - a lot of the tech firms live off of defense contracts and they have a lot of these security concerns). Different firms are going to have different priorities and strategies. So? I'm not sure why you think that's an issue for our point. We're not denying the heterogeneity of environments that firms face.
      #3C - yes, we made this point in our webinar conversation with journalists on Wednesday
      #3D - oh definitely. My co-author Hal likes to make this point a lot about the productivity of guestworkers.

      5. Search on the blog archive on the space program. I have written a lot on the importance of these investments. I'm starting to be confused by what your problem with our report is... you started with serious disagreement and now you're coming around to a lot of things we agree with you on.

      6. What do you have in mind from Monster/Linkedin, etc. - I'm having trouble seeing what the value added of those sites are. As I noted above, there are a lot of problems with things like job opening data. We actually have talked about skimming career history information off of Monster but it's a major undertaking and haven't done it yet. I'm not sure what value added that has for this report. Maybe you could elaborate.

      re: "However, if your goal was to provide an objective, thoughtful, well-rounded commentary on the state of STEM, I believe you have not done a good job."

      Well we'll agree to disagree on that.

      If you have any thoughts on how to improve the analysis, better data, etc. - I'm all ears. This is a deep research interest of all three of us and we're going to continue to work on it. Most economists don't think the sources you mention are necessarily helpful in evaluating this question. It's worth looking into why that's the case.

  6. Your #1 - Persistent shortages, yes, there are. Think of it think of it this way: For years, the term talent crisis has become a persistent on in the hiring industry. However, I propose it will, and is, acting more like a tsunami than a crisis. I could write forever on this and recently presented last week at a national recruiting conference. I call it the talent cliff. And there’s a lot of real info about this.
    Persistent… I would say it’s coming like a tsunami. At the moment it is starting in tech, oil/gas, electricity, water, and even advertising.
    Your #2 – Having experience with many of the visas, yes I think regulations are a bit too strict, convoluted and should be streamlined. That said, I also think that before that, everything should be done to encourage workers here, to have access and training to, these skills that could improve the unemployment #’s within the states. I wrote a short paper for the DOE a few years ago on how to retrain certain existing workers, skilled trades, to supply the estimated 600,000 shortfall of workers.
    5. Search on the blog archive on the space program. I have written a lot on the importance of these investments. I'm starting to be confused by what your problem with our report is... you started with serious disagreement and now you're coming around to a lot of things we agree with you on.

    This was a bit of me calming down after reading over your report the first time. Living in the industry and seeing the shortages coming for so many year, and not they’re here, I was steaming. Seeing that someone would even think there’s not a shortage. In my opinion and experience, anyone taking the ‘ there’s not a shortage’ viewpoint is either 1> Out of touch, 2> looking at select data to make their point or 3> a bit naïve and needs more info

    6. What do you have in mind from Monster/Linkedin, etc. - I'm having trouble seeing what the value added of those sites are. As I noted above, there are a lot of problems with things like job opening data. We actually have talked about skimming career history information off of Monster but it's a major undertaking and haven't done it yet. I'm not sure what value added that has for this report. Maybe you could elaborate.
    These organizations put out many papers and research. I think you would benefit from them in rounding out your viewpoint to understand what is really going on. I would also suggest you take a look at Deloitte. They have a number of good research, white papers, studies, you may find interesting. One of the first, many years ago was titles, “The talent Crisis of 2008. Where’s Your Workforce” Written in 2004, it was one of the first, recent papers chiming this bell. I could quote Allen Greenspan and others going back many decades.
    I think I’m running out of space so I’ll leave this here
    Thanks for the reply

    1. Could I get a link to your DOE paper - or is it an internal thing?

      I didn't realize the orgs put out reports - I'll definitely take a look at that. I've definitely seen Deloitte research before.

      Thanks again for commenting.

  7. Trying to find the DOE paper, it may have been on a dead drive. It was a brief description of how to offer alternative energy jobs to the skilled trades workforce, and how to do it. There are a few simple things that could probably be done in my opinion.

  8. By the way, have to ask, since you, and your team, published this, and I suspect you want to be known..... why is it that none of you has a particularly built out LinkedIn profile? Professing an expertise in the area of workforce, employment, etc.....I would think that would be one area you would want a large presence?

  9. there are 3.9 million jobs unfilled in the USA as per .. why can’t these openings be filled by the unemployed? if the currently unemployed cannot fill these gaps, who will?

  10. Free Social Media Marketing where Every thing will be Free, Facebook Likes, Twitter Followers, Twitter Tweets, Twitter Re-Tweets, Twitter Favorites, Google Plus Followers, StumbleUpon Followers, Youtube Views, Youtube Likes, Youtube Subsribes, Pinterest Followers, Pinterest Likes, Pinterest PinIt, Free Website Visitors.
    Just Join now and Free Increase your Social Media Networks.


All anonymous comments will be deleted. Consistent pseudonyms are fine.