Friday, October 23, 2015

Was Peak Chemistry Postdoc in 2011?

Busy day, but I want to follow up on a very old post. This morning, Monya Baker of Nature's news group reports that the number of biomedical postdocs is beginning to fall: 
A decades-long surge in the numbers of US biomedical postdocs may finally have ended. 
From 1979 to 2010 the number of US postdocs in the biomedical sciences has risen steadily, from just over 10,000 to more than 40,000. But in the past three years, the tide has turned, according to official statistics. 
The population of US biomedical postdocs fell 5.5% between 2010 and 2013, to just under 38,000, with losses getting bigger each year, notes a study published on 6 October — although the number of new graduates with science PhDs continues to rise...
From the same NSF survey that the paper is based on (the NSF's Survey of Graduate Students and Postdoctorates), here's the data for chemistry postdocs. I'll try to modify this over today to get something better up:

2001: 3,854
2002: 3,973
2003: 4,135
2004: 4,330
2005: 4,190
2006: 4,045
2007*: 3,952
2008: 3,943
2009: 4,219
2010: 4,179
2011: 4,005
2012: 4,016
2013: 3,850

Looks to me that there has been a steeper drop for chemistry postdocs between 2010 and 2013, with an 8% from between the two years.

It looks like I pseudo-predicted that Peak Chemistry Postdoc would be 2011 (that post was from 2012), but it looks like it was more like 2009. (In this sense, were new chemistry PhDs going to postdocs a leading indicator of problems in the #chemjobs market? I doubt it? Employment tends to be a lagging indicator of issues, although I could be wrong.)

I sincerely hope that the number of postdocs continues to fall, for the right reasons, i.e. only universities who sincerely want TT-level assistant professors to have them will be wanting them. Everyone else who wants their new entry-level PhD hires to have an extra 2 years of experience will should be required to pay for it. It seems to me that the "right" number of postdocs is in the 500-1000 range, but that's a personal preference, really.

* For this number, I chose the "2007 'new'" number. From the NSF: "In 2007, GSS-eligible fields were reclassified, newly eligible fields were added, and survey was redesigned to improve coverage and coding of GSS-eligible units. "2007new" presents data as collected in 2007; "2007old" reflects data as they would have been collected under 2006 methodology. Science fields "communication" and "family and consumer sciences/human sciences" are newly eligible. Data for these two fields are only in 2007new. Science field "multidisciplinary/interdisciplinary studies" is also newly eligible; these data may have been reported under other fields before 2007. "Neuroscience" is reported as a separate field of science in 2007new; these data were reported under health field "neurology" in 2007old and previous years. "Architecture" is reported as a separate field of engineering in 2007new; these data were reported under "civil engineering" in 2007old and previous years. See appendix A, "Technical Notes," for more detail."


  1. Because there is less NIH money out there now to support academic labs (at least in terms of inflation adjusted dollars) than there was years ago, I suspect there is just less money out there to support hiring them. Im not sure if that is a right reason.

    I would like to see this continue to fall at the expense of bringing in grad students from other countries to the USA to fill post-doc positions, but of course that is just my opinion.

  2. "Everyone else who wants their new entry-level PhD hires to have an extra 2 years of experience will should be required to pay for it."

    A nice idea, that I sympathise with, but not one that's going to happen. Doctors, lawyers, accountants, engineers all need a couple of years of training after their terminal degrees during which time they're not paid full salary, though I'm pretty sure much much more than a chemistry PDF (I'm assuming chemistry PDFs make <$50K?). All of those careers have associated meaningful associations and actual legal status. Chemists have the ACS.....In theory a meaningful chemistry organization could help chemists, but I just don't see how that could realistically be set up. I guess a union could do the same thing but, again, I don't see that happening. Ever.

    I get the impression chemists are simply to meek to stand up for themselves the way those who go to medical/law/business/engineering school do. This also explains the (largely enecdotal, though believable) poor behaviour of big (and small) name academic PIs. By gum, that's how things were in my day and we liked it!

    There is an argument for maybe a 4 to 5X excess of postdocs per TT position, i.e. as a weeding out process for academia. Do profs in other fields (engineering, social sciences) have to subjugate themselves to post-docing?

    1. Sadly, I agree with the predictions of your outcome. In the long run, I would like to see demand be so high for chemists that overtraining would not happen.

      The various residencies and fellowships that MDs get are 1) much more proportionally compensated in the long run and 2) controlled by licensure (I suspect.) We're never gonna get there, sadly.

      Students didn't need to have postdocs back in the day - not quite sure why they need them now.

    2. Ya, not sure when that shift to needing a pdf to get to academia happened. I note that KBS did a PDF, but I don't think the Boger Man or KCN did. My, unsupported by any data, observation is students from big name groups are more likely even today to move from grad school to a TT position without a PDF than Joey Erlenmeyer from "we're a top 20 program" U: Phil Baran and Eric Sorenson are good counterarguments to this. If one believes that the very best students go to work for the biggest name PIs (which I think is often true: note, I did do a PDF) then maybe this makes sense.

      I actually recall my PhD advisor telling me a PDF would the best time in my life: decent amount of freedom, no classes, no tests, no TAing, no worries about funding, just a couple of years to do science. I think he was about 75% right. I don't know that I would have the same view if I had PDFed for a jerk.

    3. I wouldn't necessarily say the best students go to work for the biggest name PIs, but maybe the people who understand how academia works/have adequate mentoring go to work for the biggest name PIs (there probably is some overlap because the "best" students are likely to go places where they have well-connected professors and good mentorship).

      Purely anecdotal, but I knew a lot of great students (at a "top" school with a mix of very strong PIs and some weaker PIs) who were good students and decent researchers, who had their pick of labs, and chose a project they found interesting with a weaker PI (many of them seriously regret their choices).

      The cynic in me thinks that the latter group are easily persuaded with advisors who oversell projects/give a lot more freedom in project choice (because the latter group don't understand that in many ways PI >> project). Because projects often fail, if/when the projects fail and these graduate students are in a weaker lab, they don't have much to fall back on.

      Conversely, in a stronger lab, there are lots of projects (because stronger labs are often supergroups), you work under a postdoc for a while, and then work on your own project. Usually your project is very strongly tied to whatever everyone else is doing, so you can fall back and change direction if need be.

  3. Yeah, I believe the Last R1 Professor Not To Have A Postdoc is Abby Doyle, but maybe I'm wrong.

    1. (caveats: *chemistry professor, *major Ph.D. granting institution)

    2. CJ, I believe that would actually be Emily Balskus (Harvard)

      It seems Jacobsen's PhD students don't need to do PDF's these days...

    3. Enough people in that one lab to fill 1/3 Chembarker's TT listings. And that is just one lab at Harvard.

    4. Seems that she did an NIH Postdoc?

    5. balskus did a 3 year postdoc in chris walsh's lab at HMS. but yeah sure, jacobsen's students get crazy preferential treatment. totally.

    6. Stimulus: Post-doc for Chris Walsh, who wrote a cool but out of print text book (which I hope will be picked up by Dover paperbacks).

      Response: get a job at R1 institution

      Status Quo, anyone? No wonder nothing changes for the better.

    7. Abby Doyle published 5 papers during her ph.d. and they were actually not that high impact. Interesting.

    8. Walsh is pretty popular as a swerve for organic chemists to get a little bio/biochem experience and get anointed as acceptable R1 material*. Lots of folks go through there but not as many papers as i would have expected.

      NMH--yes, between Berkeley and Harvard, there's easily enough 'top pedigree' people out on the market to squeeze out everyone else. Makes you think.

      (This also describes my grad school PI)

  4. Nocera, now at either Harvard or MIT, is another example of not needing a pdf.

    1. David Liu (Harvard) and Kit Cummins (MIT) are two other examples.

      Do I even want to bring up that getting a Ph.D. in three years has gone the way of the dodo? Oops, just did.

  5. I agree with NMH. There has been lots of material over the past two years on the drugmonkey blog about how the NIH paylines are very tight and anyone in a softmoney position is screwed. The first of those would be postdocs working for softmoney professors before those professors themselves are forced to leave their jobs. The moment NIH gets a doubling, postdoc numbers will grow again. There just isn't that much else to do for graduates when a lot of pharma jobs have been slashed and a lot of research even, is moving to Asia. Plus, a postdoc in the USA is still attractive to many in India and the supply of good postdocs from there has been increasing with the level of the institutes there. Many still want to come to the USA for a few years to do a postdoc as it will increase employment chances either back home, or in the USA itself.

    1. speaking of which... here's a link to drugmonkey