Monday, July 16, 2007

Physics Jobs - Part 2 (Demand)

As in my first article, the main reference for this discussion is a set of statistical reports produced by the American Institute of Physics. I recommend you read them (use the links on the left side to get the summary and full reports) and form your own conclusions, but I hope my comments here will give you some idea of what to pay attention to.

The first lesson one must get from those studies is that, at present, only about one third of all PhD physicists are employed in academia. This may be hard for a student to believe, since you may never have met a physicist working outside of a university in your entire life, but it is one of the cold hard facts you need to appreciate if your goal is to get a job as a faculty member. Nonetheless, since recent blogs have emphasized tenured faculty positions and because my only experience is as an academic (post doc, a dozen or so years as research faculty, about a decade teaching), that is what I will write about.

A word about non-academic jobs:

Some quick remarks on the other two thirds of the jobs you might find. The one advantage that a US citizen with a physics PhD has in any job search is that there are quite a few jobs that require citizenship as one element in getting the relevant security clearance. This is not just for "weapons" work, because anything (particularly aircraft) that might have a military use requires national security in addition to the usual industrial security issues. You get paid more in industry, but might have to change companies if yours loses out on a contract. You might have to move to the company that won the contract, and at least one friend was forced down the entrepreneurial road and became a private contractor.

A key bit of advice is that there is little value in taking a post doc if you plan to go into industry. They usually could care less. The sooner you get out of the ivory tower the better. However, a post doc can be a key step if you want to work in a government research lab.

Academic jobs:

I think the most important lesson is that the majority of students will be employed at an institution below (in the sense of R1, PhD, MS only, BS only, CC) the one they got their degree at. This means the work environment and professional expectations will sometimes be (very) different from what you observe as a graduate student, assuming you were paying attention to what a faculty job entails at your institution. Those will be described in part three.

The graph I made of AIP data is important enough to repeat here. Its relevance to the demand side was driven home yesterday when I ran into a retired faculty member from Wannabe Flagship University. I knew he was from the right generation, so when he mentioned how many years he had been there, I got the crucial detail for this discussion: He got his PhD in 1964 (right on the leading edge of the big peak), went directly to Wannabe Flagship (no post doc), and retired 39 years later in 2003.

That's right: Many in that huge pulse of people who got a PhD between 1960 and 1970 and got hired into a faculty job did not retire until after 2000. That would be one of the explanations for Chad's comment that the job search situation was pretty good when he got out circa 2001. At that time, fully 17% of physics faculty were over the age of 65! [See paper by Czujko linked in Note 3 below.]

But it is not quite as bad as a snake eating an elephant, even if it seemed that way to my generation. There were people hired into faculty jobs from my generation, and some of the more recent openings have been filled with "senior" hires from industry or research labs (that is, from my generation). The hiring pulse circa 1965 has been damped out somewhat, but you can still see a pronounced minimum in demographic data for physics departments. (See the paper by Neuschatz and McFarling, linked from Note 2, for some examples.)

The main thing to take away from that graph is that the current supply of PhD physicists is about 1200 per year and (probably) slowly increasing. Other data indicate that about half of these are foreign students, which might be relevant for some jobs in academia but probably not for the ones at top research universities, which get a large fraction of their faculty from overseas. [Details are in part 3 of this series.] You should assume that everyone is your competition. If in doubt, look at the faculty at your school and where they got their PhD degrees.

Data on Faculty positions:

There are a lot more faculty positions than you might think if your view is limited to the 146 or so institutions classified as Doctoral Research by the National Academy of Sciences. Those are the "PhD" category in the table below, which make up a bit more than half of the tenure-track faculty. If, as is most likely, you attend one of the universities that are classified as "very high" research (what used to be R1) in the Carnegie scheme, you might be ignoring the majority of possible academic jobs.

This table was constructed from the AIP data circa 2004 as described in Footnote 1 at the bottom of this page. All faculty refers to the total of tenured, tenure track, and temporary faculty reported by the university. Temporary faculty could be full time instructors (usually at 4-year schools) or they could be semi-permanent researchers supported by external or internal funds (in PhD programs). These numbers do not include full-time post docs or part-time adjunct instructors, which fall in a different category, but probably include multi-year research positions (glorified post docs) that run for a fixed contract period greater than one year.

Dept typeall Facultyd/dtt-t Facultyd/dt
MS only900+10730-3
BS only2700+202130+1
2 year CC  1640 -28?

The time rate of change has units of people per year. The striking detail, no surprise to us old timers, is that the total number of faculty is increasing but the number of tenure-track faculty is roughly constant. Universities are shifting resources from permanent positions to temporary ones. The rational reason for this is it gives them flexibility to deal with shifting student demand, federal research support, and student enrollment (when the baby-boom echo comes to an end). The economic reason is that they are cheaper, particularly when we consider part-time faculty at the CC level.

These are all full-time PhD positions, with one exception. The number listed for 2-year schools (community colleges) is the number of tenured faculty, but only 640 of these have a PhD degree. It happens that the data for 2-year schools also tell us the number of part-time adjunct instructors and how many of those have a PhD. The table below (explained in Footnote 2) shows only PhD faculty but includes the 330 part-time CC instructors with a PhD in the "all Faculty" column.

Dept typeall Facultyd/dtt-t Facultyd/dt
other 4 year3600+302860-2
2 year CC970 ?640 +1?

This snapshot (and the derivative) suggests that there are about 7900 PhD's employed as full time, tenured or tenure-track faculty at colleges or universities. Keeping those positions filled as people retire, die, or leave the country is what generates job openings for newly minted PhDs. There is zero or negative growth in terms of tenurable faculty positions being added to physics departments.

Annual Job Openings

This is much more speculative, but based on pretty consistent trends identified in the AIP studies. See Footnote 3 for an explanation of the numbers shown here, and take this with a big grain of salt. Once you construct estimates from estimates the values get really fuzzy, but the main trends were real as of the period (circa 2003) when these studies were done.

Dept typeopeningsUS PhD hires
other 4 year200180
2 year CC75 70?
PhD degrees 1250?

One thing jumps out of the first column: There are more openings in the BS and MS departments (what I incorrectly call "other 4 year" schools) than in the PhD departments, even though the PhD departments have 50% more faculty! Turnover is very low in major PhD programs. (See my example up top of the guy who retired after 39 years. Working for 39 years is easier to pull off if you have a 1/1 teaching load and a modest research program, if any, than if you have a 3/3 or 4/4 load at a 4-year school.) But there are other factors at work.

In the time period of the study, about 1/3 of the faculty hired in PhD departments (but only about 10% for the others) earned their PhD overseas. I used this fact to generate the estimates in the second column, which would be the number hired who earned their PhD in the U.S. (Note that these could be foreign students who earned their degree in the U.S., not just US citizens.) That significantly skews the available jobs away from the top schools.

In addition, almost a third (about 55) of the US PhD hires into universities and maybe 55 or so of the others can be expected to be "senior" hires (persons who got their PhD more than 5 years before being hired). This would reduce the bottom line to 260 (+/- ??) openings for persons who earned a PhD in the US in the past five years.

The market reality:

There are estimated to be about 1300 persons earning a PhD this year, maybe more, maybe less. I'll use 1200 under the assumption that some foreign students (who make up about half of this total) are planning to seek jobs back home. For comparison, there were 200 fewer graduates (under 1100) and probably more job openings when Chad got out.

So, comparing 260 openings to 1200 job seekers, the odds of getting some kind of tenure-track academic job are about 1 in 5 during the first few years after getting your PhD. In the longer run, if you become one of the "senior hires", the odds increase to about 30%, maybe more. Since national surveys show that about 1/3 of all physics PhD jobs are in academia, including research faculty jobs, those odds seem plausible.

However, most of those are not at a research university like the one where you earned (or are working on) your PhD. Comparing 60 to 1200 tells a different story for that part of the market. The odds of a recent PhD getting a job at a research university are only about 1 in 20. The fourth installment of this series, along with info in part 3 will address some of the things you need to pay attention to if you have this as your goal. You need a clear plan to put yourself in the top 5% after 2 or 4 years in a post doc, and be prepared to win the grants and reputation that will earn you tenure. (Getting the job is not the end of the battle!)

On the other hand, if you are an American with an interest in teaching and running a modest research program with undergraduate students, the odds might be as good as 40% (240 openings, 600 candidates) that you can find that job. I will have additional comments about seeking those kinds of jobs in the final installment of this series.

Side comment:
The long odds against getting a t-t job at a research institution are why I started this section by mentioning that most PhD students who enter academia will end up "below" (in the hierarchy of colleges) where they earned their degree. This is inevitable, and can be understood with a simple "Fermi question" analysis. How many PhD's will your major professor produce? Ten? Twenty? Fifty? Only one is needed to "replace" him or her, so all the rest are fighting for that job.


Note 1.

The AIP faculty workforce reports give the total number of FTE faculty, including t-t, temporary, and research positions (but excluding post docs). The 1994 and 2004 numbers are the source of the derivative calculated over a ten year period.

Dept type19941998200020022004
MS only800850775900900
BS only25002500260028002700
total8200835083758800 9000


The reports also give the percentage who are in temporary positions (not tabulated here, but you can see them in Table 2 of the reports linked above). From these one can calculate the following for the years 1998 to 2004. I use the two end points to get a 6-year average derivative in my tables, but you can see that these are very noisy data. The derivatives are consistent with zero.

Dept type1998200020022004
MS only748651729729
BS only2125210621842133
total727372077299 7290

The 2-year data (community colleges) are much cruder. They come from two reports, one about the 2001-2002 statistics and another about 1995-1996 statistics. In 2001-2002, there were 2560 faculty (1640 full time, 920 part time) teaching physics on 1072 campuses, with 39% (or 640) of the full timers and and 36% (or 330) of the part timers holding a PhD degree. In 1995-1996, there were 2592 faculty (1810 full time, 782 part time) teaching physics on 1052 campuses, with about 35% (or 634) of the full timers holding a PhD. In both cases, a few percent of the full time are not "tenure track", while the 1996 report stated that a significant fraction of the PhDs (corresponding to 200 of 633) were in related fields such as engineering or chemistry. I cannot tell from the methodology of the 2002 report if they corrected for the 94% response rate in their final numbers, as I did when quoting the 1996 data. If not, the number of t-t faculty in 2002 increases to 1745 and the derivate drops from -28 to -11, but I doubt if this is the case.

Note 2.

See the two papers referenced above concerning the 2-year college dataand the discussion of the data above. The increase in PhD faculty from 633 to 640 over 6 years gives a tiny (and, I am sure, statistically insignificant) positive derivative. The other results are directly from the table described in Note 1.

Internal check: This gives a total of 9970 PhD's in academia, compared to the total of 10047 (5801 teaching, 4246 research) given in the Neuschatz and McFarling "Career Outcomes for PhD Physicists ..." paper. This minor undercount is plausible given that both numbers are derived from estimates (the latter extrapolating a longitudinal study of 1850 physicists to a population of 33,729). The "teaching" number is plausible if half of the t-t faculty at PhD institutions reported their job as research rather than teaching, which is what the authors said was likely the case.

Note 3.

The estimate of foreign fraction of new faculty is based on the Czujko paper on Enrollments and Faculty in Physics, which summarizes the fraction of new faculty who got a degree overseas (34% for PhD departments in 2000, 12% in BS departments). Similar data tables can be found in the workforce papers referenced above. [double check] The 50% value for the fraction of new PhD students who are foreign comes from the 2004 AIP Enrollment and Degrees report (54% of 2004 PhD grads were foreign students, up from 45% in 2000).

The Czujko paper is also the source for the statistical data on the fraction of new hires who earned a PhD in the US more than 5 years ago (31% in PhD departments, 28% in BS departments) and within the past 5 years (35% in PhD departments, 60% in BS departments) that is used in the discussion after this table about the hiring odds for relatively recent PhD graduates.


K said...

This and the one before it are really amazing posts that every physics student should read!

Doctor Pion said...

Thanks. Physics Today has its regular articles on the subject (usually once a decade), but I think my data analysis on the demand side is totally new. I had no idea how big the 4-year market is or, for that matter, how to look for those jobs back when I was first seriously in the market.

I did get one more bit of anecdotal information this past week. A newsletter from Grad School U noted the pending retirement of two outstanding researchers. One got his PhD in 1961, hired in 1963, retiring 45 years later. The other got his PhD in 1960, hired in 1961, retiring after 47 years!

R1 universities are not a "30 and out" industry.