2008 & the sociology job market

It looks like we’re in another “economic downturn,” and many PhDs are understandably worried about what it means for the future of the sociology job market. I haven’t been to Delphi or stayed at a Holiday Inn Express, but I am knee-deep in historical ASA reports. Here’s what the data look like for how the 2008 recession affected the sociology job market in the US. Spoiler: I don’t think the market ever recovered, and more hopeful estimates say it took 4 years to recover.

The 2008 recession was bad for the sociology job market. But just how bad is actually hard to tell, because standardized data on the job market only goes back to 2009, and ASA standardized their reporting on it in 2014. Over time, not only has measurement of the job market changed, but so too have the benchmarks for “recovery” and the overarching framework/paradigm of reporting. Fortunately, ASA maintains a wealth of historical data and reports publicly on their website.

The only ASA reporting or data I can find on the counts of job listings or hiring from before 2008 is their report on the 2006 job market, published in 2008 with the provocative title “Too many or too few PhDs?” It uses ASA Job Bank listings to count jobs. The job bank was less widely used in 2006 (the report says only 80% of new PhDs consulted it while searching for a job), so this may be an undercount. They found 610 openings advertised for assistant professors, and note that this is more openings than students, because there were an average of 555 new PhDs in sociology per year from 2000-2006.

In the report on the 2008 market (from 2009) keeps the question and framing from the previous report, but extends the methodology to include a follow up survey on what came of the job searches and data from SJMR. The authors declare that 2008 is a “Down Market,” with 40% fewer Assistant Professor listings (n=370) in 2008 than 2006 (n=610). They express clear concern that this means there are only openings for 80% of new PhDs that year. The follow up survey showed even more discouraging numbers from 2008:

6.7 percent of job searches never got started because they were cancelled or suspended and another 9.6 percent were cancelled midstream for a total of 16.3 percent.

This is the last we hear of pre-2008 data, however. All subsequent reports take 2008 or 2009 as their baseline.

In the report on the 2010 job market (2011), optimistically titled “Moving toward recovery,” the authors changed both the methodology and the baseline year. Methodologically, they returned to using a survey of departments. Readers are presented with a figure showing that job openings and the number of departments hiring fell by one third from 2008 to 2009, and then rebounded about halfway back to their 2008 levels in 2010. The report on 2009-2012 (2013) extends the same charts (below). They conclude that in 2012, there were slightly more jobs than 2008. Assuming 2008 was a normal year (it wasn’t), they argue the sociology job market took four years to recover from the 2008 recession.

In addition to the job listing data showing declines in job advertisements, their survey approach allowed these reports to comment on the large number of searches that were listed but later cancelled:

A slightly higher percentage of jobs went unfilled in 2010 than in 2009 (19 percent compared to 16 percent) but the rates in both years were higher [sic] than in 2008 (with 29 percent unfilled)…. In 2009, the most common reason for unfilled positions was that the search was canceled or suspended.

We seem to be seeing another wave of cancelled searches. Karen Kelsky posted a crowdsourced list of schools with hiring freezes. The results of Dan Hirschman’s twitter poll suggest sociologists are not optimistic about the future:

The ASA report on the 2014 market standardized reporting to the format ASA still uses today. Here, focus shifts back to how many more PhDs there are than jobs. Measurement returns to counting ASA Job Bank listings, like in the 2006 report. However, unlike 2006, listings that contain multiple jobs are counted as one. (Edit: Teresa Ciabattari has pointed out my error. The 2006 and 2014-2018 reports all count individual jobs, so they are comparable numbers.) These methods produce the current chart which has been updated annually since 2014. Below, I update that chart to include the 2006 and 2008 data. (Note: several faculty have reached out to say they think 2006 was in an unusual boom period.)

In other words, the market never returned to pre-2008 hiring levels. As graduation rates show, since 2008 we have had 2-3 times more new PhDs annually than assistant professor jobs. And while great people work as lecturers and in other non-TT positions, ideally with strong unions like our @LEOUnionUmich, there were only 50 ads for lecturers and 62 ads for non tenure track assistant professors in 2018. Not nearly enough to fill the gap.

Update 28 March, 2020: I went looking for more consistent data over a longer time period. The NSF Survey of Earned Doctorates has public data on PhDs’ postgraduation plans at time of graduation for all of social science. Here’s a quick plot of that. Nothing nearly as drastic here as in the job listings data. It looks like we might have had a slight increase in people with no definite plans after 2008, and we’re in a decades-long trend of more people doing postdocs. I suspect there’s unobserved variation among the social sciences, in how many ‘no definite plans’ people later find employment, and in the nature of those positions they find.

Update 31 March, 2020: I remembered that John Robert Warren wrote an article last year on hiring and tenure requirements in top 21 departments with some useful data. Here are two figures from it, showing the number of assistant professor hires as well as the number of years between graduation and the hire. Looks like 2009-11 was a “down” period for hiring at top departments. After 2009 there’s also a clear increase in taking “gap years” (e.g. as postdocs) between graduation and their first AP appointment.

Author: Jeffrey Lockhart

Jeff is a PhD Candidate at the University of Michigan, writing his dissertation on the scientific debates around biological sex essentialism. He tweets @jw_lockhart.

7 thoughts on “2008 & the sociology job market”

  1. In my unsystematic observation of the market before and after 2010 (which to my eye was the bottom of the pit), the biggest change was from postdocs being rare and nearly quant, to being a normal part of the career path into a good R1 job, with one or two years teaching nonTT being the equivalent for smaller school /less well ranked dept hires. It remains the case today that most candidates being hired are NOT new assist profs. The nonacademic market soaks up about 25% of the new PhDs and the half that land postdocs/interim positions do mostly seem to move on into TT in the next two years. The others seem to be scrounging in the mix of academic research/teaching/admin positions at their home institutions.
    Unsystematic, qualitative impression only.


    1. Yeah, I don’t know of any national, longitudinal data on the postdoc market (NSF might have some on PhD employment in postdocs, which is related but not the same). It would be interesting to know if there’s also a short term decrease in postdoc / VAP / etc. searches, or a long term increase in them.

      Liked by 1 person

      1. I went and grabbed/plotted the NSF data on postdoc employment for new grads. It’s in the main post now. Takeaway seems to be that there’s a long term linear increase, but no big change around 2008.


  2. Thanks for this useful discussion. A quick clarification on the methodology: the reports do separate out Job Bank listings with multiple jobs. In other words, a single advertisement that lists two openings is counted as two jobs in the analysis.

    Liked by 1 person

    1. Thank you for this correction! I’ll update the post to reflect that. Do you think, then, it is fair to plot the 2006 data on the same figure as the 2009-2018 data?


  3. A scatterplot is a data visualization that shows the values ​​of two different variables as points. The data for each point is represented by its horizontal (x) and vertical (y) positions in the visualization. Additional variables can be encoded using labels, markers, color, transparency, size (bubbles), and by creating ‘small multiples’ of scatterplots. Scatter charts are also known as scatter charts, scatter charts, scatter charts, scatter plots, and scatter charts.


Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com 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

This site uses Akismet to reduce spam. Learn how your comment data is processed.

%d bloggers like this: