Via a tweet from Derek Lowe, this very interesting paper by Dr. Stephanie Cheng (a PhD economist), summarizing and extrapolating data from both the Survey of Doctorate Recipients and the Survey of Earned Doctorates. The Survey of Doctorate Recipients is the NSF's routine survey tool to look into the lifetime of doctoral graduates.
Her paper discusses three results:
It's taking longer to get out of grad school: "Over the past fifty years, mean time spent in graduate school has steadily increased by 2.2 years, from 5.8 years (s.d. = 2.1) among 1960-1980 STEM Ph.Ds. to 8.0 years (s.d. = 4.1) among 2000-2013 cohorts... Figure 5 demonstrates that fewer individuals are completing Ph.Ds. in fewer than four years and more individuals are completing Ph.Ds. in more than eight years over time."
Fewer people are getting tenure-track positions: "As doctoral training has lengthened and more STEM Ph.Ds. have pursued postdoctoral training, the probability of obtaining an academic tenure-track position has nearly halved over the past fifty years. Only 25.2% of 2000-2013 STEM Ph.D. graduating cohorts are ever observed in a tenure-track position, compared to 42.8% of 1960-1980 cohorts."
Postdocs cost PhDs lifetime earnings: "To quantify the impact of postdoctoral experience on salary at each career stage, Figure 19 gives salary regression coefficients on years of postdoctoral experience for each of the first thirty years post-Ph.D. graduation, as calculated in Equation 1. The first few years show a large negative relationship due to the salary gap between postdoctoral appointments and permanent positions. This gap closes as postdoctoral researchers move into permanent positions, but the additional training does not improve their salaries enough to overcome this early loss. As given in Equation 2, the average of these yearly coefficients can be interpreted as the postdoctoral deduction in mean lifetime earnings. Rather than provide an education premium, each additional year of postdoctoral experience reduces average lifetime earnings by $3,730."
All three of these results are not new to this discussion, almost to the point that you have to ask yourself "is this news?" And my response is a confident "yes." First, it's documented quite clearly by her analysis of SDR cohorts, and it appears that not many people have decided to do this. Second, she chose chemistry specifically as one of the PhD fields to be analyzed, and so Appendix B has some really nice data to summarize her findings. Of these, I'll choose a few of note:
- Between the 1960s and 2013, the mean years in graduate school for chemists had moved from slightly less than than 5 to slightly less than 7.
- The number of PhD chemists who took 4 years or less dropped from 40% in the 1960s to less than 10%. Meanwhile, the number who took 6 years rose from 10% to 30%.
- The percentage of PhD chemists since the 1980s who did a postdoc within 2 years of graduating has never been lower than 40%.
- The percentage of chemistry postdocs who did a 1-2 year postdoc fell around the time of the Great Recession from 30% to 20%, while those who did a 3-4 year postdoc rose from 10% to 30%.
- The percentage of PhD chemist cohorts that had a tenure-track position at anytime went from 36% in the 1960s to 10% in 2013.
- Since 1972, the modal outcome for a PhD in chemistry 10 years after graduate was a position in industry, from a high of ~60% to a low of ~42%.

