ABSTRACT
Using 16 years of NSF data, we explore determinants of self-employment by new PhDs in the US, considering the role of field of study, demographics, and a factor that has gotten much attention in recent years (though in different contexts) – the role of student debt. While a small share of the population, one would think that young doctoral recipients are an especially innovative group, and perhaps likely candidates for entrepreneurial activities. After controlling for field and demographics and quality of doctoral institution, we find some evidence consistent with an impact of student debt, especially the more exogenous undergraduate debt, in limiting the choice of self-employment as an immediate career path post-graduate school. This effect, however, is limited to graduates of schools below the most elite.
Acknwoledgement
I thank Barry Hirsch for helpful suggestions on an earlier draft. I also thank the National Science Foundation for providing access to the confidential data used here from the annual Survey of Earned Doctorates. The use of NSF data does not imply NSF endorsement of the research, research methods, or conclusions contained in this report.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 An initial – post-PhD – choice of self-employment may reflect limited job opportunities and not a long-term interest in entrepreneurship, however the limitation of the dataset is that these longer-term career patterns are not observed. We do attempt to control for this in the econometric work below.
2 They also provide an excellent review of earlier work on the topic. Even more recent is the mostly descriptive discussion in Ewing Marion Kauffman Foundation (Citation2020).
3 Clearly, the identity of the best PhD programs in the country differs across fields (and perhaps over the 16-year period studied here); nevertheless, we arbitrarily define a ‘Top 10 School’ as one of Harvard, Yale, Princeton, Columbia, MIT, Stanford, Berkeley, University of Pennsylvania, University of Michigan, and University of Chicago.
4 Probit results were quite similar to the linear probability model results presented here.
5 We see, as well, that new PhDs from top-10 institutions have about half the total debt of those attending other institutions, about one-third the graduate student debt – reflecting the fact that these institutions tend to fund most of their PhD students.