Scholarship and Inventive Activity in the University: Complements of Substitutes? By Brent Goldfarb, Gerald Marschke and Amy Smith Discussant: Nicola Lacetera Case Western Reserve University Department of Economics
The paper Question: ≶ 0? Data: novel panel from Stanford’s biochemistry and electrical engineering department , all tenure track faculty Scientific productivity: publication count + impact factor – weighted Inventive activities: Disclosed inventions with commercial potential Teaching: taught credits Statistical methods Count models (Poisson); OLS Endogeneity: FE, IV: VC disbursed, revenues of colleagues. 2sls, GMM Findings: >0 in Biochem, =0 in El. Eng.
Contribution – The question Relation b/w scientific and inventive activities: hot topic in the Economics of Science Are commercial activities compatible with the production of good science? Can universities have multiple missions? Research, teaching, commerce? Political and managerial relevance Hicks-Hamilton (1999), Agrawal-Henderson (2002), Geuna-Nesta (2003), Azoulay et al. (2004a, 2004b), Van Looy et al. (2004), Stephan et al. (2005), Markiewicz-DiMinin (2005), Breschi et al (2005), Calderini-Franzoni (2005), Calderini et al. (2005), Murray- Stern (2005); Henderson et al. (1998), Mowery et al. (2003)
Contribution – Limits in current studies Data Publications, Citations, Impact factor: Scientific value, truncation, relevant journals, reasons for citations Patents, citations: squeeze existing database, but appropriate? Most inventions not patented, citations by examiners… How about teaching? 444 Methods and Techniques Simultaneity, individual heterogeneity, unobservables. Progress, lately Theory What should we expect? How do the different incentives interact? How to model multiple missions, peer effects, career concerns, etc.?
Contribution – The data Tenure track Stanford faculty, , two depts. Publications, # and I.F.-weighted: avoid truncation, consider quality Disclosed inventions, NOT patent data – at last!! More comprehensive Teaching record: other major activity to consider! Small number. 15 scientists in Biochem How about post-docs? Big deal in Biochem. and Engineering Stanford: representative of average/median university? Can generalize complementarity? Faculty quality, resources, TLO/TTO efficiency… I.F. from ISI: keep #journals constant? Disclosed inventions with commercial potential: selected sample? Teaching credits: lab ≠ classroom? Why dummy in regressions? Variance?
Contribution – Methods and techniques Take Endogeneity seriously -- GREAT! FE, IV, GMM – Wooldrigde, Arellano-Bond. State-of-the-art techniques Identification Social interactions and peer effects: tricky first stage (Mansky 2002) Small sample bias of IV techniques (Hausman-Hahn 2002). Estimates bounce Strength of IV: show first stage (R 2 …)? Show Hausman (1978) test? Orthogonality: What if… Technological/ scientific shock, opportunity VC activity, revenues of colleagues Publications Inventions Scientist ability, arrival of a “star”, major finding VC attracted (Zucker et. al…) More inventions Big grant Buy out teaching More research
Contribution – Theory Not much theoretical discussion – not the aim of this paper, but… What are the underlying theoretical/behavioral assumptions? Is science-invention the appropriate tradeoff? Why not together (biotech…)? How about science and innovation, or entrepreneurship? Are the results surprising? Expected (especially after biotech)? Why the difference between departments?
Contribution – Summary Relevant improvement in data collection Concerns: selectivity (Stanford, valuable inventions), variable construction (teaching variable, I.F.), small sample Major advances in identification Concerns: Identification strategy, small sample Intriguing questions raised, e.g. difference among depts. and measures Major contribution! But keep an eye at concerns: reinforce your results, explain them, and find space in a quite crowded research area…