Knowledge Flows Embodied in Recent PhDs…or The News from Texas this August Paula Stephan Andrew Young School Georgia State University September 2006

Acknowledgements Support has come from the Mellon Foundation, SEWP at NBER, NSF and Kauffman Thank Brent Goldfarb, Al Link and Fiona Murray for sharing data

Motivation Knowledge flows from universities to firms play important role in fostering innovation Flows follow a variety of paths Face-to-face transmission important when tacit knowledge is involved. Placement of new PhDs with industry provides one mechanism for transmitting tacit knowledge.

“The best way to send information is to wrap it up in a person”* J. Robert Oppenheimer “The eternal apprentice,” Time Magazine, vol. 52, p. 81

Or…from the perspective of economists “The export of scientists and engineers from the academy to industrial research is potentially the most important and salutary among the mechanisms available for effecting knowledge transfers.” Dasgupta and David 1994

But Little Known About These Knowledge Flows Where do the new PhDs train? Do they go to work with firms in close proximity? Or “fly the coop”? Lack of knowledge relates to absence of data concerning placement of new PhDs going to work in industry

In Terms of Knowledge Production Function Considerable interest in sources of knowledge inputs in the knowledge production function (Griliches) Knowledge sources external to the firm have received considerable attention Proximity to knowledge sources seen to be important in facilitating innovation

Objective for today Analyze new data source concerning placements of new PhDs with firms Examine what data say about sources of new knowledge production and location of hiring firms; how important is proximity? Role of firm size Role of public initiatives

Data Source Survey of Earned Doctorates administered to all new PhDs since 1958 (NSF) 92% response rate Asks those with definite plans (63%) where they will work after graduation:  Name the organization and geographic location where you will work or study.

So What Makes It a New Data Source? Industrial placement information has never been coded But, since 1997 verbatim files have been kept of answers to “where are you going to work” question We coded verbatim for placements for those trained in S&E

Summary of Data 21,765 identified a firm where they were gong to work Represents 15% of those receiving PhDs in S&E during the period Undercounts in two serious ways:  No information for 37% who said they had plans to work in industry but had yet to be offered or accept a job in industry.  Many scientists and engineers who go to work in industry initially take a postdoctorate position.

Percent of PhDs Going to Industry

Large or Small Firms? About 40% go to a top 200 U.S. owned R&D firm in U.S.  Computer scientists, engineers and chemists most likely—around 45%; astronomers close behind.  Biology, medicine, agriculture and “other” least likely.

Drilling Down Include foreign-owned top R&D firms operating in U.S. percent up to 44% Add firms (U.S. and foreign-owned) operating in U.S. ranked in top 660 in terms of R&D and percent climbs to just shy of 50%.

Role of Small Firms Finding suggests that small firms play a larger role in innovation than R&D data would suggest  Top 200 R&D firms expend more than 70% of all R&D in U.S.  Hire only 40% of new PhDs

Role of Consulting firms and financial services Data suggest that many of new PhDs work for consulting firms and financial services, not manufacturing firms. Consistent with idea that much of innovative activity in today’s world is not restricted to products; also occurs in processes

Examples Financial Services: over 250 go to such firms as American Express, Morgan Stanley, Goldman Sachs, Merrill Lynch, Fannie Mae Consulting firms—over 350 go to firms such as Charles Rivers, Arthur Andersen, McKinsey, Price Waterhouse, Boston Consulting

Where Do They Train? New England and Middle Atlantic: 25% Pacific States: 17% Mid West: 27% South Atlantic: 15% Other: 16%

University R&D University R&D expenditures often used as a measure of knowledge available to spillover Compare placements to R&D expenditures Greater than 1 infer R&D understates availability of knowledge spillovers; Smaller than 1 overstates.

Training Relative to R&D New England and Middle Atlantic: 1.1 Pacific States: 1.0 Midwest: 1.26 South Atlantic:.81 Other:.79 Suggests that R&D data understate knowledge spillovers coming out of some universities; overstate that coming out of other universities.

Top Five Producing Universities

Next Five

High Quality Universities 6 of the top 20 training-universities are on the list of the recent ranking of the world’s top 20 universities:  Stanford (3 rd ); Berkeley (4 th ); MIT (5 th ); Cornell (12 th ); UCLA (14 th ) and Wisconsin (16 th )

Observations Heavily concentrated: Top 10 educate 25% of those going to industry Midwest plays an important role: five of top- ten are in the Midwest  Illinois  Purdue  Minnesota  Michigan  Wisconsin

Retention by Region 48% stay in the region of training  Pacific Region retains 70%  Mid Atlantic retains 51%  New England 46%  Midwest 37%

State Retention 37% stay in state of training  Midwest states retention is low Iowa retains 14% Indiana retains 12% Wisconsin retains 18%  Pacific states is high California retains 70%

State stay Rate Low Compared to Law school graduates: 57% stay in state Bachelors and Masters in science: 64% Bachelors and Masters in engineering 62%

Does it help to have a science park? Retention? Wide interest in science parks in U.S. around universities Universities around RTI place 673 PhDs with industry in six-year period; 39 of these are working in the RTI science park—represents 50% of RTI park PhD hires of new PhDs during period. Four mid-west universities placed 1559 PhDs with industry during period. Only 8 were placed at affiliated science parks—and this represented 50% of the hires by these parks.

Role of science parks Suggests minimal employment for newly minted PhDs Even the jobs they provide are as likely to go to individuals outside the area as those trained in area Firms doing the hiring are not necessarily small—but often large firms drawn to park such as BASF, Bayer, IBM, Nortel.

Role of Startups Big School: 86 startups between ;  11 of these hired one or more new PhDs for a total of 14 hires.  7 of these firms started jointly with faculty at a sister institution  12 of the 14 hires occurred in close proximity to Big School; 2 of these received PhD from Big School; 6 received PhD from co-founding sister institution.

Startups continued Majority of hires from local institutions Faculty appear to play a role in placements—plausible connection between faculty and firm in 8 of the 12 Startups provide a miniscule amount of employment in the PMSA level or among the newly minted PhDs trained at Big School

Role of licensing Case study of Stanford, period one 22.2% of newly-minted PhDs going to industry took a position with a firm that had licensed one or more Stanford technologies during the period. 28.9% of Stanford PhDs take such a position compared to 21.8% of others.

Differential Rate Licensing process often establishes formal relationships between faculty and the firm licensing the technology which may foster job placements Much of technology that is licensed is “proof of concept” and requires expertise to develop (Jensen, Thursby and Thursby and Thursby and Thursby). Students who have worked in mentor’s lab have this expertise.

Role of Geography 27% of hires made by non-licensing firms in the San Jose area are Stanford students 10% of hires by licensing firms in San Jose area are Stanford students. Lower hiring rate may reflect that in close proximity faculty are a substitute for students.

Conclusion News from Texas this August…state decided to invest $2.5 billion for science teaching and research in the University of Texas system. Primary focus building research capacity at San Antonio, El Paso and Arlington. Goal: turn these into next Austin.

Texas not alone University of California system State biotech initiatives Many examples in Europe Outside of Europe as well Not new…at least in U.S. States have a long tradition of using public universities to promote economic development and growth.

Overreacted? Possibility exists that states (and nations) have overreacted in belief that centers of knowledge excellence lead to economic development. States should be leery of investing in things that can move (Austan Goolsbee 2006)

Mobility Ideas once codified are public and mobile Faculty consult; start companies outside the area Newly minted PhDs going to industry are likely to move--only 37% stay Moreover, the “best” are more likely to leave.

Retention policies? Case studies suggest that university technology initiatives do little to encourage retention  Science Parks—no more likely to provide employment for locals than non-locals  Startups hired locally trained, but number is miniscule  Firms in San Jose that licensed Stanford technology are considerably less likely to hire Stanford PhDs than are San Jose firms who have not licensed Stanford technology

Some closing thoughts Findings remind us that an economic development strategy built on developing and retaining local knowledge may be fragile at best. Findings raise larger question of whether role of proximity to university is overemphasized in transmission of public knowledge from universities to firms.

Sources of knowledge Top source is publications and reports— proximity not important Second source informal information exchange, public meetings, conferences, consulting—proximity helps Next tier includes recently-hired graduate students. In this respect proximity--as we have shown—does not play a major role.

Questions/Comments?