The Impact of Founders’ Professional Education Background on the Adoption of Open Science by For-Profit Biotechnology Firms*

Waverly W. Ding

Haas School of Business 545 Student Services #1900

University of California – Berkeley Berkeley, CA 94720

wding@haas.berkeley.edu Tel: 510-643-4252

* I have benefited from the advice of Toby Stuart, Matt Bothner, Ron Burt, Damon Phillips, Stanislav Dobrev, the three anonymous reviewers, and seminar participants at UC Berkeley, Chicago GSB, UC Davis, Emory, HBS- Entrepreneurship, University of Illinois Urbana Champaign, London Business School and the Georgia Tech REER conference. This research is supported by the Kauffman Foundation’s entrepreneurship fellowship. Usual disclaimers apply.

The Impact of Founders’ Professional Education Background on the Adoption of Open

Science by For-Profit Biotechnology Firms

ABSTRACT

This paper investigates the effect of founders’ professional-education background on the adoption of an open-science technology strategy, using a sample of 512 young biotechnology firms. After controlling for founders’ prior work experience and other organizational and environmental factors, I find that firms with proportionally more Ph.D.-holding entrepreneurs on the founding team have a higher probability of adopting open science. In addition, founders’ educational background can mitigate the constraint of organizational environments on strategy. A crowded technological niche provides a more challenging environment for firms for implementing open science, due to higher scooping risks. The deterring effect, however, of such a high-risk environment is smaller among firms founded by proportionally more Ph.D.-holding entrepreneurs. There is also some evidence of a stronger effect of founders’ educational background on open science in an institutional environment in which open science has yet to become the industry norm. This finding is consistent with and complements the growing body of research that emphasizes the importance of entrepreneurial background in developing knowledge about new venture strategy and structure.

1

I. Introduction

A substantial body of research has shown that organizational founders’ visions and models for a

firm exert enduring impact on the development of organizational strategy and structure (Baron, Burton

and Hannan 1996, 1999; Baron, Hannan and Burton 1999, 2001; Burton and Beckman 2007; Beckman

and Burton 2008). Such research highlights the importance of understanding those factors that influence

founders’ visions and models and, in turn, early organizational strategy and structure. Recent studies have

started to investigate the role of founders’ background such as their past employment histories and

industry experience, linking them to variations in the organization of new ventures at the time of founding

(e.g., Burton 2001, Simons and Roberts 2008). This paper joins this body of research and focuses on

another important source of founders’ visions for new ventures, namely, their educational background. By

analyzing the adoption of an open-science technology strategy by 512 young biotechnology firms, I

explore the relationship between founders’ professional education and their choices of new-venture

strategies. In addition, I investigate whether and how this relationship varies across different

organizational environments. Overall, this paper aims at providing a more comprehensive understanding

of how entrepreneurs’ background shapes the organizational process of new ventures.

Prior investigations of founder background effect on start-up organizations have focused mostly

on entrepreneurs’ pre-founding work experience (Burton et al. 2002, Shane and Khurana 2003, Higgins

2005, Beckman 2006, Sorensen 2007, an exception is Boeker 1988). What has been largely ignored so far

is founders’ educational background, which may also have a significant impact on founders’ views

regarding organizations. Though education is one of the most widely studied background variables in the

entrepreneurship literature, it is most often used as a human-capital explanation for individuals’ transition

to entrepreneurship (e.g., Carroll and Mosakowski 1987, Evans and Leighton 1989, Bates 1990, Brüderl,

et al. 1992, Lazear 2005, Kim, Aldrich and Keister 2006). Few entrepreneurship scholars have examined

how professional education may influence founders’ choice of organizational strategy.

A broad set of sociological and management research emphasizes the imprinting experience of

professional education on individuals. What has been imprinted on potential entrepreneurs may, through

various mechanisms, shape their visions of a firm. First, potential entrepreneurs learn specialized

2

knowledge during the process of professional education. The knowledge structure and the way potential

entrepreneurs are trained to approach problems may constrain their future information- processing

patterns and provide a cognitive map for their evaluation of organizational strategies (Hambrick and

Mason 1984, Wiersema and Bantel 1992). Second, professional education is also a process during which

an individual internalizes core values of the profession (Larson 1977). Though far from being conclusive,

evidence from studies of professionals in industrial organizations suggests that their actions in

organizational settings could be influenced by the professional values they have internalized during the

education process, even when such values are at odds with organizational demands (e.g., Kornhauser

1962, Miller 1967). Third, assuming that organizational founders are subject to the influence of

information circulated in their networks, educational background could affect organizational strategy by

shaping a potential entrepreneur’s network of contacts. Studies of how MBA-degree-holding managers

use their business school alumni networks as sources of information provide support for this assumption

(Haunschild et al. 1999, Rider 2008).

Despite its importance, empirical evidence that relates founders’ professional-education

background to early organizational structure, strategy and practices remains scanty. Organizational

demography, particularly the research on top management teams (TMT), probably comes closest to

answering this question, but its focus is more on the team dynamics determined by the distributional

property of education among TMT members. In addition, most of the studies in this area draw from

mature organizations and are cross-sectional (e.g., Kimberly and Evanisko 1981, Bantel and Jackson 1989,

Wiersema and Bantel 1992), a research design in which identification of the managers’ educational-

background effect is often a challenge (see Hambrick 2007, for a review of the identification problems in

TMT research).

This paper seeks to empirically test the effect of founders’ professional-education background

on early-stage organizational strategy. I studied a sample of 512 for-profit biotechnology firms in the U.S.

and analyzed their adoption of open science—a policy that allows a firm’s research personnel to do basic

science research and publish the results in academic journals. One major finding is that, after controlling

for founders’ prior work experience and other organizational and environmental factors, biotechnology

3

firms with proportionally more Ph.D.-holding entrepreneurs on the founding team have a higher

probability of adopting open science. A second noteworthy finding is that founders’ educational

background can mitigate the constraint of organizational environments on strategy. A crowded

technological niche confronts firms with a more challenging environment for implementing open science,

due to higher scooping risks. The deterring effect, however, of such a high-risk environment is smaller

among firms founded by proportionally more Ph.D.-holding entrepreneurs. There is also some evidence

that founders’ educational-background effect is stronger in an institutional environment in which open

science has not yet become the industry norm.

The rest of the paper is organized as follows. Section II introduces the open-science strategy and

develops hypotheses regarding founders’ professional-education-background effects. Section III describes

the data, model and variables. Results are presented in section IV. Conclusions and discussion follow in

Section V.

II. Founders’ Professional-Education Background and Open Science

The essence of open science is summarized in the frequently quoted statement by Merton (1968,

pp. 610–11): “The substantive findings of science are…assigned to the community…The scientist’s

claims to ‘his’ intellectual ‘property’ is limited to that of recognition and esteem…Secrecy is the

antithesis of this norm; full and open communication its enactment.” At its core, the idea of open science

is at odds with the definition of technology as private property, a notion widely accepted by for-profit

firms. Hence, firms traditionally employ very different models of knowledge creation and dissemination

from those used by academic research institutions. They tend to avoid investing in basic science research,

as the economic returns to knowledge generated from basic research are difficult to appropriate (Nelson

1959, Rosenberg 1990). In addition, firms rely heavily on protective mechanisms such as patents and

trade secrets to ensure their rights to any discoveries generated from the firm’s investment (Dasgupta and

David 1994).

However, many for-profit firms in the life sciences industry have adopted research and

publication policies that, to some extent, resemble those in the academic research sector (Gambardella

4

1992, Cockburn and Henderson 1996; Lim 2004). By 2002, among the U.S. public biotech firms founded

between 1969 and 2000 studied in my sample, approximately 80% of the firms had publications in ISI-

indexed journals.

There are certainly strategic benefits associated with an open-science policy. It can help boost a

firm’s capability to absorb public sector research (Cockburn and Henderson 1998, Gittelman and Kogut

2003, Fleming and Sorenson 2004, Sorenson and Singh 2007), attract high-quality recruits (Stern 2004),

or prevent competitors from winning a patent (Lichtman, Baker and Kraus 2000, Parchomovsky 2000).

On the other hand, there is a great deal of risk associated with the open-science policy, due to the

possibility that published discoveries will be scooped by competitors. In reality, no matter how carefully a

firm screens manuscripts before publication, the risk of being scooped always exists. Though systematic

data are lacking, disputes about scooping arise routinely among firms in technology-intensive sectors. For

example, in 1984 researchers at Cistron Biotechnology submitted a paper to Nature magazine that

contained the gene sequence of an Interleukin-1 beta immune system protein. Steven Gillis, then head of

the research department at Immunex, was one of the reviewers. He allegedly copied the sequence data and

used it in Immunex’s own patent application (Kokmen 1996). After a three-year lawsuit, Cistron won a

$21 million settlement in 1996.

The case highlights the tension between adopting a practice that is designed primarily for public-

sector research and the needs of industrial firms to protect their intellectual property rights to inventions

generated with the firm’s resources. In reality, firms adopt different practices. Some are more concerned

with the risks and impose an outright ban on employee publication. Other firms weigh the strategic

benefits carefully and encourage employees to publish some of their research as university researchers do.

Some firms may even use publication activity in the evaluation of their research staff. Almost all

publishing firms screen manuscripts carefully before submission. However, the extent of screening and

the extent of publication-support offered to research staff vary across firms. Such variation is possibly

driven by how key organizational stakeholders view the benefit-risk tradeoff associated with the practice.

For these reasons, variation among newly formed biotech firms regarding the open-science practice

5

provides a suitable context for investigating the relationship between founders’ professional-education

background and early organizational strategy.

How does biotech founders’ educational background affect the adoption of open science? A

noteworthy demographic feature of biotech entrepreneurship is the presence of Ph.D.-holding founders.

Among the 1,090 biotech founders studied in my data, 540 (49.5 percent) have a Ph.D. degree, 149 (13.7

percent) hold an M.D. degree, and the rest hold other types of degrees. Compared to most other types of

educational programs (e.g., M.B.A. or B.A.), Ph.D. training stands out in the intensity of members’

socialization of professional values and identity. The core values of Ph.D. training are also consistent with

the academic norm of open science. Thus, the question related to the setting of this study is: are firms

founded by Ph.D.-holding entrepreneurs more likely to adopt open science?

There are four possible reasons why a Ph.D.-holding founder may develop a more favorable

perception of open science. One reason has to do with a founder’s exposure to the norms of science

during doctoral training. The idea of full and timely disclosure of scientific findings is woven into much

of the Ph.D. training process and reinforced by the priority-based incentive system in academia. However,

a significant body of sociological literature in the 1960s and 1970s produced evidence that challenges the

accuracy of the Mertonian description of the normative structure of science (e.g., Merton 1963, Mulkay

and Williams 1971, Cole and Cole 1973, Latour and Woolgar 1979). More recent evidence by

Blumenthal and his colleagues (1997) has shown that faculty members will deviate from the norm of

timely disclosure of scientific findings when commercial interests are at stake. Thus, although Ph.D.-

holding founders may have more exposure than non-Ph.D.-holding founders to the scientific norm of

open communication, it is questionable to what extent Ph.D.-holding founders will commit to them and

whether their commitment lasts through the time of venture creation, especially when the incentives are at

odds with the professional norms (Wallace 1995).

A second possible reason is related to Ph.D. trainees’ more science-oriented cognitive structure

developed from their doctoral training. Studies have shown that the deployment of different cognitive

models may affect a wide range of organizational outcomes, including trajectories of technological

development (Garud and Rappa 1994), formation of corporate strategies (Barr, Stimpert and Huff 1992)

6

and corporate reaction to technological disruptions (Tripsas and Gavetti 2000; Kaplan, Murray and

Henderson 2003). The basic premise is that cognitive structure or mental models developed on the basis

of historical precedents affect the ways in which a manager processes information when making decisions.

For example, equipped with a more science-based cognitive orientation, Ph.D.-holding founders may

develop a better understanding of the methods for cutting-edge scientific investigation employed in

universities and, consequently, place more importance on the strategic benefits of open science. However,

recent findings on the rise of academic entrepreneurship suggest that commercial motives have been

integrated into the cognitive framework of university scientists, particularly in the life sciences. For

example, Etzkowitz (1998) noted that with the closing gap between academic research and commercial

utilization of the discoveries from academic research, university scientists increasingly look at their

research from a dual perspective—a traditional perspective in which discoveries should enter publication

process as soon as possible to establish one’s academic reputation, and an entrepreneurial perspective in

which results are scanned for their commercial as well as intellectual potential. This finding casts doubts

on the notion that there is a significant difference between the cognitive structure of Ph.D.-trained

founders and that of founders with other types of educational background.

Professional training is also about socializing trainees into believing in the status hierarchy of the

profession. Thus, a Ph.D.-trained founder might prefer open science because of status aspirations. Status

order within a profession is positively related to the level of abstraction of knowledge (Abbott 1988).

Though a Ph.D.-trained high-tech founder is supposed to focus more on developing commercially viable

technologies, being able to publish basic science research in academic journals allows him to maintain, to

some degree, a respectable status within the larger scientific community. In fact, many scientists-turned-

biotech-founders maintain affiliations with academic institutions, not only for keeping abreast of the latest

scientific developments, but also for maintaining standing in academia. An open-science strategy is

instrumental for achieving both of these purposes.

Finally, assuming strong inertia in scientists-turned-entrepreneurs’ networks (Mauer and Ebers

2006), a Ph.D.-holding entrepreneur’s network at the stage of founding a firm will be rich in two types of

contacts. One will be with people who share a similar educational background. By the time of venture

7

creation, many of these contacts will be working in academia. Such a network effectively reinforces any

preexisting normative orientation acquired by a Ph.D.-holding founder. By the principle of homophily, a

second type of contact will be those who are also Ph.D.-trained and are now founding or managing their

own start-ups. Information regarding the open-science strategy (e.g., how to implement it and what the

benefits are) is more likely to be circulating within such networks of scientists-turned-founders. Thus,

compared to a founder with a different educational background (e.g., M.B.A.), a Ph.D.-holding founder is

more likely to learn about open science, particularly when it is still new to the industry. Drawing on both

the status and network mechanisms discussed above, I expect that

Hypothesis 1: Firms with a higher proportion of Ph.D.s on the founding team are more likely to adopt an open science strategy.

In the process of venture creation, many factors other than founders’ visions can drive a new

firm’s strategy. Whether or not an entrepreneur’s vision for a start-up can be realized is often the

negotiated outcome between the entrepreneur and the organizational environment (Johnson 2007).

Two types of environment have often been examined in previous research—technological and

institutional. The technological environment constrains the choice of organizational strategy to those that

boost the new venture’s operational efficiency. As described earlier, a key aspect of concern in biotech

firms’ technology strategy is appropriability – the extent to which a firm can protect its intellectual

property from being imitated by its competitors. This is an important dimension of a biotech firm’s

technological environment, as most firms have no product at the time of founding, and new ventures are

often built on the mere promise of a discovery. When scooping risks are high and protection of

intellectual property is difficult, a firm is operating in a high-risk environment with regard to the open-

science strategy.

There is some documented evidence that suggests resilience in professional-education effect,

even in an environment that is contradicting to professional ideals. For example, Riegal (1958) and

Kornhauser’s (1962) studies of Ph.D. scientists working for industrial firms show that in situations where

professional values are in conflict with corporate goals, professionals still show considerable commitment

8

to the values they have internalized during their graduate training. In health care, concerns over loss of

professional autonomy were cited as one of the reasons for physicians’ reluctance to participate in health

maintenance organizations during the early proliferation of HMOs (Kralewski, et al. 1987, Rosenbach, et

al. 1988, Ku and Fisher 1990).

These findings suggest that compared to founders with alternative professional education

background, Ph.D.-scientists-turned-biotech founders may be less sensitive to the risks in the

technological environment. Note that this is not to argue that scientists-turned-founders are oblivious to

technological risks in their environment. Indeed, when appropriability risk is high in the environment, all

firms may be tempted to adopt a more conservative approach to open science. However, the extent of

firms’ reaction to the appropriability risks may differ between those with more Ph.D.-trained-founder

participation and those with less Ph.D.-trained-founder participation. If the professional values that

founders have internalized during their Ph.D. training remain influential in a corporate environment, we

should find that the presence of Ph.D.-trained founders has a mitigating effect on the organizational

environment.

Hypothesis 2: The negative effect of a high-appropriability-risk technological environment on the adoption of the open-science strategy is smaller in firms with more Ph.D. founder representation.

The extent of founders’ educational-background effect may vary in different institutional

environments. Hambrick and Finkelstein’s research (1987, 1990) has shown that the predictive power of

managers’ background on organizational outcomes is contingent on the degree of managerial discretion,

which is defined as the ability of executives to conceive multiple courses of action that lie within the zone

of acceptance of powerful parties (Hambrick and Finkelstein 1987). They propose that managers’

background has a larger impact in an environment where managers are allowed more variety and change

than in an environment where managers face constrained discretion.

When examining an organization’s institutional environment, an often-considered feature is the

degree of homogeneity in the organizational field or industry (DiMaggio and Powell 1983). Homogeneity

in an organizational field is driven by isomorphism among organizations. When isomorphic pressure is

9

high, there is less variety in the organizing models that are deemed legitimate by powerful organizational

stakeholders (e.g., venture capital firms that provide funding to the start-ups). Thus strong isomorphic

pressure reduces the range of the possible courses of actions an entrepreneur may take with regard to

organizational strategy. In the biotech context, firms occupying similar competitive positions in the

industry are likely to face a similar set of organizational stakeholders. An increase in the number of firms

adopting open science helps make the practice more popular and create pressure for the non-adopting

firms to conform. Thus, an institutional environment in which there is strong isomorphic pressure to adopt

open science provides less room for managerial discretion with regard to the strategy. Drawing on the

research that supports the moderating effect of managerial discretion on the relationship between manager

background and organizational outcomes (Hambrick and Finkelstein 1987, Finkelstein and Hambrick

1990), I expect that

Hypothesis 3: The effect of Ph.D. representation on a founding team on the adoption of the open-science strategy is weaker in an environment with high isomorphic pressure for adopting open science.

III. Data, Method and Variables

III.1 Data

My data include all dedicated biotechnology firms headquartered in the U.S. that have ever filed

IPO prospectuses (Form S-1, SB-2, or S-18) with the U.S. Securities and Exchange Commission. These

firms are identified based on information from industry directories and in the “business” section of firms’

prospectuses. I used three primary sources to identify biotechnology firms: the Compustat database, the

Bioscan Directory, and the Recombinant Capital database. A total of 512 biotechnology firms founded

between 1969 and 2000 have filed IPO prospectuses. Information coded from these filings includes: (i)

biographical sketches of founders, 1 scientific advisors, and senior executives (e.g., name, gender, age,

highest degree obtained and recent jobs held); (ii) firm founding year (i.e., year of incorporation), core 1 The SEC does not have specific requirements with regard to reporting firm founders. However, founders can be identified from the prospectuses if, at the time of IPO, they serve as executive officers or scientific advisors, are principal shareholders of the firm, or hold important intellectual properties on which the firm relies. About 70% of the founders in my data can be identified from the IPO-prospectus documents. I identified the remaining 30% via search over the Internet (e.g., the Bioscan and Informagen directories).

10

technology field and mode of creation (i.e., independently created versus corporate spin-off); and (iii)

financial data up to five years prior to the IPO-prospectus filing date.

One limitation of the data is that, due to the lack of any systematic historical data on private firms,

only information about biotechnology firms that have filed IPO prospectuses in the U.S. was gathered. As

a consequence of this sampling method, I am working with a selected sample. It is likely that the firms in

this database are relatively successful compared to the average start-up firm in the biotechnology sector.

The implications of this sampling issue are discussed in the next section.

I obtained firms’ publication records from ISI’s Web of Science database. In order to minimize

problems caused by corporate name changes, I tracked all previous company names using several

business information databases (e.g., One Source, Informagen Biotech and Pharmaceutical Company

Directory, and Thomson Financial Database). In addition, I also retrieved publication records for firms’

founders, scientific advisors and executives.

I used Recombinant Capital’s (RECAP) clinical trial database to construct variables of

organizational environments. RECAP tracks over 3,400 therapeutic products that have at some time been

in the clinical trial process since 1978. Each record contains information on the product developer, target-

disease area, underlying technology, and time when the product enters each stage of the clinical trial

process. I used the data to categorize biotech firms based on their core technology areas and to construct

measures of firms’ technological and institutional environments.

III.2 Model

Because it is not feasible to date when a firm officially adopted the open-science strategy, I relied

on the outcome of adoption—a firm’s published papers in scientific journals—as an indicator of a firm’s

adoption of the strategy. The advantage of relying on the outcome is that it captures what a firm has

actually done, as opposed to symbolic adoptions (i.e., the firm makes a policy announcement but fails to

follow through). However, an important disadvantage of the outcome-based measure is that it is more

likely confounded with factors such as firms’ ability to implement the newly adopted strategy. I deal with

11

this problem by including an extensive list of variables to control for firms’ ability to publish research in

journals (more discussion is in the subsection on variables).

Traditional diffusion research would analyze a dichotomous outcome of whether or not a firm has

published papers in journals. This traditional (logit) model of adoption can be problematic in the context

of this study. First, there might be miscellaneous reasons for a biotech firm to have a research publication

even though it doesn’t have a policy of endorsing open science. For example, a paper might be reported as

a firm’s publication because of an address change of a recently graduated employee—in such a case, the

published paper says nothing about the firm’s policy. Second, a simple dichotomous measure might not

capture the full picture of how open science is carried out in a firm. As discussed in section II, the core

issue for a firm is not only announcing a policy to its staff but, more importantly, deciding the extent to

which it will support the policy. These decisions include whether to encourage staff to devote a portion of

their time to basic-science research that is not closely related to the firm’s immediate technological goal,

whether to set up incentives for staff to engage in scientific publication, whether to use publications and

reputation in the public-science sector as part of staff-performance measures, or how much restriction to

impose when censoring researchers’ publication requests. Such decisions affect the level of a firm’s

commitment to open science and are reflected in a firm’s publication count.

For the above reasons, I rely on a count of a firm’s research publications as an indicator of

adoption. I use a quasi-maximum-likelihood (QML) Poisson model. Since the Poisson model is in the

linear exponential family, the coefficient estimates remain consistent as long as the mean of the dependent

variable is correctly specified. Furthermore, in a QML Poisson, “robust” standard errors are consistent

even if the underlying data-generating process is not Poisson (Gourieroux et al. 1984). QML Poisson is

preferred because it imposes little structure on the underlying data distribution and in general is a more

conservative estimate of the coefficients due to the larger standard errors. In addition to QML Poisson, I

conduct robustness test with a logit model, which follows the conventional diffusion analysis and tests the

dichotomous outcome of a firm’s publication activity.

III.3 Variables

12

Dependent Variable

The dependent variable is firm publication count of research papers published in ISI-indexed

scientific journals by the fifth year after it was founded. Preliminary analysis of the timing of biotech

firms’ publication activity shows that most firms that eventually have adopted open science begin

publishing within the first few years of their founding. For most adopting firms, increase in the

probability of publication begins to slow down after five years from inception, at which point 65 percent

of the firms (that eventually will publish) have already started publishing. The choice of a five-year

window also allows for the time needed to set up corporate research functions and for possible publication

lag due to editorial processes. At the same time, the gap between founding and adoption is short enough

that I can reliably assess the role played by organizational founders. Sensitivity analysis conducted using

a 4-, 6-, or 7-year observation window indicates that the results are largely robust to the cutoff age.

Founder Background Variables

I computed the proportion of founders with Ph.D. as an indicator of the professional-education

background of the founders in a firm.2 I expect that firms with a higher proportion of Ph.D. founders will

be more likely to have a higher propensity to pursue open science.

To control for founders’ prior work experiences, I computed the proportion of founders with

open-science ex-employer. The open-science ex-employers include academic institutions and firms that

have published, in ISI-indexed journals, five or more papers (with five being the average publication

count of the biotech firms in my data during the first five years since firm inception). I computed the

proportion of such entrepreneurs in a firm’s founding team to measure the level of influence coming from

founders with pro-open-science work experience.

Organizational Environment Variables

2 In an unreported robustness test, I ran the same set of regressions, replacing the proportion of founders with Ph.D. with the count of founders with Ph.D. The only meaningful difference between the two sets of models is that when using the count variable, the Ph.D. founder-institutional environment-interaction effect becomes weaker. These results are available upon request.

13

Technology niche density (ND) is computed to measure the appropriability risk in an

organization’s technological environment. To the extent that two firms compete in the same technological

niche, they pursue similar knowledge and it would be easier for them to scoop discoveries from each

other. Appropriability risk is thus associated with the characteristics of a firm’s technological niche: in

densely populated niches, the risk of being scooped is higher, as there are more potential imitators. Such

an environment poses more challenge for the adoption of open science.

This measure requires a meaningful way to divide the technological space in the biotech industry.

For this purpose I used the dataset on clinical trials tracked by RECAP, in which companies’ products in

the clinical trial process are each identified with one (or, in a few cases, more) primary technological

area(s). There are a total of 34 technological areas in RECAP’s classification. I measured niche crowding

for each firm as the number of other firms competing in the firm’s main technological niche at the time of

founding. There are several points to note about this measure. (1) For a firm with more than one

technology area, I used the one in which the firm has the largest number of clinical trials as its main

technological niche. (2) I used data on firms’ clinical trials up to the fifth year after the founding date,

when there is more information available for determining a firm’s primary niche. (3) The density count

includes not only the 512 public biotech firms in my dataset, but also the pharmaceutical and private

biotechnology firms in the U.S. and abroad (the dataset includes a total of 3,728 biotech and

pharmaceutical firms, public and private). The inclusion of these firms yields a more accurate measure of

the scooping risks. It is expected that a firm’s technology niche density has a negative effect on

publication propensity.

I have also constructed a measure for the isomorphic pressure (IP) in a firm’s institutional

environment. DiMaggio and Powell (1983) suggested that firms are more likely to select their imitation

targets based on how relevant the targets are to themselves. Following previous research (Burt 1987,

Bothner 2003), I constructed a structural-equivalence-based measure of the pressure for adopting open

science in an organization’s institutional environment. This measure captures isomorphism among firms

that are close competitors to each other. For each firm i, the isomorphic pressure (IP) in the industry

when the firm was founded is:

14

€

IPi = wijPjj=1, j≠ i

J

∑ (1)

where Pj is an indicator variable that is coded 1 if competitor j has, by firm i’s founding year, already

started publishing, and wij is a weight that captures the degree of similarity between i and j in the

competitive landscape—or in network terms, the structural equivalence between i and j. To compute the

weight I relied on firms’ technological areas based on the primary technologies underlying their products

already in the clinical trial process. Firms develop their core capabilities in technological areas in which

they have generated products that reach the clinical trial stage. In these areas, firms closely monitor their

rivals and thus are more susceptible to the isomorphic pressure from other firms. Therefore, for two firms

i and j, I measured their similarity to each other by comparing their product allocation in each of these 34

technological areas. Specifically,

€

dij =niknikk∑

−n jk

n jkk∑

⎛

⎝

⎜ ⎜

⎞

⎠

⎟ ⎟

k=1

34

∑2

(2)

where nik and njk are firms i’s and j’s counts of clinical trial cases in technological area k, and nik/Σknik and

njk/Σknjk are the proportions of i’s and j’s clinical trial cases in that area by the fifth year after the firm’s

inception. Essentially, this is a Euclidean distance measure based on differences in firms’ shares of

clinical trial products in each of the technological sectors. Next, the Euclidean distance is converted into a

structural proximity weight wij using:

€

wij =max(di) − dij[max(di) − dij ]j∑

(3)

wij is then applied to the influence of all rival firms in eq. (1) to generate the covariate IP, which is used to

estimate the institutional pressure created by a firm’s similarly positioned rivals. I expect IP to have a

positive effect on the probability that a firm will adopt open science.

Control Variables

A strong alternative explanation is that the observed effects could be driven by unobserved

heterogeneity across technology subfields. For example, Ph.D.-holding entrepreneurs may be more likely

15

to sort themselves into subfields where capabilities to understand basic science knowledge are crucial and

corporate strategy is more oriented towards open science. Indeed, there are significant heterogeneities

across the technological subfields in which biotech firms do their research. Some involve more basic

research (e.g., genomics and bioinformatics), which is more frequently published in academic journals,

while others deal with more applied knowledge, which does not commonly get published in academic

journals. To control for such heterogeneity, I included a series of technology area dummies in the models.

Specifically, I first coded a firm’s technological area based on the category in which it has the biggest

share of its products in the clinical trial process according to the RECAP data. For firms that do not have

any product covered by the clinical trial database, I relied on the information about their core business and

technologies disclosed in their IPO documents. These fields were then grouped into eight primary

technology areas. Table 1 reports the distribution of biotech firms across these areas as well as the

percentage of founders with a Ph.D. degree in an area.

---- INSERT TABLE 1 ABOUT HERE ----

I also included a measure of a firm’s technological niche age, which is proxied by the total

number of biotech patent applications in the niche by a given year. The more patents filed in a technology

area, the more mature the area becomes. This variable is to control for the possibility that younger

technological areas may favor open science more than the older ones.

Firms’ research performance can be affected by the structure of their networks (Maurer and Ebers

2006, Stuart, Ozdemir and Ding 2007). To control for the influence of firms’ academic collaborators, I

used the biotech alliances data available from RECAP, in which a total of 11,000 alliance agreements

have been tracked and analyzed since 1978. For each biotechnology firm, I computed its number of

alliances with universities from inception to age five. I expect this measure to have a positive effect on a

firm’s likelihood to publish its research.

I have included six variables to control for a firm’s research capabilities. First, I obtained the

founder publication count (pre-founding) to account for the possibility that a firm’s publication count is

driven up primarily by the founders’ ability to publish. Second, the founding team size was measured to

16

control for the possibility that large founding teams secure more resources for the firm. Third, I computed

the pre-founding publication count of scientific VP, and I used this count as a proxy for the quality of the

firm’s research personnel. Fourth, I included in the models a dummy variable indicating that the firm has

a SAB and a continuous measure of the maximum publication count among SAB members. These two

measures are to account for the possibility that a firm with a SAB of star scientists can attract research

staff who are good at publishing papers. Lastly, as another indicator of a firm’s research activities, I

counted the number of patents filed by a firm by age five.

A firm’s research output also depends on the available financial resources. I obtained, from IPO

prospectuses and Compustat, five financial measures: (i) R&D expenses, (ii) total assets, (iii) net sales,

(iv) net income, and (v) total invested capital. Financials at firm ages one and two were not used because

of substantial missing data. For ages three to five, there were still 25% of firms with no data, in which

case imputation was applied using the AMELIA program (King et al. 2001).

A total of 89 firms are corporate spin-offs. These firms may have different sources of cultural

influence from independently formed ones due to influence from their parent firms (Phillips 2002,

Chatterji 2009). I controlled for such differences with a dummy variable spin-off. Finally, a series of five-

year founding-period dummies are in the models to control for any unobserved heterogeneity at the time a

firm was founded.

The descriptive statistics and the correlation matrix are provided in Table 2.

--- INSERT TABLE 2 AND 3 ABOUT HERE ---

IV. Results

Table 3 presents results from the QML Poisson regressions of firm publication count by age five.

Model 1 reports the baseline estimates with only the control variables. Among them, it is not

surprising that a firm’s number of alliances with universities, publication count of scientific VP, and the

maximum publication count among SAB members all have positive effects on a firm’s publication count.

Younger technology areas seem to favor open science more than the older ones. Large founding teams

also increase the publication count. This may have to do with the fact that the pairing of business experts

17

with scientific experts is more often observed in founding teams with two or more entrepreneurs.

Consistent with previous findings (Gambardella 1992), I find a positive relationship between a firm’s

patent and publication counts. This indicates that most firms rely on both patenting and publication in

knowledge development. Financial resources also influence the count of publications, though the impact

is small in magnitude. Holding all other variables constant, each additional million dollars a firm spends

on R&D is associated with merely 1% increase in the expected publication count. Similar effects are

found for a firm’s net income. These results are not surprising given that firms are expected to devote

more of their resources to generating patents than publications. Spinoff firms also seem to have a higher

number of publications than independently formed ones. There is evidence that a founder’s prior work

experience has a strong influence on a firm’s open-science strategy. Holding all other variables constant,

a unit (i.e., from zero to 100 percent) increase of founders with some pro-open-science work experience is

associated with an increase in the expected publication count by approximately 81% (=exp[0.592]).

Model 2 tests my Hypothesis 1 on the Ph.D. educational background effect of founders. I

included the variable that measures the proportion of Ph.D.-holding scientists on a founding team. Like

founders’ prior work experience, this variable exerts a strong impact on the likelihood that a firm will

practice open science. I graphed a firm’s expected publication count at different levels of Ph.D.- founder

participation in Figure 1. When the predicted outcomes of a non-linear model such as Poisson are

computed, the predicted outcomes vary with the values assigned to the variables in the model. Therefore,

I drew three different lines in the graph. The dotted (mean) line shows the expected publication count for

firms with various levels of Ph.D-founder presence (from 0 to 100 percent) while holding all other

variables in Model 2 at their mean. The solid (ambivalent) line stands for expected publication count

holding all other variables at zero. The dashed line is based on holding all other variables at one standard

deviation. In addition to graphical presentation, incidence ratio can be computed directly from

exponentiation of the coefficient on the proportion of founders with Ph.D. in model 2. A unit (i.e., from

zero to 100 percent) increase of Ph.D. scientists on a founding team more than doubles the firm’s

expected publication count (=exp[0.783]). Alternatively, an increase of 41 percent (one standard deviation

18

of the variable proportion of founders with Ph.D.) is associated with an increase in the expected

publication count by a factor of 1.38 (=exp[0.783*0.41]).

---- INSERT FIGURES 1, 2 AND 3 ----

The significant effect of Ph.D.-founder representation in Model 2 lends support to H1 regarding

the impact of founders’ professional-education background on a firm’s likelihood to adopt the open-

science strategy. Note that while the prior-work-experience effect is strong in Model 1, its magnitude

drops and its effect is no longer significant after educational background is included in the estimation.

The correlation of the “proportion of founders with Ph.D.” and “proportion of founders with pro-open-

science ex-employer” is 0.348, hence multi-collinearity is not a major concern between the two variables.

Model 3 tests patterns of interaction between founders’ educational background and the

organization’s technology environment at the time of founding, as hypothesized in H2. Model 3 includes

two additional variables, a firm’s technology niche density (ND) to measure the appropriability risks in

the environment, and its interaction with the proportion of founders with Ph.D. The main effect of niche

density is negative and significant. Though the level of statistical significance of the interaction effect is

relatively low, it is consistent in most of the robustness-test models reported below.

Figure 2 illustrates the moderating effect of founders’ educational background on an

organization’s technological environment. Three lines were drawn for different types of firms—those

with zero, 50 percent or 100 percent Ph.D.s on the founding team. The lines represent a firm’s expected

publication count at different levels of technology crowding when all other variables in Model 3 are held

at their mean values. Firms with different levels of Ph.D. representation on the founding team react

differently to crowding. When a firm’s founding team does not include any Ph.D.-trained scientists (the

solid line), an increase in niche density from 1 to 66 (one standard deviation) is associated with a decrease

in the firm’s expected publication by half (=[3.58-1.73]/3.58). In comparison, for a firm founded entirely

by Ph.D.-holding entrepreneurs, the same level of increase in niche density is associated with only 18

percent drop in expected publication count (= [5.91-4.89]/5.91). The finding lends support to H2 that the

negative effect of a high-appropriability-risk technological environment on open science is smaller in

19

firms with more Ph.D.-founder representation. It suggests that Ph.D.-holding founders are modifying the

impact of a challenging technology environment for open science.

In Model 4, I added a measure of isomorphic pressure for adopting open science in a firm’s

institutional environment and its interaction with the proportion of founders with a Ph.D. Isomorphic

pressure creates a more favorable institutional environment for open science and increases a firm’s

publication count. As expected in H3, an environment with strong isomorphic pressure for adopting open

science provides less room for managerial discretion, hence the effect of founders’ educational

background is weaker in such an environment.

Figure 3 illustrates how the expected firm-publication count varies with the proportion of Ph.D.

founders in three different institutional environments with high (75th percentile), medium (50th percentile)

and low (25th percentile) levels of isomorphic pressure for adopting open science. Again, all other

variables constituting model 4 are set at their mean values. When isomorphic pressure is weak (solid line),

a 40 percent increase in Ph.D.-founder representation (about one standard deviation) is associated with a

66 percent increase in the firm’s publication count (=[2.37-1.43]/2.37). In contrast, when isomorphic

pressure for adoption is high (dashed line), the same level of increase in Ph.D. founder representation is

associated with only 35 percent increase in publication count (=[3.38-2.50]/2.50).

Mechanisms Underlying Ph.D. Founder Effect

In the theoretical exposition in section II, I have drawn on previous research that offered four

possible mechanisms leading to the hypothesis on the effect of Ph.D.-holding founders on organizational

adoption of open science. Among them, the first two mechanisms, which are based on exposure to the

norms of science during doctoral training and on differences in cognitive structure resulting from the

Ph.D. training, have been challenged empirically in the literature. Hence, compared to the other

mechanisms, these two are less likely to be the primary mechanism driving the relationship between

Ph.D.-trained founders and the adoption of open science by their firms.

Though the data that I have gathered are not fine-grained enough for definitively identifying the

remaining two mechanisms, some of the results from the regression models may be informative regarding

20

which of the remaining two mechanisms appears more robust in this research context. The third

mechanism revolves around entrepreneurs’ status aspiration in science. If this mechanism is robust, I

should observe that founders in some range of the academic status hierarchy appear more status-aspiring

than others, thus more willing to use open science as a means for maintaining or improving their

academic status. Past research suggests conformity to role expectations is high in the middle and low at

either end of a status order (Phillips and Zuckerman 2001). I thus expect that founders in the middle range

of the academic status hierarchy are more likely to pursue open science as a means for enhancing their

status in science after their transition to entrepreneurship. Since academic prestige is closely associated

with productivity, the variable founder publication count (pre-founding) serves as a proxy for the overall

status profile of a firm’s founders. In the baseline regression of Table 3, no significant relationship

appears between founder publication count (pre-founding) and firm publication count. The result remains

the same if the quadratic term of the founder publication count variable is added. Thus, on the surface, the

data do not seem to support status aspiration as a primary mechanism underlying the effect of founders’

professional education on early organizational strategy of biotech firms.

This leaves us with the last mechanism discussed in section II, which is about the role of

entrepreneurs’ social network ties. Previous research has shown that a biotech firm’s ability to enter into

alliances with university partners is closely related to the academic networks of the founders and

scientific advisors of the firm (Stuart, Ozdemir and Ding 2006). Hence, the variable number of alliances

with universities in the model serves as a crude measure of the academic ties enjoyed by the founders of a

firm. The effect of this variable in the baseline Model 1 of Table 3 is positive and significant. This seems

to offer tentative evidence in support of the network ties as a primary mechanism underlying the

relationship between the proportion of Ph.D.-holding founders in a new biotech firm and its likelihood to

pursue the open science strategy.

Issues of Selection

In most studies of manager-demographic-background effect, a thorny issue is the selection

problems in the data (Hambrick 2007). For example, the top-management-team (TMT) theory posits that

21

managers with technical expertise invest heavily in R&D. However, the observed relationship between

managerial background and corporate R&D strategy could be driven by the fact that technically

sophisticated managers are more likely drawn to work for heavy-R&D-investment firms. Alternatively, it

could be driven by the fact that firms with the plan of implementing a R&D-intensive strategy select

technically sophisticated managers into top management roles because of a perceived match between the

corporate strategy and the managers’ capabilities.

Studying founders’ educational background and start-up organizational strategy to some extent

mitigates the selection problem in the above example because an organization does not exist until its

founder has created it. However, selection might still arise in my data when scientists-turned

entrepreneurs sort themselves into technology areas that clearly benefit from a strategy like open science.

I have dealt with this form of selection by including technology area fixed effects, which help control for

unobserved heterogeneity across technology areas. It should be noted that technology area dummies offer

only a limited solution to the problem. Ideally, an instrumental variable analysis provides a better way of

identifying the founder background effect. However, this method rests on the availability of a valid

instrument with which to measure the variable of interest. Unfortunately, there is no good candidate in my

data that can serve as a valid instrument for the founder-educational-background variable.

A second form of selection in my data could take place when a firm’s external stakeholders (e.g.,

VC firms) decide to support firms founded by Ph.D. entrepreneurs who wish to adopt open science -- in

the belief that the combination of Ph.D.-trained entrepreneur and open science will enhance the start-up

firm’s performance. Such selections will likely result in higher survival chances of firms with Ph.D.-

trained founders that have adopted open science.

If this form of selection exists in my data, it is more likely to occur when open science is more

widely diffused and its strategic advantages better understood (i.e., when the institutional environment is

more favorable). However, the interaction between proportion of founders with Ph.D. and isomorphic

pressure (IP) in Model 4 of Table 3 is negative and significant, suggesting that the link between founder

educational background and firm adoption of open science is actually weaker when open science has

become more widespread.

22

---- INSERT TABLE 4 ABOUT HERE ----

As another way of testing against this form of selectivity, I ran an analysis of the joint effect of

founder-educational background and firms’ open-science strategy on firms’ financial performance. If it is

true that returns to open science in the form of financial performance are higher for firms founded by

Ph.D.-holding entrepreneurs, we should see, on some measures of firm performance, a positive interaction

effect between founders’ Ph.D. education and firms’ adoption of open science. Table 4 reports analysis of

firms’ financial performance—log of sales and net income. Note that this test uses only the public biotech

firms in my sample. Models 1 and 3 include a dummy indicating a firm has adopted open science, a

dummy indicating the firm has one or more Ph.D. founders, and their interaction term. Models 2 and 4

include firm publication count, proportion of founders with Ph.D., and their interaction. There is no

significant interaction effect in any of these models. This test suggests that at least among the public

biotech firms featured in my data, selectivity due to higher return to open science for Ph.D.-founded firms

is not a serious concern.

There might be a third form of selection taking place in the industry context that I am

investigating. I found in my analysis that the relationship between Ph.D.-founder representation and firm

publication is stronger in adverse organizational environments. I attributed this to the strong influence of

entrepreneurs in reducing environmental determinism. An alternative explanation may be that, during the

early stage of the biotech industry, it was primarily academicians with a Ph.D. degree who populated the

biotech start-up firms. Non-Ph.D. entrepreneurs only successfully entered the industry during the later

stage of the industry, when open science has been more widely accepted. First of all, inclusion of a series

of period dummy variables in the model helps to minimize the possibility that my results are driven by

this form of selection. Second, I report average firm-founding years for each of the technology areas in

my data in the last column of Table 1. The difference is modest across technology areas. Genomics and

bioinformatics is the only area that is markedly different from the rest of the areas in terms of average

firm founding year. This is also an area with the highest representation of Ph.D. founders. This area,

however, emerged during the more recent period in biotech history.

23

Other Robustness Tests I conducted several alternative estimations to test the robustness of my core results from the QML

Poisson estimation. The results are summarized below.3

First, following conventional diffusion research, I used the dichotomous adoption outcome, which

is coded 1 if a firm has published at least one paper, and estimated a logit regression of firms’ adoption of

open science. The results remain consistent with those in the main estimations.

Second, a challenge to using the firm publication count as the dependent variable is that many of

the biotech founders remain academically affiliated even after they have founded their venture. I might be

over-counting a firm’s publications if its academic founder publishes under the firm’s name even though

the publications are related to his work at his university. There is also the possibility of under-counting a

firm’s publications if the academic founders publish firm-generated research under their university

affiliations. I test against such biases by excluding founders’ contribution to firm publication count and

focusing on the share of count generated by the other members of the firm. The results confirm the

findings in the main models.

Third, I test the robustness of my core models by adding a control for the proportion of a firm’s

employees with a Ph.D. degree. The concern is that the firm research capability controls currently in the

main models are not adequate in accounting for the contribution of Ph.D.-holding employees in

generating firm publications. More specifically, the impact of founder-Ph.D. background on open science

may be due to the fact that Ph.D.-trained founders could hire more employees with a Ph.D. background

and these employees are generating higher publications for the firm. I conducted the test on a subset of

the data—290 firms that have reported employee academic degree breakdown in their IPO documents.

Though the inclusion of this variable reduces the magnitude of the founders’ Ph.D.-education effect, it

does not lead to major changes in the core results.

3 Full results of the robustness tests are available from the author upon request.

24

Fourth, I replace the variable proportion of founders with Ph.D. in the main models with a

dichotomous measure of whether a firm has at least one Ph.D. on its founding team. The main effect of

Ph.D. participation on open science remains positive and significant. However, in this specification, there

is no differential effect of founders’ educational background with the change of either the firm’s

technological or institutional environment. This suggests that significant interaction effects in the main

models might be driven not just by the mere presence of Ph.D.-holding founders, but by some sort of

power balance between founders with Ph.D. background and those without.

VI. Conclusion and Discussion

This paper addresses the question whether founders’ professional-education background affects

organizational strategies of new ventures. I sampled 512 U.S. biotech firms and analyzed their practice of

open science, which, traditionally found in academia, has been gradually diffused among for-profit,

technology-intensive firms. The results from a broad range of alternative specifications appear to support

that professional education is a significant source for founders’ visions of a start-up and that it influences

the choice of organizational strategy. In the biotech context, I find that firms created by proportionally

more founders with a Ph.D. degree were much more likely to adopt open science than those created by

founders with other types of educational background. There is tentative evidence suggesting that patterns

of entrepreneurs’ social networks could be the primary mechanism underlying this observed relationship.

In addition, by and large the results suggest that founders’ educational background may

counterbalance the effect of organizational environment. When a biotech firm has more Ph.D.-trained

founders, it is less deterred by a high-risk technological environment from pursuing an open-science

strategy. When examining founders’ educational background effect in different institutional environments,

I find suggestive evidence for a stronger effect of founders’ educational background when the strategy of

interest has not yet become institutionalized in the organizational environment.

The finding of this study adds to the organizational literature by providing evidence regarding the

influence of founders’ education on the choice of new venture strategies. Such evidence complements the

growing body of work that highlights entrepreneurs’ background in our understanding of the new-

25

venture-creation process (Burton 2001, Burton, Sorensen and Beckman 2002, Shane and Khurana 2003,

Sine, Haveman and Tolbert 2005). The evidence from biotech suggests that how a new venture is

organized is not solely determined by its structural or institutional conditions -- nor by its external

stakeholders, such as venture-capital firms. Entrepreneurs draw from their educational background (or

employment history as shown in the other studies) to form their unique visions for a firm, and such

visions could further influence a new venture’s strategy and structure.

Moreover, the founder-environment interaction patterns in my study suggest that, influenced by

his professional values and orientations, an entrepreneur may choose organizational strategies and

practices that deviate from what has been considered appropriate in an organizational environment. This

has important implications for understanding the emergence of novel organizational practices in a field.

New organizations are important birthplaces of innovations. It is very often the possibility of carrying out

a new idea or a new vision that attracts individuals into entrepreneurship. If education has a nontrivial

effect in moderating the effects of an organization’s technological and institutional environments, then, by

observing demographic shifts in entrepreneurs’ educational background, we may be able to identify

sources of exogenous shocks that can potentially trigger institutional changes in an organizational field. In

this case the influx of Ph.D.-trained scientist-entrepreneurs seems to have contributed significantly to the

emergence and gradual diffusion of the practice of open science in the biotech industry. A similar

example can be found in the case of the “Chicago Boys”, a group of University-of-Chicago-trained

economists who worked under the Pinochet regime in Chile in the 1970s and helped launch the first

radical-free-market reform in that country (Valdes 1995).

The present study also speaks to the literature on the exchanges between public and private

sciences. While previous literature on the normative influence between the two sectors has focused on the

importation of commercial norms into public science (Blumenthal et al. 1997, Agrawal and Henderson

2002, Owen-Smith and Powell 2004, Krimsky 2003, Owen-Smith 2005, Azoulay, Ding and Stuart 2009),

I examined possible influences in the opposite direction, i.e., the importation of academic norms into

private science. The extent of the influence of academic norms on the organization of private science is

yet to be determined through more research. Nonetheless, this study illustrates that when exchanges take

26

place between two sectors governed by different value systems, penetration of values and norms can go

both ways. This is particularly true when there are significant personnel exchanges between the sectors, as

in the history of the biotech industry. Certain aspects of academic norms have been transmitted to the

biotech industry along with the influx of scientists-turned entrepreneurs.

One legitimate concern is the generalizability of the arguments in this paper. Biotech is hardly the

only industry that sees a significant increase in the number of firms founded by Ph.D.-trained

entrepreneurs. Other science-based industries (e.g., nano-technology and advanced materials) appear to be

witnessing increasing participation of entrepreneurs with advanced professional training. However,

biotechnology is a distinctive industry, and any claim to the contrary rests upon a shaky foundation. Thus,

caution is needed in applying the arguments of this study to other industries. For example, the strong legal

protection of corporate patents in the life sciences industry gives firms confidence to patent their

discoveries. Such confidence may also influence firms’ publishing behavior. As this example illustrates,

careful examination of the nature of the technologies and the legal environment is necessary when

extending the results of this study to other industries.

Finally, despite the broad range of tests against possible endogeneity issues, the observed finding

regarding Ph.D. founders should be interpreted as tentative evidence for the hypothesis of entrepreneurial

educational-background effect on new venture strategy. The observed effects in this study could be driven

by either founders’ influence on open-science adoption, or their influence on firms’ performance after

adoption, or a combination of both. With only public firms in the sample, I offered limited tests against

such selectivity. In addition, to rule out the possibility that Ph.D. scientists sort into the scientific fields in

which their expertise helps the execution of open science, I used scientific field dummies in combination

with an extensive list of control variables for the characteristics of scientific fields. However, this form of

endogeneity can be better tackled with an instrumental variable analysis of the data, which is not

performed due to lack of valid instruments. Hence, the findings of this paper should be interpreted with

caution. Future research on this topic may provide more conclusive evidence.

27

References

Abbott, A. 1988. The System of Professions: An Essay on the Division of Expert Labor. Chicago, IL: The University of Chicago Press.

Agrawal A., R. Henderson. 2002. Putting Patents in Context: Exploring Knowledge Transfer from MIT. Management Science 48(1) 44-60.

Azoulay, P., W.W. Ding, T.E. Stuart. 2009. The Impact of Academic Patenting on the Rate, Quality, and Direction of (Public) Research Output. Journal of Industrial Economics 57 637-676.

Bantel, K. A., S. E. Jackson. 1989. Top Management and Innovations in Banking: Does the Composition of the Top Team Make a Difference? Strategic Management Journal 10 107-124.

Baron, J. N., M.D. Burton, M.T Hannan. 1996. The Road Taken: Origins and Evolution of Employment Systems in Emerging Companies. Industrial and Corporate Change 5 239-276.

—. 1999. Engineering bureaucracy: The Genesis of Formal Policies, Positions, and Structures in High-Technology Firms. Journal of Law Economics & Organization 15 1-41.

Baron, J.N., M.T. Hannan, M.D. Burton. 1999. Building the Iron Cage: Determinants of Managerial Intensity in the Early Years of Organizations. American Sociological Review 64 527-547.

—. 2001. Labor Pains: Change in Organizational Models and Employee Turnover in Young, High-Tech Firms. American Journal of Sociology 106 960-1012.

Barr, P. S., J. L. Stimpert, A.S. Huff. 1992. Cognitive Change, Strategic Action, and Organizational Renewal. Strategic Management Journal 13 15-36.

Bates, T. 1990. Entrepreneur Human-Capital Inputs and Small Business Longevity. Review of Economics and Statistics 72(4) 551-559.

Beckman, C.M. 2006. The Influence of Founding Team Prior Company Affiliations on Firm Behavior. Academy of Management Journal 49 741-748.

Beckman, C. M., M. D. Burton. 2008. Founding the Future: Path Dependence in the Evolution of Top Management Teams from Founding to IPO. Organization Science 19(1) 3-24.

Blumenthal, D., E. G. Campbell, M. S. Anderson, N. Causino, K. S. Louis. 1997. Withholding Research Results in Academic Life Science - Evidence from a National Survey of Faculty. JAMA-Journal of the American Medical Association 277(15) 1224-1228.

Boeker, W. 1988. Organizational Origins: Entrepreneurial and Environmental Imprinting at the Time of Founding. G. Carroll, ed. Ecological Models of Organizations. Cambridge, MA: Ballinger.

Bothner, M. S. 2003. Competition and social influence: The Diffusion of the Sixth-Generation Processor in the Global Computer Industry. American Journal of Sociology 108(6) 1175-1210.

Brüderl, J., P. Preisendorfer, R. Ziegler. 1992. Survival Chances of Newly Founded Business Organizations. American Sociological Review 57(2) 227-242.

Burt, R. S. 1987. Social Contagion and Innovation - Cohesion Versus Structural Equivalence. American Journal of Sociology 92(6) 1287-335.

28

Burton, M. D. 2001. The Company They Keep: Founders’ Models for Organizing New Firms. C. B. Schoonhoven and E. Romanelli, eds. Entrepreneurship Dynamic: Origins of Entrepreneurship and the Evolution of Industries. Stanford, CA: Stanford University Press.

Burton, M. D., C. M. Beckman. 2007. Leaving a legacy: Position Imprints and Successor Turnover in Young Firms. American Sociological Review 72(2) 239-266.

Burton, M. D., J. B. Sørensen, C. M. Beckman. 2002. Coming from Good Stock: Career Histories and New Venture Formation. M. Lounsbury and M. J. Ventresca, eds. Research in the Sociology of Organizations. Oxford, UK: JAI Press.

Carroll, G. R., E. Mosakowski. 1987. The Career Dynamics of Self-Employment. Administrative Science Quarterly 32(4) 570-589.

Chatterji, A. 2009. Spawned with a Silver Spoon: Entrepreneurial Performance and Innovation in the Medical Device Industry. Strategic Management Journal 30 185-206.

Cockburn, I., R. Henderson. 1996. Public-Private Interaction in Pharmaceutical Research. Proceedings of the National Academy of Sciences of the United States of America 93(23) 12725-12730.

Cockburn, I., R. Henderson. 1998. Absorptive Capacity, Coauthoring Behavior, and the Organization of Research in Drug Discovery. Journal of Industrial Economics 46(2) 157-182.

Cole, J. R. and S. Cole. 1973. Social Stratification in Science. Chicago, IL: University of Chicago Press.

Dasgupta, P., P. A. David. 1994. Toward a New Economics of Science. Research Policy 23(5) 487-521.

DiMaggio, P. J., W. W. Powell. 1983. The Iron Cage Revisited - Institutional Isomorphism and Collective Rationality in Organizational Fields. American Sociological Review 48(2) 147-160.

Evans, D. S., L. S. Leighton. 1989. Some Empirical Aspects of Entrepreneurship. American Economic Review 79(3) 519-535.

Etzkowitz, H. 1998. The Norms of Entrepreneurial Science: Cognitive Effects of the New University-Industry Linkages. Research Policy 27 823-833.

Finkelstein, S., D. C. Hambrick. 1990. Top-management-team Tenure and Organizational Outcomes – the Moderating Role of Managerial Discretion. Administrative Science Quarterly 35(3): 484-503.

Fleming, L., O. Sorenson. 2004. Science as a Map in Technological Search. Strategic Management Journal 25(8-9) 905-928.

Gambardella, A. 1992. Competitive Advantages from in-House Scientific-Research - the United-States Pharmaceutical-Industry in the 1980s. Research Policy 21(5) 391-407.

Garud, R., M. A. Rappa 1994. A Socio-Cognitive Model of Technology Evolution: The Case of Cochlear Implants. Organization Science 5(3) 344-362.

Gittelman, M., B. Kogut. 2003. Does Good Science Lead to Valuable Knowledge? Biotechnology Firms and the Evolutionary Logic of Citation Patterns. Management Science 49(4) 366-382.

Gourieroux, C., A. Monfort, A. Trognon. 1984. Pseudo Maximum-Likelihood Methods - Applications to Poisson Models. Econometrica 52(3) 701-720.

Hambrick, D. C. 2007. Upper Echelon Theory: An Update. Academy of Management Review 32(2) 334-343.

29

Hambrick, D. C., S. Finkelstein. 1987. Managerial Discretion-A Bridge between Polar Views of Organizational Outcomes. Research in Organizational Behavior 9 369-406.

Hambrick, D. C., P. A. Mason 1984. Upper echelons: The organization as a reflection of its top managers. Academy of Management Review 9 193-206.

Haunschild, P. R., A. D. Henderson, et al. 1999. CEO Demographics and Acquisitions: Network Effects of Educational and Functional Background. Corporate Social Capital and Liability T. A. J. Leenders and S. M. Gabbay, Springer.

Higgins, M.C. 2005. Career Imprints: Creating Leaders across an Industry. San Francisco: Jossey Bass.

Johnson, V. 2007. What is Organizational Imprinting? Cultural Entrepreneurship in the Founding of the Paris Opera. American Journal of Sociology 113(1) 97-127.

Kaplan, S., F. Murray, R. Henderson. 2003. Discontinuities and Senior Management: Assessing the Role of Recognition in Pharmaceutical Firm Response to Biotechnology. Industrial and Corporate Change 12(4) 203-233.

Kimberly, J. R., M. J. Evanisko 1981. Organizational Innovation: The Influence of Individual, Organizational, and Contextual Factors on Hospital Adoption of Technological and Administrative Innovations. Academy of Management Journal 24(4) 689-713.

Kim, P., H. E. Aldrich, L. Keister. 2006. Access (Not) Denied: The Impact of Financial, Human, and Cultural Capital on Entrepreneurial Entry in the United States. Small Business Economics 27 5-22.

King, G., J. Honaker, A. Joseph, K. Scheve. 2001. Analyzing incomplete political science data: An alternative algorithm for multiple imputation. American Political Science Review 95(1) 49-69.

Kokmen, L. 1996. Seattle’s Immunex Settles Lawsuit by New Jersey’s Cistron Biotech. Seattle Times. Seattle, WA.

Kornhauser, W. 1962. Scientists in Industry: Conflict and Accommodation. Berkeley, CA: University of California Press.

Kralewski, J. E., B. Dowd, R. Feldman, J. Shapiro. 1987. The Physician Rebellion. New England Journal of Medicine 316(6) 339-342.

Krimsky, S. 2003. Science in the Private Interest—Has the Lure of Profits Corrupted Biomedical Research? Lanham, MD: Rowman & Littlefield.

Ku, L., D. Fisher. 1990. The Attitudes of Physicians toward Health-Care Cost-Containment Policies. Health Services Research 25(1) 25-42.

Larson, M. S. 1977. The Rise of Professionalism: A Sociological Analysis. Berkeley, CA: University of California Press.

Latour, B., S. Woolgar. 1979. Laboratory Life: The Construction of Scientific Facts. Beverly Hill, CA, Sage.

Lazear, E.P. 2005. Entrepreneurship. Journal of Labor Economics 23(4) 649-680.

Lerner, J. 1995. Patenting in the Shadow of Competitors. Journal of Law, Economics 38(2) 463-495.

Lichtman, D., S. Baker, K. Kraus. 2000. Strategic Disclosure in the Patent System. Vanderbilt Law Review 53(6) 2175-2217.

30

Lim, K. 2004. The Relationship between Research and Innovation in the Semiconductor and Pharmaceutical Industries (1981-97). Research Policy 33 287-321.

Maurer, I., M. Ebers. 2006. Dynamics of Social Capital and Their Performance Implications: Lessons from Biotechnology Start-ups. Administrative Science Quarterly 51(2) 262-292.

Merton, R. K. 1963. Resistance to the Systematic Study of Multiple Discoveries in Science. Archives Europeennes de Sociologie 4(2) 237-282.

Merton, R.K. 1968. Social Theory and Social Structure. New York: Free Press.

Miller, G. A. 1967. Professionals in Bureaucracy - Alienation among Industrial Scientists and Engineers. American Sociological Review 32(5) 755-768.

Mulkay, M. J., A. T. Williams. 1971. Sociological Study of a Physics Department. British Journal of Sociology 22(1) 68-82.

Nelson, R. R. 1959. The Simple Economics of Basic Scientific-Research. Journal of Political Economy 67(3) 297-306.

Owen-Smith, J. 2005. Trends and Transitions in the Institutional Environment for Public and Private Science. Higher Education 49(1-2) 91-117.

Owen-Smith, J., W. W. Powell. 2004. Careers and Contradictions: Faculty Responses to the Transformation of Knowledge and its Uses in the Life Sciences. Sociologie Du Travail 46(3) 347-377.

Parchomovsky, G. 2000. Publish or Perish. Michigan Law Review 98(4) 926-952.

Phillips, D. J. 2002. Genealogical Approach to Organizational Life Chances: The Parent-Progeny Transfer among Silicon Valley Law Firms, 1946-1996. Administrative Science Quarterly 47(3) 474-506.

Phillips, D. J., E. W. Zuckerman. 2001. Middle-Status Conformity: Theoretical Restatement and Empirical Demonstration in Two Markets. American Journal of Sociology 107(2) 379-429.

Podolny, J. M., T. E. Stuart, M. T. Hannan. 1996. Networks, Knowledge, and Niches: Competition in the Worldwide Semiconductor Industry, 1984-1991. American Journal of Sociology 102 659-689.

Rider, C. I. 2008. Don't We Know You? Embedding Inter-organizational Relations in Organizational Members' Prior Education and Employment Networks. Working paper.

Riegel, J. W. 1958. Administration of Salaries and Intangible Rewards for Engineers and Scientists. Ann Arbor, MI: University of Michigan.

Rosenbach, M. L., B. S. Harrow, S. Hurdle. 1988. Physician Participation in Alternative Health Plans. Health Care Financing Review 9 63-80.

Rosenberg, N. 1990. Why Do Firms Do Basic Research with Their Own Money. Research Policy 19(2) 165-174.

Shane, S., R. Khurana. 2003. Bringing Individuals Back In: the Effects of Career Experience on New Firm Founding. Industrial and Corporate Change 12(3) 519-543.

Simons, T., P.W. Roberts. 2008. Local and Non-local Pre-founding Experience and New Organizational Form Penetration: the Case of Israeli Wine Industry. Administrative Science Quarterly 53(2) 235-265.

31

Sine, W.D., H.A. Haveman, P.S. Tolbert. 2005. Risky Business? Entrepreneurship in the New Independent-Power Sector. Administrative Science Quarterly 50(2) 200-232.

Sorensen, J.B. 2007. Bureaucracy and Entrepreneurship: Workplace Effects on Entrepreneurial Entry. Administrative Science Quarterly 52 387-412.

Sorenson, O., J. Singh. 2007. Science, Social Networks and Spillovers. Industry and Innovation 14(2) 219-238.

Stern, S. 2004. Do Scientists Pay to be Scientists? Management Science 50(6) 835-853.

Stuart, T. E., W. W. Ding. 2006. When Do Scientists Become Entrepreneurs? The Social Structural Antecedents of Commercial Activity in the Academic Life Sciences. American Journal of Sociology 112(1) 97-144.

Stuart, T. E., S. Z. Ozdemir, W. W. Ding. 2007. Vertical Alliance Networks: The Case of University-Biotechnology-Pharmaceutical Alliance Chains. Research Policy 36(4) 477-498.

Tripsas, M., G. Gavetti. 2000. Capabilities, Cognition and Inertia: Evidence from Digital Imaging. Strategic Management Journal 21 1147-1161.

Valdés, J.G. 1995. Pinochet's Economists: The Chicago School of Economics in Chile. Cambridge: Cambridge University Press.

Wallace, J. E. 1995. Organizational and Professional Commitment in Professional and Non-professional Organizations. Administrative Science Quarterly 40(2) 228-255.

Wiersema, M. F., K. A. Bantel 1992. Top Management Team Demography and Corporate Strategic Change. Academy of Management Journal 35 91-121.

32

Table 1: Distribution of Biotech Firms across Technology Areas

Technology Areas Number (percent)

of Firms Fraction of Founders

with Ph.D.* Mean Firm

Founding Year Synthetics and semi-synthetics 124 (24.2%) 48.7% 1989.3 Generics and other 101 (19.7%) 56.4% 1986.4 Monoclonals 100 (19.5%) 45.6% 1987.1 Drug delivery and vaccines 73 (14.3%) 41.8% 1988.6 Enzyme inhibitors, peptides, antisense and

ribozymes, interferon and interleukins 40 (7.8%) 55.6% 1986.6

Recombinant DNA, Cell and Gene Therapy 34 (6.6%) 39.2% 1987.7 Genomics and bioinformatics 29 (5.7%) 66.7% 1993.8 Factors – Growth, Blood and Hematopoietic 11 (2.2%) 37.5% 1987.8

* The fraction is computed by dividing the number of Ph.D.-holding founders of firms in a technology area by the total number of founders with all degree backgrounds in the technology area.

33

Table 2 Descriptive Statistics and Correlation Matrix

Mean Standard

Deviation (1) (2) (3) (4) (5) (6) (7) (8) (9)

(1) Firm publication count 5.643 11.854 1

(2) Firm publication dummy 0.594 0.492 0.394 1

(3) Founding year 1988.1 5.541 0.161 0.207 1

(4) Technological niche age 191.1 174.9 0.101 0.136 0.674 1

(5) Num. of alliances with univ. 1.186 2.048 0.220 0.273 0.153 0.161 1

(6) Founding team size 2.111 1.276 0.136 0.094 0.086 0.132 0.020 1

(7) Pub. count of scientific VP 24.55 47.61 0.182 0.184 0.111 0.088 0.185 -0.015 1

(8) Firm has a SAB 0.635 0.482 0.054 0.116 0.098 0.116 0.142 -0.011 0.161 1

(9) Max. pub. count among SAB members 70.15 116.6 0.173 0.159 0.097 0.110 0.077 0.043 0.220 0.457 1

(10) Founder publication Count (pre-founding) 9.346 35.34 0.021 0.058 0.014 0.068 0.091 0.254 0.068 -0.023 0.001

(11) R&D expenses (in million $) 7.619 14.30 0.348 0.173 0.331 0.201 0.109 0.180 0.060 -0.042 0.057

(12) Spin-off 0.164 0.371 0.119 0.012 0.098 0.111 0.022 0.266 -0.028 -0.091 -0.050

(13) Number of patents 4.320 8.862 0.318 0.179 0.099 0.152 0.046 0.036 0.049 0.006 0.166

(14) Total assets (in million $) 26.95 55.13 0.189 0.193 0.246 0.217 0.077 0.123 0.046 0.023 0.067

(15) Net sales (in million $) 5.024 11.18 0.195 0.135 0.096 0.068 -0.028 0.023 0.009 -0.071 0.001

(16) Net income (in million $) -7.281 12.35 -0.147 -0.184 -0.387 -0.261 -0.113 -0.117 -0.087 0.001 -0.062

(17) Total invested capital (in mm$) 36.09 92.14 0.270 0.215 0.324 0.256 0.100 0.154 0.033 0.011 0.089

(18) Proportion of founders with open science ex-employer 0.640 0.405 0.100 0.181 0.120 0.047 0.111 -0.150 0.037 0.061 0.034

(19) Proportion of founders with Ph.D. 0.505 0.410 0.167 0.207 0.132 0.033 0.073 -0.050 0.073 0.181 0.047

(20) Tech. niche density (ND) 42.68 66.07 -0.075 -0.033 0.486 0.699 -0.005 0.174 0.015 0.049 0.046

(21) Isomorphic pressure (IP) 0.700 0.614 0.202 0.235 -0.103 0.009 0.082 -0.032 0.016 0.062 0.065

34

Table 2 Descriptive Statistics and Correlation Matrix (Continued)

(10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21)

(10) Founder publication Count (pre-founding) 1

(11) R&D expenses (in million $) -0.003 1

(12) Spin-off -0.061 0.159 1

(13) Number of patents -0.008 0.192 0.064 1

(14) Total assets (in million $) -0.008 0.427 0.129 0.123 1

(15) Net sales (in million $) -0.006 0.359 0.016 0.253 0.445 1

(16) Net income (in million $) 0.010 -0.622 -0.102 -0.129 -0.486 -0.317 1

(17) Total invested capital (in mm$) -0.015 0.662 0.176 0.168 0.852 0.489 -0.616 1

(18) Proportion of founder with open science ex-employer 0.065 -0.017 -0.268 0.066 0.052 -0.024 0.008 0.012 1

(19) Proportion of founders with Ph.D. 0.016 0.017 -0.057 0.076 0.028 -0.036 -0.012 0.024 0.348 1

(20) Tech. niche density (ND) 0.058 0.169 0.050 -0.012 0.149 0.004 -0.270 0.163 -0.030 -0.031 1

(21) Isomorphic pressure (IP) 0.009 0.121 0.021 0.152 0.119 0.244 -0.087 0.101 0.059 0.024 -0.070 1

35

Table 3 Quasi-Maximum-Likelihood Poisson Estimates of Firm Publication Count By Age Five  

(1) (2) (3) (4) -0.001 -0.001 0.0003 -0.0002 Technological niche age (0.001)* (0.001) (0.001) (0.001) 0.105 0.113 0.103 0.099 Number of alliances with universities (0.027)** (0.024)** (0.024)** (0.025)** 0.228 0.247 0.255 0.277 Founding team size (0.064)** (0.065)** (0.065)** (0.066)** 0.005 0.004 0.004 0.004 Publication count of scientific VP (0.001)** (0.001)** (0.001)** (0.001)** -0.055 -0.132 -0.187 -0.242 Firm has SAB (0.209) (0.206) (0.202) (0.207) 0.001 0.001 0.001 0.001 Max. pub. count among SAB members (0.000)** (0.000)** (0.000)** (0.000)** -0.001 -0.001 -0.001 -0.001 Founder publication count (pre-founding) (0.002) (0.002) (0.002) (0.002) 0.024 0.020 0.019 0.019 Number of patent applications (0.006)** (0.006)** (0.006)** (0.006)** 0.009 0.010 0.009 0.009 R&D expenses (mm$) (0.004)* (0.005)* (0.005)* (0.005)† -0.002 -0.002 -0.002 -0.002 Total assets (mm$) (0.002) (0.002) (0.002) (0.002) 0.011 0.011 0.008 0.003 Net sales (mm$) (0.007) (0.007) (0.007) (0.008) 0.015 0.015 0.011 0.011 Net income (mm$) (0.005)** (0.005)** (0.005)* (0.006)† 0.002 0.002 0.002 0.002 Total invested capital (mm$) (0.001) (0.001) (0.001) (0.001) 0.873 0.895 0.789 0.746 Spin-off (yes = 1) (0.277)** (0.278)** (0.254)** (0.269)** 0.592 0.285 0.226 0.078 Proportion of founders w/ publishing ex-employers (0.263)* (0.261) (0.252) (0.259)

0.783 0.489 1.350 Proportion of founders with Ph.D. (0.214)** (0.274)† (0.360)** -0.011 -0.005 Technology niche density (ND) (0.004)** (0.002)* 0.008 Proportion of founders with Ph.D. ND (0.005)† 0.630 Isomorphic pressure (IP) (0.209)** -0.585 Proportion of founders with Ph.D. IP (0.261)*

0.731 0.389 0.606 0.026 Constant (0.537) (0.535) (0.518) (0.482) L -2554.53 -2471.85 -2398.91 -2353.87 Wald Chi2 458.58 459.02 484.77 509.09 d.f. 26 27 29 30

Notes: (1) All models control for founding period dummies (grouped in five years) and technology area dummies. (2) Number of Observations = 512. (3) Robust standard errors in parentheses. (4) † significant at 10%, * significant at 5%, ** significant at 1%.

36

Table 4 OLS Regression of Firm Financial Performance at Age Five   (1)

Log of Net Sales ($mm)

(2) Log of Net

Sales ($mm)

(3) Net Income

($mm)

(4) Net Income

($mm)

-0.0001 -0.0005 0.007 0.007 Technological niche age (0.002) (0.002) (0.005) (0.005)

-0.052 -0.008 -0.275 -0.486 Number of alliances with universities (0.068) (0.069) (0.218) (0.215)*

0.003 0.004 -0.005 -0.009 Publication count of scientific VP (0.003) (0.003) (0.009) (0.009)

-0.002 -0.002 -0.001 -0.003 Max. pub. count among SAB members (0.001)* (0.001)† (0.004) (0.004)

0.006 0.006 0.005 0.005 Founder publication count (pre-founding) (0.004) (0.004) (0.012) (0.012)

0.018 0.242 0.878 0.629 Percent founders with publishing ex-employers (0.350) (0.367) (1.129) (1.139)

0.048 0.047 -0.062 -0.085 Number of patent applications (0.016)** (0.017)** (0.051) (0.052)†

0.027 0.018 -0.431 -0.464 R&D expenses (0.010)** (0.011)† (0.033)** (0.034)**

0.343 0.226 0.681 0.307 Spinoff (yes = 1) (0.371) (0.380) (1.195) (1.180)

-0.005 -0.004 -0.028 -0.023 Technology niche density (ND) (0.003) (0.003) (0.010)** (0.010)*

0.162 0.369 -0.933 -1.344 Isomorphic pressure (IP) (0.224) (0.225) (0.722) (0.699)†

1.557 -0.008 Has adopted open science (pub. Count > 0) (0.487)** (1.569)

-0.501 1.639 Has one or more Ph.D. founder (0.481) (1.347)

0.402 -2.055 Has adopted open science Has Ph.D. founders (0.573) (1.847)

0.047 0.359 Firm publication count (0.030) (0.092)** -0.463 1.614 Proportion of founders with Ph.D. (0.389) (1.207) -0.005 -0.349 Firm pub. count Proportion of founders with

Ph.D. (0.037) (0.116)**

-0.404 -0.293 0.146 0.204 Constant (0.677) (0.680) (2.183) (2.112)

R2 0.184 0.143 0.466 0.480 Notes: (1) All models control for founding period dummies (grouped in five years) and technology area dummies. (2) Number of Observations = 512. (3) Standard errors in parentheses. (4) † significant at 10%, * significant at 5%, ** significant at 1%.

37

38