Academic Entrepreneurs: Social Learning and Participation in ...

Academic Entrepreneurs: Social Learning and Participation in University Technology Transfer

Janet Bercovitz The Fuqua School of Business Duke University Box 91020, Durham, NC 27708-0120 Phone: 919.660.7993 Fax: 919.681.6244 Email: [email protected] Maryann Feldman Rotman School of Management University of Toronto 105 St. George Street, Room 529 Toronto, ON M5S 3E6 Canada Phone: 416 946 5511 Fax: 416 978 5433 Email: [email protected]

Abstract: This paper examines faculty participation in university technology transfer using data on individual researchers from the medical schools of Duke University and Johns Hopkins University. The decision to file an invention disclosure is used to signal faculty members’ willingness to engage in a new organizational strategic initiative related to technology transfer. Our results suggest that the adoption of initiatives is a function of the norms at the institutions where the individual trained, the observed behavior of those in leadership positions, and the observed behavior of similar individuals. Keywords: organizational initiatives and change, social learning, university-industry technology transfer Acknowledgements: We wish to acknowledge support for this paper from the Andrew W. Mellon Foundation as part of a larger project on evolving university–industry relationships. We are indebted to the technology transfer personnel, research administrators and medical school faculty at Duke University and Johns Hopkins University for generously sharing their time and expertise in identifying salient issues. This paper has benefited from discussions with Irwin Feller and Rich Burton.

“Changing a university’s culture takes time, like turning a tanker takes time, there’s a lot of inertia to overcome” Theodore Poehler, Vice Provost for Research, Johns Hopkins University (as quoted by Lynch, 1988) The pursuit of new initiatives is essential to organizational survival (Van de Ven, 1986). However, there are myriad examples of organizations that failed to embrace change and adapt to environmental shifts (Kotter, 1996; Christensen and Bowers 1996). Even when top management makes public pronouncements about new initiatives, modifies incentives and diverts significant resources to develop supportive organizational structures, the persistence of existing organizational routines and norms may act as an impediment to organizational change (Hannan and Freeman, 1984; Nelson and Winter, 1982; Leonard-Barton, 1992). Successful realization of strategic initiatives by an organization depends, in a large part, on the acceptance of individuals within the organization (Sullivan, Sullivan, and Buffton, 2002). Major organizational change, which requires modification of routines and norms, rests on individual employees choosing to adopt behaviors that support the desired change (Whelan-Berry, Gordon, and Hinnings, 2003). Individual decisions in an organizational context are influenced by both economic and social factors (Manski, 2000; M.S. Feldman, 2000). Actions may be the product of the financial incentives offered, the salience of the opportunity perceived, and prior and ongoing socialization processes. In this paper, we focus on the third factor, investigating the link between social imprinting and social learning on individual decisions that are at the root of organizational change. Sociologists and organizational theorists have long studied the importance of social learning on the creation and diffusion of organizational norms (DiMaggio and Powell, 1983). Economists have recently incorporated social learning as an influence on decision-making (Duflo and Saez, 2000; Glaeser, Sacerdote and Scheinkerman, 1996; Manski 2000; Sorenson, 2002). This paper adds to the literature by explicitly considering the effects of social imprinting and social learning on the adoption of new strategic organizational initiatives by individuals within the organization, while controlling for financial incentives and opportunity drivers. The context we study is the adoption of an initiative to promote technology transfer by American research universities. Research commercialization has emerged as a new mission for American universities that differs from the older norms favoring the open dissemination of research discoveries

2

(Nelson, 2000). Faced with budgetary difficulties, universities now attempt to actively market their discoveries to industry, use their inventions to form new companies and engage in commercial activity related to economic development. These initiatives are described by Slaughter and Leslie (1997) as marking a new era of academic capitalism while Etzkowitz (1983) uses the term entrepreneurial universities. The process of change occurred in three distinct phases. First, while there was limited early institutional experimentation with technology transfer, the passage of the 1980 Bayh-Dole Act provided a broad legislative mandate and legitimized these activities.1 This was followed by a second stage of implementation as institutions set up dedicated technology transfer offices, adopted the policies and procedures to minimize conflict of interest, and establish royalty-sharing incentives, among other 2

supportive activities. Still, even with these institutional changes there was a noted performance gap in the realization of these new initiatives (Siegel, Waldman and Link, 2003). Becoming an entrepreneurial university active in technology transfer requires the participation and commitment of the faculty. The entire technology transfer process is predicated on individual faculty members disclosing their inventions to the university’s technology transfer office (Owen Smith and Powell, 2000). To the extent that disclosing signals the acceptance of the university’s new mission we have a rare opportunity to gauge acceptance of an organizational initiative. Using a social imprinting and social learning lens, this study examines participation by individual faculty members in technology transfer as an organizational initiative. The analysis focuses on faculty at the medical schools of two prominent research universities, Johns Hopkins University and Duke University. Neither institution had significant technology transfer activity before the 1980 passage of the Bayh-Dole Act. Both medical schools established dedicated technology transfer and licensing offices in the mid 1980s and made substantial progress in developing supportive organizational structures and carrying-out this new mission (Feller et al. 2000; Bercovitz et al 2001). We demonstrate variation among faculty members in terms of participation in technology-transfer among academic departments. Moreover, both within and across departments there is variation in participation among faculty members of different academic ranks. Through an empirical investigation of the choice by individual faculty members to disclose inventions to the technology transfer office we find evidence that, in addition to technical opportunity, social interactions and peer expectations influence the decision to participate in technology transfer and support the strategic initiative set-forth by the university administration.

1

The Bayh-Dole Act (PL 96-517) allows for the transfer of exclusive control over government funded inventions to universities for the purpose of further development and commercialization. Universities are then permitted to license the inventions to other parties and retain any licensing fees that may result. 2 By 1998, every Carnegie I and II American Research University had a dedicated technology transfer office.

3

This paper makes three contributions. First, we focus attention on the role of individual agents in organizational change. The heterogeneity in organizational performance that is unexplained by models that rely on firm and industry level analysis highlights the relevance of this approach (Hawawini, Subramanian and Verdin 2003). Second, our analysis places the individual in a social context and considers the other individuals that they interact with. One interesting artifact of our data is that participation in the strategic initiative we study is not evenly distributed within the organization but concentrated in specific specialized subunits. We follow the suggestion of McGrath, MacMillian and Venkataraman (1995) that the analysis of competitive advantage needs to examine a process-centered paradigm at the level of the organizational sub-unit. Sorenson (2002), Glaeser et al. (1996), among others, note that a similar concentration of activity suggests localized learning within the subunit. Third, we conducted in-depth interviews to develop hypotheses which we empirically test. Our results suggest that the decision to participate in strategic initiatives is influenced by social learning both prior to joining the organization as imparted though training and afterwards by observing others in the organization. The paper proceeds as follows. The first section draws upon the literature and the results of interviews with technology transfer managers and faculty members to develop a set of propositions about the individual faculty member’s decision to disclose new inventions. The third section of the paper introduces the data and methodology and the fourth section provides empirical results. Conclusions and discussion are offered in section five.

Invention Disclosures and Faculty Participation The process of technology transfer involves at least three different stages, with invention disclosure as the initiating stage. These disclosures are then evaluated as to their patentability and subsequently licensed to firms who develop the technology and pay licensing revenues (Thursby and Thursby, 2002). While outcome measures such as the number of patents and licensing revenues are important metrics, such outcomes are predicated on faculty disclosing their inventions. At face value, the decision to disclose research results should be straightforward. First, increased technology transfer activity has become an articulated goal of the university administration and is espoused as a strategic initiative at most academic institutions. Royalty-sharing incentives have been adopted and technology transfer offices actively encourage faculty participation. Second, disclosing research results to the technology transfer office is a stipulation of federal research grants, which constitute the largest source of university research funding. Third, the costs associated with disclosing an

4

invention are low with the required forms available on-line. 3 Fourth, there are limited quality barriers as there are no objective standards that faculty discoveries are required to meet to warrant filing an invention disclosure to the technology transfer office.4 Technology transfer managers are trying to encourage faculty to disclose since the number of faculty disclosures is one criterion used to evaluate the performance of the technology-transfer office.5 Thursby et al. (2001) argue that invention disclosures represent only a subset of university research with commercial potential and later Thursby and Thursby (2002) suggest three reasons why faculty would choose not to disclose research results. However each reason may be countered. First, it is claimed that faculty who specialize in basic research may not disclose because they are unwilling to spend time on the applied R&D required to interest businesses in licensing the invention. This is perhaps countered by the trend towards patenting basic scientific results from projects like the human genome which, though basic, have immediate commercial potential. Second, faculty may not disclose inventions because they are unwilling to risk publication delays that may be required to interest industrial partners in licensing the technology. Our interviews with both faculty members and TTO officials, however, indicate that this is more a perceptual problem than a reality. There are strategic ways to accommodate both academic and commercial interests but to do this requires a sophisticated understanding of the technology transfer process. Trusted peers who are familiar with the process can communicate strategies to accommodate both academic and commercial interests. The third reason faculty members may not disclose because they believe that commercial activity is not appropriate for an academic scientist. This view certainly represents the older norm of open academic science. However, to the extent faculty members do disclose their inventions these norms appear to be changing. Our contribution is to examine the disclosure process and to empirically test factors that may motivate faculty to disclose their inventions and participate in university technology transfer. Table 1 demonstrates variation in disclosing behavior by academic departments within the medical schools at Duke University and Johns Hopkins University. These two universities are comparable: both universities have well established and renowned medical schools. Both universities had little experience with patenting and licensing prior to the passage of the Bayh-Dole Act. Both established dedicated technology 3

The cost of filing a patent is at least $10,000 while the monetary costs associated with disclosure are negligible. Jensen, Thursby and Thursby (2003) title a paper “The Disclosure and Licensing of University Inventions: Doing the Best We Can with the S**t We Get to Work With” – the title is from an interview with a tech transfer administrator who was bemoaning the quality of faculty disclosures. Historically, there has been great variation in the types of inventions that were seen as patentable. For example, the University of Wisconsin founded the first technology transfer organization, the Wisconsin Alumni Research Foundation, around their vitamin patents, while Johns Hopkins University decided that their vitamin discoveries belonged in the public domain (Feldman and Desrochers 2004). 5 Mowery, Sampat and Ziedonis (2000) note that about 20% of disclosures were patented after six years, indicating that greater scrutiny accompanies the post-disclosure stage of the technology transfer process. 4

5

transfer offices in the mid-1980s, adopting similar royalty-sharing incentives for faculty inventions. Moreover, beginning in the late 1980s and accelerating in the early 1990s, increasing technology-transfer activities was articulated as a strategic initiative by the administration of both these institutions. [Table 1 here] Since medical schools account for the majority of university invention disclosures they are the focus of our analysis. The medical school departments at the two universities have slightly different names. We matched fifteen departments within the same field of inquiry and verified our matches with faculty members in the different departments. As a first comparison, the number of faculty in each department for the year academic year 1997-1998 is presented in Table 1.6 These are the individuals most likely to be in a position to disclose inventions.7 In most cases, the universities have a similar number of faculty within each department except for Cell Biology (Cell Anatomy and Biology) with 53 faculty members at Duke and 19 at Johns Hopkins, Ophthalmology with 34 at Duke and 119 at Hopkins and Neurobiology (Neuroscience) with 42 at Duke and 70 at Hopkins. Technology transfer activity is concentrated within certain departments at the medical schools, as demonstrated by the number of faculty members filing disclosures. We might expect that technological opportunity would be greater in some fields than in others and that these high opportunity departments would have a similar share of faculty who disclose inventions.8 This does not appear to hold. What is rather striking is the variation in the number of disclosures normalized by department size or stated as the number of invention disclosure events per faculty member.9 In aggregate, there were 0.384 disclosures per faculty member at Duke and 0.414 at Hopkins. However, there was substantial variation between similar departments across the two universities. For example, on average, ophthalmology faculty at JHU were involved in 1.2 disclosures each while individual Duke ophthalmology faculty, on average, contributed 0.38 disclosures. Conversely, surgery faculty members at Duke were approximately twice as likely to be involved with disclosing as the JHU surgery faculty. Considering the eight departments across both universities, where disclosure events per faculty members are relatively high (greater than 0.75), we see that only 50% of the time do the similar departments at both universities fall within this set. Further,

6

These include Full Professors, Associate Professors, and Assistant Professors. Other individuals, such as staff, graduate students and post-docs may disclose inventions but this is a small percentage of the activity. It is more likely that disclosures that involve non-faculty have at least one faculty member listed as an inventor. 8 The number of faculty members who have filed invention disclosures captures those who have disclosed in the three-year period, 1996-1998. This does not correspond directly to the absolute count of disclosures since more than one faculty member may be listed on a single disclosure. If a faculty member appeared on an invention disclosure they are counted as having filed a disclosure. 9 We use the term invention disclosure events to capture the number of times that an individual was listed on an invention disclosure. 7

6

only 30% of the disclosure events in our data set originate from these four, high opportunity departments. While technical opportunity matters it is clearly not the only driver of faculty disclosure behavior. The question thus becomes who discloses in the faculty, what are their characteristics and to what types of incentives do they respond? To develop hypotheses we rely on interviews with technology transfer officials and faculty members.10 Given that the outcome of disclosing the invention in terms of profitability is uncertain, individual motivation appears to matter. We assume that every individual in the medical school has potential to disclose inventions. While we may expect that certain fields of research would be more amenable to disclosure, the lack of objective standards indicates disclosure is an individual decision. We expect that faculty would be responsive to financial incentives and that there would be a direct relationship between licensing royalty distribution rates and the amount of technology transfer activity across universities.11 Our focus on departments within institutions holds these rates relatively constant since there is a convergence in incentives. Both universities we examine have a similar distribution rate with one-third of future revenue going to the individual faculty member, one third going to the central administration and one-third to the department. Departments may “sweeten the deal” by distributing a share of their third of the royalties to the inventing faculty members’ lab. This practice was first used to encourage technology transfer; however, it is now well established across departments in both universities. In forming expectations about the benefits of disclosing, a faculty member may be influenced by perceived opportunity, financial incentives or social interactions. Given that opportunity to disclose appears not to be solely dictated by technological field and that financial incentives are relatively constant, this paper tests the role social interaction plays on the preference of faculty members to disclose their inventions. The decision to disclose appears to be influenced by three categories of social interaction and organizational learning that we term training effects, leadership effects and peer effects. Each of these is described below. Training effects There is a diverse body of literature on social imprinting that gives background on how norms associated with training influence subsequent behavior and drive the adoption and diffusion of new practices. Many authors have argued that social institutions, with educational institutions being a key subset of this group, mold individual perspective by promoting, both implicitly and explicitly, a particular set of norms and/or values of “how things ought to be done” (Schein, 1985; Locke, 1985; Haas, 1992; 10

To date, we have conducted over 70 interviews with technology transfer officials, university administrators and faculty members as background for this study. 11 Lach and Schankerman (2003) investigate how economic incentives affect the number and commercial value of inventions generated by university scientists.

7

Calori et al. 1997; Biglaiser, 2002). DiMaggio and Powell (1983: 153) emphasize the role universities play in this socialization process, stating that those “drawn from the same universities and filtered on a common set of attributes, . . . will tend to view problems in a similar fashion, see the same policies, procedures and structures as normatively sanctioned and legitimated, and approach decisions in much the same way.”12 Support for these arguments may be found across academic domains. For example, in a series of studies, Frank, Gilovich, and Regan (1993) find evidence that economics students, particularly those trained by an instructor with research interests in game theory, are more likely to adopt selfinterested behavior than their peers. Similarly, recent work in political economy shows that the presence of U.S.-trained economists is a key predictor of the adoption of various types of neoliberal reform (e.g., tariff rate reduction, capital account liberalization) in emerging markets (Biglaiser and Brown, 2003; Chwieroth and Fellow, 2003).13 In sum, professional training can instill a particular set of norms and ideas and in acting according to these norms, students serve as a critical conduit for the diffusion of new ideas and practices. In our context, the logic of imprinting implies that individuals who trained at institutions where participation in technology transfer was accepted and actively practiced will be more likely to adopt these practices in their own careers. Interviews and anecdotal evidence are supportive of this conjecture. For example, one professor active in technology-transfer indicated that his graduate school mentors had disclosed and licensed their technology. He learned about disclosing by observing their experiences and this dictated his expectations for a professional career. While he recognized that when he joined the Hopkins faculty the culture did not support technology transfer, he believed that disclosing would provide a vehicle for implementing his ideas. Similarly, William Brody, current president of Johns Hopkins, started as Assistant Professor of Radiology in 1972. Brody learned about technology transfer during his graduate study at Stanford University’s Medical School and Department of Electrical Engineering, a very active department in terms of involvement with industry. Once at Hopkins, he continued to actively disclose inventions and subsequently started a company. His expectation was that technology transfer would be part of his career. In contrast, faculty who received their medical school training at institutions where technology transfer was not perceived as a legitimate activity often questioned the long term impacts of this activity both on their careers and on the broader pursuit of science. Several, including the chair of the Duke radiology department who trained at Cornell, had no intention to disclose and expressed strong sentiments against technology transfer pursuits even though such activity was now strongly

12

In a similar vein, Finnemore and Sikkink (1998: 905) argue: “Professional training does more than simply transfer technical knowledge; it actively socializes people to value certain things above others.” 13 Neoliberal ideas, that emphasize the virtures of free markets, are widely accepted and extensively taught in U.S. economic departments (Pinera, 1991).

8

supported by the university administration. This foundation of theoretical logic and anecdotal evidence, leads to the following testable hypothesis:14 H1: Individuals whose graduate training incorporated technology transfer objectives will be more likely to disclose innovations. The length of time an individual has been out of training is also likely to influence their disclosure decisions. As Ryder (1965) notes, individuals are imprinted, to some degree, by the major events that occur and accepted norms that are prevalent during their formative stages of development. The doctoral and/or medical training experience is clearly a formative period for academic researchers (Cartwright, 1979). Levin and Stephan (1991) confirm the importance of timing in a scientist’s training, finding that vintage, based upon year of doctorate degree award, affected scientific productivity in their study of the research productivity of academic scientists.15 Though never static, views regarding the proper role of academic scientists in commercialization activities have evolved considerably in the recent past, both as drivers of, and in response to, the 1980 Bayh-Dole Act. While historically, the norm of open science, with the goal of publication and open dissemination of findings, dominated in academic communities, there has been growing acceptance of academic institutions and scientists as entrepreneurs who have a responsibility to contribute to economic growth by working with industry to leverage academic research for commercial purposes. Recent studies document this trend, showing it is becoming increasingly common for academic administrators and individual scientists to consider commercialization as a key part of their missions (Zucker, Darby, and Brewer, 1998; Thursby and Thursby, 2002; Murray, 2003). The training and exposure received by medical and doctoral students were likely colored by this evolution in thought. The earlier an individual completed her training, the more likely she is to have been exposed to, and adopted, the traditional norms of science that do not favor disclosing. Conversely, the more recently trained the scientist, the more likely she encountered an environment supportive of commercialization activity. Our interviews revealed that new faculty recruits frequently inquire about university tech-transfer capabilities and even ask to meet with the technology transfer office before accepting a position. Thus, we hypothesize: H2: The likelihood that an individual will engage in disclosure activities will increase the more recent the vintage of their last graduate degree. 14

Selection bias may be a potential concern however we think that the reputation of the two medical schools and the resources offered will be more important to an academic scientist and that academic culture relative to technology transfer would not be a primary criteria for accepting a position. 15 In another context, Schewe and Meredith (1994) highlight vintage effects showing that an individual’s attitude towards jobs, money, and savings is significantly influenced by the conditions encountered when the individual first becomes an “economic adult”.

9

Leadership Effects Leaders influence behavior in organizations both by building culture and by acting as role models. The visible behavior of those in leadership roles drives organizational culture by signaling what actions are expected and valued (House, 1977; Schein, 1985). Further, given that individual behavior is shaped by the observation and imitation of others in a social context, subordinates vicariously learn what activities are deemed legitimate and thus worthy of emulation by observing the actions of the leader (Bandura, 1977; 1986). Specifically, the stance taken by a leader motivates a particular set of behaviors by influencing the self-concepts, or value systems, of the followers (Shamir, House, and Arthur, 1993). Culture and role-modeling cues are most pertinent in environments beset with ambiguity. For example, when the criteria for advancement is not clearly specified and delineated, individuals tend to use heuristics and adjust their expectations relative to the values highlighted by the behavior of those in leadership roles. In academic departments, the leadership role is generally held by the department chair. In medical schools the power and influence of the chair is particularly strong and appointments are of long tenure. The chair plays a direct and powerful role in reviewing and evaluating an individual’s performance and making recommendations about promotion and tenure. One contentious issue is how technology transfer activity is treated in promotions and tenure.16 The rules appear to be subjective and the problem for individual faculty members is to discern how their activity will be evaluated with limited information.17 One signal that the chair is predisposed to consider disclosing as legitimate faculty activity would be the observed behavior of the chair. Thus, if the chair is active in technology transfer as demonstrated by his prior disclosures, then he sends a signal that technology transfer is a valid activity. In this case, other members of the department would seemingly be more likely to disclose. We may expect that this signal would be stronger for junior faculty members who face greater uncertainty about expectations regarding promotion and tenure. However, our interviews suggest that senior faculty members also benchmark their performance against the department chair.

16

This was mentioned as a problem in several interviews. While the university may promote technology transfer if the department has not embraced it then the individual will face difficulties. There is no hard and fast rule for evaluating technology transfer activity relative to academic work. We have been told that the MIT electrical engineering department values a patent as much as an academic article in a high quality journal although there does not appear to be any quantification of these trade-offs at the two universities examined here. 17 The Chair has a role in hiring decisions and could select individuals with similar technology transfer attitudes. Interview evidence revealed that criteria other than technology transfer is more important in hiring decisions. To overcome this potential source of bias we eliminate new hires and examine the group of individuals who had been at the university for the entire study period.

10

H3: Individuals in departments where the chair is actively involved in technology transfer are more likely to engage in technology transfer activities.

Peer Effects When faced with uncertainty about the proper course of action, social learning theory posits that individuals will model the behaviors of referent others (Bandura, 1986). In addition to leaders, as detailed above, one’s peer group acts as a relevant source of information. Numerous prior studies provide evidence that learning activity occurs within a cohort of peers as individuals draw inferences about the value of alternative behaviors by observing the choices of “similar” others (Glaeser et al., 1996; Duflo and Saez, 2000; Sorensen, 2001).18 For scientists, both industrial and academic, local group norms have been shown to play a significant role in determining individual behavior (Louis, Blumenthal, Gluck, and Stoto, 1989; Pelz and Andrews, 1976). We expect that individuals will be more likely to engage in technology transfer activities when they observe individuals with similar characteristics in their departments are also disclosing. In academic communities, peer groups commonly form based on professional rank. Having entered the institution at a similar time, and facing similar issues in managing their career, those of similar rank tend to look to each other for information and direction. H4: Individuals are more likely to disclose if their peers engage in technology transfer activities.

Data, Variables and Methods Our empirical analysis uses an original database of individual faculty member compiled from administrative records at Duke University and Johns Hopkins University. We have data for faculty members across 15 departments in the two medical schools for the academic years 1991-1999. We selected to examine medical school departments because the majority of technology transfer activity originates within medical schools. Our selection of departments was constrained by the degree to which departments were present in both universities. Under the advice of medical school faculty, we selected closely aligned departments – that is, places where similar work was being done despite slight differences in the name of the academic departments. [Table 2 here]

18

In economic terms, there is the expectation that individuals with similar characteristics will face similar payoffs given similar choices (Ellison and Fudenberg, 1993)

11

The fifteen departments used in our analysis are presented in Tables 1 and 2. Our selection included medical school departments such as anesthesiology, pathology, radiology and surgery that are oriented toward providing patient services and are ancillary to other departments. These departments are termed nexus departments. Our interviews suggest that faculty in these departments may be in a position to engage in greater inventive activity. First, these faculty members consult with multiple departments and may learn about technology transfer from other faculty members. In addition, our interviews suggested that nexus departments also have the type of practical problem-solving focus that promotes user-defined invention. Our analysis also included a set of clinical departments that provide primary patient care oriented toward a specific specialty such as cardiology, ophthalmology, pediatrics or psychiatry. These departments include the largest numbers of medical school faculty. While these department names suggest routine patient care, the expectation is that faculty members in these departments have an active research program. Finally, we also examined basic science departments such as cell biology, genetics and immunology. These are high opportunity departments in which we expect basic scientific discoveries may easily lead to invention disclosures. We used a three year window to track disclosures. This was chosen to capture a reasonable time period during which an individual faculty member might have results to disclose. Thus, we examine faculty disclosure behavior for the academic year 1996 -1997 through academic year 1998-1999. Personnel records, university course catalogues and archival data were used to build records for faculty members. Data on the disclosures are from the records of the technology transfer offices at the two universities. Table 2 demonstrates the great variation in the rank of faculty members who disclose in each department. For example, at Duke Medical School nearly half of the junior faculty members in pharmacology disclosed in the three-year period, while all of the full and associate professors in the genetics department disclosed. The table also provides information on which department chairs had disclosures in the prior time period. At Duke Medical School, nine of the fifteen chairs had a history of disclosing, while seven of the fifteen chairs have a history of disclosing at Johns Hopkins University. A paired sample t-test of the equivalence of the percentage of faculty reveals that there are statistically significant differences at the two institutions (t=-2.441, 44 d.f.) To test our hypothesis, we estimate the training, leadership, and cohort effects on the observed filing of disclosures. We are using a two period model with the decision to file disclosures at time t as a function of individual attributes and the behavior observed in the time period, t-1.

12

Training Effects We use three variables to investigate training effects on faculty propensity to disclose. The first two variables capture the technology transfer culture of the institution where the faculty member received their graduate training. Certain universities have historically had greater receptiveness to, and greater involvement in, technology-transfer activities than others. The intensity of a university’s technologytransfer culture can be proxied using the institution’s patenting history (Mowery and Ziedonis, 2002). For our measure, we calculate the total number of patents applied for by an individual’s graduate institution in the five years preceding the grant date of the faculty member’s graduate degree. This is intended to capture the activity at the time that the individual was in training at the institution. To the extent that student and alumni attitudes toward technology transfer are influenced by norms they are exposed to during training, we expect that individuals educated at universities active in patenting will have a greater predisposition towards technology transfer, ceteras paribus. Stanford, mentioned several times in faculty interviews, stands out as a well-known example of a pro-technology transfer university with a strong history of patenting activity. As such, we code a Stanford dummy variable equal to one if an individual has an advanced degree from Stanford, as a simple alternative measure of pro-technology transfer imprinting. Our third measure of pro-technology transfer exposure explicitly captures the era in which in the faculty member was trained. As noted earlier, the acceptance of technology transfer activities at most universities has increased substantially over the past few decades. The more recent the faculty’s training, the more likely it is that the idea of active commercialization of research would have been included as an expectation. To capture the timing effect associated with the individual’s training, or experience years, we measure the number of years since the faculty member received their last advanced degree. Leadership Effects To explore the influence of key leadership on faculty’s propensity to disclose we create a measure of the chairmen’s involvement in technology transfer. A dummy variable indicating whether or not the chair has disclosed an invention to the technology transfer office in the prior five-year period, 1991-1995 is used. This variable is coded 1 for yes, and 0 for no. It should be noted that the chairmen of the medical school departments typically hold this leadership position for extended periods of time. In our data set, there was no turn over in the chair in any of the departments during the period studied.

13

Cohort Effect We define an individual’s cohort as those individuals of the same rank in the same department. The cohort effect is measured as the percentage of faculty at the same rank within the department who disclosed in the 1991-1995 time period.

Table 3: Summary of Predictions Greater Propensity to Disclose if

Variable Description:

Training Effects

Where: Institution with a ProTechnology Transfer Culture (University Patenting Activity)

+

Stanford Degree (1 = of graduate from Stanford; 0 = Otherwise)

+

When: Experience Years (Years Since Last Grad. Degree)

-

Chair has Disclosed Previously (1 = yes; 0 = no)

+

Percentage of Faculty within Rank/Department Cohort that Have Disclosed

+

Individuals are Trained with an Acceptance of Technology Transfer

Leadership Effects Cohort Effects

Department Chair has a History of Disclosing Similar Faculty Actively Disclose

Expected Sign

Table 3 summarizes the variables used to test our propositions and their predicted signs. The unit of observation is the individual faculty member and we are interested if the individual engaged in technology transfer by filing a disclosure. The dependent variable is equal to zero if the individual did not file an invention disclosure in the three-year window for 1996-1998. The dependent variable is equal to one if the individual filed one or more disclosures during this period. The probability of disclosing is estimated using a PROBIT model (Maddala, 1983). Control Variables We include several control variables in the estimation. First, disclosure behavior may be influenced by the amount of resources available for scientific inquiry. Further we expect a lag between the receipt of research funding and the type of discovery that precedes an invention disclosure. To control for any such influence, we include a dollar measure of the NIH awards received by each faculty member

14

in the previous five-year period. NIH funding is the most prominent source of medical school funding and carries the provision that invention disclosures be filed on the resulting discoveries. In addition to resource availability, we also include two controls for inventive capacity. Bisociation, the ability to relate two seemingly unrelated concepts, is argued to be at the root of creativity and innovation (Koestler, 1990). Considering breadth of knowledge as a necessary condition for bisociation, entrepreneurial research has shown that individuals with interdisciplinary educational backgrounds and expansive prior knowledge are better positioned to recognize, and then act upon, innovation opportunities (Venkataramen, 1997; Shane 2000). One sign of a breadth of knowledge is the subsequent appointment of a faculty member to multiple departments. As such, we include a boundary spanning dummy variable, coded as one if the faculty member is associated with more than one department. A second indication of breadth of knowledge is the attainment of multiple graduate degrees. Individuals who hold both a Ph.D. degree and a M.D. are expected to have training that encompasses research and practical application. In this respect, they may be in an advantageous position to develop new science with an eye to the commercial potential for such innovations. Thus, we include a dummy variable, coded as 1, to capture those individuals having both MDs and PhDs. We also control for the number of previous disclosures (1991-1995) for each faculty member. We expect that those individuals that have disclosed in the past are likely to continue this behavior. As a rough measure of technical opportunity, we include dummy variables to control for department type, basic science and nexus, with clinical as the omitted variable. We add rank dummy variables to control for faculty rank and a non-US dummy to capture the effects of having been trained at a foreign institution. Finally, we include a university dummy variable to control for institutional differences between Duke and Johns Hopkins. Descriptive statistics are presented in Table 4 and correlations are in Table 5. [Tables 4 and 5 here]

Results Table 6 provides results for all medical school faculty members in the selected departments at the two medical schools.19 Model (1) provides a baseline model. The number of disclosures in the prior time period has a strong and statistically significant effect on the propensity to disclose in the current time period. This is, of course, to be expected as individuals tend to repeat established behaviors. Those who have previously disclosed are likely to continue this behavior, if the experience was reasonable. 19

The sample size varies from the total reported in Table 1 due to missing data and movement of faculty. The sample size drops between Model 1 and Model 2 due to missing degreee/graduate school information.

15

Interestingly, the amount of individual NIH funding is not statistically significant. The availability of resources does not appear to play a deterministic role in promoting disclosure activity. Faculty trained outside the U.S. are significantly more likely to disclose inventions than those trained in the U.S. This finding is consistent with other studies that document high levels of entrepreneurial activity in many immigrant communities (Aldrich, 1999). [Table 6 here] Measures of training breadth also contribute explanatory power. First, the coefficient on the dual degree dummy variable is positive and significant. Individuals having earned both an MD and a PhD show a greater propensity to disclose than colleagues with single degrees. Holding both degrees increases the probability of disclosing by 14%. We also find a strong positive relationship between occupying a boundary position at the medical school and the likelihood of disclosure. Boundary spanning individuals, those with appointments in more than one academic department, are 4% more likely to disclose. Faculty in basic science and nexus departments are significantly more likely to disclose than faculty in clinical departments, the omitted category. This may reflect the more patient-oriented nature of departments like pediatrics and psychiatry, however it should be noted that individuals in these departments do disclose inventions and the expectation at prominent medical schools is that all faculty conduct research. Academic rank, using the rank of associate professor as the omitted category, is not statistically significant in this specification. Finally, we find no significant differences between the two universities. Model (2) builds on the basic specification by adding the training (social imprinting) variables. The explanatory power of the model increases significantly with the addition of the independent variables of theoretical interest. A likelihood ratio test comparing Model (2) to Model (1) is significant with a pvalue less than 0.01. Experience years, calculated as the number of years since the last graduate degree, is negatively and significantly related to participation in technology transfer: the probability of disclosing decreases by 1% for each year since the completion of graduate study. This result, which supports H2, indicates that the earlier an individual completed their training, the less likely they are to pursue commercialization opportunities. Model 2 also adds in the influence of completing graduate training at historically pro-technology transfer institutions. The coefficient on the institution patent activity variable is positive and significant. We find that a 1% increase in patenting activity at the institution where an individual received their graduate training is associated with a 1% increase in the probability that the individual will disclose. In Model 3, we use the alternative measure of pro-technology culture imprinting, having received one’s graduate degree from Stanford which our interviews revealed to be particularly pro-technology transfer. The coefficient on the Stanford degree dummy is positive and significant.

16

Holding a Stanford degree increases the probability of engaging in technology transfer by 21%, all other things being equal. Thus, hypothesis 1 is also supported. Though generally consistent, there is one noteworthy change to the control variables between Model (1) and Models (2) and (3). With the addition of the social imprinting variables, the rank controls become significant with full professors more, and assistant professor less, likely to disclose as compared to associate professors. These results are in-line with the human capital argument that those individuals who are well established in their academic careers will be more likely to leverage their reputations for commercial gain (Stephan and Levin, 1992). The results of Model (2) and (3) are robust to departmental fixed effects as demonstrated in the appendix tables. Model (4) investigates the effect of the chair’s disclosing behavior and finds evidence of a significant leadership effect. It appears that, to a significant degree, individual faculty model their technology-transfer behavior on the example set by their department chair. As shown in Model (4), if the chair has disclosed any inventions to the technology transfer office in the past five years, then the probability that the faculty member will disclose increases by 6%. Thus, hypothesis 3 is also supported. Model (5) considers the influence an individual’s cohort has on disclosure activity. The coefficient on the cohort variable is positive and significant, suggesting that an individual’s disclosure choice is swayed by the actions of those with similar rank within their department as predicted by hypothesis 4. We find that a 1% increase in the percentage of faculty disclosing within the relevant cohort increases the probability of an individual disclosing by 17%. Though these results are supportive of a peer effect, it is possible that these findings arise due to shared unobserved characteristics of the department. Common unobservables could lead to commonality of behavior without imitation (Bikhchandani, Hirshleifer, and Welch, 1998). Manski (2000) notes that one problem that plagues social learning research is distinguishing endogenous interactions from correlated effects. To delve into this issue, we ran an additional model that includes an “other” cohort variable, those individuals in the same department but of a different rank, as well as the own cohort variable. Specifically, in Model (6) we investigate whether an individual’s disclosure decision is differentially influenced by the actions of individuals inside versus outside the focal cohort. Under this specification, the coefficient on own-cohort remains positive and significant, while the coefficient on the other cohort variable, while positive, does not reach significance. Finding that only own-cohort effects are important indicates that social learning rather than departmental unobserved variables are driving our results. A significant subset of the faculty members who disclosed inventions in the 1996-1998 window were disclosing for the first time. Disclosure in these cases represents a significant change in behavior by the individual scientists. To investigate the factors influencing this decision to change behaviors and join the ranks of academic entrepreneurs, we estimated a model including only those faculty members in

17

moderate to low opportunity departments that had not previously disclosed in the 1991-1995 period by removing the 336 faculty members with previous invention disclosures from the data set. We also eliminated the 166 faculty members remaining in the basic science departments and include department dummy variables, with surgery as the missing category in the estimation. Descriptive statistics for this smaller data set are provided in Table 4. [Table 7 Here] Table 7 presents the results of the model estimation on those individuals from moderate to low opportunity departments (nexus and clinical departments) who are new participants to technology transfer. Model (1) begins with the basic specification. Again, we find that academic rank and individual NIH funding are not statistically related to disclosure. As before, we also find that individuals having dual degrees and those with non-U.S. degrees are more likely to disclose an invention. Interestingly, only one of the departmental control variables shows significance. Those individuals in the department of psychiatry are significantly less likely to disclose inventions. Results from Model (2), which adds the training variables to the baseline model, are consistent with the results found in the analysis in Table 5 for all faculty members. In this specification, we find a significant negative relationship between experience years and probability of disclosure. The longer the period since training, the less likely an individual will change their behavior and become involved in technology-transfer activities. Though significant in isolation, our measure of patenting activity at the individual’s graduate institution falls just below significance when the previous experience variable is added to the model. This may be due to the moderately high correlation between these two variable (r = 0.474) as our measure of patenting activity at the individual’s graduate institution is calibrated to the time when the individual was in attendance. An alternative explanation is that graduate school training may have a more immediate impact with individuals realizing their career aspirations with prior disclosures. Individuals in this subset of the data from high patenting graduate institutions appear not to have internalized this norm from their graduate training. Model (3) shows that the chair effect is similar whether we consider the disclosure activity of all faculty members or solely the subset of faculty with no previous disclosure activity. In Model (4), we again find evidence of a cohort effect as the coefficient is both positive and significant. Note that we find this significant effect in an estimation that includes departmental fixed effects. In these lower opportunity or more traditional medical school departments, first-time disclosures are significantly more likely when there is higher disclosure activity within one’s cohort. In essence, the actions of peers seem to be a key trigger of behavior change in these environments. While it is difficult to attribute causality to this finding

18

to either an increased familiarity with the process which then either reduces the uncertainty of adopting the new behavior or a change in perspective as the activity become legitimized, the results suggest that the actions of one’s peers matter in influencing the propensity to participate in tech transfer. In Model (5) we add the outside cohort variable which is not statistically significant. Individuals learn about disclosing from observing their peers within the department. Discussion and Conclusion In sum, the results suggest that the decision to participate in technology transfer through the process of disclosing inventions is strongly influenced by training effects, leadership effects and cohort effects. Individuals are more likely to disclose inventions if they trained at institutions at the forefront in terms of technology transfer benchmarking. Individuals who trained at institutions that have long established and relatively successful technology transfer operations are more likely to disclose their inventions. In addition, we find a negative career experience effect: the longer the time that had elapsed since graduate training, the less likely the faculty member was to actively embrace the new commercialization norm. We also find that where the chair of the department is active in technology transfer other members of the department are also likely to disclose. Finally, we find that technology transfer behavior is mediated by the experience of those in a similar position, in terms of academic rank and departmental affiliation. If an individual can observe others at their academic rank disclosing then they are more likely to follow, other things being equal. One alternate explanation that could be offered is that selection, rather than socialization, drives the results related to departmental effects. Specifically, it could be argued that instead of being influenced by the action of leaders and peers, individuals pre-disposed to disclose are differentially hired-in to departments supportive of technology transfer activities. To test for selection effects, we looked at the 190 individuals that were hired by either of the two universities in 1991, the first year of our panel. If selection was in fact a dominant determinant of the disclosure dynamic, we would expect to find evidence that department chairs with a history of disclosure, indicating an acceptance of this activity, would be more likely to subsequently appoint individuals who would be predisposed to disclosing. For example, chairs who wanted to encourage technology transfer activity might hire individuals whose breadth of training suggested that they might be more inclined to engage in commercial activity or from graduate schools where this activity was accepted. An independent samples t-test showed that there was no statistically significant difference (t=1.605; p>0.10) in the likelihood that an individual having a dual (both PhD and MD) degree would be hired into a department in which the chair actively disclosed versus a department headed by a chair who did not participate in technology transfer activities. Similarly, in a second independent samples t-test, the hypothesis that the means of the patenting activity of the new-hires’

19

graduate institutions were equal across departments led by technology-transfer active versus non-active chairs could not be rejected (t=1.638; p>0.10). These results lend credibility to our argument that social learning influences an individual’s decision to follow strategic initiatives and choose to participate in new activities. However, we do not interpret these t-test results as justifying the conclusion that selection plays no role in the technology transfer dynamic observed. Rather, it appears that both factors – selection and socialization – may be material in organizations. The challenge for future research is to disentangle the contributions of these alternative drivers to better understand in what circumstances one is likely more pertinent than the other. Of course, organizations that introduce new initiatives face the decision of replacing employees or motivating current employees to adopt requisite new behaviors. This paper has explored factors that motivate current employees to change their behaviors. The challenges associated with driving individuals to accept and embrace strategic initiatives clearly exist and are pertinent in most large organizations. As such, though the academic environment is unique in many respects, the findings of this study may have value to other organizations attempting to embrace new initiatives. As public institutions, universities have greater transparency, making them easier to study. Moreover, the acceptance of new policies is generally difficult to observe. In our case, the filing of an invention disclosure signals acceptance of the new initiative in an objective and quantifiable manner. Introducing strategic initiatives requires thinking creatively about the process of organizational change. Three key insights that we derive from our study are the multi-level characteristic of the change process, the importance of departmental (or sub-unit) composition in promoting change, and the role of individuals in adopting and diffusing the acceptance of strategic initiatives. First, change occurs as a nested process at different organizational levels. The realization of the objectives of increasing university interaction with industry and facilitating the commercialization of academic research has been an ongoing process for close to 25 years, required change at the legislative level, the adoption of new policies and resources at the organizational level, the diffusion of appropriate policies to the within-organizational group-level, and finally, the individual-level. Further, the effectiveness as well as rate of change appears to be a product of the sequencing and linking of actions taken at different levels. CEOs and change leaders are asking academicians and consultants why change cannot happen faster and with greater efficiency. Implementing organizational level policies and procedures may not be sufficient to orchestrate change processes. Our results suggest that change can be facilitated when appropriate resources are specifically focused on group- and individual-level change processes. This is suggestive of a more bottoms-up approach than appears to be the norm and may mean rethinking how we have traditionally allocated resources in the implementation of organizational change (Whelan-Berry,

20

Gordon and Hinings, 2003). Successfully adopting strategic initiatives may be predicated on fully understanding the inherent group- and individual-level dynamics and the factors that favor the acceptance of the new initiatives. The results presented here suggest the importance of a sub-unit dynamics in the acceptance of organizational initiatives. Most critically, participation in the strategic initiative was concentrated in a few departments. A department head that actively supports the initiative was important however the lack of this type of leadership was not an insurmountable obstacle. Even if sub-unit leaders do not participate in the strategic initiatives individuals will adopt the new behavior when they are able to observe their peers participating. This suggests that a few well-placed individuals who have been trained to embrace the new initiative may facilitate adoption by others. ………………………………………………….

21

References Aldrich, Howard E. (1999). Organizations Evolving. London: Sage. Bandura, Albert. (1986). Social Foundations of Thought and Action: A Social Cognitive Theory. Englewood Cliffs, NJ: Prentice-Hall. Banerjee, A.V. (1992). “A Simple Model of Herd Behavior,” Quarterly Journal of Economics, 107(3): 797-817. Bantel, Karen A. and Susan E. Jackson. (1989). “Top Management and Innovations in Banking: Does the Composition of the Top Team Make a Difference?” Strategic Management Journal, 10: 107124. Bercovitz, J. E. L, M. P. Feldman, I. Feller, and R. M. Burton. (2001). “Organizational Structure as a Determinant of Academic Patent and Licensing Behavior: An Exploratory Study of Duke, Johns Hopkins, and Pennsylvania State Universities.” Journal of Technology Transfer, 26: 21-35. Biglaiser, G. (2002). Guardians of the Nation? Economists, Generals, and Economic Reform in Latin America, Notre Dame: University of Notre Dame Press. Biglaiser, G. and D. Brown. (2003). “What’s the Big Idea? An Ideational Explanation for Tariff Reform in Latin America, 1981-1995,” Paper presented at the American Political Science Association, Philadelphia, PA. Bikhchandani, Sushil, David Hirshleifer, and Ivo Welch. (1992). “A Theory of Fads, Fashions, Custom, and Cultural Change as Informational Cascades,” Journal of Political Economy, 100: 992-1026. Brock, W.A. and S.N. Durlauf (2001). Discrete Choice with Social Interactions. Review of Economic Studies 68(2): 235-260. Brown, Shona L. and Kathleen M. Eisenhardt. (1995). “Product Development: Past Research, Present Findings, and Future Directions,” Academy of Management Review, 20(2): 343-378. Burt, Ronald S. (1992). Structural Holes: The Social Structure of Competition. Cambridge, MA: Harvard University Press. Calori, R., M. Lubatkin, P. Very, and J. Veiga. (1997). “Modelling the Origins of Nationally-Bound Administrative Heritages: A Historical Institutional Analysis of French and British Firms,” Organization Science, 18(6): 681-696. Cartwright, D. (1979). “Contemporary Social Psychology in Historical Perspective,” Social Psychology Quarterly, 42: 82-93. Christensen, C.M. and J.L. Bowers (1996). “Customer Power, Strategic Investment and the Failure of Leading Firms,: Strategic Management Journal ,17: 197-219. Chwieroth, J. (2003). “Neoliberal Norms and Capital Account Liberalization in Emerging Markets: The Role of Domestic-Level Knowledge-Based Experts,” Paper presented at the American Political Science Association, Philadelphia, PA. Clark, K. B. and T. Fujimoto. (1991). Product Development Performance. Boston: Harvard Business School Press. Cohen, W. M. and D. A. Levinthal, (1990). “Absorptive Capacity: A new perspective on learning and innovation,” Administrative Science Quarterly, v. 35 (1990), p. 128 – 152. DiMaggio, P. and W. Powell. (1983). “The Iron Cage Revisited: Institutional Isomorphism and Collective Rationality in Organizational Fields,” American Sociological Review, 48(2): 147-160.

22

Dougherty, D. (1992). “Interpretive Barriers to Successful Product Innovation in Large Firms,” Organization Science, 3: 179-202. Duflo E.and E. Saez (2000). Participation and Investment Decisions in a Retirement Plan: The Influence of Colleagues Choices. Mimeo. Etzkowitz, H., (1983). ‘Entrepreneurial Scientists and Entrepreneurial Universities in American Academic Science’, Minerva, 21: 1-21 . Etzkowitz, H. (1998). “The norms of entrepreneurial science: cognitive effects of the new university– industry linkages,” Research Policy. 27: 823-833. Ellison, G. and D. Fudenberg. (1993). “Rules of Thumb for Social Learning,” Journal of Political Economy, 101(4): 612-643. Feldman, M. P. and P. Desrochers. (forthcoming) “Truth for Its Own Sake: Academic Culture and Technology Transfer at the Johns Hopkins University.” Minerva. Feldman, M. P., I. Feller, J. E. L. Bercovitz and R. M. Burton. (2002) “Equity and The Technology Transfer Strategies of American Research Universities. Management Science. 48: 105-121. Feldman, Martha S. 2000. Organizational routines as a source of continuous change. Organization Science 11, no. 6:611. Feller, I., M. Feldman, and J. Bercovitz and R. Burton (2000). A Disaggregate Analysis of University Patent and Licensing Behavior. mimeo. Finnemore, M. and K. Sikkink. (1998). “International Norm Dynamics and Political Change,” International Organization, 52(4): 887-918. Frank, R. T. Gilovich, and D. Regan. (1993). “Does Studying Economics Inhibit Cooperation?” The Journal of Economic Perspectives, 7(2): 159-171. Geiger, R. L.,(1993). Research and Relevant Knowledge : American research universities since World War II (New York : Oxford University Press). Geroski, P.( 2000). Models of technology diffusion. Research Policy, 29 (4-5): 603-625. Glaeser, E.L., B. Sacerdote, and J.A. Scheinkerman (1996). Crime and Social Interactions. Quarterly Journal of Economics 111(2): 507-548. Haas, P. (1992). “Introduction: Epistemic Communities and International Policy Coordination,” International Organization 46(1): 1-35. Hannan, Michael and John Freeman. (1984). “Structural Inertia and Organizational Change.” American Sociological Review, 48: 149-164. House, Robert J. (1977). “A 1976 Theory of Charismatic Leadership,´in J.G. Hunt and L.L. Larson (Eds.), Leadership: The Cutting Edge. Carbondale: Southern Illinois University Press. Jensen,R.A. and J.G. Thursby and M.C. Thursby. (2003). The Disclosure and Licensing of University Inventions: “the Best we ca do with the s**t we get to work with” April 15, 2003. Koestler, Arthur. (1990). The Act of Creation. New York: Viking. Lach, Saul and Mark Schankerman. 2003. “Incentives and Invention in Universities,” NBER Working Paper 9727. Leonard-Barton, Dorothy. (1992). “Core Capabilities and Core Rigidities,” Strategic Management Journal, Summer: 111-125.

23

Levin, Sharon G. and Paula E. Stephan. (1991). “Research Productivity Over the Life Cycle: Evidence for Academic Scientists,” The American Economic Review, 81(1): 114-132. Locke, R. (1985). “The Relationship Between Higher Educational Management Cultures in Britain and West Germany: A Comparative Analysis of Higher Education from a Historical Perspective, “ in P. Joynt and M. Warner (Eds.), Managing in Different Cultures, Norwich: Page Bros. Ltd., 96127. Louis, Karen S., David Blumenthal, Michael Gluck, and Michael Stoto. (1989). “Entrepreneurs in Academe: An Exploration of Behaviors Among Life Scientists,” Administrative Science Quarterly, 34: 110-131. Lowe, R.A. (2001). “The Role and Experience of Start-ups in Commercializing University Inventions,” in G. Libecap (ed.) Entrepreneurship Inputs and Outcomes. JAI Press: Amsterdam, The Netherlands. Maddala, G.S. (1983). Limited-Dependent and Qualitative Variables in Econometrics. Cambridge University Press: Cambridge, UK. Mazzoleni, Roberto, and Bhaven N. Sampat. (2002). University Patenting: An Assessment of the Causes and Consequences of Recent Changes in Strategies and Practices. Revue d'Economie Industrielle 0, no. 99:233-248. Mansfield, E. (1961), Technical change and the rate of imitation. Econometrica 29(4):741-66. Manski, C.F. (2000). Economic Analysis of Social Interactions. Journal of Economic Perspectives 14(3): 115-136. Mowery, D., R. Nelson, B. Sampat and A. Ziedonis (1999) “The Effects of the Bayh-Dole Act on U.S. University Research and Technology Transfer” in Industrializing Knowledge, edited by L. Branscomb, F. Kodama, and R. Florida, (Cambridge, MA: MIT Press), 269-306. Mowery, David C. and A. Ziedonis (1999), “The Effects of the Bayh-Dole Act on U.S. University Research and Technology Transfer: Analyzing Data from Entrants and Incumbents” Paper presented at the Science and Technology Group, NBER Summer Institute (Cambridge, MA: National Bureau of Economic Research). Mowery, David C., and Arvids A. Ziedonis. (2002). Academic Patent Quality and Quantity before and after the Bayh-Dole Act in the United States. Research Policy 31, no. 3:399-418. Mowery, David C., and Bhaven N. Sampat. (2001). “University Patents and Patent Policy Debates in the USA, 1925-1980,” Industrial and Corporate Change 10, no. 3:781-814. Nelson, Richard R. (2001). “Observations on the Post-Bayh-Dole Rise of Patenting at American Universities,” Journal of Technology Transfer 26, no. 1-2:13-19. Nelson, Richard, and Sidney Winter. (1982). An Evolutionary Theory of Economic Change. Cambridge, MA: Harvard University Press. Nygaard, A and R. Dahlstrom. (2002). “Role Stress and Effectiveness in Horizontal Alliance,” Journal of Marketing, 66(2): 61-82. Owen-Smith, J. and W. W. Powell. (2001). “To Patent or Not: Faculty Decisions and Institutional Success at Technology Transfer”, The Journal of Technology Transfer, 26:99 – 114. Rogers, E. M. (1983). Diffusion of innovations, 3rd Ed. New York: Free Press. Rosenberg, Nathan and Richard Nelson. 1994. “American Universities and Technical Advance in Industry,” Research Policy, 23: 323-348.

24

Ryder, N. (1965). “The Cohort as a Concept in the Study of Social Change,” American Sociological Review, 30: 843-861. Schein, Edgar H. (1985). Organizational Culture and Leadership,: A Dynamic View, San Francisco, CA: Jossey Bass. Schewe, Charles D. and Geoffrey E. Meredith. (1994). “Cohort Effects and Life Insurance,” Marketing Management, 3(3): 25. Shamir, Boas, Robert J. House and Michael B. Arthur. (1993). “The Motivational Effects of Charismatic Leadership: A Self-Concept Based Theory,” Organization Science, 4(4): 577-594. Shane, Scott (2000). “Prior Knowledge and the Discovery of Entrepreneurial Opportunities,” Organization Science, 11(4): 448-469. Siegel, Don, D. Waldman and A. Link (2003) “Assessing the Impact of Organizational Practices on the Productivity of University Technology Transfer Offices: An Exploratory Study”, Research Policy.32:3-30. Slaughter, S., and L. Leslie. 1997. Academic Capitalism: Politics, Policies and the Entrepreneurial University. Baltimore: Johns Hopkins University Press. Sorensen, A.T. (2002). Social Learning in the Demand for Employer-Sponsored Health Insurance. Mimeo. Stephan, P.E. and S.G. Levin (1992). Striking the mother lode in science: the importance of age, place, and time (New York : Oxford University Press). Sullivan, W., R. Sullivan, and B. Buffton. (2002). “Aligning Individual and Organizational Values to Support Change,” Journal of Change Management, 2(3): 247-254. Thursby, Jerry G., and Sukanya Kemp. (2002). Growth and Productive Efficiency of University Intellectual Property Licensing. Research Policy 31, no. 1:109-124. Thursby, J.G. and M.C. Thursby. (2002). “Who is Selling the Ivory Tower? Sources of Growth in University Licensing,” Management Science, 48: 90-104. Thursby, J.G. and M.C. Thursby. (2003). “Patterns on Research and Licensing Activity of Science and Engineering Faculty”, April 2003. Thursby, Jerry G., Richard Jensen, and Marie C. Thursby. (2001). Objectives, Characteristics and Outcomes of University Licensing: A Survey of Major U.S. Universities. Journal of Technology Transfer 26, no. 1-2:59-72. Van de Ven, A. H. (1986). Central problems in the management of innovation. Management Science 32: 59-607. Venkatarman, S. (1997). “The Distinctive Domain of Entrepreneurship Research: An Editor’s Perspective,” in Advances in Entrepreneurship, Firm Emergence, and Growth. J. Katz, R. Brockhaus, eds. JAI Press, Greenwhich, CT. Whelan-Berry, Karen, Judith Gordon, and C.R. Hinings. (2003). “Strengthening Organizational Change Processes,” The Journal of Applied Behavioral Science, 39(2): 186-207. Zucker, Lynne G., Michael R. Darby, and Marilynne B. Brewer. 1998. “Intellectual Human Capital and the Birth of US Biotechnology Enterprises.” The American Economic Review, 88: 290-306.

25

Table 1: Invention Disclosures are Concentrated in Different Departments within Medical Schools Number of Faculty Members Hopkins Department

Duke Department Anesthesiology

Anesthesiology

Cell Biology

Cardiovascular Division/Medicine Cell Anatomy and Biology

Genetics

Molecular Biology and Genetics

Cardiology

Immunology Microbiology

Immunology Biological Chemistry

Number of Faculty Members who have filed Invention Disclosures JHU

Percentage of Disclosing Faculty

Number of Invention Disclosure Events

Duke

JHU

Duke

Percentage of Disclosing Faculty

Duke

Disclosure Events per Faculty Member

JHU

Disclosure Events per Faculty Member

76

88

12

7.36%

13

7.43%

27

0.355

23

0.261

43

39

11

6.75%

10

5.71%

26

0.605

48

1.231

53

19

13

7.98%

5

2.86%

49

0.925

9

0.474

12

25

7

4.29%

4

2.29%

23

1.917

53

2.12

40

16

14

8.59%

9

5.14%

42

1.05

18

1.125

37

37

5

3.07%

4

2.29%

13

0.351

11

0.297

Ophthalmology

Ophthalmology

34

112

5

3.07%

17

9.71%

13

0.382

130

1.161

Pathology

Pathology

62

91

11

6.75%

20

11.43%

25

0.403

70

0.769

Pharmacology

Pharmacology and Molecular Science

38

25

13

7.98%

13

7.43%

43

1.132

40

1.6

Radiolology

Radiolology

61

84

8

4.91%

10

5.71%

13

0.213

27

0.321

Neurobiology

Neuroscience

42

70

14

8.59%

22

12.57%

32

0.762

53

0.757

OB/Gyn

OB/Gyn

85

85

5

3.07%

1

0.57%

8

0.094

8

0.094

Pediatrics

Pediatrics

121

246

11

6.75%

19

10.86%

14

0.116

28

0.114

Psychiatry

Psychiatry

158

191

9

5.52%

5

2.86%

19

0.12

17

0.089

Surgery

Surgery Total

194

258

25

15.34%

23

13.14%

58

0.299

39

0.151

1056

1386

163

100.00%

175

100.00%

405

0.384

574

0.414

26

Table 2: Difference in Disclosures by Medical School Faculty by Department Duke Faculty Disclosing: 1996-1998 Percentage of faculty disclosing Associate Assistant Professors Professors Chair discloses Full Professors

Percentage of Hopkins Faculty Disclosing 1996-1998 Chair Percentage of faculty disclosing Associate Assistant Discloses Full Professors Professors Professors

Department

Type

Anesthesiology Cardiology Cell Biology Genetics Immunology Microbiology

Nexus Clinical Basic Basic Basic Basic

No Yes Yes No Yes Yes

0% 33% 27% 100% 54% 38%

24% 31% 0% 100% 20% 0%

15% 17% 39% 29% 25% 25%

No Yes No Yes Yes Yes

20% 29% 17% 54% 60% 45%

19% 46% 25% 50% 57% 17%

13% 25% 60% 50% 75% 100%

Ophthalmology Pathology

Clinical Nexus

Yes Yes

38% 25%

0% 10%

16% 13%

No Yes

28% 27%

14% 26%

15% 35%

Pharmacology Radiolology Neurobiology OB/Gyn Pediatrics Psychiatry Surgery

Basic Nexus Basic Clinical Clinical Clinical Nexus

Yes No Yes No Yes No No

23% 24% 30% 0% 21% 10% 13%

25% 5% 38% 33% 12% 7% 29%

47% 12% 33% 5% 2% 3% 8%

Yes Yes No No No No No

69% 31% 41% 29% 13% 5% 26%

100% 23% 58% 0% 17% 5% 15%

66.67 10% 22% 2% 7% 3% 7%

27

Table 4: Descriptive Statistics Mean Std. Dev. Min All Faculty Disclosure Filed in Current Time Period 0.169 0.375 0 Graduate Institution Patent Activity 5.308 9.937 0 Stanford Graduate Degree 0.014 0.117 0 Chair Discloses, Prior Time Period 0.372 0.485 0 Percentage of Cohort with Prior Disclosures 0.176 0.153 0 Years Since Last Graduate Degree 23.216 9.401 4 Non-US Degree 0.114 0.316 0 Individual NIH Awards, Prior Period ($1,000) 620.11 1868.82 0 Number of Previous Disclosures 0.516 1.816 0 Boundary-Spanning Individual 0.338 0.473 0 Dual Degree, Holds Both PhD and MD 0.076 0.266 0 Full Professor 0.255 0.436 0 Assistant Professor 0.457 0.498 0 Nexus Department 0.412 0.492 0 Basic Department 0.141 0.348 0 Percentage of Outside Cohort with Prior Disclosures 0.191 0.135 0.022 University (0 = Hopkins; 1 = Duke) 0.494 0.500 0 Faculty In Non-Basic Medical School Departments Without Prior Disclosures Disclosure Filed in Current Time Period 0.079 0.270 0 Chair Discloses, Prior Time Period 0.248 0.432 0 Percentage of Cohort with Prior Disclosures 0.132 0.111 0 Graduate Institution Patent Activity 4.486 9.203 0 Years Since Last Graduate Degree 23.633 9.588 4 Non-US Degree 0.123 0.329 0 Individual NIH Awards, Prior Period (000) 294.888 1221.371 0 Boundary-Spanning Individual 0.260 0.439 0 Dual Degree, Holds Both PhD and MD 0.052 0.222 0 Full Professor 0.195 0.396 0 Assistant Professor 0.529 0.499 0 Percentage of Outside Cohort with Prior Disclosures 0.168 0.116 0.022 University (0 = Hopkins; 1 = Duke) 0.495 0.500 0 Variable

28

Max 1 114 1 1 1 60 1 19600 31 1 1 1 1 1 1 0.833 1 1 1 0.769 114 60 1 15400 1 1 1 1 0.833 1

Table 5: Correlations All Faculty

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

Disclosure Filed, Current Period Graduate Institution Patenting Stanford Graduate Degree Chair Discloses, Prior Period % Cohort with Prior Disclosure Years Since Last Graduate Degree Non-US Degree Individual NIH Awards ($1,000) Number of Previous Disclosures Boundary-Spanning Individual Dual Degree, Holds PhD and MD Full Professor Assistant Professor Nexus Department Basic Department % Outside Cohort, Prior Discl. University (0=Hopkins; 1=Duke)

1

2

3

4

5

6

7

8

9

10

11

12

1 0.15 0.15 0.19 0.26 -0.15 0.05 0.16 0.39 0.17 0.16 0.10 -0.11 0.01 0.20 0.17 -0.04

1 0.21 0.01 0.03 -0.49 -0.18 -0.04 0.06 0.04 0.12 -0.19 0.17 -0.01 0.09 0.06 -0.14

1 0.08 0.11 -0.05 -0.04 0.01 0.11 0.04 0.11 -0.00 -0.03 -0.04 0.13 0.13 -0.03

1 0.48 0.01 -0.01 0.12 0.14 0.13 0.12 0.14 -0.14 0.12 0.48 0.56 0.14

1 0.08 -0.00 0.24 0.28 0.26 0.16 0.40 -0.43 0.08 0.37 0.32 -0.14

1 0.01 0.10 -0.04 -0.01 -0.10 0.40 -0.31 -0.04 -0.02 -0.03 0.08

1 -0.02 -0.02 -0.00 -0.01 -0.03 -0.00 0.10 -0.06 -0.03 -0.05

1 0.24 0.15 0.06 0.33 -0.23 -0.04 0.17 0.05 -0.03

1 0.15 0.08 0.16 -0.15 -0.00 0.15 0.13 -0.06

1 0.16 0.18 -0.23 0.08 0.17 0.10 -0.23

1 0.01 -0.08 0.00 0.09 0.10 -0.07

1 -0.54 -0.03 0.17 -0.03 -0.02

29

13

14

15

16

1 0.02 1 -0.16 -0.34 1 0.03 0.05 0.29 1 0.09 -0.2 0.06 0.03

17

1

Table 6: All Faculty Members Probit Estimation: DV = Disclosure Filed (0, 1) Model (1) Years Since Last Graduate Degree Graduate Institution Patent Activity

Model (2) -0.034 *** (0.006) 0.010 ** (0.004)

Stanford Graduate Degree

Model (3) -0.040 *** (0.006)

Model (4) -0.035 *** (0.006) 0.010 ** (0.004)

Model (5) -0.026 *** (0.006) 0.009 ** (0.004)

Model (6) -0.026 *** (0.006) 0.009 ** (0.004)

0.888 *** (0.285)

0.763 *** (0.298) 0.451 (0.304) 0.347 *** (0.033) 0.040 * (0.020) 0.410 *** (0.120) 0.206 ** (0.088) 0.412 *** (0.132)

0.721 ** (0.294)

Chair Discloses, Prior Period

0.258 *** (0.093)

Cohort Disclosures, Prior Period Outside Cohort Disclosures, Prior Period Number of Prior Disclosures

0.384 *** 0.359 *** 0.353 *** 0.349 *** 0.348 *** (0.032) (0.033) (0.033) (0.033) (0.033) Individual NIH Awards, Prior Period 0.023 0.027 0.028 0.026 0.039 * (0.021) (0.021) (0.021) (0.021) (0.020) Non-US Degree 0.316 *** 0.436 *** 0.380 *** 0.423 *** 0.412 *** (0.112) (0.120) (0.116) (0.121) (0.119) Boundary-Spanning Individual 0.183 ** 0.180 ** 0.181 ** 0.172 * 0.204 ** (0.086) (0.089) (0.088) (0.088) (0.088) Dual Degree, Both PhD and MD 0.507 *** 0.437 *** 0.435 *** 0.409 *** 0.421 *** (0.127) (0.130) (0.131) (0.131) (0.131) Full Professor -0.056 0.243 ** 0.248 ** 0.239 ** (0.107) (0.117) (0.117) (0.117) Assistant Professor -0.062 -0.252 ** -0.238 ** -0.251 ** (0.095) (0.101) (0.101) (0.102) Basic Science Department 0.610 *** 0.517 *** 0.507 *** 0.362 *** 0.428 *** 0.392 *** (0.114) (0.118) (0.118) (0.130) (0.125) (0.127) Nexus Service Department 0.191 ** 0.180 * 0.183 ** 0.167 * 0.116 0.103 (0.089) (0.092) (0.092) (0.092) (0.093) (0.093) University Dummy Variable -0.038 0.049 0.031 0.025 0.055 0.053 (0.081) (0.085) (0.084) (0.085) (0.085) (0.085) Constant -1.514 *** -0.852 *** -0.657 *** -0.896 *** -1.192 *** -1.253 *** (0.102) (0.181) (0.159) (0.183) (0,161) (0.166) N Log Likelihood Pseudo R2 * p < 0.10

1793 -642.831 0.212 ** p < 0.05

1779 -610.824 0.248

*** p < 0.01

30

1779 -610.514 0.248

1779 -607.044 0.252

1779 -613.949 0.244

1779 -612.860 0.245

Model (1)

0.049 (0.040) 0.333 ** (0.144) 0.187 (0.124) 0.703 *** (0.178) -0.111 (0.161) -0.092 (0.127) 0.213 (0.182) 0.270 (0.240) 0.247 (0.206) 0.203 (0.200) -0.084 (0.242) -0.331 (0.237) -0.278 (0.179) -0.622 *** (0.221) -0.025 (0.111) -1.453 *** (0.159)

0.051 (0.040) 0.480 *** (0.153) 0.147 (0.127) 0.621 *** (0.185) 0.204 (0.173) -0.283 ** (0.137) 0.122 (0.189) 0.181 (0.246) 0.267 (0.218) 0.241 (0.205) -0.205 (0.249) -0.242 (0.251) -0.228 (0.186) -0.727 *** (0.233) 0.042 (0.116) -0.663 ** (0.264)

0.050 (0.040) 0.438 *** (0.156) 0.131 (0.128) 0.612 *** (0.185) 0.198 (0.174) -0.286 ** (0.137) 0.169 (0.191) -0.135 (0.298) 0.188 (0.222) -0.070 (0.264) -0.367 (0.271) -0.196 (0.252) -0.336 * (0.196) -0.679 *** (0.234) -0.034 (0.123) -0.653 *** (0.265)

0.067 * (0.038) 0.429 *** (0.153) 0.190 (0.125) 0.676 *** (0.184)

Model (5) -0.028 *** (0.008) 0.006 (0.006) 0.106 (0.226) 1.124* (0.584) 0.914 (0.591) 0.065 * (0.038) 0.398 ** (0.155) 0.201 (0.126) 0.681 *** (0.184)

0.188 (0.191) 0.119 (0.250) 0.319 (0.217) 0.117 (0.212) -0.189 (0.248) -0.154 (0.255) -0.177 (0.188) -0.629 *** (0.230) 0.041 (0.115) -1.15*** (0.244)

0.273 (0.196) -0.011 (0.303) 0.293 (0.225) -0.020 (0.270) -0.245 (0.269) -0.100 (0.257) -0.261 (0.200) -0.486 ** (0.243) -0.024 (0.124) -1.311 *** (0.278)

1291 -327.031 0.083

1280 -308.845 0.132

1280 -307.103 0.137

1280 -310.463 0.128

1280 -308.285 0.134

Years Since Last Graduate Degree Graduate Institution Patent Activity

Model (2) -0.038 *** (0.009) 0.008 (0.006)

Chair Discloses, Prior Period

Model (3) -0.038 *** (0.009) 0.007 (0.006) 0.356 * (0.190)

Cohort Disclosures, Prior Period Outside Cohort Disclosures, Prior Period Individual NIH Awards, Prior Period Non-US Degree Boundary-Spanning Individual Dual Degree, Both PhD and MD Full Professor Assistant Professor Anesthesiology Cardiology Ophthalmology Pathology Radiology OB Pediatrics Psychiatry University Dummy Variable Constant N Log Likelihood Pseudo R2 * p < 0.10

** p < 0.05

Model (4) -0.030 *** (0.008) 0.006 (0.006)

1.272 ** (0.561)

*** p < 0.01

31

Appendix Table: Empirical Results: PROBIT Model: All Faculty with Department Fixed Effects Dependent Variable = Disclosure filed (0,1) Surgery Omitted Years Since Last Graduate Degree

-0.025*** (0.005)

Graduate Institution Patent Activity

0.008*** (0.004)

Chair Discloses, Prior Period

0.259 (0.162)

Cohort Disclosure, Prior Period

0.604* (0.328)

Outside Cohort Disclosure, Prior Period

0.231 (0.374)

Individual NIH Awards, Prior Period

0.039 (0.020)*

Non-US Degree

0.421*** (0.122)

Boundary Spanning Training

0.210*** (0.090)

Holds both MD and PhD degrees

0.409*** (0.131)

Number of individual disclosures, prior period

0.344 (0.34)**

Anesthesiology Department

0.216 (0.169)

Cardiology Department

0.289 (0.227)

Ophthalmology Department

0.246 (0.190)

Pathology Department

0.109 (0.202)

Radiology Department

-0.014 (0.196)

Obstetrics Department

-0.349* (0.241)

Pediatrics Department

-0.0.68 (0.147)

Psychiatry Department

-0.400* (0.188)

Pharmacology Department

0.461 (0.271)**

Cell Biology Department

0.102 (0.249)

Genetics Department

0.443 (0.290)

Immunology Department

0.561 (0.261)**

Biology Department

0.406 (0.285)

Neurology Department

0.424** (0.209)

Constant N Log Likelihood Pseudo R2

-1.168*** (0.207) 1779 -607.540 0.2576

Appendix Table ALT: Empirical Results: PROBIT Model: All Faculty with Department Fixed Effects Dependent Variable = Disclosure filed (0,1) Surgery Omitted Years Since Last Graduate Degree

-0.039*** (0.006)

Graduate Institution Patent Activity

0.010*** (0.004)

Number of Previous Disclosures

0.344 *** (0.034)

Individual NIH Awards, Prior Period

0.039 (0.021)*

Non-US Degree

0.421*** (0.122)

Boundary Spanning Training

0.211*** (0.091)

Holds both MD and PhD degrees

0.409*** (0.126)

Full Professor

0.292 *** (0.111)

Assistant Professor

-0.352 *** (0.099)

Anesthesiology Department

0.125 (0.165)

Cardiology Department

0.454 (0.186)

Ophthalmology Department

0.304 (0.180)

Pathology Department

0.342** (0.161)

Radiology Department

0.042 ** (0.185)

Obstetrics Department

-0.403* (0.242)

Pediatrics Department

-0.108 (0.147)

Psychiatry Department

-0.551 *** (0.179)

Pharmacology Department

0.674 (0.021)***

Cell Biology Department

0.338 (0.202)*

Genetics Department

0.348 (0.292)

Immunology Department

1.027 (0.227)***

Biology Department

0.632 (0.250)**

Neurology Department

0.516*** (0.170)

University dummy variable

0.061 (0.087)

Constant

-0.469 (0.250)

N

1779

Log Likelihood

-607.540

Pseudo R2

0.258

33

34