Gender ethnicity and entrepreneurship RP

Page 1

R

Gi

MC

a

ARAA

KIBMGE

1

f(wttilpi

dMt

(

h0

ARTICLE IN PRESSG ModelESPOL-3103; No. of Pages 12

Research Policy xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Research Policy

jo ur nal ho me page: www.elsev ier .com/ locate / respol

ender, ethnicity and entrepreneurship in initial public offerings:llustrations from an open database*�

artin Kenney ∗, Donald Pattonommunity and Regional Development Unit, University of California, Davis, CA, USA

r t i c l e i n f o

rticle history:eceived 4 January 2015ccepted 8 January 2015vailable online xxx

eywords:nitial public offeringsoard of directorsanagement teamsenderthnicity

a b s t r a c t

This paper describes the variables in a freely available database of all emerging growth firms (EGF) thatmade an initial stock offering (IPO) on US public markets from 1990 through 2010. Our expectation isthat researchers from a variety of disciplines can use this data to answer a wide variety of social sciencequestions and combine it with other databases. To illustrate how the data can be used, we describe thegender and nationality of the top management teams (TMTs) and board of directors (BoDs) of thesefirms. We confirm that women are under-represented in all functional positions, but, in contrast tomuch of the popular press, we find that statistically Silicon Valley firms perform better than the nationalaverage. Gender ratios differ by function with women most prevalent at the CFO position and are mostprevalent in the biomedical industry. Using undergraduate education, as an identifier for nationality, wefind that, contrary to the popular press, there are more European than Asian immigrants in the TMTs. Thissuggests that European immigrants are more likely to immigrate with advanced degrees, while the Asianimmigrants have only Bachelor’s degrees. In the immigration literature, it has been observed that specificimmigrant groups concentrate in particular occupations. To test for this effect, we study the backgrounds

of all identifiable Taiwanese immigrants. A remarkably high concentration of Taiwanese TMT memberswere from two Taiwanese universities’ electrical engineering departments, then received U.S. graduatedegrees, particularly from UC Berkeley, and entered semiconductor-related industries. This databasewill contribute to reproducible social science as the same quality-controlled data is now available to allresearchers.

© 2015 Elsevier B.V. All rights reserved.

. Introduction

In the U.S. political economy, the life-cycle of the most success-ul entrepreneurial firms usually contains an initial public offeringIPO). From a macro-political economy perspective these firms,hile a highly select population, are inherently significant, because

hey include nearly all of the largest firms created in the lastwo decades. From an analytical perspective, studying IPO firmss advantageous because U.S. legal requirements mandate that the

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016/

isting firm provide detailed information on its operations and keyersons. Their importance, combined with the level of detailed

nformation revealed at their public offering, make these firms of

� The authors gratefully acknowledge the support of the National Science Foun-ation Science of Science Policy Program award #08-004743. The authors thankaryann Feldman and three anonymous reviewers for their comments and sugges-

ions. We also thank Casey Castaldi for her research assistance.∗ Corresponding author. Tel.: +1 5307520328.

E-mail addresses: [email protected] (M. Kenney), [email protected]. Patton).

ttp://dx.doi.org/10.1016/j.respol.2015.01.003048-7333/© 2015 Elsevier B.V. All rights reserved.

great inherent interest to scholars and policy-makers, alike. Newentrepreneurial firms, many of which are funded by venture cap-ital, are an important component of the U.S. national innovationsystem (on U.S., see Mowery 1992 on venture capital, see Kenney,2011). In this innovation model, particularly in information, com-munication, and biomedical technology-related entrepreneurship,an initial public stock offering is an important step in the firm’s life.With a few notable exceptions, such as SAS and Epic Systems, overthe last five decades successful U.S. technology-based firms haveeither been acquired or have undertaken IPOs.

icity and entrepreneurship in initial public offerings: illustrationsj.respol.2015.01.003

This paper describes a freely available database of allentrepreneurial firms undertaking an initial public stock offeringduring the twenty-one year period from 1990 through 2010.1 We

1 Since 2010 we have provided data to more than 150 researchers. Most of therequests have come from doctoral students in business schools, including UC Berke-ley, MIT, Harvard, Yale, and Wharton, but we have also had requests from universityfaculty and research staff as well as numerous research institutes such as Brookingsand the World Bank. Data has been provided to the venture capital-financed firmFindthebest.com and The Private Capital Research Institute.

Page 2

ING ModelR

2 earch

stsabp

itsioeLcpvesa

staaovutaedwsm

2

sbsapaNObCKdeefvccw“

imi(2

the 2028 firms that conducted an IPO between 1996 and 2010 andthat were under 30 years of age at the time of their IPO. We are nowexpanding the database through 2014. For the years from 1990 to

2 Jay Ritter has eased this comparison process as his website

ARTICLEESPOL-3103; No. of Pages 12

M. Kenney, D. Patton / Res

elect a few salient variables regarding gender and national origino illustrate potential uses of the database to answer a number ofcholarly questions, especially in regards to entrepreneurship in

variety of sectors, but particularly the science and technology-ased firms that make up more than 50 percent of the entire firmopulation.

Because the database is an Excel format, the ability to combinet with other databases, public or private, should accelerate fur-her research on entrepreneurship in the social and managementciences. The value of this database is not only intrinsic, but alson the fact that it can be rather easily linked to a wide variety ofther databases including Starmetrics database described by Lanet al. (2015) (in this issue) and the patent database described byi et al. (2014). The database can also be extended by using otherommonly available databases such as VenturExpert, CRISP stockrice data, the NETS database, and other commercial databases pro-iding data extracted from financial filings. Other uses could be toxamine the effects of policies such as the SBIR Program in creatinguccessful firms or the fate of university spinoffs. In essence, this is

resource that others can build upon in their research.This brief contribution describes the database and undertakes

ome simple illustrations of how it might be used. The first sec-ion discusses IPOs and previous usage of the data from IPOs incademic papers. This is followed by a description of the publiclyvailable variables in the database. The first illustration of the usesf the database is an exploration of gender issues in these firms byarious managerial and directorial functions and differences basedpon industry or geography. The second illustration is a descrip-ion of the national origin of the managers of these firms, including

discussion of the education of these managers with particularmphasis on foreign PhDs. To further demonstrate the value of theatabase we examine aspects of the career trajectories of all Tai-anese university graduates serving on the TMTs. The concluding

ection suggests some benefits that such an open-source databaseay provide to the scholarly community

. IPOs

An IPO is a singular event in a new firm’s life-cycle when its tran-itions from being private to being publicly held. Because of the highar for undertaking a public offering, IPO firms are among the mostuccessful and for this reason alone are of great interest. The otherttraction is the level of detailed information that encompasses keyersonnel, firm promoters, significant business contacts (such aslliances with other organizations) and various other disclosures.aturally, its uses for scholarly purposes have not gone unnoticed.f particular interest has been the attributes of top managers andoard members (e.g., Amason et al., 2006; Beckman et al., 2007;hen et al., 2008; Cohen and Dean, 2005; Higgins and Gulati, 2006;im and Higgins, 2007). Another field of research has exploited theisclosures to examine the affiliations of IPO firms and their back-rs with other organizations (Pollock and Gulati, 2007; Hochbergt al., 2007). For example, Gulati and Higgins (2003) extracted datarom a sample of biotechnology IPO firms to explain the signalingalue of different types of interorganizational partnerships in theontexts of different types of uncertainty and found that ventureapital endorsements were most important during “cold” markets,hile investment bank endorsements were more beneficial during

hot” markets.Founder retention in entrepreneurial firms is of interest both

n finance and general management. The employment of one or

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016

ore of the founders at the time of the IPO has been found toncrease a firm’s value by sending a positive signal to investorsCerto et al., 2001; Fischer and Pollock, 2004; Jain and Tabak,008; Nelson, 2003). Little is known about whether function affects

PRESSPolicy xxx (2015) xxx–xxx

founder retention. There is evidence that retention may also behigher for the technical founders than for the general managers,though this has not been tested on a large sample of IPOs (Boekerand Karichalil, 2002). Pollock et al. (2009) found that CEO-founderswere more likely to be present at the IPO when there was greaterbusiness certainty in the particular sector. Put differently, when thesector was competitive and uncertain the founder was less likelyto be CEO at the time of the IPO. The addition of high-prestige CEOreplacing the founder may be the result of the firm wishing to cre-ate the perception of legitimacy among public investors (Chen et al.,2008; Cohen and Dean, 2005).

Despite the intense interest in entrepreneurship, and the factthat many of best known U.S. technology firms including Microsoft,Intel, Oracle, Google, and Yahoo! retained their founders in keypositions through the IPO and beyond, there has been minimalresearch on the persistence of founders from the firm’s inceptionto a liquidity event such as an IPO and its impact on firm perfor-mance (for exceptions, see Beckman and Burton 2008; Beckmanet al., 2007; Liu et al., 2012).

IPOs have also been used in agency theory studies, because thefirms are undergoing a transition from privately held firms withprincipals such as venture capitalists closely monitoring the firm’sbehavior, to a situation where they are being monitored by lessengaged actors such as securities analysts. This interest unleasheda wave of studies regarding the board composition at the time of theIPO (Beatty and Zajac, 1994; Certo, 2003; Filatotchev et al., 2006;Bruton et al., 2010).

While there has been significant use of IPO data to ask questionsspecifically related to IPOs, there is a much larger research commu-nity that should be able to use portions of the database in ongoingprojects in which IPOs only play a small part. For example, indus-try studies students might find it useful to extract data on IPOs inspecific industries, or just identify the IPOs in a particular industry.Studies of the spatial agglomeration or specializations of venturecapital firms could utilize IPOs as a success measurement. A cross-national comparison of high-impact entrepreneurship is receivingmore attention through venues such as the Global Entrepreneur-ship Monitor, but there have been fewer comparisons between IPOsin different markets (Doidge et al., 2013).2 The information in ourdatabase could simplify conducting such comparisons because theU.S. data is now available in an easily usable form. Because U.S. stockmarkets are the mecca for foreign EGFs, it is possible to compareU.S., Chinese, and Israeli EGFs that list on U.S. markets.3

there is a list and description of the data that we have extractedand is already provided or in the future will be provided on request.In addition to the firm data, we have assembled a database of all themanagers and directors of these firms as listed in IPO prospectus,where each unique individual is given an ID as some individualswere involved in more than one IPO (this is a unique feature of thedatabase and will save substantial time for researchers).

Because this is a database of EGF IPOs, it is a subset of all firmsthat have gone public. The database currently contains 3939 firmsand 43,695 individuals who were managers and directors of thesefirms. However, because some portions of the 1990–1996 data arenot yet fully completed, we confine the substantive illustrations to

icity and entrepreneurship in initial public offerings: illustrations/j.respol.2015.01.003

http://bear.warrington.ufl.edu/ritter/ipodata.htm has information on wheredata for IPOs in other nations can be accessed.

3 This may be particularly interesting in the case of China, as many key ChineseInternet firms have listed on US markets. For example, Sina.com is an analog ofYahoo!, Baidu resembles Google, 51 Jobs resembles Monster.com, etc.

Page 3

ARTICLE IN PRESSG ModelRESPOL-3103; No. of Pages 12

M. Kenney, D. Patton / Research Policy xxx (2015) xxx–xxx 3

Table 1Variables included in emerging growth firm IPO database.

Data currently available Future data availability

Firm data 1990–2010Company ID/CIK number Law firm name and address, 1996–2010Company name Investment banks, 1996–2006Year of the IPO. Number of employees data textState of incorporation Number of employees at IPOCompany street address After-IPO employment by year 1996–2010Stock exchange After-IPO revenue by year 1996–2010a

Stock symbol IPO firm fate through 1996–2010–continuing operation, merger, delistingShares offeredShares outstandingInitial share priceOffer sizeUnderwriter discountCompany auditorSEC 4-digit SICFirm 4-digit SICYear firm foundedUnit offering Y/NInternet firm Y/NIPO prospectuses after OCR, 1990–1996 (by request)IPO TMT and BoD variables, 1990–2010Individual unique identifier Venture capitalist Y/NFull name University education by degreeb

Age at the time of the IPO. University addressCompany Position Address for BoD members, 1996–2010b

TMT and BoD member’s biography Firm founder or co-founderb

Gender Founder or co-founder of another for-profit firmb

Legend: bold are totals and unbolded are sub-totals.a This data is from CRSP and thus cannot be open-sourced, but will be provided on an individual request basis.b When reported. For certain variables, the SEC data was augmented by Internet searches.

firm I

2fitb2t

s4b

a

Fig. 1. Emerging growth

010, the EGF database accounts for around 77% of all domesticrms that went public. There were 2247 EGF IPOs from June 1996

o 2010. Fig. 1 shows that the number of IPOs differed dramaticallyy year. The drop of IPOs after the end of the Dot.com Bubble in000 was dramatic and for the decade was lower than anytime inhe 1990s.8

The industrial composition of these emerging growth IPOs is

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016/

hown in Table 2. All but 42 firms can be classified based on their-digit Standard Industrial Code.9 We group these into one of sixasic industry sectors along with those identified as Internet firms.

8 For an explanation of this drop, see Gao et al. (2013).9 The data for these 42 firms were extracted from older pdf documents that lacked

n SIC assignment.

POs by Year 1990–2010.

The largest single sector is Information and Computer Technology(ICT), which was responsible for 22.7% of all IPOs and, if combinedwith the Internet (11.3%), comprised exactly 34% of all EGF IPOs.

Because the address of each firm is included, the geographic dis-tribution of IPOs can be analyzed at any level needed (Kenney et al.,2012: 14–16; Patton and Kenney 2005). EGFs are concentrated in afew states (see Table 3

3. Database description and overview

Any firm intending to sell its stock to the public on U.S. markets

icity and entrepreneurship in initial public offerings: illustrationsj.respol.2015.01.003

must file a publicly available prospectus with the U.S. Securitiesand Exchange Commission prior to the offering. Prior to the offer-ing, the issuing firm must make significant disclosures about itsoperations, management, and governance. It was not until June

Page 4

ARTICLE IN PRESSG ModelRESPOL-3103; No. of Pages 12

4 M. Kenney, D. Patton / Research Policy xxx (2015) xxx–xxx

Table 2Emerging growth IPOs by industry sector and component, 1990–2010.

Basic Sector Sector Components Count Count Percent of Total

Bio-medical 464 11.9Biotechnology 281 7.2Medical Instruments 183 4.7

ICT 883 22.7Communications 112 2.9Computer Systems 76 2.0Computers 125 3.2Electronic Equipment 116 3.0Semiconductors 104 2.7Software 259 6.6Telephone and Telegraph 91 2.3

Internet 440 11.3Retail 437 11.2

Retail Trade 278 7.1Wholesale Trade 159 4.1

Services 506 13.0Business Services 82 2.1Computer Programming 69 1.8Computer Services 59 1.5Health Services 116 3.0Services 180 4.6

Manufacturing 435 11.2General Instruments 64 1.6Machinery 97 2.5Manufactured Goods 274 7.0

Other 732 18.83, 897 100.0

423, 939

1IfJttMa

ittiaOoe

atetemetfia

pIRSa

p

Table 3EGF IPOs by industry class among the top five states, 1990–2010.

Basic Class California N. Y. Mass Texas Florida Top 5

Internet 184 40 40 18 14 296% of Class 42% 9% 9% 4% 3% 67%Bio-medical 174 24 66 14 15 293% of Class 38% 5% 14% 3% 3% 63%ICT 368 37 92 46 32 575% of Class 42% 4% 10% 5% 4% 65%Retail 80 41 11 42 37 211% of Class 18% 9% 3% 10% 8% 48%Services 99 56 33 39 28 255% of Class 20% 11% 7% 8% 6% 50%Manufacturing 106 33 21 34 17 211% of Class 24% 8% 5% 8% 4% 49%Other 119 76 24 94 52 365% of Class 16% 10% 3% 13% 7% 50%No SIC 11 1 5 2 2 21

Total

No SIC

Database Total

996 that public firms were required to make all of their documentsnternet accessible. Therefore, online access to all IPO documentsor firms going public only begins in June 1996. For IPOs prior toune 1996, we relied on IPO registration statements and prospec-uses that were scanned into either PDF or TIFF format. It is fromhese documents that the data for IPOs from January 1990 through

ay 1996 were obtained. Because the data is derived from publiclyvailable sources we can freely distribute it.

The database is comprised of all emerging growth firm (EGF)nitial public offerings on American stock exchanges and filed withhe Securities and Exchange Commission (SEC) from January 1990hrough December 2010. In assembling the set of firms to bencluded we relied on Thomson Financial Venture Expert, SDC datand other sources to generate a list of all IPOs over this time period.4

ur interest in IPOs is centered on new firm formation. Becausef this interest we have chosen to restrict our data collection tomerging growth, or de novo, firms that have gone public.5

Emerging growth refers to newly established firms, or firms thatre not based on older firms by being a spinoff or subsidiary opera-ion. The emerging growth status of firms was established throughxamination of the prospectus, particularly in the summary wherehe firm describes its activities, history, and business. Usually themerging growth status of a firm going public can be readily deter-ined. Companies that are the product of a merger are considered

merging growth, if they have had some history of operation prior

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016

o the IPO, while companies that are reorganizations of existingrms are not. Any firm that originated as a new, independent entitynd remained independent throughout its history is considered to

4 There are many lists of IPOs, none identical to the other. Although we reliedrimarily on SDC data, we worked closely with Jay Ritter to develop a list of EGF

POs (Kenney et al., 2012). In the process of matching the lists both our database anditter’s was improved. Ritter had 31 IPOs that we had missed confining ourselves toDC data (approximately 1.5%), and misclassifications in both databases were foundnd corrected.

5 Jay Ritter has recently updated tables and statistics on IPOs from 1960 to theresent on his website http://bear.warrington.ufl.edu/ritter/.

Total 1,141 308 292 289 197 2,227% of Total 29% 8% 7% 7% 5% 57%

be an EGF.6 The database, itself, has significant value because itis a curated list of emerging growth IPOs, thereby eliminating thenecessity for scholars to examine prospectuses themselves – a sub-stantial time savings.7 The SEC filings provide an enormous amountof information on the listing firm. In Table 1) with over half located

icity and entrepreneurship in initial public offerings: illustrations/j.respol.2015.01.003

in California, New York, Massachusetts, and Texas. California aloneis home to 29% of the total. The distribution is particularly concen-trated in the Internet, ICT, and bio-medical sectors where California

6 As does Ritter, we exclude the following types of firms and filings: mutual funds,real estate investment trusts (REITs), asset acquisition or blank check companies,ADRs, and foreign F-1 filers. These are common exclusions in studies of IPOs, butwe include some types of filings (unit offerings) and firms (banks) that are oftenexcluded (see Ritter and Welch 2002: 1797). For these firms, we include a dummyvariable so that researchers can exclude them, if they wish.

7 Ritter provides lists of growth capital-sponsored and roll-up IPOs, for furtherinformation, see http://bear.warrington.ufl.edu/ritter/ipodata.htm.

Page 5

ING ModelR

earch

aa

ctttwtiptSepcbbaStiti

4

tWwwr

5

cTetltofnHbsspbm

id

tai

omeo

ARTICLEESPOL-3103; No. of Pages 12

M. Kenney, D. Patton / Res

ccounted for approximately 40% of all EGFs. The top five statesccount for well over 60% of the total IPOs in these three sectors.10

The geographical data provided above is available from someommercial IPO databases, though we have more thoroughly vet-ed the EGFs. More significantly, for geographers we have identifiedhe location of all of the venture capitalists on the board of direc-ors. One significant contribution here is that each venture capitalistas searched by name allowing their attribution to the correct loca-

ion for multilocational venture capital firms. This is a significantmprovement because nearly all previous research on the geogra-hy of linkages between firms and venture capitalists has attributedhe venture capital activity to the headquarters (see, for example,orenson and Stuart 2001 among many others), when, as Chent al. (2010) show, most larger venture capital firms have multi-le offices.11 We also locate the EGF’s law firm. With the ventureapitalists on the board, the firm’s law firm, and the investmentank’s law firm (a proxy for the location of the lead investmentanker), one can understand the location of some key intermedi-ries in an EGF’s growth (for the use of this data, see Buhr andmith, 2010; Kenney and Patton, 2005). Of particular interest inhe future will be to examine the evolution of these intermediariesn an emerging industry such as the Internet, for which we havehe entire population.12 In the next two sections, we turn to somellustrative applications

. Illustrative applications – gender and national origin

In the following applications of this data we restricted our atten-ion to EGFs that were less than 30 years old at the time of their IPO.

e also limited our attention to IPOs from June 1996 to 2010 forhich we extracted the gender of all managers and directors, asell as the educational background of all managers and directors

eporting their education.

. Gender and IPOs

The topic of gender and entrepreneurship has received signifi-ant media and academic attention (e.g., Robb and Watson 2012).he 2013 IPO of the social media firm Twitter heightened inter-st in the topic of gender in the management and direction ofechnology firms, especially those headquartered in Silicon Val-ey (e.g., Kristof, 2013). Despite the now voluminous research onhe impact of top management team (TMT) diversity, the resultsn its effect are inconsistent (Cannella et al., 2008). Similarly, thusar in terms of IPO performance the evidence for gender discrimi-ation is mixed (Mohan and Chen 2004; Welbourne et al., 2007).owever, other studies have suggested that discrimination maye occurring. To illustrate, a recent experimental study asking MBAtudents to rate IPO prospectuses found that “despite identical per-

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016/

onal qualifications and firm financials, female founders/CEOs wereerceived as less capable than their male counterparts, and IPOs ledy female founders/CEOs were considered less attractive invest-ents,” (Bigelow et al., 2012). While not using gender as a variable,

10 It is notable with the greater granularity that is possible with the database,ndustrial clusters such as Houston in oil-related activities and Minnesota in medicalevices do become visible.11 This problem of attribution is impossible to correct without having the name ofhe venture capitalists. Given that this entire genre of papers is about social networksnd in venture capital the networks are between individuals, i.e., entrepreneurs andndividual venture capitalists, this can lead to spurious findings regarding distance.12 As mentioned earlier, the definition of the population depends on the definitionf an Internet firm. In a broad definition that includes data communications equip-ent and semiconductor firms such as Cisco and Broadcom, we would not have the

ntire population. If Internet refers to worldwide web-based firms only, then, givenur coverage from 1990 to 2010, we have the entire industry.

PRESSPolicy xxx (2015) xxx–xxx 5

Zimmerman (2008) found that general TMT diversity had a positiveeffect on the amount of capital raised.

In 2014 the law firm, Fenwick and West (2014) reported therepresentation of women in the management teams or board ofdirectors of 150 largest public Silicon Valley technology firms andcompared it to that of the top S&P 100 firms.13 Their study includesa number of now mature firms such as Hewlett Packard, Intel, andothers where the executives no longer have any relationship to thefirm’s entrepreneurial period. This makes it possible to comparethe gender composition by function in the EGFs with firms in theS&P 100 and the SV 150. During the time period under study boththe S&P 100 and the SV 150 public firms experienced percentageincreases in women in all positions (see Table 4). For example, at theCEO level for the SV 150 there was an increase from zero in 1996 to2.7 percent in 2013, while the S&P 100 increased from 0 in 1996 to 6percent in 2013. For our EGFs, women CEOs comprised 3.1 percent.For all groups of firms, women were between three and four timesmore likely to be CFOs than CEOs. Both S&P 100 and SV 150 sawdramatic increases in the percentage of CFOs (from approximately2% to 10.1% and 13.3%, respectively), while for the EGFs the CFOswere 9.5%.

When the entire TMT is considered, women made up approx-imately 5% in 1996 for both the S&P 100 and SV 150, but bothincreased markedly in 2013 to 14.7% and 11.5%, respectively (seeTable 4). In contrast, women in the TMT increased only slightlyin the IPO firms. The women board members among the existingfirms increased significantly from 1996, 11% and 1%, respectively,to 19.9% and 9.1% in 2013, while for IPOs only 5% of the direc-tors were women. The proportion of women directors, though,did increase among the IPO firms suggesting that political pres-sure for increased representation may have been effective. As ageneral rule, in the 1990s, despite having low numbers of womeninvolved in the TMT or BoD, EGFs initially compared favorably toboth the S&P 100 and SV 150, but by 2013 lagged the public firms.This suggests that firms subject to public disclosure and pressurethey are more likely to respond, but when private may make deci-sions based on other criteria, such as, homophily and strong-tienetwork constraints (McPherson et al., 2001; Ruef et al., 2003).14

The changes in Table 4 suggest that the increased concernregarding the representation of women in TMT positions and onBoDs had an impact. For this reason, it might be expected that therewere changes overtime in the percentages of women at variouspositions in our firm population. However, Fig. 2 shows that therewas, at best, only a slight change over time, with the single excep-tion that the proportion of female directors did have a slight, butsignificant increase from 1996 to 2010.15

In 2014, popular press articles and even the U.S. presidentfaulted Silicon Valley, in particular, for having an under-representation of women in various top management and directorpositions (see, for example, Huddleston, 2014). Contrary to expec-tations, our data finds that Silicon Valley EGFs have a higherpercentage of female CEOs than in any other region and the national

icity and entrepreneurship in initial public offerings: illustrationsj.respol.2015.01.003

average (5.2% versus 3.1%) (See Table 5). The relatively open andperformance-based networks in the region may be more accept-ing of women executives and directors (Cohen and Fields 2000).16

13 Fenwick and West adopt the San Jose Mercury News definition of Silicon Valleyas consisting of Santa Clara, San Mateo, San Francisco, Alameda, and Contra Costacounties.

14 It is difficult to measure or define “competence,” as a criteria for appointingindividuals to these various positions.

15 The increase in the proportion of non-executive women directors from 1996 to2010 was significant at the .05 level. All other changes in the proportion of womenover time in other capacities were not significantly different from zero.

16 On general network openness in Silicon Valley, see Cohen and Fields (2000). Fora more formal effort to model these networks, see Fleming and Frenken (2007).

Page 6

ARTICLE IN PRESSG ModelRESPOL-3103; No. of Pages 12

6 M. Kenney, D. Patton / Research Policy xxx (2015) xxx–xxx

Table 4Number and Percent Women by Function in All EGF IPOs from 1996–2010, Silicon Valley 150 and S&P 100 Public Firms in 1996 and 2013.

IPOpersonneltotal

IPOpersonnelwomen

Silicon Valley 1501996

Silicon Valley 1502013

Standard and Poor’s1001996

Standard and Poor’s1002013

Total 24,509 2,080% Women 8.5%Directors 9,906 499% Women 5.0% ∼1% 9.1% ∼11% 19.9%TMT 14,603 1,581% Women 10.8% ∼5% 11.5% ∼5% 14.7%CEO 2,028 63% Women 3.1% 0% 2.7% 0% 6.0%CFO 2,004 191% Women 9.5% ∼2% 13.3% ∼2% 10.1%TMT w/t CEO and CFO 10,571 1,327% Women 12.6%C-level technical 582 6% Women 1l0% 0% 10.0% 0% 5.0%

Twenty four individuals had a joint function of CEO/CFO. All of these were counted as CEOs only. Percentages of women directors and TMT for the SV 150 and the S&P 100were the average percentage of women directors and TMT with each firm. Only chief-level technical, scientific, and medical officers (CTO, CSO, and CMO) were counted asC-level technical personnel. For the SV 150 and S&P 100 these percentages were the percentage of companies with a woman CTO or other top technology, engineering orR&D executive.

Fig. 2. Percentage women by function in year of firm’s IPO, 1996–2010.

Table 5Number and percent women for various management positions and directors by regions, 1996–2010.

CEOs CFOs Directors TMT TMT w/t CEO and CFO Totalpersonnel

Silicon Valley 22 49 124 391 320 515% Women 5.2% 11.7% 5.8% 12.0% 13.3% 9.5%Southern California 3 24 43 171 144 214% Women 1.4% 11.6% 4.3% 11.7% 13.7% 8.7%New York State 7 15 45 149 127 194% Women 4.6% 10.1% 6.6% 13.7% 16.2% 11.0%Massachusetts 3 10 39 118 105 157% Women 2.0% 6.6% 4.9% 10.3% 12.5% 8.1%AllOther 28 93 248 752 631 1,000

Fctdtpi

% Women 2.6% 8.6% 4.7%

Total 63 191 499

% Women 3.1% 9.5% 5.0%

or the second most important position, the CFO, once again, Sili-on Valley had higher percentages female than other regions andhe national average (11.7% versus 9.5%). Contrary to common wis-om, in every category but one, Silicon Valley placed higher than

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016

he national average and had either the highest and next highestercentage of female involvement. Silicon Valley led other regions

n women in the two top corporate positions, CEO and CFO. From

9.8% 11.5% 7.7%1,581 1,327 2,080

10.8% 12.6% 8.5%

this we can conclude that while leadership in Silicon Valley EGFs isabsolutely not “representative” of the overall U.S. population, theperformance appears to be somewhat better than other regions.Chi-square analysis suggests that these observed differences in

icity and entrepreneurship in initial public offerings: illustrations/j.respol.2015.01.003

women’s participation in Silicon Valley are not random – the TMTdifferences from the entire population are significant at the .001level in every case except for the directors.

Page 7

ARTICLE IN PRESSG ModelRESPOL-3103; No. of Pages 12

M. Kenney, D. Patton / Research Policy xxx (2015) xxx–xxx 7

Table 6Percent women for various management positions and directors by industry, 1996–2010 (Number of firms in parentheses).

Basicsector Sector components CEO CFO Directors TMT TMT w/t CEO and CFO Total

Bio-medical (255) 4.3% 10.3% 6.7% 13.1% 15.4% 10.3%Biotechnology (165) 5.5% 11.7% 7.3% 14.2% 16.4% 11.1%Medical Instruments (90) 2.2% 7.9% 5.6% 11.3% 13.6% 8.8%

Internet (431) 4.9% 9.8% 5.7% 12.1% 13.7% 9.7%ICT (419) 1.9% 10.0% 3.5% 7.8% 8.5% 6.2%

Communications (44) 6.8% 13.6% 2.3% 9.2% 8.8% 6.7%Computer Systems (43) 0% 14.0% 1.9% 9.1% 9.9% 6.6%Computers (46) 2.2% 4.3% 3.0% 5.8% 6.8% 4.7%Electronic Equipment (35) 0% 8.8% 2.8% 5.2% 5.5% 4.2%Semiconductors (69) 0% 7.2% 1.9% 2.3% 1.8% 2.1%Software (126) 2.4% 10.3% 4.8% 9.6% 10.9% 7.8%

5%

0%

5%

eitsrgIttraIaTfittgf

nliprhttl2wlewtves

t

poittd

Telephone and Telegraph (56) 1.8% 12.All Other (923) 2.5% 9.Total (2028) 3.1% 9.

It is almost axiomatic that industry “recipes” affect a wide vari-ty of practices (Spender 1989), and in nearly all studies wherendustry is used as a control variable it is significant. In Table 6he percentages of women executives and board members is pre-ented. First, in the biomedical field, women were more stronglyepresented than in the other fields, though there presence wasreatest in certain positions such as finance and human resources.n the engineering-based fields of information and communicationechnologies, the representation among women was low except forhe small group of firms in communications. We expected a largerepresentation of women in the Internet firms and this was the cases 4.9% of the CEOs were female as opposed to 3.1% for all CEOs.n some sectors, including semiconductors, electronic equipment,nd computer systems, there were few women and no female CEOs.hese results suggest that in the firms whose managers come fromelds such as biology in which there are more women graduates,

here is a relatively higher presence of women.17 In semiconduc-ors, there are few women in any capacity – an interesting resultiven that Intel, the largest semiconductor firm in the world, has aemale president.

While varying by industry and region, our data finds that theumber of women affiliated with EGFs at the time of the IPO is

ow. Our data cannot explain the reasons for these outcomes, but,n combination, with other data such as firm survival, it could beossible to measure the value of gender diversity. If underrep-esentation in the STEM disciplinary fields is the first selectionurdle, gaining experience in an existing firm may be second selec-ion hurdle, raising venture capital or other funding might be ahird hurdle, and surviving the startup process until an IPO isikely to be yet another hurdle (Bellucci et al., 2010; Brush et al.,002; Cooper et al., 1994; Greene et al., 2001). The final hurdleould be any biases on the part of investment bankers and pub-

ic investors.18 The lack of initial flow of women seems plausiblenough as women are underrepresented in the STEM fields fromhich the high-technology firms emerge, but our data indicates

hat the pattern of under-representation extends to retail and ser-

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016/

ices where under-enrollment in STEM fields should have littleffect. Our data can only show the individual’s status at the IPOtage, but our work can be extended by examining whether, at the

17 The differences in the proportion of women across industries among TMT, Direc-ors, and personnel as a whole were all significant at the .001 level.18 In this research note, while the data can be derived, we do not analyze theropensity to add women directors or managers in the lead-up to the initial publicffering or what has been termed “window dressing” (Chen et al., 2008). This is an

nteresting question because there may be two opposing forces at work. Namely,he increasing social demand that women be part of a public firm’s managementeam and board of directors and mimetic isomorphic pressure to conform with theominant logic of male managers and executives.

5.7% 10.5% 11.5% 8.7%4.8% 11.0% 13.2% 8.4%5.0% 10.8% 12.6% 8.5%

time of the IPO, the women are more likely to have been with firmfor a comparatively long-time or have been added immediatelyprior to the IPO. Those added immediately prior to the IPO (an actionthat has been termed “window-dressing”) would presumably be,in some measure, added to respond to the predilections of invest-ment bankers and public investors (e.g., Chen et al. (2008). Finally,because we have established the educational background, whenreported, for all the individuals associated with the firm, it is possi-ble to use education as a status measure. In this way, the databaseshould be valuable for scholars seeking to understand gender andentrepreneurship in EGFs.

6. National origin of IPO managers

There is significant academic and policy interest in the role ofimmigrants in U.S. entrepreneurship (Alarcón, 1999; Portes et al.,2002; Saxenian, 2006) and in U.S. science and engineering startups(Kerr 2010, 2013). One drawback is that we have the manage-ment team at the time of the IPO, rather than the team at the timeof the firm’s founding, and thus the founder question cannot beaddressed directly. A number of methodologies have been used forthe identification of the national origin of founders of Americanentrepreneurial firms. The first and most common is direct ques-tioning through a survey. Survey methodology examples includeWadhwa et al. (2007) that used telephone interviews where a firmrepresentative was asked if the firm’s founder was from a foreigncountry, and if so, from what country (see, also, Hart and Acs, 2011for survey research results). This can be further refined by choosingto survey respondents that belong to ethnic associations, such asThe Indus Entrepreneurs or Monte Jade Association (Dossani, 2002;Saxenian, 1994, 2002).

A second methodology is determination of national origin basedon inference from individual names. Name inference can be quitecrude, namely anyone with a particular name is categorized asbeing of the nationality with than name – a technique that is usu-ally applied only to Asians or Russians and makes the dubiousassumption that all individuals with an Asian or Russian name arenot US-born. Saxenian (1994, 2002) is an example of the secondapproach, as she infers national origin by examining CEO data andscoring individuals as immigrants if they had a “foreign” name.A more sophisticated technique is used by Kerr (2010), wherein

icity and entrepreneurship in initial public offerings: illustrationsj.respol.2015.01.003

census data for the individual’s location is used to attribute a prob-ability of being born abroad.19

The approach taken here is to identify the national origin of man-agers on the basis of their post-secondary educations as reported

19 (See, Kerr, 2008 for a description of the algorithm).

Page 8

ARTICLE IN PRESSG ModelRESPOL-3103; No. of Pages 12

8 M. Kenney, D. Patton / Research Policy xxx (2015) xxx–xxx

Table 7IPO managers by highest degree obtained, 1996–2010.

PhD and MD JD MA MBA Other Total

USA 901 655 1,092 2,043 3,466 8,157% of column 71.9% 99.2% 78.7% 94.5% 90.0% 87.6%Canada 37 3 19 27 76 162% of column 3.0% 0.5% 1.4% 1.2% 2.0% 1.7%Other Americas 9 1 3 3 3 19% of column 0.7% 0.2% 0.2% 0.1% 0.1% 0.2%United Kingdom 77 1 45 21 141 285% of column 6.1% 0.2% 3.2% 1.0% 3.7% 3.1%Other Europe 107 62 23 63 255% of column 8.5% 4.5% 1.1% 1.6% 2.7%India 36 86 19 23 164% of column 2.9% 6.2% 0.9% 0.6% 1.8%Taiwan 30 32 6 3 71% of column 2.4% 2.3% 0.3% 0.1% 0.8%Other Asia 41 30 9 44 124% of column 3.3% 2.2% 0.4% 1.1% 1.3%Israel 9 10 7 19 45% of column 0.7% 0.7% 0.3% 0.5% 0.5%Africa and other Mideast 6 4 1 2 11% of column 0.5% 0.3% 0.1% 0.1% 0.1%Total Foreign 352 5 296 118 385 1156% of column 28.1% 0.8% 21.3% 5.5% 10.0% 12.4%Total 1,253 660 1,388 2,161 3,851 9,313

T gree.

p

iaitseathsqvmidpd

atseawi

odraKA

aBt

dt

Table 8 presents the composition of managers with a PhD. In allbut a few cases the identity of the institution that awarded a PhDis reported in the biographies of the managers so the nationality of

he other category includes all managers whose highest degree is a Bachelor’s deharmacy and veterinary science.

n their biographies in the IPO prospectus.20 This methodology hasdvantages and drawbacks. The first advantage is that this reports in a legal document and presumably is accurate. Conversely,his creates a disadvantage that we attribute nationality to post-econdary education, which misses immigrants that spent theirntire post-secondary career in the U.S. The other methodologiesttribute nationality to name or place of birth. Consider Jerry Yang,he founder of Yahoo!, who moved to the U.S. at the age of 10 withis mother. Technically, he is an immigrant, but should he be con-idered an immigrant entrepreneur? There is no best answer to thisuestion. Also, we attribute nationality to the location of the uni-ersity that awarded their Bachelor’s degree.21 For the remaininganagers the earliest degree determines their nationality start-

ng with a master’s degree, followed by one of the doctoral levelegrees. We believe that this is justified in the overwhelming pre-onderance of individuals, but may introduce some error into theatabase.

An advantage of our methodology over others is that we areble to identify Europeans and Canadians; a source of immigrantshat are not easily identified by name recognition programs. This isignificant because the discussion of skilled immigration is almostntirely focused upon South and East Asians. Remarkably littlettention is given to those from other nations; a tendency thate show below leads to serious misunderstanding of the role of

mmigrants in the U.S. high-technology economy.Reporting of education is voluntary and thus not complete. In

ur population of 14,603 managers, only 9313 managers (63.7%)isclosed their education (see Table 7). Overall, using our crite-ia, one in eight managers was educated abroad. The top countries

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016

fter the United States, in order of their importance, are the Unitedingdom (285), India (164), Canada (162), Taiwan (71), France (50),ustralia (49), Germany (49), and Israel (45). Not surprisingly, the

20 Those not reporting education are dropped in this analysis. 221,147 managersnd directors reported their education in their biography, but did not reveal theirachelor’s degree. In these cases an Internet search was made to fill in their educa-ion history.21 1,147 managers and directors reported their education in their biography, butid not reveal their bachelor’s degree. In these cases an Internet search was madeo fill in their education history.

It also includes 35 US managers with professional degrees in fields like dentistry,

top three are English-speaking nations, and the remainder havebeen closely allied with the United States for decades. Most strik-ing is the large number of UK-educated managers. Similarly, thelarge number of Canadians may not be surprising, but has beenabsent in most discussions. In total, Europeans make up almost 6%of all managers and are more numerous than those from any othercontinent. Simply put, this result questions the intense concentra-tion on East and South Asians in the discussions of immigration andhigh-technology business growth. Our data does differ from some-what from other studies. For example, Hart and Acs (2011) foundthat among high-impact, high-tech firm founders, Indians (15.9%)were the largest group followed by UK (10%), and Canada and Japantied for third with 6 percent.22

In addition to using the degree to identify the national origin ofthese managers, we also recorded all reported degrees. This per-mitted classification of the managers by their highest degree.23 Forthose reporting education, 13.5%, 1253 out of 9313, held a PhD orMD. Moreover, managers with at least a Bachelor’s degree froma foreign university comprised 28.1% percent of the total num-ber of PhDs and MDs (see Table 7). This is well above the 12.4%identified as foreign nationals in the entire managerial population.The standard explanation is that U.S. universities with their highquality postgraduate programs are a powerful attractant for for-eign students. This is undoubtedly the case, but there are significantdifferences in the population.

icity and entrepreneurship in initial public offerings: illustrations/j.respol.2015.01.003

22 DesRoches et al., 2007 in a similar study confirmed that 16 percent of theirsample were immigrants.

23 The individuals in this population were quite accomplished, with many individ-uals having multiple higher degrees such as both an MD and a PhD or a PhD anda JD. So as to not double count individuals, we followed the following scheme forclassifying individuals by highest degree. A PhD was higher than MD and MD washigher than a JD. At the masters level, an MA or MS was considered to be higher thanan MBA. This relatively arbitrary classification could be changed depending upon aresearcher’s goals. It is likely that in terms of chronological career order, the PhD andMD were roughly simultaneous, with the MBA coming the last among those indi-viduals with advanced engineering and scientific degrees. Unfortunately, the yearof the degree was not reported in most cases.

Page 9

ARTICLE ING ModelRESPOL-3103; No. of Pages 12

M. Kenney, D. Patton / Research

Table 8TMT PhDs by industry, national origin, and percent of total PhDs granted by U.S.institutions, 1996–2010.

Bio-medical ICT and Internet All other Total

USA 329 176 230 73598.5% 98.9% 98.3% 98.5%

Canada 13 10 11 3446.2% 50.0% 45.5% 47.1%

Other America 4 1 1 650.0% 100% 0% 50.0%

UK 41 7 24 727.3% 0% 8.3% 6.9%

Other Europe 36 26 27 8911.1% 19.2% 7.4% 12.4%

India 9 23 4 3677.8% 87.0% 50.0% 80.6%

Taiwan 6 22 2 30100% 95.5% 100% 96.7%

Other Asia 12 15 11 3833.3% 80.0% 36.4% 52.6%

Israel 5 2 2 940.0% 50.0% 0% 33.3%

Africa and other Mideast 1 4 1 6100% 50.0% 0% 50.0%

All Foreign 127 110 83 32027.6% 60.9% 20.5% 37.2%

Total 456 286 313 1,055

Ti

dotd

nccdaPtdufm2bitiic

asogBth

cd

cme

he entry below each cell is the percentage of PhDs in that cell granted by U.S.nstitutions.

egree granting institution can be identified. In Table 8 the nationalrigin of managers holding a PhD is compared to the industrial sec-or of the newly public firm. The second entry is the percentage ofoctorates that were awarded by an American university.

Examination of the university granting the Ph.D. for the variousationalities uncovers some remarkable differences. The overallonclusion is that the European and Canadian managers haveome to the U.S. with PhDs, while the Asians had only Bachelor‘’egrees.24 To illustrate the differences, only 5 out of 72 UK man-gers with a PhD received their degree in the U.S. For all EuropeanhDs, only 12.4 percent had received their PhD in the U.S. In con-rast, 29 out of 30 Taiwanese managers with a PhD received theiregree in the U.S. There are also industrial differences, in our pop-lation U.S. universities award the highest proportion of PhDs to

oreigners in the ICT industries. Even as the bio-medical sector reliesore on PhDs than any other industry by far, the U.S awards just

7.6% of these degrees to foreigners. Taken as a whole the differenceetween Europe and Asia is notable and suggests that the European

mmigrants in these types of industries may be more likely to comeo the U.S. with significantly greater human capital than the Asianmmigrants.25 Put differently, in this population at the level of thendividual, Europe transfers greater amounts of invested humanapital to the U.S. than do the East and South Asian nations.

Immigration studies have considered with migratory pathwaysnd it is possible to partially trace them. In addition to univer-ity, often discipline is also reported, so it is possible to traceur respondents to particular degrees and departments. Fig. 3raphically represents of the career paths of those receiving their

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016/

achelor’s degree in Taiwan. There are some noteworthy observa-ions regarding these 71 immigrants. First, nearly half of them, 44%,eld PhDs and served in 43 unique IPO firms, of which 37% were

24 Unfortunately, our data cannot tell us if those immigrants with foreign PhDsame to the U.S. for post-doctoral training (see Miller and Feldman 2014 for furtheriscussion of postdoctoral degrees as an under-paid labor force).25 Not surprisingly, Hunt (2011) finds that the gross contribution by immigrantsan be ranked from highest to lowest in the following order: postdoctoral fellows andedical residents, graduate students, temporary work visa holders, college students

tc.

PRESSPolicy xxx (2015) xxx–xxx 9

in semiconductors or semiconductor design software – a remark-able concentration. The two leading Taiwanese source universitieswere National Taiwan University (NTU) in Taipei that sent 26 per-sons and National Chiao Tung University in Hsinchu that sent 19; intotal they were responsible for 61 percent of all TMT members. Allof them received advanced degrees in the U.S. Of the 26 NationalTaiwan University graduates, 69 percent received U.S. PhDs. Therewas concentration in electrical engineering graduates that had adistinct preference for UC Berkeley electrical engineering and com-puter science (4), though, in total, there were an almost equal totalnumber with Stanford PhDs (3). NTU graduates were remarkablyconcentrated in the information and communication technologyfields and were affiliated with 42 percent of all the firms and 75% ofall semiconductor-related firms. The second important source uni-versity was National Chiao Tung in Hsinchu. Interestingly, while itselectrical engineering and computer science departments were keysource departments (63%), only 26 percent of its graduates receivedU.S. PhDs and the highest concentration was in pharmaceuticals.

The concentration of Taiwanese in semiconductors is not sur-prising, as Saxenian (1994, 2002) and Dossani (2002) found similarresults in their studies of Taiwanese entrepreneurs in Silicon Val-ley. The UC Berkeley Department of Electrical Engineering andComputer Science has a long history of admitting Taiwanese toits graduate programs, and, particularly, in semiconductor engi-neering. There appears to be a path from electrical engineering atNational Taiwan University.26 Because of the relatively small N,we cannot definitively answer questions about immigrants, butit is possible to illustrate pathways and the data reinforces theobservation about the importance of graduate degrees at researchuniversities as recruitment and vetting institutions for talentedAsians. These results suggest that these highly skilled immigrantsmay be reproducing recruitment networks similar to those docu-mented among low-skilled immigrants (e.g., Bashi 2007; Harvey2008; Sanders et al., 2002).

Immigration to the U.S. by highly trained individuals and par-ticularly the role of U.S. universities in the attraction of skilledindividuals has received significant interest (e.g., Stephan 2010;Stephan and Levin 2001). In recent research, Franzoni et al. (2014)find that immigrant scientists out-perform indigenous researchersin terms of impact factor. Using the 2003 National Survey ofCollege Graduates, Hunt (2011) found that immigrants who firstentered on a student/trainee visa or a temporary work visa out-performed natives with similar education in terms of wages,patenting, commercializing licensing patents, publishing and start-ing firms. While our population is far smaller than that of theseprevious researchers, the database contains the names of indi-viduals and thus allows far more fine-grained tracking. Further,because this is an IPO, it is possible to determine whether firmswith immigrant managers immigrant managers or directorshavesuperior financial outcomes.

7. Discussion

The first contribution of the database is apparently trivial, but, inreality, quite important. Namely, we have produced a comprehen-sive, curated list of all emerging growth firms that have listed on US

icity and entrepreneurship in initial public offerings: illustrationsj.respol.2015.01.003

markets during the last two decades. Prior to building this database,scholars have had to build their own databases, but have not sharedthem making reproducibility difficult.27 This also addresses one of

26 In particular, see Fig. 4.2 in Kenney et al., 2014. Also, two UC Berkeley professors.Chenming Calvin Hu and Ming C. Wu, were National Taiwan University graduatesthat received their PhDs at UC Berkeley and then became professors.

27 The importance of sharing data was, perhaps, best illustrated in the recent casewhere Kenneth Rogoff and Carmen Reinhart shared their data with Thomas Herndon

Page 10

ARTICLE IN PRESSG ModelRESPOL-3103; No. of Pages 12

10 M. Kenney, D. Patton / Research Policy xxx (2015) xxx–xxx

26 Na�ona l Taiwan University10 Electr ical engineer ing 1 Compu ter Science

71 individuals

18 PhDs 4 UCB 3 Stanford 1 MI T 4 UCB EE&CS 1 Stanford EE&CS

19 Na�onal Chiao Tun g University 9 Electr ical engineering 3 Compu ter Science

5 PhDs1 UCB EE&CS 1 Ha rvard

30 PhDs

43 firms16 semi con-relate d 7 pha rmaceu� cals 6 net work equip.

18 firms12 semi con-relate d 4 so�ware 1 pharma 1 net work equip.

12 firms 4 pharma 3 internet 2 semi con-relate d

aths w

tmairo

rdswiitsttasut

iUBfiobfieeiHfie

ei

aod

2 net work equip.

Fig. 3. Taiwanese-educated TMT members academic p

he drawbacks to relying upon commercial databases, namely thereay be copyright issues prohibiting sharing.28 Normally, articles

re published without providing an appendix detailing the firmsncluded in the study. This stymies other researchers intent uponeproducing or improving upon previous studies. Of course, the lackf availability protects authors from critique.

While the IPO database itself offers ample opportunity for directesearch, its greatest promise is through linkage with yet otheratabases to test new propositions in geography, entrepreneur-hip and innovation studies, sociology, and finance. For example,hen linked to the patent database described by Li et al. (2014),

t could provide a comprehensive history of the IPO firm’s patent-ng behavior and the patenting by the TMT members. The data onhe firm’s age, location, whether VC-financed or not, and SIC codehould simplify matching with similar firms that did not under-ake an IPO. This could be particularly useful for those interested inhe fate of roughly comparable VC-financed firms. Another possiblevenue for further research is linking the IPO database to univer-ity technology licensing, SBIR grants, and even R&D expendituressing the Starmetrics database described in Lane et al. (2015) inhis Issue.

IPOs, and the new industries catalyzed in the firms undertak-ng them, are one of the most salient expressions of how the.S. innovation economy has grown during the last five decades.ecause IPOs are so prevalent for new rapidly growing firms, theserms offer an opportunity for substantial insight into the impactf entrepreneurship on the U.S. economy. For example, it is possi-le to measure the new employment and revenues created by theserms (Kenney et al., 2012). Using the NETS database, it is possible toxplore geographical changes in firm employment. This is of inter-st, because the concentration of the headquarters of these firmsn a few locations suggests a highly uneven geographical impact.

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016

owever, it is possible that the employment impact diffuses asrms grow. For regional scientists, the patterns of this regionalxpansion might be particularly interesting, as is the question of the

t al. (2014) who were then unable to reproduce the original results, thereby callingnto question their conclusions.28 There are other important initiatives underway that should interest schol-rs. For example, the Association for Corporate Growth http://growtheconomy.rg/iegc.lasso provides limited academic access to the PitchBook and NETSatabases. This can provide access to deep and quite expensive data.

ith PhDs, key universities, and industries, 1996–2010.

ways this expansion differs by industry. Despite its policy impor-tance, there has been little study of job growth by EGFs (Kenneyet al., 2012).

Another field of research that this database can facilitate is inter-national comparisons o (Doidge et al., 2013; Loughran et al., 1994;Lewellyn and Bao, 2014). This is especially timely as a number ofnations have established stock markets specifically for encouraginginvestment exits for domestic and international technology firms.Our database simplifies the establishment of comparison popula-tions and furnishes a template for comparison.

8. Conclusion

This article describes a database available to the greater aca-demic community, illustrates a few of its potential uses, andprovides a few descriptive examples of database applications.The results from our examples suggest important avenues forfuture research. For geographers, it can contribute to better under-standings of the spatial aspects of successful entrepreneurship.In particular, far more can be done concerning the location ofentrepreneurial support networks constituents including, not onlythe law firms, local venture capitalists, independent directors, andinvestment bankers, but also local educational institutions. Forthose interested in diversity in entrepreneurship or corporate man-agement team formation, the ethnicity and gender data should beof use. The brief discussion of the Taiwanese managers suggeststhat far more research can be done on ethnic networks and immi-gration paths. Other researchers should be able to use this data, inconjunction with data from other sources.

In the interest of brevity, this paper did not analyze the data onthe independent directors that are included in the database. Thesame gender, founder, and educational data has been collected forall the directors. One obvious use of the data is to examine whetherco-ethnicity ties exist between the TMT and BoD members and theeffect on IPO outcomes. The contribution by Bengtsson and Hsu(2014) found that shared ethnicity increases “the likelihood thata VC firm invests in a startup, strengthens the degree of the ven-

icity and entrepreneurship in initial public offerings: illustrations/j.respol.2015.01.003

ture capital firm’s involvement, increases the size and scope of theinvestment, and makes the financial contract more entrepreneur-friendly,” but that the firm had inferior performance. With our data,it is possible to test these relationships within the TMT and between

Page 11

ING ModelR

earch

te

Rnmwadnvscftopmiasbss

hbFsvtdooot

R

A

A

A

B

B

B

B

B

B

B

B

ti

ARTICLEESPOL-3103; No. of Pages 12

M. Kenney, D. Patton / Res

he TMT and the BoD, particularly because there is no need to inferthnicity linkages.29

There are some more general issues that should be raised here.eplicability has been an important component of progress in theatural sciences. The physical and biological sciences have alreadyoved a long distance toward the norms of an open science inhich data is provided to invite and ease replication and encour-

ge cumulative advance. This is more difficult in the social sciencesue to privacy issues, the use of commercial databases that doot permit data-sharing, and reluctance among researchers to pro-ide their data to others (for a recent examination of this issue,ee Andreoli-Versbach and Mueller-Langer 2014). And yet, in theurrent environment, data-hoarding is becoming increasingly inde-ensible. To illustrate the depth of this problem, even relativelyrivial requirements such as the publication of firm names or therganizations used in quantitative studies is uncommon. So it isossible that one researcher studying the “biotechnology” industryay use different firms than another researcher studying the same

ndustry. Effectively, the results are not comparable. This lack ofdherence to the norms of science must be slowing the progress ofocial science. Lack of comparability and reproducibility may haveeen justified or understandable in the days prior to standardizedoftware and the ability to costlessly send electronic files, but todayeem antiquated.

This article described an existing database, portions of whichave already been released for general use. Other portions areeing completed and will be released over the next two years.inally, a new extension is to use The Wayback Machine to recon-truct management, firm description, and address changes for 500enture capital-backed firms that during the last decade under-ook an IPO and 500 similar venture capital-backed firms thatid not. This openly available database will provide greater detailn the evolution of venture capital-backed firms. The new eraf open-sourced software and increased information availabilityffers scholars opportunities to strengthen the scientific basis ofhe social sciences.

eferences

larcón, R., 1999. Recruitment processes among foreign-born engineers andscientists in Silicon Valley. Am. Behav. Sci. 42, 1381–1389.

mason, A.C., Shrader, R.C., Tompson, G.H., 2006. Newness and novelty: relatingtop management team composition to new venture performance. J. Bus.Venturing 21 (1), 125–148.

ndreoli-Versbach, P., Mueller-Langer, F., 2014. Open access to data: an idealprofessed but not practiced. Res. Policy 43 (9), 1621–1633.

ashi, V.F., 2007. Survival of the Knitted: Immigrant Social Networks in a StratifiedWorld. Stanford University Press, Stanford, CA.

eatty, R.P., Zajac, E.J., 1994. Managerial incentives, monitoring, and risk bearing: astudy of executive compensation, ownership, and board structure in initialpublic offerings. Admin. Sci. Q. 39 (2), 313–335.

eckman, C.M., Burton, M.D., 2008. Founding the future: path dependence in theevolution of top management teams from founding to IPO. Org. Sci. 19 (1),3–24.

eckman, C.M., Burton, M.D., O’Reilly, C., 2007. Early teams: the impact of teamdemography on VC financing and going public. J. Bus. Venturing 22 (2),147–173.

ellucci, A., Borisov, A., Zazzaro, A., 2010. Does gender matter in bank–firmrelationships? Evidence from small business lending. J. Bank. Finance 34 (12),2968–2984.

engtsson, O., Hsu, D.H., 2014. Ethnic matching in the U. S. venture capital market.J. Bus. Venturing.

igelow, L., Lundmark, L., Parks, J.M., Wuebker, R., 2012. Skirting the issues:

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016/

experimental evidence of gender bias in IPO prospectus evaluations. J. Manage.40 (6), 1732–1759.

oeker, W., Karichalil, R., 2002. Entrepreneurial transitions: factors influencingfounder departure. Acad. Manage. J. 45 (4f), 818–826.

29 Bengtsson and Hsu (2014) infer ethnicity on the basis of names. Their assump-ion is that, for example, a Chinese–American or Jew would be more likely to investn a Chinese from China or Jew than to an individual from another race or religion.

PRESSPolicy xxx (2015) xxx–xxx 11

Brush, C.G., Carter, N.M., Greene, P.G., Hart, M.M., Gatewood, E., 2002. The role ofsocial capital and gender in linking financial suppliers and entrepreneurialfirms: a framework for future research. Venture Capital: Int. J. Entrepreneur.Finance 4 (4), 305–323.

Bruton, G.D., Filatotchev, I., Chahine, S., Wright, M., 2010. Governance, ownershipstructure, and performance of IPO firms: the impact of different types ofprivate equity investors and institutional environments. Strateg. Manage. J. 31(5), 491–509.

Buhr, H., Smith, J.O.-, 2010. Networks as institutional support: law firm andventure capitalist relations and regional diversity in high-technology IPOs. Res.Sociol. Work 21, 95–126.

Cannella, A.A., Park, J.H., Lee, H.U., 2008. Top management team functionalbackground diversity and firm performance: examining the roles of teammember colocation and environmental uncertainty. Acad. Manage. J. 51 (4),768–784.

Certo, S.T., 2003. Influencing initial public offering investors with prestige:signaling with board structures. Acad. Manage. Rev. 28 (3), 432–446.

Certo, S.T., Covin, J.G., Daily, C.M., Dalton, D.R., 2001. Wealth and the effects offounder management among IPO–stage new ventures. Strateg. Manage. J. 7,641–658.

Chen, H., Gompers, P., Kovner, A., Lerner, J., 2010. Buy local? The geography ofventure capital. J. Urban Econ. 67 (1), 90–102.

Chen, G., Hambrick, D.C., Pollock, T.G., 2008. Puttin’on the Ritz: pre-IPO enlistmentof prestigious affiliates as deadline-induced remediation. Acad. Manage. J. 51(5), 954–975.

Cohen, B.D., Dean, T.J., 2005. Information asymmetry and investor valuation ofIPOs: top management team legitimacy as a capital market signal. Strateg.Manage. J. 26 (7), 683–690.

Cohen, S.S., Fields, G., 2000. Social capital and capital gains: an examination ofsocial capital in Silicon Valley. In: Kenney, M. (Ed.), Understanding SiliconValley Stanford. Stanford University Press, pp. 190–217.

Cooper, A.C., Gimeno-Gascon, F.J., Woo, C.Y., 1994. Initial human and financialcapital as predictors of new venture performance. J. Bus. Venturing 9 (5),371–395.

DesRoches, D., et al., 2007. Kauffman Firm Survey: Final Report. MathematicaResearch, Princeton.

Doidge, C., Karolyi, G.A., Stulz, R.M., 2013. The US left behind? Financialglobalization and the rise of IPOs outside the US. J. Financ. Econ. 110 (3),546–573.

Dossani, R., 2002. Chinese and Indian Engineers and Their Networks in SiliconValley. Asia/Pacific Research Center, Stanford, CA.

Fenwick., West., (2014). Gender Diversity in Silicon Valley: A Comparison of SiliconValley Public Companies and Large Public Companies.

Filatotchev, I., Wright, M., Arberk, M., 2006. Venture capitalists, syndication andgovernance in initial public offerings. Small Bus. Econ. 26 (4), 337–350.

Fischer, H.M., Pollock, T.G., 2004. Effects of social capital and power on survivingtransformational change: the case of initial public offerings. Acad. Manage. J.47 (4), 463–481.

Fleming, L., Frenken, K., 2007. The evolution of inventor networks in the SiliconValley and Boston regions. Adv. Complex Syst. 10 (01), 53–71.

Franzoni, C., Scellato, G., Stephan, P., 2014. The mover’s advantage: the superiorperformance of migrant scientists. Econ. Lett. 122 (1), 89–93.

Gao, X., Ritter, J.R., Zhu, Z., 2013. Where have all the IPOs gone? J. Financ. Quant.Anal. 48 (6), 1663–1692.

Greene, P., Brush, C., Hart, M., Saparito, P., 2001. Patterns of venture capital funding:Is gender a factor? Venture Capital: Int. J. Entrepreneur. Finance 3 (1), 63–83.

Gulati, R., Higgins, M.C., 2003. Which ties matter when? The contingent effects ofinterorganizational partnerships on IPO success. Strateg. Manage. J. 24 (2),127–144.

Hart, D.M., Acs, Z.J., 2011. High-tech immigrant entrepreneurship in the UnitedStates. Econ. Dev. Q. 25 (2), 116–129.

Harvey, W.S., 2008. Strong or weak ties? British and Indian expatriate scientistsfinding jobs in Boston. Global Networks 8 (4), 453–473.

Herndon, T., Ash, M., Pollin, R., 2014. Does high public debt consistently stifleeconomic growth? A critique of Reinhart and Rogoff. Cambr. J. Econ. 38 (2),257–279.

Higgins, M.C., Gulati, R., 2006. Stacking the deck: the effects of top managementbackgrounds on investor decisions. Strateg. Manage. J. 27 (1), 1–25.

Hochberg, Y.V., Ljungqvist, A., Lu, Y., 2007. Whom you know matters: venturecapital networks and investment performance. J. Finance 62 (1), 251–301.

Huddleston, T. Jr., (2014). Obama criticizes Silicon Valley’s gender gap. FortuneJune 10, 2014. http://fortune.com/2014/06/10/obama-tumblr-gender-gap

Hunt, J., 2011. Which immigrants are most innovative and entrepreneurial?Distinctions by entry visa. J. Labor Econ. 29 (3), 417–457.

Jain, B.A., Tabak, F., 2008. Factors influencing the choice between founder versusnon-founder CEOs for IPO firms. J. Bus. Venturing 23 (1), 21–45.

Kenney, M., 2011. How Venture capital became a component of the U.S.NSI. Ind.Corp. Change 20 (6), 1677–1723.

Kenney, M., Mowery, D., Patton, D., 2014. Electrical engineering and computerscience at UC Berkeley and Silicon Valley: modes of regional engagement. In:Kenney, M., Mowery, D. (Eds.), Public Universities and Regional Development:

icity and entrepreneurship in initial public offerings: illustrationsj.respol.2015.01.003

Insights from the University of California. Stanford University Press, Stanford.Kenney, M., Patton, D., 2005. Entrepreneurial geographies: support networks in

three high-technology industries. Econ. Geogr. 81 (2), 201–228.

Page 12

ING ModelR

1 earch

K

K

K

K

K

KL

L

L

L

L

M

M

M

M

N

P

in IPOs: an examination of gender diversity in top management teams. Group

ARTICLEESPOL-3103; No. of Pages 12

2 M. Kenney, D. Patton / Res

enney, M., Patton, D., Ritter, J., 2012. Post-IPO Employment and Revenue Growthfor U.S. IPOs, June 1996–2010. Ewing Marion Kauffman Foundation, KansasCity, Missouri http://www.kauffman.org/uploadedfiles/post ipo report.pdf

err, W.R., (2008). The Ethnic Composition of U.S. Inventors. Harvard BusinessSchool Working Paper, No. 08–006, May 2007. (Permanent working paperdescribing ethnic-name patenting data, revised December 2008.).

err, W.R., 2010. Breakthrough inventions and migrating clusters of innovation. J.Urban Econ. 67 (1), 46–60.

err, W.R., 2013. US High-skilled Immigration, Innovation, and Entrepreneurship:Empirical Approaches and Evidence (No. w19377). National Bureau ofEconomic Research.

im, J.W., Higgins, M.C., 2007. Where do alliances come from? The effects of upperechelons on alliance formation. Res. Policy 36 (4), 499–514.

ristof, N.D., (2013). Twitter, Women and Power. New York Times (23. 10. 13.).ane, J., Owen-Smith, J., Rosen, R., Weinberg, B., 2015. New Linked Data on Science

Investments, the Scientific Workforce and the Economic and Scientific Resultsof Science. Res. Policy.

ewellyn, K.B., Bao, S.R., 2014. A cross-national investigation of IPO activity: therole of formal institutions and national culture. Int. Bus. Rev. 23 (6), 1167–1178.

i, G.C., Lai, R., D’Amour, A., Doolin, D.M., Sun, Y., Torvik, V.I., Fleming, L., 2014.Disambiguation and co-authorship networks of the US patent inventordatabase, 1975–2010. Res. Policy 43 (6), 941–955.

iu, K., Li, J., Hesterly, W., Cannella Jr., T., 2012. Top management team tenure andtechnological inventions at post-IPO biotechnology firms. J. Bus. Res. 65 (9),1349–1356.

oughran, Ritter, J., Rydqvist, K., 1994. Initial public offerings: internationalinsights. Pac.-Basin Finance J. 2, 165–199.

cPherson, M., -Lovin, L.S., Cook, J.M., 2001. Birds of a feather: homophily in socialnetworks. Annu. Rev. Sociol. 27, 415–444.

iller, J.M., Feldman, M.P., 2014. The sorcerer’s postdoc apprentice: uncertainfunding and contingent highly skilled labour. Cambr. J. Reg. Econ. Soc. 7 (2),289–305.

ohan, N.J., Chen, C.R., 2004. Are IPOs priced differently based upon gender? J.Behav. Finance 5 (1), 57–65.

owery, D.C., 1992. The US national innovation system: origins and prospects forchange. Res. Policy 21 (2), 125–144.

Please cite this article in press as: Kenney, M., Patton, D., Gender, ethnfrom an open database*. Res. Policy (2015), http://dx.doi.org/10.1016

elson, T., 2003. The persistence of founder influence: management, ownership,and performance effects at initial public offering. Strateg. Manage. J. 24 (8),707–724.

atton, D., Kenney, M., 2005. The spatial configuration of the entrepreneurialsupport network for the semiconductor industry. R&D Manage. 35 (1), 1–16.

PRESSPolicy xxx (2015) xxx–xxx

Pollock, T.G., Fund, B.R., Baker, T., 2009. Dance with the one that brought you?Venture capital firms and the retention of founder–CEOs. Strateg. Entrepr. J. 3(3), 199–217.

Pollock, T.G., Gulati, R., 2007. Standing out from the crowd: thevisibility-enhancing effects of IPO-related signals on alliance formation byentrepreneurial firms. Strateg. Org. 5 (4), 339–372.

Portes, A., Guarnizo, L.E., Haller, W.J., 2002. Transnational entrepreneurs: analternative form of immigrant economic adaptation. Am. Sociol. Rev. 68,278–298.

Ritter, J., Welch, I., 2002. A review of IPO activity, pricing and allocations. J. Finance52, 1795–1828.

Robb, A.M., Watson, J., 2012. Gender differences in firm performance: evidencefrom new ventures in the United States. J. Bus. Venturing 27 (5), 544–558.

Ruef, M., Aldrich, H.E., Carter, N.M., 2003. The structure of founding teams:homophily, strong ties, and isolation among US entrepreneurs. Am. Sociol. Rev.2, 195–222.

Sanders, J., Nee, V., Sernau, S., 2002. Asian immigrants’ reliance on social ties in amultiethnic labor market. Social Forces 81 (1), 281–314.

Saxenian, A., 2006. The New Argonauts: Regional Advantage in a Global Economy.Harvard University Press.

Saxenian, R., 1994. Regional advantage: culture and competition in Silicon Valleyand Route, 128. Harvard University Press.

Saxenian, A., 2002. Silicon Valley’s new immigrant high-growth entrepreneurs.Econ. Dev. Q. 16, 20–31.

Sorenson, O., Stuart, T.E., 2001. Syndication networks and the spatial distributionof venture capital investments. Am. J. Sociol. 106 (6), 1546–1588.

Spender, J.C., 1989. Industry Recipes. Basil Blackwell, Oxford.Stephan, P.E., 2010. The I’s have it: immigration and innovation, the perspective

from academe. In Innovation Policy and the Economy, 10. University ofChicago Press, pp. 83–127.

Stephan, P.E., Levin, S.G., 2001. Exceptional contributions to US science by theforeign-born and foreign-educated. Popul. Res. Policy Rev. 20 (1–2), 59–79.

Wadhwa, V., Saxenian, A., Rissing, B., Gereffi, G., 2007. America’s New ImmigrantEntrepreneurs. Kauffman Foundation Report.

Welbourne, T.M., Cycyota, C.S., Ferrante, C.J., 2007. Wall street reaction to women

icity and entrepreneurship in initial public offerings: illustrations/j.respol.2015.01.003

Org. Manage. 32 (5), 524–547.Zimmerman, M.A., 2008. The influence of top management team heterogeneity on

the capital raised through an initial public offering. Entrepr. Theory Pract. 32(3), 391–414.