This article was downloaded by: [University of Saskatchewan Library]On: 06 October 2012, At: 08:17Publisher: RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registeredoffice: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Economics of Innovation and NewTechnologyPublication details, including instructions for authors andsubscription information:http://www.tandfonline.com/loi/gein20

Universities’ trademark patterns andpossible determinantsMariagrazia Squicciarini a , Valentine Millot a b & Hélène Dernis aa OECD Directorate for Science, Technology and Industry, 2, rueAndré Pascal 75775, Paris Cedex 16, Franceb BETA, Université de Strasbourg, France

Version of record first published: 26 Apr 2012.

To cite this article: Mariagrazia Squicciarini, Valentine Millot & Hélène Dernis (2012): Universities’trademark patterns and possible determinants, Economics of Innovation and New Technology,21:5-6, 473-504

To link to this article: http://dx.doi.org/10.1080/10438599.2012.656526

PLEASE SCROLL DOWN FOR ARTICLE

Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions

This article may be used for research, teaching, and private study purposes. Anysubstantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,systematic supply, or distribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representationthat the contents will be complete or accurate or up to date. The accuracy of anyinstructions, formulae, and drug doses should be independently verified with primarysources. The publisher shall not be liable for any loss, actions, claims, proceedings,demand, or costs or damages whatsoever or howsoever caused arising directly orindirectly in connection with or arising out of the use of this material.

Economics of Innovation and New TechnologyVol. 21, Nos. 5–6, September 2012, 473–504

Universities’ trademark patterns and possible determinants†

Mariagrazia Squicciarinia*, Valentine Millota,b and Hélène Dernisa

aOECD Directorate for Science, Technology and Industry, 2, rue André Pascal 75775,Paris Cedex 16, France; bBETA, Université de Strasbourg, France

(Received 8 September 2011; final version received 23 December 2011 )

Academic institutions may register trademarks (TMs) to protect and exploit keyintangible assets (e.g. reputation), to better market current and prospective initiatives,and to better appropriate the output of innovative activities. TM registration by aca-demic institutions – so far overlooked by the literature addressing the third function ofuniversities – is investigated here. The analysis relies on a novel panel data set con-taining information about US universities, their main characteristics, and their TM andpatent activities over the period 1997–2007. Our contribution is exploratory in natureand descriptive in aim and uncovers a number of relationships worth being investigatedfurther, among them are the persistence of Intellectual Property Rights activities by aca-demic institutions and the existence of positive and significant relationships betweenTM registration and universities’ characteristics such as being private institutions, thenumber of students enrolled and the share of graduate students, the share of federal fundsreceived, and the presence of medical schools.

Keywords: academic entrepreneurship; universities; trademarks; patents

JEL Classification: I2; L26; O34

1. IntroductionThe US Patent and Trademark Law Amendment Act of 1980, commonly known as theBayh–Dole Act, constitutes the legal framework enabling the transfer of federally fundedinventions generated by universities, small businesses, and non-profit organizations. Overthe last three decades, many scholars have engaged in assessing the effect of such animportant piece of legislation over universities’ ‘third function’, that is, entrepreneurialand economic development activities (see Etzkowitz 1998, 2003; Kutinlahti 2005, fora survey of the literature).1 University patenting and licensing, academic patent quality,university–industry technology transfer, and university entrepreneurship have been investi-gated extensively (e.g. Mowery et al. 2001, 2004; Thursby and Thursby 2003; Shane 2004;

*Corresponding author. Email: mariagrazia.squicciarini@oecd.org†The opinions expressed in this paper are those of the authors and do not necessarily reflect theofficial views of the Organisation for Economic Co-operation and Development (OECD) or of thegovernments of its member countries.

ISSN 1043-8599 print/ISSN 1476-8364 online© 2012 Taylor & Francishttp://dx.doi.org/10.1080/10438599.2012.656526http://www.tandfonline.com

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

474 M. Squicciarini et al.

Åstebro and Bazzazian 2010), whereas other aspects have been overlooked. To the best ofour knowledge, no study in fact has investigated universities’ trademarking behaviors anddeterminants. Trademarks (TMs) nevertheless represent one of the knowledge appropria-tion and commercialization devices that universities may wish to exploit. Universities actingentrepreneurially and behaving rationally may in fact rely on the full array of IntellectualProperty Rights (IPRs) conferred by the law and hence use TMs as well as patents.

TM data may be very useful to inform the debate on academic activities andentrepreneurship and their determinants, as they are linked to innovative and marketingactivities (Greenhalgh and Rogers 2007), and proxy non-technological innovations andinnovation in services (Schmoch 2003; Mendonça, Santos Pereira, and Godinho 2004;Hipp and Grupp 2005). Universities’ TMs (and patents) are thus at the center of the presentwork aiming to understand whether and to what extent universities use TMs, how TM userelates to academic patenting, and which university-specific variables contribute to explainIPR use by academia.

The analysis relies on a novel panel data set containing information about universitieslocated in the USA. It covers the period 1997–2007 and combines TM data obtained fromthe United States Patent and Trademark Office (USPTO); patent data extracted from theEuropean Patent Office (EPO)-managed patent statistical (PATSTAT) database; and dataof universities’ characteristics published by the US ‘Center for Measuring University Per-formance’ (henceforth MUP).2 The latter are used as control variables to investigate theway university-specific characteristics may influence academic patenting and trademarkingbehaviors, for example, the amount of funds devoted to research and the number of studentsat undergraduate and postgraduate levels.

Our contribution is exploratory in nature and primarily descriptive in aim. Data avail-ability constraints and the absence of a broad and encompassing theoretical backgrounddrive the simple first-hand analysis that we propose. The estimates given rely on countdata models and address relationships rather than causal links. Selection and endogeneityconcerns are at present overlooked, as our main goal is to gain some broad knowledgeabout the way academic institutions use IPR and TMs in particular. Despite its simplicity,the analysis proposed contributes to a better understanding of universities’ strategic behav-iors related to the appropriation of the knowledge they generate and to the way they dealwith their intangible assets – especially their ‘innovative property’ (Corrado, Hulten, andSichel 2009).

Our results suggest the existence of a significant and positive relationship between TMactivities engaged in by academic institutions and universities’ characteristics such as thenumber of students enrolled, the share of graduate students, the presence of medical schools,the share of federal funds received, and being a private institution. In addition, TM behaviorsappear to be persistent, as having registered TMs in the past is positively related to applyingfor TMs over the period considered. Significant and negative relationships instead emergebetween TM behaviors and the number of universities located in the same state. Althoughobserved only in cross-sectional estimates relying on pooled data, the relationship thatemerges between TMs and patents is also negative. This might signal possible substitutioneffects, whereby universities may focus on patents or TMs in a certain year, but not onboth, or mirror the different stages of the product life cycle over which patents and TMsare used. This issue, as others our greenfield exploration leaves unaddressed, is left forfuture research.

The remainder of this paper first surveys the literature concerned with TM use by firms.It discusses the possible reasons for universities applying for TMs and the relationships thatmay exist between patent and trademark activities. It then presents the first-hand data set

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 475

built for the analysis and presents some descriptive statistics about US universities’ patentingand trademarking behaviors. These descriptive statistics are followed by an outline of theanalytical strategy followed and of the variables used for the purpose. The results obtainedfor the possible drivers of academic trademarking activities and the way patents and TMsare used by universities are hence proposed. A short discussion on key findings and someissues for future research conclude the paper.

2. Why relying on TMs?TMs are distinctive signs – that is, names, words, symbols, or images or a combinationof these elements – used to identify goods or services produced or provided by a specificperson, enterprise, or institution. TMs should help customers identify and purchase productsor services that meet their needs and expectations in terms of, for example, nature, quality,and price. Several categories of marks exist and all grant the owners the exclusive rightto use such signs to identify the goods or services produced or to authorize another partyto use them in return for payment. TM protection is enforced by courts, and the length ofprotection may vary, although TMs can typically be renewed indefinitely upon payment ofadditional fees.

The only requirement for TMs to be registered is that the sign(s) or the combination ofsigns they are made of should be distinctive. Contrary to patents, TMs do not need that theproduct, service, or entity they aim to identify be novel, useful, or the result of an inventivestep. Registered TMs are often part of brands3 and branding strategies, since brands can belegally protected only in so far as (some of) their parts are protected by IPRs.

The literature on TM use is relatively recent and much scarcer than the one looking atfirms’ and universities’ patenting activities, and no contribution appears to have addresseduniversities’ trademarking patterns. In what follows, we briefly highlight the main issuesdealt with by the literature concerned with TM use and firm performance and the relationshipthat may exist between innovation,4 patents, and TMs. While focusing on firms and theirbehaviors, these papers offer insights that might help understanding the use of TMs byinstitutions such as universities. For a survey of university patenting, licensing, and relatedissues, the reader might refer to Verspagen (2006), Thursby and Thursby (2007), Crespi etal. (2010), and Åstebro and Bazzazian (2010).

2.1. Firms’ TMsThe literature on TMs, TM use, the link innovation–patents–TMs, and the performanceof firms is very recent and relatively scarce. Schmoch (2003) has been the first to empiri-cally investigate the relationship between the use of TMs and firms’ innovative activities.Relying on 2001 Community Innovation Survey data of German firms, he found a signif-icant correlation between innovation and the share of turnover associated with innovativeproducts in knowledge-intensive business services (KIBS). He concluded that TMs can beused as innovation indicators for KIBS. Mendonça, Santos Pereira, and Godinho (2004)relied on survey results to address the IPR behavior of Portuguese firms. In all the sectorsconsidered, they found a significant and positive correlation between the use of patents andthe use of TMs. They also found TMs to be used more intensively by high-tech manufactur-ing firms and knowledge-intensive services. They too concluded that TMs could be used asnovel sources of information on innovation. Malmberg’s (2005) analysis exploited Swedishlongitudinal data to study TM registrations and new product introduction by firms in var-ious sectors. He found the results to depend on sectors: TM registrations appear to be not

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

476 M. Squicciarini et al.

related to the number of new products introduced in the electromechanical and automotivesectors, but TMs do correlate with the introduction of new products in the pharmaceuti-cal industry. Similar to his predecessors, Malmberg concluded TMs to be well related toinnovation, especially in industries producing goods and services that directly target finalconsumers.

Greenhalgh and Rogers (2007) and Sandner (2009) focused on the relationship betweenTM use and the economic performance of firms. The former exploited UK- and EuropeanUnion (EU)-level IPR data linked to companies’ financial information and found TM activ-ity to be positively associated with stock market values, especially at the EU level and inservice firms. TM activity further appears to be positively related to productivity, and trade-marking firms exhibit a significantly higher value added than non-TM ones. Greenhalghand Rogers (2007) concluded that TMs may serve as good proxies of firm-level unobserv-ables, including innovative effort. Finally, Sandner (2009) studied the IPR behavior of largepublicly traded companies and mapped them into sets of possible strategic behaviors. Hefound TM registrations to be positively related to market value when firms follow a branddevelopment strategy, whereas this relationship does not appear to be significant in the caseof brand creation strategies. He concluded that the expected cash flows are higher whenfirms rely on existing brands to introduce new products than when creating (numerous)new brands.

Overall, the above literature suggests that TM registration can be used to proxy (sometypes of) innovative activities, that the use of TMs and other IPRs – especially patents – iscorrelated, and that TMs are positively related to firm performance. Millot and Squicciarini(2011) built on this literature and on the patent one and discussed the timing and type of IPRsthat firms use vis-à-vis the strategic goal they pursue. With respect to TMs, they noted thatfirms may rely on TMs already when getting established to signal their very existence. Lateron, TM registration can be triggered by product and process innovations (or both) – whethertechnological or not – as well as by market innovations. The latter may take the form of newfirm focus, intended as new or different bundle(s) of products and processes that firms aimto focus on, launch, or make known, and new target markets, intended geographically or interms of any other dimension used to segment markets (price, quality, services associated,etc.). Relying on French firm microdata, Millot and Squicciarini (2011) found TMs toalways be positively related to sales. Older firms appear to trademark proportionally lessthan younger ones, but both patent and TM activities are found to be persistent (Cefis 2003),with high-tech manufacturing and knowledge-intensive service firms being comparativelymore likely to trademark. Finally, firms in highly concentrated industries as well as exportersseem to rely proportionally more on TMs.

2.2. Universities’ TMsWe propose that the main findings of the literature investigating TM use vis-à-vis patents,innovative activities, and firm performance and the hypotheses put forward in Millot andSquicciarini (2011) could be applicable – at least to a certain extent – to academic insti-tutions. In particular, we believe that US universities committed to effectively accomplishtheir three main functions5 might rely on TM registration to signal, protect, or better exploitsome of their key intangible assets. Through TMs, universities may legally protect theirreputation, market their current and prospective activities, and better appropriate and selltheir innovative output.

Reputation, a key asset for academic institutions, is the equivalent to what brandsrepresent for enterprises. Reputation drives the selection and self-selection of students and

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 477

professors alike; may raise the likelihood of obtaining external funds, with these beingprivate or public; facilitates networking with other top institutions and with the privatesector; and more generally grant academic institutions a number of competitive advantagesover their local and global competitors. It may hence be reasonable to expect that whengetting established or at later stages, universities might trademark their names or logos tobetter manage their ‘brand’.

Similar to the way firms would behave, entrepreneurial universities might further registerTMs when launching new educational products (e.g. new master courses); when openingnew departments or research centers/units (e.g. ‘Spacewatch’, University of Arizona);or when offering new services and products (e.g. American Customer Satisfaction Index(ACSI), University of Michigan)6. When the latter result from research and innovationactivities, TM registration may be observed in conjunction with patent filing (e.g. Bioglass,University of Florida7).

In what follows, we shed some light on the entrepreneurial behavior of academic institu-tions and investigate the relationships that may exist between TM registration and a numberof institutional characteristics (e.g. universities being private or public) and performancevariables, such as the number of students enrolled, the funding obtained, and the patentsowned. Our study opens a greenfield site, and future work might wish to deal with some ofthe shortcomings and concerns our experimental study leaves unaddressed – in particular,selectivity and endogeneity. The original data set built for the present study neverthelessallows us to offer a number of interesting descriptive statistics and to draw attention toissues deserving future investigation.

3. DataData on academic Intellectual Property (IP) were extracted from the USPTO ‘TrademarkCasefile Dataset (1884–2010)’ and the EPO worldwide patent statistical database (‘PAT-STAT’, April 2011). PATSTAT contains data on USPTO patents granted since 1974 andpatent applications since 2000. While featuring a wealth of information about TMs andpatents, respectively, these data sets do not include any indication about the type of appli-cant or owner of the IPR considered – whether a private individual, a firm, or a university.Therefore, the allocation of patents and TMs to academic institution has to be inferred fromthe very name of the applicant(s). To this end, a slightly modified version of the patent-based algorithm developed by Van Looy, Du Plessis, and Magerman (2006)8 was usedfor the identification of academic applicants owning patent and TM rights at USPTO. TMdata were then double-checked to correct for the possible allocation of rights to entitiesmistakenly identified as universities (e.g. ‘University book store’).

The sample used in our study included applicants categorized as universities – bothprivate and public ones – university hospitals, and the different types of offices and orga-nizations entrusted with the commercial exploitation of university IPRs.9 The pieces ofinformation contained in our data set had undergone two name harmonization and con-solidation procedures, one automatic and the other manual. Data were first regrouped andharmonized to account for possible name variations – including misspelling – using a newname harmonization algorithm purposely developed by Idener© for the OECD (2011).10

IP portfolios were then manually consolidated at the individual university level to avoidparts of the very same institution being mistakenly considered as different bodies (e.g.‘Georgetown University’ and ‘Georgetown University Medical Center’ being consolidatedinto ‘Georgetown University’).11

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

478 M. Squicciarini et al.

Table 1. The number of distinct US academic IP owners identified after each consolidation step.

USPTO USPTONumber of distinct academic IP owners patents TMs

Identified using Van Looy, Du Plessis, and Magerman’s (2006) algorithm 5490 9941Identified after removing entities mistakenly identified as universities 6798Identified after using OECD (2011) name harmonization algorithm 2440 2838Identified after manual consolidation 785 1423

Source: The authors’ own compilation on OECD (2011), Patent Database, and USPTO (2011), ‘The USPTOTrademark Casefile Dataset (1884–2010)’.

Table 2. The number and share of distinct academic IP owners and IPapplications matched to institutions in the MUP panel.

Number of Patents TMs

Academic IP owners 359 (46%) 434 (30%)Applications/registrations 60,545 (93%) 10,173 (73%)

Source: The authors’ own compilation on OECD (2011), Patent Database, USPTO(2011), ‘The USPTO Trademark Casefile Dataset (1884–2010)’, and MUP data.

Table 1 presents the number of academic IP applicant names obtained after each dataconsolidation step. As can be seen, the name harmonization and consolidation procedurefollowed substantially reduced the noise contained in the data, with the number of distinctTM and patent owners contained in the final sample being only one-seventh of the initialone.12

These IP data were matched to university-related data compiled by the MUP of ArizonaState University. MUP annually ranks universities on the basis of a number of variablesobtained from public sources (e.g. National Science Foundation). MUP data are availableonline,13 cover the period 1997–2007, and encompass more than 600 academic institu-tions (the so-called Top American Research Universities). The MUP data are typicallycollected at the geographic campus level, whereas IP ownership data may sometimes notbe as detailed. Therefore, in very few (although important) cases, data were further consol-idated at the aggregate entity level.14 Table 2 presents the overall number and proportion ofacademic IP owners and of patent applications/TM registrations that have been matchedto the institutional data contained in the MUP panel.

In the case of patents, 359 academic IP owners were matched to universities in theMUP sample (i.e. 46% of our academic patent sample). These universities account for 93%of all patent applications that we found to belong to academic institutions. In the case ofTMs, MUP proved to only provide data concerning 30% of the US academic institutionsthat have registered TMs. They nevertheless account for 70% of all TMs registered by theuniversities in the USA.

4. Descriptive statistics4.1. IP activities of US universities, 1974–2010The yearly number of patents and TMs granted by the USPTO to US universities has beengrowing steadily over the last three decades.15 Although the overall number of patentsgranted to universities (60,545 in total during the period considered) is significantly higherthan the number of the academic TM applications (10,173), the utilization of both types

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 479

0

100

200

300

400

500

600

700

0

500

1 000

1 500

2 000

2 500

3 000

3 500

4 000

Patents Trademarks (right axis)

Figure 1. The yearly number of academic IP applications at USPTO.Source: The authors’ own compilation on OECD (2011), Patent Database, and USPTO (2011), ‘TheUSPTO Trademark Casefile Dataset (1884–2010)’.Note: Patent data correspond to patents granted. Patents are allocated to years according to the dateof application. USPTO patent data and TM data are available and complete from 1974 and 1983,respectively.

0.00%

0.05%

0.10%

0.15%

0.20%

0.25%

0.30%

0.35%

0.40%

0.00%

0.50%

1.00%

1.50%

2.00%

2.50%

3.00%

3.50%

4.00%

4.50%

Patents Trademarks (right axis)

Figure 2. The yearly proportion of academic IPRs in all US applications at USPTO, 1974–2010.Source: The authors’ own compilation on OECD (2011), Patent Database, and USPTO (2011), ‘TheUSPTO Trademark Casefile Dataset (1884–2010)’.Note: Patent data correspond to patents granted. Patents are allocated to the various years accordingto the date of application. USPTO patent data and TM data are available and complete from 1974 and1983, respectively.

of IPRs increased after the passing of the US Patent and Trademark Law Amendment Actof 1980.

As can be seen from Figure 1, which shows the yearly number of academic IP appli-cations at USPTO, the peak in patenting activities occurred in 1995 due to a change inthe patent regime.16 Figure 2 compares the figures related to patents and TMs owned byuniversities with the total number of patent and TM applications made at USPTO over theperiod considered (by all US applicants). The proportion of patents granted to academicapplicants shows an overall increasing trend, although not a constant one, and a seeminglydecreasing pattern since the 2000s (as was also found by Leydesdorff and Meyer 2010).

4.2. Universities’ main characteristicsThe sample used in the present study relies on data from 621 US universities, whose maincharacteristics and IP activities (if any) were observed over the period 1997–2007. Table 3

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

480 M. Squicciarini et al.

Table 3. Universities’ main characteristics, 1997–2007.

Total Share of University Students Share of Number ofresearch federal age in 2006 (in graduates universities

expenditures funds (in 2006) thousands) (%) in stateNumber of Medicalobservations μ σ μ σ school μ σ μ σ μ σ μ σ

Overall sample 621 83,493 192,679 0.6 0.23 0.18 115 49 12.32 14.62 0.27 0.3 22.33 15.5Public universities 369 82,885 210,867 0.58 0.22 0.19 106 46 15.86 16.73 0.24 0.2 19.62 14.5Private universities 252 84,418 161,187 0.63 0.24 0.18 128 50 6.44 6.92 0.34 0.3 26.3 16.2

Source: The authors’ own compilation on MUP data (2011).Notes: μ, mean; σ , standard deviation. Fund figures are given at constant prices (base year 2007).

summarizes some key features of the institutions considered (the corresponding detailedyearly statistics are given in Table A1). The first two variables, namely, ‘total researchexpenditures’ and ‘share of federal funds’, are time variant; conversely, variables such asthe presence of a ‘medical school’, ‘university age’ since establishment, the number of‘students in 2006’, the ‘share of graduates’, and the ‘number of universities in state’ aretime invariant. In particular, the number of students and the share of graduates are relatedto the year 2006.17

Almost 60% of the universities in the sample are public institutions. These are, onaverage, younger than their private counterparts – with age being defined as the numberof years that have elapsed from establishment until 2006. Public universities are also, onaverage, much bigger than private ones in terms of the number of students (on average,16,000 in public universities and 6,000 in private ones), although their share of graduatestudents is much smaller than the one featured by private academic institutions (24% and34%, respectively).

When it comes to overall research expenditures, private and public institutions invest,on average, a very similar amount of money per year, although given the difference in size,the amount of funds per student differs notably. Finally, on average, around 19% of publicuniversities and 18% of private universities feature medical schools.

The MUP data further provided an indicator of the way federal research funds wereallocated (in the fiscal year 2004) across disciplines, grouped into nine main categories:Computer Sciences, Engineering, Environmental Sciences, Life Sciences, MathematicalSciences, Physical Sciences, Psychology, Social Sciences, and Other Sciences. Institutionswith 95% or more funds concentrated in a certain area j were identified as ‘All j’; those with75–94% in one field as ‘Heavy j’; those with 50–74% as ‘Strong j’; and those with 25–49% as‘Moderate j’. Several research focuses may coexist, for example, ‘Strong Life Sciences andModerate Computer Sciences’, and those institutions whose expenditures are distributedevenly across all disciplines were identified as ‘Mixed’. Based on this MUP indicator,we built a university-specific variable called ‘main federal research focus’, indicating thediscipline in which the proportion of research expenditures is highest. Table 4 displaysthe frequencies of the pairs subject area/intensity of the research focus and of the mainfederal focuses.

4.3. Our sample: academic institutions and use of IPRTable 5 presents some academic IP-related statistics. Data suggest that the majority of theuniversities in the sample have not made use of IPRs during the years 1997–2007, withmedian values (not displayed here) equaling zero for all variables considered. The only

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 481

Table 4. Universities’ research focus, 2004.

Research focusMain federal

All Heavy Strong Moderate Low Mixed research focus

Computer sciences 0 3 5 2 8 10 8Engineering 7 8 29 33 35 10 68Environmental sciences 2 8 20 20 19 10 42Life sciences 72 64 102 127 1 10 334Mathematical sciences 0 1 3 3 4 10 6Physical sciences 6 7 27 45 28 10 62Psychology 0 3 9 6 11 10 14Social sciences 0 2 5 8 6 10 11Other sciences 3 0 5 2 5 10 9

Source: The authors’ own compilation on MUP data (2011).

exception is represented by the median value of the number of TMs registered before 1997,which takes the value of 1 – that is, that most universities have registered at least one TMin the past.

Public and private universities exhibit very similar TM application figures, as well asaverage patent application and grant data. Big differences conversely exist in the distributionof private and public institutions when it comes to patenting: the distribution of privateuniversities looks much less dispersed than that of public ones, as highlighted by standarddeviations and the maximum yearly number of patents applied and granted.

Figure 3 shows the number of TMs registered and patents applied for by academicinstitutions in the various states over the period 1997–2007. The area of the circles isproportional to the overall number of applications – with the darker circles denoting patentsand the lighter ones denoting TMs. The horizontal axis displays the number of universitieslocated in each state, while the vertical axis indicates the average number of IP applicationsmade by universities in the considered state. The dotted 45 line depicted accounts for thedifferent x- and y-axis scales used.

During the period considered, academic patenting occurred more frequently than aca-demic trademarking. While differences emerge across States, the number of patents almostalways exceeds that of TMs. New York, California, Texas, Pennsylvania, and Massachusettsare the States in which universities have been most IP active during the years 1997–2007.With a few exceptions – notably New York and a few other States appearing below the45 line – on average, universities own at least one IPR each, with universities in Utahseeming to rely proportionally more on IP protection than academic institutions located inother States.

4.4. Distribution of patents and TMs by IPC and Nice classesThe patenting activity of US universities appears to be mainly related to material science,medicine, microbiology, and electrics and electronics. This can be found by looking at themain International Patent Classification (IPC)18 classes that the academic patents belong to(see Table 6).

TM registrations detail the list of goods or services (or both) to which the TM wouldapply. These are classified according to the International Classification of Goods and Ser-vices for the Purposes of the Registration of Marks, known as the ‘Nice Classification’. Thecurrent edition – ninth, entered into force on 1 January 2007 – contains 34 classes of goods

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

482M

.Squicciarinietal.

Table 5. Universities’ TM and patent activities, 1997–2007.

TMS Patent applications Patent grantedNumber of

Year Observations Mean P75 SD Max Mean P75 SD Max Mean P75 SD Max

Overall sample 1997 621 0.65 0 2.72 38 0.07 0 0.42 7 5.29 1 22.77 4521998 0.76 0 2.5 27 0.17 0 0.85 14 5.33 2 23.53 4661999 0.89 0 2.42 22 0.29 0 1.55 30 5.74 2 24.99 4992000 0.89 0 2.5 22 0.52 0 2.38 48 6.24 2 26.16 5162001 0.86 0 3.04 45 5.33 1 25.81 518 6.08 2 27.56 5492002 0.94 0 2.54 21 5.51 1 25.39 508 5.96 2 25.3 4932003 0.91 0 2.68 33 5.62 1 25.2 487 5.71 2 23.59 4302004 0.98 0 3.06 30 5.41 1 21.69 400 5.25 1 19.55 3412005 0.96 1 3 37 5.47 2 21.03 366 4.88 2 17.92 3002006 0.97 1 2.83 34 5.14 2 19.82 325 4.04 2 15.24 2392007 1.09 1 2.9 29 4.17 1 15.65 259 2.71 1 9.99 145

Public universities 1997 369 0.7 0 2.74 38 0.05 0 0.42 7 5.34 2 26.19 4521998 0.74 0 2.42 23 0.13 0 0.82 14 5.18 2 26.33 4661999 0.96 0 2.69 22 0.29 0 1.75 30 5.87 2 28.39 4992000 0.86 0 2.26 18 0.52 0 2.69 48 6.53 3 29.87 5162001 0.77 0 2.56 35 5.34 2 29.54 518 6.16 3 31.35 5492002 0.88 0 2.5 21 5.6 2 29.36 508 6.17 2 28.86 4932003 0.81 0 2.28 23 5.79 2 28.93 487 5.92 2 26.32 4302004 1.03 0 3.46 30 5.58 2 24.51 400 5.5 2 21.67 3412005 1.03 0 3.1 37 5.6 2 22.79 366 5.18 2 19.46 3002006 1.07 1 3.23 34 5.26 3 20.82 325 4.3 2 16.05 2392007 1.12 1 3.02 29 4.29 2 16.72 259 2.91 2 10.31 145D

ownl

oade

d by

[U

nive

rsity

of

Sask

atch

ewan

Lib

rary

] at

08:

17 0

6 O

ctob

er 2

012

Econom

icsofInnovation

andN

ewTechnology

483

Private universities 1997 252 0.58 0 2.68 38 0.09 0 0.4 3 5.21 1 16.59 1401998 0.8 0 2.63 27 0.23 0 0.9 7 5.56 1 18.75 1541999 0.79 1 1.97 15 0.29 0 1.21 12 5.55 1 18.98 1612000 0.94 0 2.82 22 0.51 0 1.85 17 5.83 1 19.55 1642001 1 0 3.63 45 5.32 1 19.13 179 5.95 1 20.86 1892002 1.03 1 2.6 19 5.37 1 18.12 161 5.65 1 18.98 1772003 1.06 1 3.16 33 5.38 0 18.48 155 5.4 1 18.94 1742004 0.92 1 2.38 22 5.16 0 16.77 138 4.88 0 16 1352005 0.87 1 2.86 34 5.29 1 18.19 169 4.45 1 15.43 1432006 0.82 0 2.11 16 4.95 1 18.29 187 3.66 1 13.98 1412007 1.04 0.5 2.72 20 3.99 1 13.98 120 2.43 1 9.51 109

Before 1997 Tms Overall 621 5.95 5 15.25 165 0.04 0 0.34 6 40.91 11 164.12 2639Public unis 369 6.38 5 17.7 165 0.02 0 0.19 3 39.64 14 166.56 2639Private unis 252 5.31 5 10.69 66 0.06 0 0.48 6 42.76 7.5 160.79 1890

Source: The authors’ own compilation on OECD (2011), Patent Database, USPTO (2011), ‘The USPTO Trademark Casefile Dataset (1884–2010)’, and MUP data.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

484 M. Squicciarini et al.

ALAR

AZ

CA

CO

CT

DC

DE

FL

GA

HI

IA

ID

IL

IN

KS

KY

LA

MA

MD

ME

MI

MN

MO

MS

MT

NC

ND

NE

NH

NJ

NM

NV

NY

OH

OKOR

PA

PR

RI

SCSD

TN

TX

UT

VA

VI

VT

WA

WI

WV

WY

0

20

40

60

80

100

120

140

160

0 10 20 30 40 50 60

Number of academic TM

Number of academic PAT

Number of academic institutions

Average number of patent and TM applications by academic institution

Figure 3. Academic TM and patent applications by State (1997–2007).Source: The authors’ own compilation on OECD (2011), Patent Database, USPTO (2011), ‘TheUSPTO Trademark Casefile Dataset (1884–2010)’, and MUP data.

Table 6. Top 10 IPC classes in academic patent applications.

% of total academicIPC IPC code description applications

A61K Preparations for medical, dental, or toilet purposes 14G01N Investigating or analyzing materials by determining their chemical

or physical properties8

C12N Micro-organisms or enzymes; compositions thereof; propagating,preserving, or maintaining micro-organisms; mutation or geneticengineering; culture media

7

C07K Peptides 5H01L Semiconductor devices; electric solid-state devices not otherwise

provided for5

C12Q Diagnosis; surgery; identification 4A61B Measuring or testing processes involving enzymes or micro-

organisms; compositions or test papers therefore; processes ofpreparing such compositions; condition-responsive control inmicrobiological or enzymological processes

4

G06F Electric digital data processing 3C07D Heterocyclic compounds 2G02B Optical elements, systems 2

Source: The authors’ own compilation on OECD (2011) and Patent Database.Note: Statistics presented at the IPC subclass level.

and 11 classes of services, with good TMs being detailed in classes 1–34 and service TMscorresponding to classes 35–45.19

As expected, TM registration by US academic institutions mainly pertain to education-related services, followed by the kind of items that are related to the universities’ name

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 485

Table 7. Top 10 classes designated in academic TM applications.

Share in academic Share in allClass Description applications TM applications

41 Education services; providing of training; entertainment;sporting and cultural activities

32.34% 7.69%

25 Clothing, footwear, and headgear 13.32% 5.52%16 Paper, cardboard, and goods made from these materials,

not included in other classes; printed matter;bookbinding material; photographs; stationery;adhesives for stationery or household purposes;artists’ materials; paint brushes; typewriters andoffice requisites (except furniture); instructionaland teaching material (except apparatus); plasticmaterials for packaging (not included in otherclasses); printers’ type; printing blocks

11.44% 5.69%

42 Scientific and technological services and researchand design relating thereto; industrial analysisand research services; design and development ofcomputer hardware and software

6.31% 6.96%

21 Household or kitchen utensils and containers; combsand sponges; brushes (except paint brushes); brush-making materials; articles for cleaning purposes;steelwool; unworked or semi-worked glass (exceptglass used in building); glassware, porcelain, andearthenware not included in other classes

4.99% 1.54%

9 Scientific, nautical, surveying, photographic,cinematographic, optical, weighing, measuring,signaling, checking (supervision), life-saving andteaching apparatus and instruments; apparatus andinstruments for conducting, switching, transforming,accumulating, regulating, or controlling electricity;apparatus for recording, transmission, or reproductionof sound or images; magnetic data carriers, recordingdiscs; automatic vending machines and mechanismsfor coin-operated apparatus; cash registers,calculating machines, data processing equipment andcomputers; fire-extinguishing apparatus

4.33% 11.26%

35 Advertising; business management; businessadministration; office functions

3.20% 8.57%

14 Precious metals and their alloys and goods in preciousmetals or coated therewith, not included in otherclasses; jewellery, precious stones; horological andchronometric instruments

2.59% 1.28%

28 Games and playthings; gymnastic and sporting articlesnot included in other classes; decorations forChristmas trees

2.35% 2.91%

44 Medical services; veterinary services; hygienic andbeauty care for human beings or animals; agriculture,horticulture, and forestry services

2.26% 1.39%

Source: The authors’ own compilation on USPTO (2011), ‘The USPTO Trademark Casefile Dataset (1884–2010)’.

or brand and that are sold as gadgets, for example, clothing and stationery. These types ofTMs are typically registered in classes 41, 25, 16, 21, 14, and 28 (see Table 7). Examplesare ‘MIT’, which is a registered TM of the Massachusetts Institute of Technology (MIT),and is protected in classes 41, 16, and 25, and the ‘CALIFORNIA STATE UNIVERSITYSTANISLAUS VOX VERITAS VITA MCMLX’ TM, protected in the very same classes.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

486 M. Squicciarini et al.

A second group of TM classes often used by universities is related to high-tech productsand to research and scientific services – namely, classes 42, 9, and 44. Examples of academicTMs registered in the various Nice classes that are related to the output of university researchactivities are given in Table A2, in Appendix 1.

The proportion of academic TMs registered in the classes seemingly related to researchoutput has been steadily increasing over the last three decades, reaching around one-thirdof all applications in 2008 (Figure 4). This may suggest that academic institutions haveinitially relied on TMs to protect or better sell their ‘brand’, and in later years – from the1990s – they have begun to use them in relation to their research activities.

As can be seen in Table 8, the vast majority of academic IPs – especially patents –belong to institutions mainly engaged in life science research, followed by engineering andphysical sciences.

In all disciplines, most institutions file at least one TM and only a minority file patentsonly. In life sciences and engineering, joint use of TMs and patents is relatively frequent,whereas in disciplines such as mathematics, social sciences, and psychology, it is commonto observe more TM registrations than patent applications (Figure 5). The list of the top 50

0%

5%

10%

15%

20%

25%

30%

35%

40%

Figure 4. The proportion of academic TMs filed in classes that may relate to research output (classes1, 5, 9, 10, 35, 42, and 44).Source: The authors’ own compilation on USPTO (2011), ‘The USPTO Trademark Casefile Dataset(1884–2010)’.

Table 8. IP use by main federal research focus, 1997–2007.

TMs in classes1, 5, 9, 10, 35,

Patents TMs 42, and 44

Computer Sciences 131 102 38Engineering 3146 816 153Environmental Sciences 246 236 35Life Sciences 17,739 4014 964Mathematical Sciences 0 11 0Physical Sciences 1578 572 57Psychology 5 67 5Social Sciences 165 42 2Mixed 344 177 26Other Sciences 3 15 0

Source: The authors’ own compilation on OECD (2011), Patent Database, USPTO (2011),‘The USPTO Trademark Casefile Dataset (1884–2010)’, and MUP data.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 487

0

.2

.4

.6

.8

1_C

ompu

ter

Sci

ence

s

2_E

ngin

eerin

g

3_E

nviro

nmen

tal S

cien

ces

4_Li

fe S

cien

ces

5_M

athe

mat

ical

Sci

ence

s

6_P

hysi

cal S

cien

ces

7_P

sych

olog

y

8_S

ocia

l Sci

ence

s

9_M

ixed

or

NA

9_O

ther

Sci

ence

s

TM only PAT onlyTM and PAT

Figure 5. The proportion of academic institutions with joint or separate use of patents and TMs bymain federal research focus (1997–2007).Source: The authors’ own compilation on OECD (2011), Patent Database, USPTO (2011), ‘TheUSPTO Trademark Casefile Dataset (1884–2010)’, and MUP data.Note: The complement to 100% is represented by institutions without any IP.

academic applicants of patents and TMs during the period 1997–2007 is given in Table A3in Appendix, which further displays the top 50 universities in terms of TMs registered inclasses seemingly related to research activities.

5. Empirical analysis and main resultsWe rely on count data models to shed some light on the possible institutional charac-teristics and factors that may explain TM registration by US universities. Data availabilityconstraints and the thinness of the literature investigating TM behaviors constrain our green-field exploration. This explains why we limit ourselves to uncover possible relationships andraise issues for future research, while leaving selectivity and endogeneity concerns unad-dressed, as are causal links. Our dependent variable, tmit , is the number of TMs appliedfor by institution i at time t (in years). The control variables that we use account for someuniversity-specific characteristics and for the innovative behaviors of universities. They arelisted in Table 9, together with a short explanation of their content.

Patappit denotes the number of patents applied for by institution i at time t ; patgrantitrefers to the number of patents granted to institution i and applied for at time t ; and before1997tmi is a time-invariant discrete variable indicating the total number of TMs applied forby university i at the onset of our observation period. Controli is a dichotomous variabledenoting whether university i is a public one (controli = 0) or a private one (controli = 1).Tot_fund_constit is a positively determined continuous variable revealing the total amountof funds (in 1000 dollars) that university i has invested in research at time t (i.e. in yeart), whereas fedfund_shit denotes the share of university i’s research expenditures financedby the federal government in year t. Five additional variables are university specific andtime invariant: agei mirrors university i’s age in 2006, that is, the number of years sinceestablishment; stud2006i, the number of students enrolled (in thousands); gstud_sharei, theshare of graduate students (out of all students); first_profstudi, the share of first professionalstudents;20 and med_schooli, the presence of a medical school. Stud2006i, gstud_sharei,and first_profstudi are related to the fall 2006 period. Uni_state is a time-invariant discrete

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

488 M. Squicciarini et al.

Table 9. Variables’ name and content.

Variable Description

tmit Number of TMs applied for by university i at time tpatgrantit Number of patents granted to university i and applied for at time tpatappit Number of patents applied for by university i at time tbefore 1997tmi Total number of TMs applied for by university i before 1997controli Dichotomous variable denoting university ownership: 1, private; 0, publictot−fund−constit Research expenditures of university i at time t (in 1000 dollars, at constant

values)fedfund−shit Share of university i’s research expenditures financed by federal government

at time tagei Age is the number of years from establishment of university i until 2006stud2006i Number of students enrolled in university i in fall 2006 (in thousands)gstud−sharei Share of graduate students (out of all students) enrolled in university i in fall

2006first−profstudi Share of first professional students enrolled in university i in fall 2006med−schooli Dichotomous variable denoting the presence of a medical school: 1, yes; 0,

nouni−statei Number of universities in the staten−researchareasi Number of disciplines federal funds are allocated to in university i (see

Section 4.2)engineeringi, . . .,

social sciencesi

Proportion of federal research funds allocated to each discipline (see Section4.2)

variable accounting for the number of universities located in the state where university iis located. N_researchareasi is a count variable ranging from 0 to 9, denoting the numberof scientific fields in which university i has received federal funds for research activities in2004. Nine variables account for the intensity of federal research funds across disciplines.The value of these variables ranges from 0 to 1, in a stepwise fashion21 (see Section 4.2 formore details).

Some of the variables appear to be strongly correlated with each other, as can be seenfrom Table 10. This is the case, for instance, for the number of students, which appears tobe very much linked to the number of patents granted, the stock of TMs in 1997, the typeof ownership – that is, being private or public – and the overall amount of research funds.Estimates nevertheless prove to be not significantly affected by the inclusion or exclusion ofthese highly correlated variables – in terms of either significance or sign of the regressors.

As for the relationship between TMs and patents, it might be reasonable to suppose theexistence of a ‘one-to-many’ link or of a ‘many-to-many’ relationship. Hypothesizing a one-to-many link would entail expecting that one TM registration would occur for every x patentapplications. Conversely, expecting a many-to-many relationship would entail expectingthe number of TMs to be similar to the number of patents applied for over a certain period.To this end, different time spans can be considered, depending on whether lags might bethought to exist between the time of patent applications and that of TM registrations. Asthe literature is silent on both timing and quantity issues, we remain agnostic and try tomaximally exploit the informational content of our data. This leads us to estimating countdata models – the alternative being to rely on discrete choice models – and to hypothesizepatents and TM registrations to occur simultaneously.22

Given that we only have a few time-varying regressors available, the analysis relies onboth panel and pooled cross-section estimates, as shown in Table 11. Both Models 1 and 2follow a Poisson specification, with Model 1 being estimated in the form of a fixed-effects(FE) panel and Model 2 as a pooled cross-section model. Model 3 is conversely specified as

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Econom

icsofInnovation

andN

ewTechnology

489

Table 10. Pairwise correlation matrix.

before tot−fund− gstud− first− med− n−tm patapp patgrant 1997tm control const fedfund−sh age stud2006 share profstud school uni−state researchareas

tm 1patapp 0.27∗ 1patgrant 0.33∗ 0.75∗ 1before1997tm 0.51∗ 0.34∗ 0.41∗ 1control −0.03∗ 1tot−fund−const 0.35∗ 0.71∗ 0.74∗ 0.36∗ 1fedfund−sh 0.03 0.10∗ 0.09∗ 1age 0.15∗ 0.13∗ 0.15∗ 0.18∗ 0.22∗ 0.22∗ 1stud2006 0.32∗ 0.48∗ 0.59∗ 0.35∗ −0.31∗ 0.65∗ −0.09∗ 0.06∗ 1gstud−share 0.06∗ 0.08∗ 0.10∗ 0.05∗ 0.19∗ 0.19∗ 0.04∗ −0.12∗ −0.09∗ 1first−profstud −0.03 −0.02 −0.06∗ 0.18∗ 0.04 −0.13∗ −0.19∗ 0.80∗ 1med−school 0.21∗ 0.20∗ 0.24∗ 0.19∗ 0.40∗ 0.11∗ 0.13∗ 0.17∗ 0.49∗ 0.41∗ 1uni−state 0.06∗ 0.07∗ 0.21∗ 0.03 −0.05∗ 0.03 0.17∗ 0.09∗ −0.04∗ 1n−researchareas 0.04∗ 0.04∗ 0.05∗ 0.11∗ −0.03∗ −0.06∗ 0.05∗ 0.16∗ −0.11∗ −0.12∗ −0.07∗ −0.05∗ 1

Note: Correlation coefficients displayed if significance level ≥0.10.∗Significance level ≥ 0.01.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

490M

.Squicciarinietal.

Table 11. Regression results.

Model 1: Poissona (panel FE) Model 2: Poissona (pooled) Model 3: Tobit (pooled) Model 4: ZIP (pooled)

(1) (2) (3) (4) (5) (6) (7) (8)Variables tm tm tm tm tm tm tma inflate tma inflate

Patgrant 1.000 0.997∗∗∗ −0.010∗ 0.999∗∗ −0.019∗∗∗(0.001) (0.000) (0.006) (0.000) (0.006)

Patapp 0.998∗∗∗ 0.998∗∗∗ −0.014∗∗ 0.998∗∗∗ −0.009∗(0.000) (0.000) (0.006) (0.000) (0.005)

before1997tm 1.020∗∗∗ 1.020∗∗∗ 0.127∗∗∗ 0.127∗∗∗ 1.011∗∗∗ −0.048∗∗∗ 1.011∗∗∗ −0.051∗∗∗(0.000) (0.000) (0.006) (0.006) (0.000) (0.005) (0.000) (0.005)

Control 1.524∗∗∗ 1.449∗∗∗ 2.030∗∗∗ 1.983∗∗∗ 1.160∗∗∗ −0.811∗∗∗ 1.147∗∗∗ −0.834∗∗∗(0.058) (0.054) (0.307) (0.304) (0.050) (0.116) (0.049) (0.116)

tot−fund−const 1.000 1.000 1.000∗∗∗ 1.000∗∗∗ −0.000∗ −0.000 1.000 0.000 1.000 0.000(0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000) (0.000)

fedfund−sh−const 1.912∗∗∗ 1.866∗∗∗ 1.467∗∗∗ 1.433∗∗∗ 1.014 0.993 1.481∗∗∗ 0.025 1.468∗∗∗ 0.008(0.396) (0.387) (0.126) (0.123) (0.622) (0.621) (0.144) (0.221) (0.142) (0.220)

Age 1.002∗∗∗ 1.002∗∗∗ 0.009∗∗∗ 0.009∗∗∗ 1.001∗∗ −0.001 1.001∗∗ −0.002(0.000) (0.000) (0.002) (0.002) (0.000) (0.001) (0.000) (0.001)

stud2006 1.022∗∗∗ 1.017∗∗∗ 0.124∗∗∗ 0.119∗∗∗ 1.007∗∗∗ −0.055∗∗∗ 1.006∗∗∗ −0.057∗∗∗(0.001) (0.001) (0.012) (0.011) (0.001) (0.006) (0.001) (0.005)

gstud−share 5.035∗∗∗ 4.533∗∗∗ 5.664∗∗∗ 5.560∗∗∗ 1.126 −1.241∗∗∗ 1.098 −1.397∗∗∗(0.606) (0.543) (0.926) (0.921) (0.175) (0.353) (0.165) (0.349)

first−profstud 0.080∗∗∗ 0.086∗∗∗ −6.160∗∗∗ −6.073∗∗∗ 0.560∗∗ 0.482 0.575∗∗ 0.686(0.018) (0.019) (1.504) (1.501) (0.144) (0.562) (0.145) (0.561)

med−school 1.966∗∗∗ 1.991∗∗∗ 2.028∗∗∗ 2.006∗∗∗ 1.195∗∗∗ −0.372∗∗∗ 1.184∗∗∗ −0.400∗∗∗(0.088) (0.089) (0.369) (0.369) (0.057) (0.141) (0.056) (0.139)

uni−state 0.988∗∗∗ 0.987∗∗∗ −0.042∗∗∗ −0.042∗∗∗ 0.996∗∗∗ 0.012∗∗∗ 0.996∗∗∗ 0.012∗∗∗(0.001) (0.001) (0.008) (0.008) (0.001) (0.003) (0.001) (0.003)

nb−researchareas 1.090∗∗∗ 1.097∗∗∗ 0.026 0.029 1.027 0.037 1.033∗ 0.045(0.018) (0.019) (0.102) (0.102) (0.020) (0.044) (0.019) (0.043)

Engineering 0.777∗∗ 0.722∗∗∗ −0.388 −0.412 0.980 −0.173 0.940 −0.158(0.082) (0.076) (0.810) (0.809) (0.122) (0.285) (0.115) (0.285)

Environmental sciences 0.285∗∗∗ 0.276∗∗∗ −2.211∗∗ −2.229∗∗ 0.401∗∗∗ −0.374 0.378∗∗∗ −0.342(0.049) (0.047) (1.030) (1.029) (0.080) (0.385) (0.076) (0.388)D

ownl

oade

d by

[U

nive

rsity

of

Sask

atch

ewan

Lib

rary

] at

08:

17 0

6 O

ctob

er 2

012

Econom

icsofInnovation

andN

ewTechnology

491

Life sciences 0.726∗∗∗ 0.697∗∗∗ −1.027 −1.060 1.053 0.085 1.021 0.114(0.062) (0.060) (0.649) (0.649) (0.100) (0.233) (0.097) (0.233)

Other sciences 0.270∗∗∗ 0.240∗∗∗ −2.232 −2.288 0.312∗∗ −0.513 0.287∗∗ −0.518(0.102) (0.093) (1.826) (1.826) (0.163) (0.759) (0.150) (0.765)

Physical sciences 0.643∗∗∗ 0.608∗∗∗ −1.638∗ −1.678∗ 1.403∗∗ 0.576∗ 1.316∗ 0.538∗(0.079) (0.075) (0.869) (0.867) (0.210) (0.312) (0.193) (0.309)

Psychology 0.272∗∗∗ 0.278∗∗∗ −3.095∗∗ −3.082∗∗ 0.589∗∗ 0.241 0.565∗∗ 0.349(0.060) (0.060) (1.300) (1.298) (0.147) (0.467) (0.141) (0.469)

Computer sciences 1.603∗∗∗ 1.626∗∗∗ 1.214 1.194 1.732∗∗∗ −0.340 1.679∗∗∗ −0.313(0.249) (0.251) (1.376) (1.375) (0.293) (0.486) (0.283) (0.486)

Mathematical sciences 0.004∗∗∗ 0.004∗∗∗ −4.522∗ −4.599∗ 0.014∗∗∗ −2.545 0.012∗∗∗ −2.611(0.004) (0.004) (2.499) (2.496) (0.013) (1.676) (0.012) (1.696)

Social sciences 0.848 0.791 −0.766 −0.777 2.118∗∗ 0.173 2.035∗∗ 0.188(0.205) (0.192) (1.671) (1.669) (0.654) (0.560) (0.626) (0.559)

Constant 0.185∗∗∗ 0.201∗∗∗ −10.335∗∗∗ −10.323∗∗∗ 1.382∗∗∗ 3.080∗∗∗ 1.394∗∗∗ 3.046∗∗∗(0.019) (0.021) (0.843) (0.840) (0.165) (0.307) (0.162) (0.307)

Year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes YesObservations 2911 2911 4232 4232 4232 4232 4232 4232 4232 4232Log likelihood −4754 −4747 −7622 −7635 −5235 −5234 −5625 −5625 −5626 −5626DF 13 13 31 31 31 31 31 31 31 31Chi2 99.63 112.8 6101 6075 1375 1377 1224 1224 1234 1234

Note: Standard errors are given within parentheses.∗∗∗p < 0.01.∗∗p < 0.05.∗p < 0.1. aExponentiated coefficients (IRRs).

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

492 M. Squicciarini et al.

a pooled cross-section Tobit model to deal with the possible censoring of the non-negativedependent variable.23 Finally, Model 4 corresponds to a zero-inflated Poisson (ZIP) overpooled data, accounting for the prevalence of zeros observed in the data. ZIP regressionssplit the process, modeling the outcome as zero or non-zero: a logit model first predictswhether or not the dependent variable is zero – in our case, whether university i is likelyto have at least one TM; a Poisson model then predicts the counts for those universitiesfor which the dependent variable is likely to be positive. Two columns of estimates aredisplayed for each type of model specified, depending on the use of patent applications orgrants among the explanatory variables: columns 1, 3, 5, and 7 rely on patents granted;columns 2, 4, 6, and 8 on patent applications.

Table 11 presents the coefficients of the Poisson models (i.e. Models 1, 2, and 4) asincidence rate ratios24 (IRRs), that is, the ratios by which the dependent variable changesfor a unit change in the explanatory variable (see e.g. Long 1997 or Long and Freese 2006).The coefficients of the Tobit model (Model 3) instead correspond to marginal effects. Finally,note the two parts in which the outcome of the ZIP model is presented. Column 1 (i.e. the oneon the left-hand side) corresponds to the Poisson part of the model, whereas the results of thelogit zero-inflated estimates are reported in column 2 (called ‘inflate’). Positive coefficientsin the ‘inflate’ part imply a higher chance of zeros, that is, of not applying for TMs (seeLong 1997; Greene 2008, for a discussion of zero-modified count models). Finally, variablesubscripts have been omitted for the sake of simplicity.

In the analysis that follows, we mainly focus on the sign and significance of the coeffi-cients rather than on their size. We do so being aware of the limitations of our analysis andespecially of the characteristics of our variables and the fact that only a few of them are timevarying. For instance, we lack the precise information about how research expenditures orstudents are allocated across the different disciplines and about their variation over time.Characteristics such as these are very much likely to shape or depend upon universities’IP behaviors and would improve the analysis if available. To further simplify the analysis,we present the sign of significant coefficients in Table 12. As can be seen, results appearto be very much robust across the different model specifications (the results of the Akaikeinformation criterion (AIC) and Bayesian information criterion (BIC) are given in Table A4in Appendix 1). In particular, a slight negative relationship emerges between the differenttypes of IPs considered: the more the patents applied for or granted, the less the TMs appliedfor over the same year. However, additional discrete choice panel regressions not shownhere25 suggest that, when significant, a positive relationship is related to TM registrationand patenting as such, that is, independently of the number of applications considered. Thisis also suggested by the results of the ‘inflate’ part of the ZIP model, which shows the like-lihood of TM registration to be positively related to patenting activities – both applicationsand grants.

TM registration nevertheless appears to be persistent, as the stock of TMs applied forbefore our observation period is positively related to TM applications during the period1997–2007.

As expected, private universities prove to be more entrepreneurial in nature and exhibit astronger likelihood to register a TM. TM activities further prove to be somewhat negativelycorrelated with the amount of research expenditures, whereas the share of federal fundsreceived is positively related to TM applications. Old (and possibly more established) uni-versities also seem to be more keen on relying on TMs. Taken together, the student-relatedvariables and institutional variables included in the model suggest that bigger universities,universities with a proportionally higher number of graduate students and universities car-rying out research in many fields, as well as those that see the presence of medical schools,

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 493

Table 12. Regressions results: sign of significant coefficients.

Model 1: Poisson Model 2: Poisson Model 3: Tobit Model 4: ZIP(panel FE) (pooled) (pooled) (pooled)

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

Variables tm tm tm tm model model tm inflate tm inflate

patgrant − − − −patapp − − − − −before1997tm + + + + + − + −control + + + + + − + −tot−fund−const − − −fedfund−sh + + + + + +Age + + + + + +Stud2006 + + + + + − + −gstud−share + + + + − −First−profstud − − − − − −Med−school + + + + + − + −uni−state − − − − − + − +nb−researchareas + + +Engineering − −Environmental

sciences− − − − − −

Life sciences − −Other sciences − − − −Physical sciences − − − − + + + +Psychology − − − − − −Computer

sciences+ + + +

Mathematicalsciences

− − − − − −Social sciences + +Note: Sign displayed for p < 0.1.

register more TMs. In particular, a stronger focus in computer sciences leads to observingmore TM activity. The share of first professional students conversely lowers the likelihoodto rely on this type of IP. Overall, the results seem to support the idea that universities mayregister TMs when launching new educational products, opening new units or departments,or in relationship to innovation activities.

Finally, the smaller the number of universities located in a certain state, the higherthe number of TMs applied for. These somewhat counterintuitive results – as it wouldbe reasonable to expect that competition at the local level would increase the need foruniversities to differentiate themselves from their competitors, possibly using TMs – maymirror the necessity of launching new initiatives and of being more dynamic in order toattract funds or students.

6. ConclusionsOf the many analyses focusing on universities’ third function, none seem to have addressedTM registration by academic institutions. We conversely believe that universities may relyon TMs to signal, protect, or better exploit some of their key intangible assets. Through TMs,universities may in fact legally protect their reputation, market their current and prospectiveactivities, and better appropriate and sell their innovative output.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

494 M. Squicciarini et al.

This paper is, to the best of our knowledge, the first to address this gap in the academicentrepreneurship literature. It does so by investigating whether and to what extent uni-versities in the USA use TMs; how TM registration is related to patenting by academicinstitutions; and the way some university-specific variables contribute to explain universi-ties’ TMs. The analysis relies on a novel data set containing information about universities’characteristics and their IPR activities over the period 1997–2007. Our contribution, whichis exploratory in nature and primarily descriptive in aim, is nevertheless constrained bythe availability of a wider set of time-varying regressors and by the absence of a broadand encompassing theoretical framework backing the formulation of the hypotheses thatwe tested.

Despite these shortcomings, our analysis sheds light on a number of interesting featuresrelated to universities’ IP behaviors and TM activities in particular. Our results suggest theexistence of a significant and positive relationship between TM activities engaged in by aca-demic institutions and universities’ characteristics such as the number of students enrolled,the share of graduate students, the presence of medical schools, the share of federal fundsreceived, and being a private institution. In addition, TM behaviors appear to be persistent,since having registered TMs in the past is positively related to applying for TMs duringour observation period. Significant and negative relationships instead emerge between TMbehaviors and the number of universities located in the same state. The relationship thatemerges between TMs and patents is also negative. Additional panel regressions carriedout in the form of discrete choice models anyway question such results and suggest that,when significant, a positive relationship exists between TM activity and patent applicationsor grants. The discussion about the type of relationships possibly linking TMs to patentsthus remains very much open. Future research might wish to address this ‘one-to-many’ or‘many-to-many’ IP link, as well as the timing that characterizes TM registration vis-à-vispatent application, in order to shed light on the possible existence of complementarities orsubstitution effects.

A way to further refine our greenfield exploration would be to acquire precise andtime-varying data about some key within-university characteristics, such as the number ofstudents, the number of graduates, and the amount of money spent in research by eachdepartment. The universities in our sample are very well known to differ, for example, inthe relative importance of their departments, their national ranking, and their organization,and these might have consequences on IP use and strategies (see e.g. Bercovitz et al. 2001,in this respect).

Another issue deserving investigation is whether and – if so – to what extent the rela-tionship TMs–patents is influenced by the quality of the innovations patented. Moreover,it would also be interesting to exploit some of the information contained in TM data, suchas the Nice classes in which TMs have been registered, in order to better investigate therelationship innovation output–TM registration. In this way, it would be possible to treatdifferently those TMs registered in classes typically (and almost exclusively) related toreputation and brand protection from those likely related to research activities.

Of course, all the above can be achieved after addressing possible selection andendogeneity concerns and gathering more extensive data.

AcknowledgementsWe are grateful to Stuart Graham for granting us access to the USPTO Trademark Casefile Dataset andto Chiara Criscuolo, Vincenzo Spiezia, and Patrick Llerena for their helpful comments. All mistakesremain our own.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 495

Notes1. The entrepreneurial and economic development activity function is supposed to be carried out in

addition to the traditional education and research functions.2. Visit http://mup.asu.edu/index.html for further details about the center and its data collection

methods.3. A brand is a combination of tangible and intangible elements, such as TMs, designs, and logos

and the concept, image, and reputation these elements transmit about products and/or services.See Lom (2004) for a discussion. Firms can use TMs alone or in association with other means ofprotection such as secrecy, lead time, or patents (Davis 2009).

4. We define innovation following the OECD and Eurostat (2005) Oslo Manual as the implemen-tation of a new or significantly improved product (good or service) or process, a new marketingmethod, or a new organizational method in business practices, workplace organization, or externalrelations.

5. Teaching, research, and economic development. See Etzkowitz (2003) for a discussion.6. A brief description of the examples cited, as well as additional ones, can be found in Appendix 1.7. USPTO patent numbers 4,478,904 and 4,103,002.8. This keyword-based algorithm relies on patent data to identify the different types of patent owners.

It subdivides IP users into five non-mutually exclusive categories: individual applicants; firmsor business enterprises; government agencies and (private or public) non-profit organizations;universities and higher education institutions (i.e. academic applicants); and hospitals. See VanLooy, Du Plessis, and Magerman (2006) for more details.

9. Examples are ‘Board of Trustees’, ‘Research Foundation’, ‘Research Services and DevelopmentCompany’, ‘President and Fellows’, and ‘Board of Regents’.

10. A spin-off of the University of Seville (Spain), www.idener.es.11. The university-related names considered and the consolidated list are available from the authors

upon request.12. The number of patent applications and TM registrations does not change throughout the consoli-

dation procedure. The name harmonization carried out aims at avoiding that patents and TMs besplit between institutions mistakenly identified as different entities.

13. http://mup.asu.edu/research_data.html (last accessed 30 June 2011).14. Campus-related consolidation has been carried out in the following cases: City University of New

York – 11 campuses; University of California – 9 campuses; University of Hawaii – 2 campuses;and University of Houston – 3 campuses.

15. Patent grants data are available from 1974 until 2010; TM applications data relate to the period1983–2008.

16. On 8 June 1995, the term of patent was modified and made dependent on the priority date insteadof on the issue date. No more submarine patents could thus be kept (i.e. a patent whose issuanceand publication are intentionally delayed).

17. Data about the number of students are available for the years 2004 and 2006, but the number ofgraduates is available only for the year 2006.

18. The IPC established by the Strasbourg Agreement 1971 (and continuously revised since then –the current version entered into force on 1 January 2011) provides for a hierarchical system oflanguage-independent symbols for the classification of patents and utility models according tothe different areas of technology to which they pertain. The IPC divides technology into eightsections with approximately 70,000 subdivisions, with different levels of aggregation, namelysection, class, subclass, group, and subgroup. See www.wipo.int for more details.

19. Most countries allow for multi-class filings and a few countries allow for only single-class appli-cations (e.g. China). The period of protection may vary – it is typically 10 years, but TMs canbe renewed indefinitely upon payment of fees, which are generally proportional to the numberof classes designated. For more information about TMs, see the OECD 2011 STI Scoreboard(2011).

20. First professional students are students seeking degrees in medical fields, such as chiropracticmedicine, dentistry, and pharmacy, as well as those seeking degrees in law and theology.

21. Possible values are 0, 0.11, 0.12, 0.37, 0.62, 0.85, and 1, corresponding, respectively, to ‘mixed’,‘low’, ‘moderate’, ‘strong’, ‘heavy’ and ‘all’ research intensities in each of the disciplinesconsidered.

22. In a similar fashion to the simultaneous relationship between patents and R&D uncovered byHall, Griliches, and Hausman already in 1986.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

496 M. Squicciarini et al.

23. In a Tobit model, the dependent variable is equal to zero if the latent variable is negative andequal to the latent variable otherwise. See Greene (2008) for details.

24. IRRs are given by the exponentiated coefficients of the Poisson regression.25. In particular, conditional logit regressions, known to suffer less from the identification problems

affecting panel discrete choice models (see Greene 2008, in this respect). The results of theseestimates are available from the authors upon request.

ReferencesÅstebro, T., and N. Bazzazian. 2010. Universities, entrepreneurship and local economic development.

In Handbook of research on entrepreneurship and regional development, ed. M. Fritsch, 252–333.Cheltenham: Edward Elgar.

Bercovitz, J., M. Feldman, I. Feller, and R. Burton. 2001. Organizational structure as a determinantof academic patent and licensing behavior: An exploratory study of Duke, Johns Hopkins, andPennsylvania State Universities. The Journal of Technology Transfer 26, no. 1–2: 21–35.

Cefis, E. 2003. Is there any persistence in innovative activities? International Journal of IndustrialOrganization 21, no. 4: 489–515.

Corrado, C., C. Hulten, and D. Sichel. 2009. Intangible capital and U.S. economic growth. Review ofIncome and Wealth 55: 661–85.

Crespi, G., A. Geuna, Ö. Nomaler, and B. Verspagen. 2010. University IPRs and knowledge transfer: Isuniversity ownership more efficient? Economics of Innovation and New Technology 19: 627–48.

Davis, L. 2009. Leveraging trademarks to capture innovation returns. Paper presented at theDRUID Summer Conference 2009, June, in Copenhagen, Denmark. Available at http://www2.druid.dk/conferences/viewpaper.php?id=5655&cf=32 (last accessed August 8, 2011).

Etzkowitz, H. 1998. The norms of entrepreneurial science: Cognitive efforts of the new universityindustry linkages. Research Policy 27: 823–33.

Etzkowitz, H. 2003. Research groups as quasi-firms: The invention of the entrepreneurial university.Research Policy 32: 109–21.

Greene, W. 2008. Econometric analysis. 7th ed. Englewood Cliffs, NJ: Prentice-Hall.Greenhalgh, C., and M. Rogers. 2007. The value of intellectual property rights to firms and society.

Oxford Review of Economic Policy 23, no. 4: 541–67.Hall, B.H., Z. Griliches, and J.A. Hausman. 1986. Patents and R&D: Is there a lag? International

Economic Review 27: 265–83.Hipp, C., and H. Grupp. 2005. Innovation in the service sector: The demand for service-specific

innovation measurement concepts and typologies. Research Policy 34, no. 4: 517–35.Kutinlahti, P. 2005. Universities approaching market. Intertwining scientific and entrepreneurial

goals. Espoo: VTT Publications 589.Leydesdorff, L., and M. Meyer. 2010. The decline of university patenting and the end of the Bayh–Dole

effect. Scientometrics 83, no. 2: 355–62.Lom, H. 2004. Branding: How to use intellectual property to create value for your business? World

Intellectual Property Organization (WIPO). http://www.wipo.int/sme/en/documents/branding.htm (last accessed August 8, 2011).

Long, J.S. 1997. Regression models for categorical and limited dependent variables. Thousand Oaks,CA: Sage.

Long, J.S., and J. Freese. 2006. Regression models for categorical dependent variables using stata.2nd ed. College Station, TX: Stata Press.

Malmberg, C. 2005. Trademark statistics as innovation indicators? – A micro study. WorkingPaper No. 2005/17, Centre for Innovation, Research and Competence in the Learning Economy(CIRCLE), Lund University. http://www.circle.lu.se/upload/CIRCLE/workingpapers/200517_Malmberg.pdf.

Mendonça, S., T. Santos Pereira, and M.M. Godinho. 2004. Trademarks as an indicator of innovationand industrial change. Research Policy 33, no. 9: 1385–404.

Millot, V., and M. Squicciarini. 2011. Patent and trademark use by firms: An empirical investigation.Mimeo.

Mowery, D.C., R.R. Nelson, B.N. Sampat, and A.A. Ziedonis. 2001. The growth of patenting andlicensing by U.S. universities: An assessment of the effects of the Bayh–Dole act of 1980. ResearchPolicy 30: 99–119.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Economics of Innovation and New Technology 497

Mowery, D.C., R.R. Nelson, B.N. Sampat, and A.A. Ziedonis. 2004. Ivory tower and industrialinnovation: University–industry technology transfer before and after the Bayh–Dole act in theUnited States. Stanford, CA: Stanford University Press.

OECD. 2011. Science, technology and industry scoreboard. Paris: OECD.OECD and Eurostat. 2005. Oslo manual: Proposed guidelines for collecting and interpreting

innovation data. 3rd ed. Paris: OECD.Sandner, P. 2009. The valuation of intangible assets: An exploration of patent and trademark

portfolios. Wesbaden: Gabler.Schmoch, U. 2003. Service marks as novel innovation indicator. Research Evaluation 12, no. 2:

149–56.Shane S.A. 2004. Academic entrepreneurship: University spinoffs and wealth creation. Cheltenham:

Edward Elgar.Thursby, J.G., and M.C. Thursby. 2003. University licensing and the Bayh–Dole act. Science 301,

no. 5636: 1052.Thursby, J.G., and M.C. Thursby. 2007. University licensing. Oxford Review of Economic Policy 23,

no. 4: 620–39.Van Looy, B., M. Du Plessis, and T. Magerman. 2006. Data production methods for harmonized patent

statistics: Patentee sector allocation. Open Access publications from Katholieke UniversiteitLeuven; urn:hdl:123456789/228565, Katholieke Universiteit Leuven.

Verspagen, B. 2006. University research, intellectual property rights and European innovationsystems. Journal of Economic Surveys 20, no. 4: 607–32.

Appendix 1. Examples of universities’ TMs related to research activity (inalphabetical order).

ACSI, University of Michigan. The ACSI is an economic indicator measuring the satisfaction ofconsumers in the USA, based on both customer interviewing and econometric modeling. The projectwas started in 1994 by researchers at the National Quality Research Center, a research unit withinthe University of Michigan, in cooperation with partners at the American Society for Quality inMilwaukee, Wisconsin, and Claes Fornell International Group in Ann Arbor. In December 1995,‘ACSI’ was deposited as a TM at USPTO by the Michigan University. It is registered in class 35,referring more precisely to ‘business and market research and analysis services, namely, the researchand periodic measurement, publication and distribution to others of customer evaluation of the qualityof goods and services purchased in the United States in major industry sectors’. In 2009, a privatecompany was formed called ‘ACSI’ based in Ann Arbor, Michigan. The index is now produced by thiscompany and no longer by the University of Michigan, although Claes Fornell, who was and still isthe principal researcher behind the ACSI, remains a professor at the university (for more information,visit http://www.theacsi.org/index.php).

Bioglass, University of Florida. Bioglass is a biomaterial that was developed at the University ofFlorida at the end of the 1960s. During the years of the Vietnam war, Larry Hench, material engineerat the University of Florida, started research on the possible use of glass as a prosthesis materialfor the soldiers who had their limbs amputated. This research, which received funding from theUS Army Medical R&D Command, led to the creation of Bioglass. Bioglass, initially usedfor bone regeneration, later proved to have multiple useful applications for clinical use, notably inperiodontics (PerioGlas). ‘Bioglass’ was deposited as a TM by the University of Florida, registeredat USPTO in August 1982 in classes 10 (referring to ‘Biologically Active Glass and BiologicallyActive Glass-Coated or Laminated Ceramic or Metal, Formed into Bone Screws, Rods and Pins andOther Dental and Surgical Implants’), 9 (referring to ‘Laboratory Equipment – Namely, BiologicallyActive Glass and Glass-Coated or Laminated Tissue Culture Discs’), and 42 (referring to ‘Services –Namely, Biomedical Engineering Services – Namely, Designing and Fabricating Biologically ActiveGlass and Biologically Active Glass-Coated or Laminated Ceramic or Metallic Dental and SurgicalImplants’) (for more information, visit http://new.novabone.com/history.html).

Brainmap, University of Texas. BrainMap is an online database of published functional neuroimagingexperiments, created and developed at the Research Imaging Institute of the University of TexasHealth Science Center at San Antonio. Its aim is to provide a tool to share methods and results ofstudies in specific research domains, such as language, memory, attention, emotion, and perception.‘Brainmap’ was deposited as a TM by the University of Texas at USPTO first in class 9 in November

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

498 M. Squicciarini et al.

1995 (referring to a ‘computer software for use in the field of human neuroscience, namely, softwarefor creating digitized representations of the human brain’), later in class 16 in March 2004 (referringto ‘printed educational materials, namely, testing booklets in the field of self-assessment and thinkingskills’), and lastly in December 2006 in class 44 (referring to a service ‘providing an online computerdatabase in the field of neurology and brain function; providing a database in the field of neurology andbrain function that also allows inputting and collection of data and information for research purposes’)(for more information, visit http://brainmap.org/).

COEUS, MIT. Coeus is an electronic research administration system for award management, devel-oped by the MIT in the early 1990s. Its purpose is to assist the research community for grant proposalsand pre- and post-award management. ‘Coeus’ was registered as a TM at USPTO by the MIT in July1997 in class 9 (referring to ‘computer software that manages sponsored program databases andprovides for the electronic transfer and management of research grant proposals and data by anymethod, i.e., CD ROM, diskettes, magnetic tape, or downloadable from a remote site, and relateddocumentation’) (for more information, visit http://osp.mit.edu/coeus/).

Loom and Powerloom, University of Southern California. Loom and Powerloom are knowledgerepresentation languages for constructing intelligent applications, developed by researchers in theArtificial Intelligence research group at the University of Southern California’s Information SciencesInstitute. ‘Loom’ and ‘Powerloom’ are two registered TMs at USPTO owned by the University ofSouthern California, both filed in July 2003, and both in class 9 (referring to ‘computer softwareand downloadable computer software, namely, computer program modules and data files that encodeknowledge and ontologies for exploitation by artificial intelligence and other automated reasoningapplications; downloadable electronic user manuals therefore, and electronic user manuals thereforerecorded on computer media’) (for more information, visit http://www.isi.edu/isd/LOOM/).

Octave, Carnegie Mellon. Octave (Operationally Critical Threat, Asset, and Vulnerability Evalu-ationSM) corresponds to a set of methods and tools for risk-based information security strategicassessment and planning. It was developed by the Carnegie Mellon Software Engineering Instituteand launched in the beginning of the year 2000. The TM ‘Octave’ was deposited by the CarnegieMellon University in March 2002 in three different classes 35, 16, and 41, referring, respectively,to ‘business consulting services in the field of information security, risk evaluation and risk man-agement’, ‘printed publications, namely, books and reports in the field of information security, riskevaluation and risk management’, and to ‘education and training services, namely, group workshopsand self-paced classes, and cd-rom presentations, in the field of information security, risk evaluationand risk management’ (for more information, visit http://www.cert.org/octave/).

OMIM, Johns Hopkins University. Online Mendelian Inheritance in Man (OMIM) is a timely com-pendium of human genes and genetic phenotypes, intended for use by researchers, advanced students,and other professionals concerned with genetic disorders. This database was initiated in the early1960s by Dr Victor A. McKusick. It was initially a catalog of Mendelian traits and disorders, entitledMendelian Inheritance in Man (MIM). Several book editions of MIM were published later on, andthe online version, OMIM, was created in 1985 by a collaboration between the National Library ofMedicine and the William H. Welch Medical Library at Johns Hopkins. ‘OMIM’ is a TM deposited byJohn Hopkins University at USPTO in December 2001 in class 44 (described as a service ‘Providinghealth related information, namely, information via an online medium regarding Mendelian relatedinheritance’) (for more information, visit http://www.omim.org/).

Spacewatch, University of Arizona. Spacewatch is a project founded in 1980 by Tom Gehrels andMcMillan at the University of Arizona’s Lunar and Planetary Laboratory. Its purpose is to explorethe various populations of small objects in the solar system and study the statistics of asteroids andcomets in order to investigate the dynamical evolution of the solar system. It is besides involved infinding potential targets for interplanetary spacecraft missions, as well as objects that might presenta hazard to the Earth. The TM ‘Spacewatch’ was deposited at USPTO by the University of Arizonain July 2000. It is protected in classes 41 and 42, referring, respectively, to ‘educational research,namely, research in the field of discovering, identifying, investigating and monitoring asteroids andcomets; development and dissemination of methodology and educational materials for others in thefield of discovering, identifying, investigating and monitoring asteroids and comets’ and to ‘scientificresearch; research in the field of identifying, investigating and monitoring asteroids and comets’ (formore information, visit http://spacewatch.lpl.arizona.edu/).

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Econom

icsofInnovation

andN

ewTechnology

499

Table A1. Universities’ main characteristics, 1997–2007.

Total funds Share of federal Students in Share of Number ofreceived funds Age 2006 (in 1000) graduate universities in State

MedicalYear Nobs. Mean SD Mean SD school Mean Sd Mean SD Mean Sd Mean SD

Overall sample 1997 621 63,829 132,270.37 58% 0.23 18% 105.9 49.36 12 14.63 27% 0.25 22.3 15.531998 68,561 138,995.88 58% 0.23 19%1999 74,447 148,808.17 58% 0.24 18%2000 77,003 159,349.60 57% 0.24 18%2001 77,932 165,267.89 58% 0.24 19%2002 77,210 167,263.73 59% 0.23 19%2003 75,275 170,660.76 60% 0.23 18%2004 79,511 172,162.75 62% 0.23 18%2005 77,167 175,211.13 62% 0.22 18%2006 81,656 176,785.26 62% 0.22 18%2007 162,504 370,335.96 61% 0.22 18%

Public universities 1997 369 62,753 144,902.65 55% 0.22 19% 97 46.71 16 16.76 24% 0.21 19.6 14.461998 67,265 151,957.05 55% 0.22 19%1999 72,549 163,059.31 55% 0.23 19%2000 75,068 172,840.22 55% 0.23 19%2001 75,817 178,664.38 56% 0.23 19%2002 76,340 182,485.71 57% 0.23 19%2003 76,427 189,153.40 59% 0.22 19%2004 79,466 187,391.35 60% 0.23 19%2005 76,655 191,491.30 62% 0.21 19%2006 78,373 187,863.42 61% 0.21 19%2007 165,352 407,669.27 60% 0.21 18%

Private universities 1997 252 65,342 112,580.29 62% 0.23 18% 119 50.31 6 6.94 34% 0.3 26.3 16.21998 70,434 118,249.00 62% 0.24 18%1999 77,261 125,193.88 62% 0.25 18%2000 79,892 137,230.12 62% 0.25 18%2001 81,212 142,509.52 62% 0.24 18%2002 78,532 141,583.00 62% 0.23 18%2003 73,516 138,255.91 63% 0.24 18%2004 79,581 146,368.46 64% 0.24 18%2005 77,968 146,787.88 63% 0.24 18%2006 87,046 157,379.45 65% 0.22 18%2007 158,026 303,660.14 63% 0.24 18%

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

500M

.Squicciarinietal.

Table A2. Examples of academic TMs in the various classes relating to the output of the university’s research (name of the TM and description of the productscovered).

Examples

Class Description Word Mark Applicant Description

1 Chemical products ADEASY Johns Hopkins University, The Biological materials, namely, DNA vectors used to produce recombinant adenovirusesfor scientific and medical research

PANTHERSKIN Florida International University Corrosion- and fire-inhibiting chemicals for use in the manufacture of steel structures,transport vehicles, and industrial machinery

5 Pharmaceuticalproducts

CHICAGO BCGVACCINE

University of Illinois Bacillus Calmette Guerin vaccine

DEUTRANE University of Iowa Inhalational anesthetics for surgical use9 Computer software 3D-AUTOLAYOUT Carnegie Mellon University Computer programs for solving three-dimensional layout and design problems

ACAPELLA University of Washington DNA and protein sample analysis instrumentation for research purposeMAMMO-FAX University of Pittsburgh Hardware and software for use in analyzing mammograms and transmitting them by

telefacsimileSWIMSOUND Virginia Commonwealth

UniversityWaterproof apparatus for listening to radios or mp3 players via bone conduction

during recreational and competitive swimming or while bicycling, jogging, orhiking in rain or high-moisture environments

10 Medical apparatus ELECTROGENE University of Pennsylvania Electronic medical apparatus used in the treatment of arthritis and othercartilage-related diseases and cartilage damage, for human and veterinary use

SMART DRAIN Alfred E. Mann Institute forBiomedical Engineering atthe University of SouthernCalifornia

Surgical drain with optical fibers for viewing tissue conditions, which do not featuremicroprocessors, computing capabilities, or electronic control capabilities

31 Plants and seeds,organic products

ACALYPHABOURBONSTREET

University of Georgia Live plants, namely, Acalypha godseffiana plants

CROSBREED Rutgers, The State Universityof New Jersey

Disease-resistant oysters

ORGENIC Auburn University Living animals, namely, transgenic animals and transgenic catfishDow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

Econom

icsofInnovation

andN

ewTechnology

501

35 Business man-agement,business services,advertising

DEMONTECH DePaul University Business consultation services in the field of start-up high-technology businesses;consultation in the field of human resources and employee recruitment

FLORIDA SCAN Florida State University Forecasting information concerning the economic condition of the state of FloridaNASA TECHLINK Montana State University Business consultation services, namely, technology transfer assistance, technology

transfer agreement formation assistance, technology licensing assistance andbusiness assistance in areas of technology commercialization, planning, andstrategic marketing

36 Finance, insurance,real estate

CARNEGIEMELLONCYLAB

Carnegie Mellon University Consultation and research for others in the field of risk management

CREIGHTONPORTFOLIOINDEX

Creighton University Financial research and information services; financial consulting services, namely,expert analysis in finance

42 Research services HUMAN CAPITALLAB

Bellevue University Research and development and consultation related thereto in the field of humancapital investment and management

MINE-TO-MILL University of Queensland Engineering consultancy services; consultancy services relating to the extraction,handling, refining, and processing of ores, metals and minerals

SHARPBRAIN University of South Carolina Research and development of products that enable senior citizens to maintainintellectual activity, brain health, and memory in order to facilitate independentliving

SPEED RX University of Vermont andState Agriculture College

Laboratory services, namely, predicting the speed and efficacy of blood-clottingagents

UNHCEMS University of New Hampshire Providing an online computer database in the field of chemical and environmentalmanagement

44 Medical services ... ON THEIR WAY The Vanderbilt University Health care services; hospital services; and medical services;COMPLETELIFE Indiana University Psychological counseling; dietary and nutritional guidance; music therapy for

physical, psychological, and cognitive purposes; massage; and oncology pharmacyconsulting

TRAUMA BURNCENTER

University of Michigan Medical services, namely, patient care and medical research

Source: USPTO (2011), ‘The USPTO Trademark Casefile Dataset (1884–2010)’.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

502M

.Squicciarinietal.

Table A3. Top 50 academic applicants of patents and TMs (1997–2007).

TMs in classes 1, 5, 9, 10,Patents TMs 35, 42, and 44

University of California 2962 University of Pennsylvania 169 University of Pennsylvania 75MIT 1075 University of Texas, Austin 164 University of Chicago 46California Institute of Technology 887 Georgia State University 162 University of Texas, Austin 45University of Texas, Austin 821 Georgia Institute of Technology 162 University of Washington, Seattle 33Stanford University 747 University of Central Florida 120 University of California 33University of Wisconsin, Madison 732 University of California 117 Carnegie Mellon University 32Johns Hopkins University 674 Harvard University 105 Harvard University 29University of Michigan, Ann Arbor 597 Tulane University 91 Colorado State University 28University of Florida 469 University of Washington, Seattle 82 Vanderbilt University 25Cornell University 439 University of Wisconsin, Madison 78 Johns Hopkins University 23Columbia University 405 Rice University 76 Oregon Health & Science University 23Harvard University 396 University of Chicago 74 University of Central Florida 21University of Washington, Seattle 379 Carnegie Mellon University 71 University of Michigan, Ann Arbor 21Georgia Institute of Technology 374 Ohio State University, Columbus 70 University of Southern California 21University of Illinois, Urbana-Champaign 372 Rutgers the State University of NJ, New Brunswick 69 Georgia Institute of Technology 17University of Pennsylvania 362 University of Pittsburgh, Pittsburgh 68 Georgia State University 17Michigan State University 339 Vanderbilt University 66 University of Illinois, Urbana-Champaign 16Duke University 336 University of Florida 65 University of North Dakota 15Pennsylvania State University, University Park 310 University of Michigan, Ann Arbor 64 Yale University 15North Carolina State University 265 Auburn University 61 Duke University 15New York University 264 Florida International University 61 Tulane University 14University of Utah 228 University of Southern California 59 Ohio State University, Columbus 14Princeton University 222 University of Nebraska, Lincoln 57 Georgetown University 13University of South Florida 211 University of Illinois, Urbana-Champaign 52 University of Nebraska, Lincoln 12Purdue University, West Lafayette 208 Syracuse University 51 Stanford University 12University of North Carolina, Chapel Hill 204 Brigham Young University, Provo 51 University of Rochester 12Northwestern University 202 University of Akron, Akron 48 University of Florida 12D

ownl

oade

d by

[U

nive

rsity

of

Sask

atch

ewan

Lib

rary

] at

08:

17 0

6 O

ctob

er 2

012

Econom

icsofInnovation

andN

ewTechnology

503

University of Chicago 201 Arizona State University, Tempe 48 Auburn University 11University of Iowa 195 Johns Hopkins University 47 University of Arkansas, Little Rock 11Yale University 192 American University 43 University of Colorado, Boulder 11Rice University 190 University of Hawaii 42 Brigham Young University, Provo 11Ohio State University, Columbus 180 Dartmouth College 42 University of New Hampshire, Durham 10Rutgers the State University of NJ, New Brunswick 174 Skidmore College 41 Cornell University 10University of Rochester 169 University of Oklahoma, Norman 40 Indiana University, Bloomington 10Texas A&M University 168 University of Arkansas, Little Rock 40 Baylor College of Medicine 10University of Massachusetts, Boston 162 Yale University 40 Wake Forest University 9Baylor College of Medicine 147 University of Missouri, Columbia 38 University of Iowa 9University of Arkansas, Little Rock 146 Northwestern University 36 University of South Carolina, Columbia 9Emory University 145 Baylor College of Medicine 36 University of Missouri, Columbia 9University of Kentucky 142 Indiana University, Bloomington 36 Arizona State University, Tempe 8University of Maryland, College Park 139 Duke University 35 Rutgers the State University of NJ, New

Brunswick8

University at Albany 138 West Virginia University 35 University of Utah 8Vanderbilt University 136 Case Western Reserve University 34 Columbia University 8University of Central Florida 132 Drexel University 34 Washington University in St. Louis 8Carnegie Mellon University 128 University of New Mexico, Albuquerque 34 University of Alabama, Birmingham 8University of Missouri, Columbia 127 University of South Dakota 33 University of Akron, Akron 8Washington University in St. Louis 126 Tufts University 32 Northwestern University 8University of Colorado, Boulder 122 University of Iowa 32 University of Maryland, Baltimore 7University of Virginia 118 Oregon Health & Science University 32 Medical College of Wisconsin 7Iowa State University 117 University of North Carolina, Chapel Hill 32 Medical University of South Carolina 7

Source: The authors’ own compilation on OECD (2011), Patent Database, USPTO (2011), ‘The USPTO Trademark Casefile Dataset (1884–2010)’, and MUP data.

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2

504 M. Squicciarini et al.

Table A4. Comparing models: AIC and BIC results.

Log LogLikelihood Likelihood

Obs (null) (model) dF AIC BIC

Model 1: Poissonpanel FE

patgrant 2911 . −4754.49 13 9534.97 9612.661

patapp 2911 . −4747.32 13 9520.647 9598.338Model 2: Poisson

pooledpatgrant 4232 −10,672.2 −7621.592 32 15,307.18 15,510.4

patapp 4232 −10,672.2 −7634.653 32 15,333.31 15,536.52Model 3: Tobit

pooledpatgrant 4232 −5922.415 −5234.822 33 10,535.64 10,745.21

patapp 4232 −5922.415 −5233.738 33 10,533.48 10,743.04Model 4: ZIP pooled patgrant 4232 −6236.922 −5625.121 64 11,378.24 11,784.67

patapp 4232 −6243.409 −5626.396 64 11,380.79 11,787.22

Dow

nloa

ded

by [

Uni

vers

ity o

f Sa

skat

chew

an L

ibra

ry]

at 0

8:17

06

Oct

ober

201

2