Contents lists available at SciVerse ScienceDirect
Research Policy
j our nal ho me p ag e: www.elsev ier .com/ locate / respol
niversity effects on regional innovation
obin Cowana,b,∗, Natalia Zinovyevac
BETA, University of Strasbourg, 61 avenue de la Forêt Noire, 67085 Strasbourg, FranceUNU-MERIT, Maastricht University, Keizer Karelplein 19, 6211 TC Maastricht, The NetherlandsInstitute of Public Goods and Policies (IPP-CSIC), Calle Albasanz 26-28, Madrid 28037, Spain
r t i c l e i n f o
rticle history:eceived 30 June 2011eceived in revised form9 September 2012ccepted 11 October 2012
a b s t r a c t
This paper analyzes empirically whether expansion of a university system affects local industry innova-tion. We examine how the opening of new university schools in Italy during 1985–2000 affected regionalinnovation. We find that creation of new schools increased regional innovation activity already withinfive years. On average, an opening of a new school has led to a seven percent change in the number ofpatents filed by regional firms. The evidence suggests that the effect is mainly generated by high quality
Between 1960 and 2000, there was a large expansion in univer-ities in the industrialized countries. Early expansion was to dealith the baby boom coming of university age; later expansion wasriven by the desire to increase the proportion of the populationeceiving tertiary education.1 The clearest effect was just that: anncrease in the general education level of the labor force. Naturally,
rise in student numbers tended to be accompanied by a rise in theize of the professoriate and an increase in the sizes and numbersf universities.
University expansion coincided with spectacular rise of innova-ion activity in industrialized world. In 1963 the US Patent Officeranted around 45 thousand patents; by the end of the nineties theearly number of granted patents approached 160 thousand (Hallt al., 2001). How to maintain this competitiveness and get morennovation out of a knowledge system has become a hotly debated
ssue. Following the line taken in the literature on innovation sys-ems, it is often suggested that stimulating academic research andlose interactions between academia, industry and government are
E-mail addresses: [email protected] (R. Cowan),[email protected] (N. Zinovyeva).1 According to the Global Education Digest 2009 by UNESCO Institute of Statistics,
he share of students in North America and Western Europe that enroll in tertiaryducation during five years after the end of secondary education increased by 41ercentage points from 30% in 1970 to 71% in 2007.
necessary to promote knowledge flows and innovation. These pol-icy suggestions are often based on the idea that universities havewithin them some of the keys to increasing innovative activity.2
The fact that the increase of innovation activity during pastdecades coincides with the increase in the size of the university sec-tor might suggest that the innovation performance of an economyis determined in part by the supply of universities in the innovationsystem. This hypothesis motivates our analysis.
There have been many studies of the relationship betweenuniversities and industrial innovation, particularly at the regionallevel (see Section 2.1). The vast majority of these studies analyzecross-sectional data, focusing on either the presence or size ofuniversities and the relationship with local innovation activity.Generally, they document a strong relationship between universityresearch activity and industrial innovation. But there are well-known difficulties in drawing conclusions from cross-sectionalanalysis about phenomena that take place over time, so while theresults are suggestive, one must be cautious in drawing the “obvi-
ous” policy conclusions from them, particularly in terms of whetheropening new universities is a good idea. Additionally, endogene-ity problems are rife in this kind of work – some of the effects of
2 An OECD 2007 report “Higher Education and Regions: Globally Competitive,Locally Engaged” estimates that only 10% of UK firms currently interact with uni-versities with most university–industry links focusing on big business and a fewhi-tech fields. The report concludes that “the potential of higher education institu-tions to contribute to the economic, social and cultural development of their regionsis far from being fully realized”.
niversity–industry interaction are driven by supply of knowledge,ome by demand for it; external factors may drive both public andrivate research output simultaneously; the location of universi-ies and firms is often endogenously determined (Mairesse and
ohnen, 2010) – all imply that identifying empirically the effectf universities would ideally rely on exogenous shocks to univer-ity supply. Such shocks are rare in real life and most studies relyn strong assumptions to claim the existence of the supply-sideffects.
There was, however, a period of several years in the 1980s and990s in which Italy opened many new university schools in differ-nt regions of the country.3 University expansion was centralizednd, as was acknowledged later by policy makers, the distribu-ion of new schools across regions was largely independent of theroperties of the regional economy. In fact no significant corre-
ation can be observed between the number of new schools in region and regional characteristics including population, sharef graduates in the labor force, private and public investment inesearch and development, and value added produced by differ-nt economic sectors. We use this episode to ask directly whetherxpanding university activity by opening new universities has andentifiable effect on local industrial innovation. This is the firstssue we address in this paper.
The second issue has to do with the nature of the relationshipetween universities and industrial innovation. There have beeneveral studies on the “channels” of interaction between univer-ity and industry (see Section 2.2). By and large, these studies areased on firm surveys, asking firms about their external sources ofnowledge or information. As one might expect, firms use many dif-erent channels for accessing university expertise: academic papersr patents, conferences, seminars, consulting, and so on. But oneould frame the question in a slightly different way. What measuresf university activity help explain their effects on local innova-ion? Scientific publications are thought to represent advances inasic knowledge. Patents represent advances in applied knowl-dge. Both of these activities indicate human capital capable ofroducing novel knowledge, basic and applied respectively. Weonstruct measures of these activities using data from Thompson ISInd the European Patent Office. Additionally though, universitiesight possess other competences harder to quantify or describe,
or example skills or accumulated knowledge that can be appliedo issues other than creating novelty. These too could be of valuen industrial innovation activities. In the latter part of the paper weerform an accounting exercise in an attempt to assess whetherhe human capital associated with creating new basic knowledge,reating new applied knowledge, or something different is whatrives the university effect on industrial innovation.
For two reasons we focus on the short-term effects of academicesearch. First, it is likely that regional collaboration networks growastest in the first few years after opening of new university schools.econd, considering the short-run effect of universities allows us
o identify the direct influence of academic research on innovationctivity and to exclude other channels. In particular, it permits us tovoid the issue of how graduates contribute to innovation.4 So by
3 In the Italian system teaching is organized into schools (facoltà) and researchs organized into departments (dipartimenti). Departments and schools may or mayot coincide. To simplify presentation, we refer only to “schools”, and our measure ofhe date of opening of a new school is the year in which the first class was registeredithin a newly formed school. This should not be read to imply that university
xpansion affected only teaching. A new school in most cases implied creation of aew department. This conflation of schools and departments, teaching and researchnits, is not an issue for our analysis, as both measure university presence in theegion.
4 The effects of an increased quantity and quality of graduates in a region areikely to be very diffuse and hard to identify. However, they do not emerge within
Policy 42 (2013) 788– 800 789
focusing on the short term effects, we can identify direct knowledgespillover effects from university faculties to local industries.
Our results suggest that there is indeed a significant effect ofthe creation of new university schools on regional research andinnovation activity. Industrial patenting increases following theintroduction of a new school to a region: on average, one newschool has led to about a seven percent increase in the numberof patents filed by regional firms five years later. But the qualityof patents produced as a consequence of university supply shockis not different from the rest of regional patents. Given that thelevel of development of a region affects its absorptive capacity,one might expect that more developed regions with more inten-sive R&D activity benefit more from interactions with universities.However, contrary to this hypothesis, we find that less devel-oped regions benefit more from university–industry interactions.Regarding the second issue, we find that the number of academicpatents explains essentially none of the effect of universities oninnovation. Publications corrected for quality explain most of theeffect of universities on local industrial innovation. This suggeststhat in order to increase regional innovation the intermediate pol-icy goal should be to increase the amount of high quality academicresearch carried out in the region.
The rest of the paper is organized as follows. Section 2 reviewsthe existent empirical findings concerning the role of academicresearch in innovation systems. Section 3 describes the data. Sec-tion 4.1 introduces the empirical model and comments on the mainidentification assumptions. The results of the empirical analysis areprovided in Sections 4.2 and 4.3. Finally, Section 5 concludes.
2. Background literature
2.1. Identifying the effect of university R&D
There exists a large literature analyzing the relationshipbetween academic research and industrial innovation activity. Thatuniversity effects on industrial innovation might be localized stemsfrom the nature of knowledge. While to a great extent the businessof universities is to produce codified knowledge, tacit knowledgeremains central in the diffusion process (see for example Cowanet al., 2000). While codified knowledge can be diffused very widely,and now very rapidly, tacit knowledge, by its nature, cannot. Jaffeet al. (1993) showed that diffusion of the knowledge contained inpatents, which are by definition highly codified, has a strong geo-graphical pattern – diffusion is very much local, and access to theknowledge spreads geographically over time. Breschi and Lissoni(2009) revisited this issue and showed that in fact it is social ratherthan geographic distance over which the diffusion takes place. Thatis, inventors learn about the existence of a patent (and presum-ably the knowledge it contains) through their direct social contacts.Since most social contacts are local, we can expect (geographically)localized knowledge diffusion.
As early as the 1980s it was suggested that technology clusterssuch as those in Massachusetts and California would be impossiblewithout the technology transfer from universities in these areas(Saxenian, 1985; Dorfman, 1983). It was not long though, beforeseveral case studies questioned the generality of the role of uni-
versity as an accelerator of regional innovation (Feldman, 1994a;Rogers and Larsen, 1984) and suggested that various character-istics of regional technological infrastructure (business services,
five years after a school opens: the official duration of most degrees in Italy (inthe period analyzed) is five years. But fewer than 20% of graduates complete theireducation on time and, on average, students take two more years to graduate afterthe end of the official program (Bagues et al., 2008).
7 search
to
rastsi(
oaaiqbtaksRbtriTslspa2c
etaiapdso
tdPaepmcomSttaava(
tv
edge transfer from academia, as perceived by survey respondents(Cohen et al., 2002; Cassiman et al., 2008; Bekkers and Bodas Freitas,2008). While publication is necessarily strongly correlated with the
90 R. Cowan, N. Zinovyeva / Re
echnologically related firms, etc.) are necessary for developmentf university research outputs.
To understand the magnitudes of possible effects of universityesearch on industrial innovation, Jaffe (1989) provided a moreggregate econometric analysis. He used data for 8 years for 29 UStates to test whether there is an impact of university R&D on indus-rial patenting, and found a significant positive effect. Several latertudies confirmed Jaffe’s finding, using firms’ product and processnnovations instead of patents as a measure of innovative activityAcs et al., 1992; Feldman, 1994b; Feldman and Florida, 1994).
The challenge that runs throughout the empirical literaturen technology transfer is the problem of identifying causation in
system rife with endogeneity. A positive association betweencademic research and industrial innovation may not necessar-ly imply that universities increase local innovation activity. It isuite possible that increases in university outputs are “causedy” increases in industrial R&D and associated with easy accesso industrial inputs such as equipment or materials. A morective industrial R&D sector may facilitate and stimulate universitynowledge production. Thus the causation may run in the oppo-ite direction, and if there is inertia in the variables (industrial&D and university papers and patents) as seems very likely toe the case, then extracting the causal direction is difficult statis-ically. Increases in university research production might also beeinforced by the self-selection of academics able to benefit fromnteraction with industry into highly innovative industrial districts.his will introduce a bias into the types of activities of univer-ities, changing the types of outputs depending on the nature ofocal innovation activities. This would be consistent with resultshowing a positive correlation between professors’ scientific out-ut (as measured by published papers), and their applied outputs measured by patents (Carayol and Matt, 2004; Stephan et al.,007). Again this causes problems for statistical understanding ofausation.
In order to address those endogeneity problems, Jaffe (1989)stimated a system of three equations: the first equation charac-erizing the effect of industrial and university R&D on patenting,nd two equations describing the determinants of, respectively,ndustrial R&D and university R&D. To identify the model, Jaffessumed that industry R&D does not depend on the number ofrivate and public institutions and that university R&D does notepend on manufacturing value added, once, respectively, univer-ity and industrial R&D are taken into account. Thus the consistencyf Jaffe’s findings depends on the validity of these assumptions.
Econometric analysis of university effects on industrial innova-ion at the regional level relies on the assumption that knowledgeiffusion and spillovers are geographically localized. Bottazzi anderi (2003) analyze the effect of total regional R&D on innovationnd find that in Europe the effect of R&D is very localized andxists only within a distance of 300 km. Andersson et al. (2009)rovide evidence suggesting that spillovers from university invest-ent might be even more localized. They analyze the effects of
hanges in the Swedish university system and find that roughly halff the productivity gains from aggregate university investments areanifest within 5–8 km of the community in which they are made.
ome authors have claimed that the evidence of firms’ dispropor-ionate location in areas close to universities already suggests thathe potential positive interactions between industry and universityre likely to be quite localized (Abramovsky et al., 2007; Audretschnd Stephan, 1996). Still, there is also evidence suggesting that uni-ersity R&D might be related to patenting activity much further,way following the collaboration networks of university professors
Ponds et al., 2010).
Many studies at the firm level have confirmed that those firmshat collaborate with universities have higher propensities to inno-ate (Loof and Brostrom, 2008; Zucker et al., 1998), especially in
Policy 42 (2013) 788– 800
those technological areas that require frontier scientific knowledge(Hall et al., 2003).
Once again, even if the empirical evidence suggests that indus-trial innovation and university research tend to cluster in the samelocations and university–industry collaborations are more frequentin highly innovative firms, it is difficult to claim empirically thatthe intensity of university research influences the innovativeness ofindustrial sector, and not vice versa. Typically, one needs to rely onobservable characteristics, which could be used as controls, and toassume that they are sufficient to exclude any correlated effect.
To be policy relevant, it is also important that the observed uni-versity effect is not actually a crowding out effect of private R&D,i.e. that university supply does not lead to a substitution of privatefunding for public funding (Mairesse and Mohnen, 2010).
2.2. Transfer of knowledge and expertise
University research is conducted under a very different sys-tem of incentives than is that in the private sector (Stephan, 1996;Dasgupta and David, 1994). This could easily lead to the accumula-tion of very different types of human capital and knowledge in thetwo sectors, and thus raises the possibility of synergies betweenthem in innovation activity.
The most obvious competence associated with universityresearchers is that aimed at expanding the knowledge frontier. Thishas been the traditional research function of universities, and isseen as perhaps the main activity of the professoriate. Publicationis traditionally the activity most highly valued in the univer-sity setting. More recently, though, academic scientists have beenencouraged to produce applied knowledge. Here again, since theaim has been to express this knowledge in terms of patents, noveltyis paramount. The types of human capital needed to produce thesetypes of novelty are likely to be well-proxied by publication andpatenting measures, and indeed, the literature on the channels ofknowledge communication between industry and university findsthat publications and patents, particularly the former, tend to beimportant channels of communication.
Academic patents are often discussed, especially by policy mak-ers, as one of the main channels of knowledge and technologytransfer from university. In part, this belief motivated the U.S. Bayh-Dole Act (1980), which gave permission for US universities to patenttechnology developed with federal funds. The underlying rationalewas that this should speed up technology transfer by bringing newcommercialization opportunities to the market.5 In Europe, manyuniversities have also recently adopted technology transfer poli-cies. But at the same time, many academics expressed concernsabout potential detrimental effects of incentives to patent on thetype and the quality of the research output produced (Lundvall,1992; Henderson et al., 1998). Contrary to the apparent belief ofpolicy makers, the empirical evidence tends to suggest that aca-demic patenting per se is not a key channel of technology transfer(Agrawal and Henderson, 2002; Arundel and Geuna, 2004; Cohenet al., 2002; D’Este and Patel, 2007).
It is often argued that the transfer of university knowledge couldalso be spread through the more traditional academic channelssuch as scientific publications, seminars or face-to-face interac-tions. In fact, publications (as well as co-publishing with industry)are often found to be among the most important channels of knowl-
5 Market failure theory suggests that due to the public good nature of knowledge,private companies have little incentive to invest in developing an invention that isnot protected by a patent.
search Policy 42 (2013) 788– 800 791
aicoapsko
gdi((apaoa
2
riacrci
weas1asaptt
clIE
ea
eap(m
vbelaosu
0
2
4
6
8
10
12
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
Num
ber
of
new
schools
Year
R. Cowan, N. Zinovyeva / Re
bility to produce frontier knowledge, other channels discussedn that literature may be much less so. Conferences, workshops,onsulting and so on may be avenues for transmitting somethingther than cutting-edge knowledge. Academics may have generalnalytic or synthetic skills that can be used for things other thanublishable research; they have (in principle at least) pedagogickills gathering, synthesizing, codifying and delivering existingnowledge. All of these could be transferred or facilitate transferf useful knowledge.
One challenge arising from this literature stems from the sug-estion that subjectiveness of survey responses can lead to veryifferent opinions about channels’ importance, depending on who
s asked to evaluate it. For instance, Bekkers and Bodas Freitas2008) compare perceptions of the importance of academic patentsamong other things) as a channel of technology transfer amongcademics and private sector R&D workers and report that therivate sector considers them to be twice as important as doescademia. In this paper we avoid this problem and do not focusn channels of transfer, rather we use an accounting procedure toddress the issue of different types of knowledge flow.
.3. This study in the literature, and hypotheses
The extent to which knowledge flows from university affectegional innovation and which type of academic expertise mattersn this respect, are important empirical questions. Still, as discussedbove, empirical analysis of these issues faces several methodologi-al problems related to the identification of the effects of universityesearch. The presence of those problems often opens a window forriticism of the recent empirical findings on university effects onndustry.
In what follows we perform a statistical analysis to examinehether a rapid growth of the university system in Italy had an
ffect on local industrial innovation activity. University schools cre-tion in the 1980s and 1990s was part of a plan to expand educationupply, adopted by the Italian government in the beginning of the980s.6 The plan sought to unburden over-crowded universitiesnd to improve graduation rates.7 We exploit the fact that, as wehow in Section 4.1, the rationale of this rapid university expansionnd consequently the timing of new university openings was inde-endent of the demands of local innovation systems. This allows uso avoid many of the endogeneity problems typically attendant onhis type of study.
The focus on Italy is especially relevant since it is one of theountries that has to catch up with other European countries in theevel of innovation activity in firms. According to the Communitynnovation Survey (CIS) 2008, Italy still lags behind the majority ofuropean countries in the percentage of firms that innovate.
Finally, we explore which type of academic knowledge andxpertise is effectively transferred to industry. In particular, were interested in understanding the extent to which industrial
6 Some new university openings were already approved in the late 1970s. How-ver, the substantial reform came with law n. 382 11/7/1980, which provided thatny variation in the existing university supply should be included in a developmentlan (piani triennali), to be approved by the Minister of Education every three yearsLaw no. 590 14/8/1982). Some autonomy was introduced starting from 1995. For
ore details see Bratti et al. (2008).7 In most cases the new schools were opened within previously existing uni-
ersities, but often located in different towns. With time some of these schoolsecame independent universities (for example, what is now the University of East-rn Piedmont was founded in 1998 on the basis of schools of the University of Turinocated in Vercelli, Alessandria and Novara). In a few cases the new units appeared as
result of the split of over-crowded universities in big megalopolises (the Universityf Rome III was founded in 1992 simply by taking part of the staff from Univer-ity of Rome La Sapienza). There are very few examples of opening completely newniversities from scratch (one such is the University of Teramo, founded in 1993).
Fig. 1. Annual number of new university schools (excluding schools in humanitiesand social sciences), 1985–2000.
innovation is affected by the activities and human capital associatedwith professors’ publishing as opposed to those associated withthe more applied activities associated with patenting. The answerto this question can help universities and policy makers to makejudgements about the alignment of existing incentive structures inacademia to the necessities of local industries.
3. Data and variables
The analysis is performed using Italian data at the regional level.The database includes characteristics of the university system, indi-cators of industrial and academic innovation activity and economicindicators observed for 20 Italian regions between 1984 and 2000.
Our main indicator of the university presence in the region isthe number of university schools in science, medicine and engineer-ing. We consider the date students were first enrolled in the degreeprogram of the school as the date of the creation of this school. Infor-mation about the number of first-year students at the school levelwas obtained from different issues of the Italian National StatisticalBureau bulletins (L’università in cifre and Lo Stato dell’Università).8
According to this definition, 65 schools in science, medicine andengineering were opened for enrollment between 1985 and 2000.(Fig. 1 describes the dynamics of university expansion across timeand Fig. 2 shows the geographical distribution of new schools.) Outof the total of 65 new schools, 29 schools were in civil and industrialengineering, 12 in sciences, 11 in agriculture and veterinary, and13 in medicine, pharmaceutics and chemistry. The average Italianregion has nine schools and every fifth region in a given year openeda new school (Table 1).
The number of schools might seem too aggregate an indica-tor of university presence, since schools can vary considerablyby size. However, we have sufficient evidence to believe that thevariation in the number of schools was exogenous to regionalinnovation activity and the dynamics of other factors affectinginnovation. By contrast, the number of professors hired by thesenew schools is likely to have been determined by the demand foreducation. This demand could be correlated with the innovation
activity in the region. So the number of professors is unlikely tobe an exogenous shock. Similarly to the case of school size, anyindicator of university–industry collaborations (such as the num-ber or size of university technology-transfer offices) would suffer
8 The data are available from 1984 from the printed annual editions of L’universitàin cifre and Lo Stato dell’Università (available in most university libraries in Italy).Data for years from 1988 on were accessed at http://ionio.cineca.it, in November2007.
Private R&D investment, mln euros 229 421 439 582 147 205 48 59Public Non-University R&D investment, mln euros 90 188 76 71 199 341 27 29Public University R&D investment, mln euros 150 128 154 125 186 140 119 117Population, mln 2.4 1.8 2.7 2.3 2.1 1.4 2.1 1.6Population of 19-olds in total population, % 15.7 4.2 13.3 3.4 15.0 3.7 18.9 3.3University graduates in the labor force, % 7.6 2.2 6.9 2.1 8.3 2.5 7.9 1.8VA per capita, thousand euros: 14.8 5.5 18.1 5.5 14.9 4.6 11.0 3.3
Industrial VA in total VA, % 21.9 7.5 25.7 7.7 23.1 6.7 16.7 4.3Services VA in total VA, % 67.2 6.7 65.2 7.3 67.4 7.3 69.4 4.7Agriculture VA in total VA, % 4.3 2.0 3.1 1.2 3.6 1.3 6.1 1.6Construction VA in total VA, % 6.6 2.1 6.1 1.9 5.9 1.7 7.8 2.1
Notes: (*) Total number of regions is 20. Regions classified as “Northern” are Piedmont (PIE), Aosta Valley (AOS), Lombardy (LOM), Friuli-Venezia-Giulia (FVG), Trentino-AltoA s “Cenr , Cam1
fcpaa
ptpemt(uupri
ta
w2
iw
dige (TAA), Veneto (VEN), Emilia-Romagna (EMR), Liguria (LIG); regions classified aegions classified as “Southern” are Abruzzo (ABR), Basilicata (BAS), Calabria (CAL)984–2000.
rom endogeneity.9 An additional problem of indicators of formalollaborations is that they unavoidably miss the important com-onent of informal collaborations (Link et al., 2007).10 All theserguments justify our focus on the number of university schools asn indicator of university presence in the region.
We measure regional innovation productivity by the number ofatents registered in the European Patent Office, using the loca-ion of inventors to determine the region where the patent isroduced.11,12 In order to disentangle the knowledge spilloverffect of universities from the direct effect of university R&D invest-ent on patenting and the crowding out effect, we split patents into
wo groups: those that are produced with university participationor academic patents), and the rest (industrial patents). Note thatntil recently it has been difficult to attach patenting activity toniversity research. In fact, in contrast to the US case, up to the
resent, in Italy universities did not generally retain the propertyights on inventions done by their researchers. Often “IPRs overnventions derived from sponsored research programmes were left
9 In Italy, there were almost no universities that adopted a patent or technologyransfer policy until 1996 (see Baldini et al., 2006), so this is not an option for ournalysis.10 However, it is likely that the intensity of informal collaboration is correlatedith formal collaboration activities due to potential complementarities (Cohen et al.,
002; Grimpe and Hussinger, 2008).11 Specifically, the database includes all patent applications that passed a prelim-nary examination in the EPO. The assigned date of the patent is the priority date,
hich is the date of the first filing world-wide.12 Patents with inventors from two different regions are counted twice.
tral” are Tuscany (TUS), Umbria (UMB), Marche (MAR), Lazio (LAZ), Sardinia (SAR);pania (CAM), Molise (MOL), Apulia (APU), Sicily (SIC). Mean values for the period
to the sponsors” (Balconi et al., p. 133). The recent KEINS EP-INVdatabase on academic patenting (Lissoni et al., 2006) matches thenames of the inventors of the patents with a list of university pro-fessors. Thanks to this methodology, the KEINS database includesnot only any patent owned by universities, but also all patents thatinvolve university scientists, whether the patents are owned byfirms, public research organizations, universities, or the scientiststhemselves. In the following we use the KEINS database to identifyindustrial and academic patents. We observe that an average regionproduced annually 142 patents, seven of which were produced withthe participation of academic inventors.
We measure the quality of innovation by the average num-ber of citations received by these patents before 2004. Naturally,series on patent citations are subject to a truncation bias sincethe number of citations any patent receives grows with time, andour data include citations received only until 2004.13 We correctfor truncation bias following the method developed by Jaffe andTrajtenberg (1996) in the version where the diffusion process isassumed to have the same shape in all technological sectors.14 Fig. 3
shows the evolution of the number of patents, patent citations andcorrected citations in industrial and academic sectors. Patents, both
13 More precisely, citation variables count the number of citations received byregional patents from all Italian patents until 2004.
14 Results of our analysis are not affected if instead we simply use a 5-year windowfor patent citations.
R. Cowan, N. Zinovyeva / Research
(4,8](2,4][0,2]
Fig. 2. Location of schools created during 1985–2000 across Italian regions. Thec8g
it
fhcio
kpwt
osdsucif
na
A
presence and innovation activity could be also justified since theeffects of institutional changes could take time to realize (see theabove findings in Table 2). Below we consider the right-hand-side
olors indicate numbers of schools opened: dark gray indicates that between 5 and schools were opened in the region over the entire period; light gray, 3 or 4; paleray, between 0 and 2.
ndustrial and academic, and citations to them grew steadily overhe period.15
Information on professors’ publication records are obtainedrom ISI Web of Science. We use all publications from 1984 to 2000aving at least one coauthor with an Italian affiliation. We observeitations received by these publications up to 2009. An average Ital-an region has produced more than a thousand publications a year;n average a publication has received about 17 citations.
The extent to which technological innovations rely on scientificnowledge is measured by the propensity of patents to cite non-atent literature (NPL). Not surprisingly, we observe that patentsith inventors from academia draw more on scientific knowledge
han pure industrial patents (column 1, Table 1).We use information from the Italian National Statistical Bureau
n several regional characteristics including private and publicpending on research and development, value-added produced byifferent economic sectors (industry, services, agriculture and con-truction), population, population aged 19, and the proportion ofniversity graduates in the labor force. We apply the depreciationoefficient used by Gordon (1990) (19.3%) to the time-series of R&Dnvestment in order to construct an indicator of the stock of R&Dacilities available in a region.
Note that the intensity of innovation activity is very heteroge-eous across Italian regions (columns 3–8, Table 1). Between 1984nd 2000, regions in the North of Italy were investing almost ten
15 Other ways to treat the truncation bias in citations are discussed in the Appendix.
Policy 42 (2013) 788– 800 793
times more in R&D than Southern regions. These differences arealso reflected in the number of patent applications made by inven-tors from these regions. The gap in the innovation activity acrossthese regions is not due to the size effect: there are no impor-tant population differences across the regions. There are also nosubstantial differences in university presence or in the educationallevel of the labor force. Broadly put, the differences in the innova-tion activity could be attributed to a relatively low income level inthe South and to differences in industrial structure: in the Northmanufacturing has a larger weight in the economy than in theSouth, whereas in the South the service sector and agriculture arerelatively more important.16
4. Empirical analysis
We start by observing simple correlations between the num-ber of new schools opened in a region and the variation in variousindicators of research and innovation activity (Table 2). First, weanalyze whether the opening of new schools is associated withan increase in academic research activity. We find that universityexpansion is associated with the growth of university R&D stockin the region observed about two-five years later (column 1). Sim-ilarly, higher university presence is associated with the rise of thenumber of scientific publications even within the next three years(column 2). We do not observe any clear (significant at standardlevel of 5%) relationship between university expansion and thegrowth of academic patenting within first five years (column 3).Second, we analyze the relationship between university expan-sion and the growth of industrial innovation activity. We find thatthere is a positive correlation between university expansion and thegrowth of industrial patenting a few years later (column 4), eventhough new university units are not associated with the growth ofprivate R&D and income per capita (columns 5 and 6).17
Though suggestive, the results in Table 2 should be consideredwith caution. In principle, the creation of new schools might be notindependent of regional innovation activity. In addition, the abovecorrelations could be confounded by characteristics of the economywhich vary simultaneously with university expansion and regionalinnovation. In the following we make explicit the assumptions ofour identification strategy and analyze the above findings in moredetail.
4.1. Empirical model and identification strategy
The main problem we seek to address is the possibility of circularcausation between university research and industrial innovation.In order to address it, we analyze and build upon the standardreduced-form relationship between industrial innovation output,Pi,t, and the number of schools in the region, Ui,t,
Pi,t = ̨ + ˇUi,t + Xi,t� + ct + ci + �i,t (1)
The simplest way to avoid potential simultaneity problem in model(1) is to consider the right-hand-side variables – including univer-sity presence – with a time lag. The time lag between university
16 Evangelista et al. (2001) also observe that, according to the 1995 wave of theCommunity Innovation Survey, there are very few science-based firms in the South.Regionally disaggregated data are not available in later CIS waves.
17 Note that the difference between the effect of new schools on industrial patent-ing observed in years 3, 4, and 5 is not statistically significant. Throughout the paper,we adopt the conventional level of significance – p-value less than 5%. The estimateswith p-value between 5% and 10% are considered as marginally significant.
794 R. Cowan, N. Zinovyeva / Research Policy 42 (2013) 788– 800
Table 2Correlation between new school opening and variations in the indicators of private and public research activity.
Number of years afterschool opening
(1) (2) (3) (4) (5) (6)
University R&D stock Publications Academic patents Industrial patents Private R&D stock VA per capita
ariables with a 5-year lag. Nevertheless, given the presence oftrong autocorrelation in the data, lagging independent variabless likely to be insufficient to avoid endogeneity.
In order to capture some differences across regions, we includen extensive list of observable regional characteristics among theontrols, Xi,t . The size of the region – both in terms of population,specially younger population, and in terms of economic produc-ion – may reflect inherent local demand for higher education asell as the propensity to innovate. Therefore we control for popu-
ation, the proportion of 19 year olds, and aggregate value-added.ndustrial composition affects the propensity of a region to patent.his in turn may affect the value of certain types of high-skilleduman capital and so the ability of a region to lobby the centralovernment for more university resources. Thus, we control forroportions of regional value-added in industry, construction, ser-ices and agriculture. An additional control for industrial structures the share of graduates in the local labor force. Public non-niversity R&D and private R&D both affect industrial patenting,nd may reflect a general attitude towards the value of knowl-dge production and training, thus again affecting the ability toobby for more university resources. We include both controls inur estimations.
University expansion was stronger in the 1990s than in the980s, coinciding with the rapid growth of innovation activity (seeigs. 1 and 3). In other words, the timing of university expansionenerally might be not independent of the time trend affectingnnovation activity, ct. In order to account for time effects influenc-ng all regions simultaneously, we introduce a set of year dummiesmong the controls.
Notwithstanding the inclusion of the above controls, one might
uspect that regions differ on other perhaps non-observable dimen-ions, and these differences might explain both the universityresence in the region and the development of innovation activ-
ty. In other words, unobserved regional effects, ci, might be
Fig. 3. Evolution of academic (left panel) and industrial (ri
correlated with university presence. Given the panel structure ofour data, we can account for ci in two ways: using a fixed effectestimator (or, equivalently, including the regional dummies amongcontrols) or using a difference estimator. Below we report resultsobtained using the first-difference estimator. We note, though, thatfixed effect estimation produces results that are statistically simi-lar to the ones presented here. We prefer the difference estimatorto the fixed effect specification since the former does not requirestrictly exogenous regressors (that is, it does not require that indus-trial innovation has no impact on future right-hand-side variablesincluded in Xi,t , such as value added and R&D) (Wooldridge, 2002).
With regard to the count nature of our dependent variable, in thefollowing we adopt the traditional approach of modeling regionalinnovation activity using a log–log relationship between universitypresence and regional patents (Jaffe, 1989; Feldman and Florida,1994). We prefer this model to a negative binomial specificationfor two reasons. First, given that the mean number of regionalpatents is quite high (134 patents, see Table 1 for more details),the negative binomial distribution is essentially normal and thelog–log model provides a good approximation. Second, the linearmodel permits capturing region-specific effects without imposingthe strong exogeneity assumption. Performing a negative binomialestimation with predetermined regressors requires a GMM esti-mation for which our sample size is not sufficiently large (Blundellet al., 2002).
Ultimately, our identifying assumption is that the error term,�i,t, is uncorrelated with university presence in the region onceregional time-invariant effects, time effects and observable time-varying characteristics are taken into account. In other words, weassume that during the analyzed period no variations in regional
characteristics (apart from the ones included in Xi,t) affected boththe variation in the number of university schools and the varia-tion in the regional innovation 5-years later. Is this assumptionjustified?
ght panel) innovation outcomes in Italy, 1985–2000.
R. Cowan, N. Zinovyeva / Research
Ff
imwooeas
sctccwscp
t
but it can nevertheless reflect any substantial changes in inventors’reliance on scientific publications and basic knowledge. Still, wedo not observe any significant effect of new schools on the nature
18 New school openings are for the period from 1985 to 2000. The mismatch ratiouses data from the triennial Italian National Statistical Bureau (ISTAT) representativesurvey of graduates, the 1995 edition, which surveys students graduating in 1992.
ig. 4. Number of new schools opened between 1984 and 2000 by regional demandor corresponding professions.
To answer this question we need to understand the factors thatnfluenced university expansion. As was acknowledged by policy
akers ex post, the distribution of new units across the regionsas largely independent of regional labor market demands. The
penings seemed to be associated with an indiscriminate allocationf funds across the regions. In this regard, the Observatory for thevaluation of the university system in the Ministry of Educationnd Research (MURST) after analyzing the expansion of universityystem in the beginning of 90s concludes:
The rules by which new institutions were created does not seemto have followed any logic tailoring university development toterritorial specificities. It seems not to have made reference toa demand for university education (that is, responding to thepotential scope of use of the new initiatives), nor does it seemto have made reference to the demand for graduates (the for-mative needs of the country) or to existing infrastructure. Insubstance, no rigorous evaluations of the initiatives were done,either in absolute terms, or concerning compatibility with therest of the system. The criterion actually favored was geograph-ical re-equilibrium, which aimed to bring the offer of universityeducation and subjects near to the demand, ignoring not onlythe “real” size of this demand (which sometimes turned out tobe less than the minimum requirements for the initiative to beefficient and effective), but also the importance of the trans-portation system, the receptive capacity of the population ofstudents and students’ financial support in determining accessto university establishments. So, [. . .] at least to a large extent,the prevalent logic was the one of incremental expansion anddistribution “by drops of rain”, without evaluating other ini-tiatives that were suppressed [. . .]. (p. 3, Verifica dei piani disviluppo dell’universita 1986–90 e 1991–93, Osservatorio perla valutazione del sistema universitario, MURST, 1997; authors’translation).
This evaluation of the Italian Ministry of Education and Researchupports our identifying assumption. In Table 3 we also analyze theontemporaneous correlation between observable regional charac-eristics included in Xi,t and university expansion. Consistent withharacterization done by the Italian Ministry, observable regionalharacteristics in the analyzed period seem to be poorly correlatedith university expansion. Correlation between the number of new
chools and the contemporaneous dynamics in observed regionalharacteristics, including the growth of regional patents, is very
oor as well.
Similarly, the opening of new schools does not seem to be relatedo the demand for particular professions. In Fig. 4 we plot the degree
Policy 42 (2013) 788– 800 795
of the fit of educational supply to local demand for skilled laborin a region-discipline (measured as the number of new graduatesincorporated in the labor force over the number of new gradua-tes from local universities) versus the number of new schools ineach region-discipline.18 An economically driven policy might aimto locate schools in regions that were importing skilled labor, lead-ing to a positive correlation between university expansion and ourmeasure of the fit of educational supply. Visually, no positive corre-lation is apparent, and indeed, the correlation between educationalfit and the number of new schools in a corresponding disciplineand region is on aggregate −0.055. Again, this is consistent withthe MURST analogy between school creation and drops of rain.
Overall, the above evidence implies that exploiting the variationin the number of schools within regions across time allows con-sistent estimation of the effect of university presence on regionalinnovation.
4.2. Regional innovation activity
Quantity. The estimation results for model (1) with innovationactivity being measured by the (log) number of industrial patentsare presented in Table 4. We find that opening a new univer-sity school significantly increases regional innovation activity. Thecoefficients here are elasticities, so an increase of one percent in thenumber of schools in a region increases industrial patenting in thatregion by 0.68 percent. The mean number of schools per regionis nine, and the mean number of patent applications per regionper year is 134 (Table 1), so on average, one new university schoolbrings about ten new patent applications by regional non-academicinventors five years later.19
Quality. The number of patents might be an appropriate indi-cator to capture the quantity of innovation, but it might hideimportant changes in quality. An observed increase in the quantityof patents as a result of a new university school in a region doesnot guarantee that the overall value of regional innovation activitygrows. Therefore it is important to analyze the effect of universitieson the quality of the patents produced.
We check for effects on average patent quality, measured asthe average number of citations received by regional patents. Evi-dence exists suggesting that patent citations represent a valid wayto capture patent importance. (See Jaffe et al. (2000) for exam-ple.) Specifically, we estimate Eq. (1) using the average number ofcitations to regional industrial patents as the dependent variable.Results are presented in column 2 of Table 4. Opening a new schoolhas no significant effect on the number of citations per regionalpatent.
One might also ask whether not just the quality, but also otherpatent characteristics have changed. For instance, one might beinterested to know whether the presence of a university affects therate with which industrial innovation draws information directlyfrom scientific publications. We use non-patent-literature (NPL)citations done by industrial patents to capture this patent char-acteristic. This is a very noisy measure (not least because manycitations are actually added by patent examiners (Akers, 2000)),
It covers information concerning graduates’ university-to-work transition, asking,inter alia, where and in what discipline they graduated, and where they work. Thedescription of the data could be found in Bagues et al. (2008).
19 See the Appendix A for robustness checks.
796 R. Cowan, N. Zinovyeva / Research Policy 42 (2013) 788– 800
Table 3Cross-correlation table.
Table 4The effect of the university expansion on industrial patents.
(1) (2) (3)Log patents Citations per patent NPL citations per patent
innovation process. Universities can serve as knowledge sourceswhen industry underinvests in private R&D.20
otes: First-difference model estimates. Independent variables are 5-year lagged. In* p-Value <0.100.
** p-Value <0.050.*** p-Value <0.010.
f industrial patenting as measured by NPL citations (column 3,able 4).
Interestingly, while private R&D tends to increase the quan-
ity and the quality of produced industrial patents, it is negativelyelated to the degree to which industrial inventors rely on scientificublications. One interpretation of the negative effect is that pri-ate R&D and local academic research can serve as substitutes in the
ntheses standard errors clustered by region.
20 Again this must be treated cautiously since many citations are entered in a patentnot by the inventors but by examiners.
R. Cowan, N. Zinovyeva / Research Policy 42 (2013) 788– 800 797
Table 5The effect of the university expansion on the number of regional industrial patents.
(1) (2) (3) (4) (5) (6) (7) (8) (9) (10)
Geographic location Value added per capita R&D stock Graduates in thelabor force
IndustrialVA/agricultural VA
Center and South North Low High Low High Low High Low High
Number of observations 88 132 110 110 110 110 110 110 110 110
Notes: First-difference model estimates. Regional characteristics and year dummies are included in all regressions. Independent variables are 5-year lagged. In parenthesesstandard errors clustered by region. Regions with high level of VA per capita (abbreviations are the same as in Table 1, in parentheses – the year when a region first reachesthe median): ABR (1997), EMR (1991), FVG (1994), LAZ (1991), LIG (1995), LOM, MAR (1995), MOL (2000), PIE (1991), TAA, TUS (1993), UMB (1995), VEN (1995), AOS (1989).Regions with high level of private R&D: ABR (1992), CAM, EMR, FVG (1990), LAZ, LIG, LOM, PIE, SIC (1998), TUS, VEN. Regions with high ratio of industrial VA to agriculturalVA: ABR (2000), EMR (1989), FVG (1988), LAZ (1986), LIG (1986), LOM, MAR (1991), PIE, TUS, UMB (1988), VEN, AOS.
*
efe
tmnaewisN
fnrRutrn
baTolawgtr
4
vt
goasu
p-Value <0.100.** p-Value <0.050.
*** p-Value <0.010.
Regional heterogeneity. A natural question is whether theseffects hold uniformly across regions, or whether regions with dif-erent economic development respond differently. In Table 5 wexplore whether university spillovers differ by the type of region.
The results suggest that the strongest spillover effect occurs inhe center and in the south of Italy (columns 1 and 2). As we have
entioned before, southern and central regions differ from theorthern regions in the level of income, private R&D investmentnd industrial structure. To test whether these characteristics of theconomy are in fact conditioning the strength of the spillover effect,e analyze how the effect differs across regions with different load-
ngs along these dimensions. Specifically, for each dimension, weplit the regions according to the median values of each variable.ote that regions can move from one group to another across time.
We observe that regions with low per capita income benefitrom knowledge spillovers from universities, whereas, on average,o positive university effect could be observed for high incomeegions (columns 3 and 4). Regions that have relatively low levels of&D and those with a less educated labor force benefit more fromniversities (column 5–8). In less industrialized regions universi-ies have a significant effect on innovation. In more industrializedegions the effect of universities is potentially larger, but is veryoisy and not statistically significant (columns 9 and 10).21
To summarize, we have observed that an increase in the num-er of schools is followed by an increase in industrial innovationctivity in the region: the number of industrial patents increases.he effect of new schools depends on the economic characteristicsf the region: poorer regions with relatively low human capital andow investment in R&D benefit most from university presence. Theverage characteristics of industrial patents seem not to changeith university presence, at least in the very short run. This sug-
ests that in the short run the new industrial patents induced byhe creation of new schools are on average not different from theest of industrial patents.
.3. Type of knowledge transfer
In the previous section we found that at the regional level, uni-ersity presence has a positive influence on the quantity of indus-rial innovation. Recall that in our analysis we define industrial
21 The number of new schools is not sufficiently large to perform detailed hetero-eneity analysis of the effect of different types of schools. However, according tour data, the strongest effects were generated by schools in medicine, chemistrynd pharmacy, veterinary, and agriculture. Science and engineering schools havetronger effects only in relatively more industrialized regions. Results are availablepon request.
innovation as innovation done without academic inventors. There-fore the observed increase in industrial patenting after new schoolcreation captures a spillover effect of university on industrial sector.
In this section we ask whether it is possible to identify the typeof university expertise that generates this spillover effect on indus-try. Specifically, we are interested in understanding whether firmsare benefiting from university capability to produce frontier scien-tific research, from a more applied inventive potential of academicresearchers or from other types of human capital or endowments.Understanding the type of expertise that is effectively transferredfrom university to local industry may be important for the designof university incentive structures.
University capability to produce frontier scientific researchcan be measured by the current publication records of profes-sors employed in a university. A quality-adjusted measure wouldweight the importance of each publication by its impact in termsof received citations. This is a common way to characterize thequality of academic researchers and there seems to be a trend nowtowards making evaluation and incentives for researchers formallyrelated to their (quality-adjusted) publication records. We definethe quality weighted measure of scientific publications as the totalcitations received between date of publication and 2009. Recently,in addition, there has been an emphasis on academic patenting:in a variety of ways academic researchers have been encouragedto patent their findings, possibly in collaboration with firms. Pro-fessors’ patents (and the number of citations received by thesepatents) is our measure of academics’ inventive potential. These areboth topical measures, since today publications and especially aca-demic patents are often used to measure university contributionto economic activity. In this section we ask to what extent thesemeasures of academic research activity can explain the effects wehave found in the previous section.
We undertake the following accounting exercise. To the pre-vious model, we add variables representing our measures ofprofessors’ expertise, and ask how their inclusion affects the esti-mated coefficient of the number of schools. Specifically we includein model (1) publications and academic patents as well as theircitations, Puni
Note that scientific articles and academic patents might take upto several years to be published or listed in EPO. Additionally, it
might take some time for schools to hire the necessary staff. So tocapture correctly the increase in the regional human capital due tothe opening of new university schools one would need to considerthe change in publications and academic patents over several years.
798 R. Cowan, N. Zinovyeva / Research Policy 42 (2013) 788– 800
Table 6Explaining the effect of schools on industrial patents.
Notes: First-difference model estimates. Regional characteristics and year dummies are included in all regressions. Independent variables are 5-year lagged. In parenthesesstandard errors clustered by region. All lags of publication and patent indicators are included. In parentheses standard errors clustered by region. The variance inflation factor(VIF) of Log Schools in column 4 is 1.45.
Ci
sosclrauipo
ptasiwfa
tothtsmb4tpsiiadb
5
tbco
We observe that the human capital associated with traditionaluniversity production, as measured by scientific publications and
* p-Value <0.100.** p-Value <0.050.
*** p-Value <0.010.
onsequently, we allow the effect to be distributed in time andnclude all lags of our measures in Puni
i,t .A positive ̌ in Eq. (1) signals the existence of a causal relation-
hip between the opening of a new school and regional innovationutput. However, if the inclusion of Puni
i,t in Eq. (2) reducesignificantly the size of ̌ relative to its value in Eq. (1), we canlaim that the corresponding university research output proxies ateast one type of human capital which is effectively translated intoegional innovation. If, controlling for professors’ publications andcademic patents, we still observe a significant residual effect ofniversities on industrial patenting, we might conclude that there
s something beyond the academic human capital captured byublications and academic patents that generates positive effectsn industrial innovation.
Note that university publications and academic patents areotentially endogenous to industrial patenting. An active indus-rial R&D sector could generate spillover effects inducing universityctivity, perhaps directly as industry seeks partners, or throughome less direct, spillover mechanism. On the other hand, for var-ous cost efficiency reasons, firms might engage in collaboration
ith university crowding out their independent research. There-ore we do not interpret the estimated direct effects of publicationsnd academic patents here.
The results are presented in Table 6. The first, very clear observa-ion is that academic patenting (columns 5, 6, and 7) captures nonef the effects of university presence on industrial innovation. Exper-ise in applied knowledge generation inside a university seems toave little effect on local industrial innovativeness. Second, exper-ise measured by a simple count of publications (column 2) alsoeems completely ineffective. However, if we include quality in theeasure of expertise (column 3) the coefficient on Log Schools falls
y roughly 30 percent. Including both quantity and quality (column) reduces the coefficient by about 40 percent, and it becomes sta-istically insignificant. This suggests that what matters from theoint of view of industrial innovation is to have a local univer-ity producing a significant amount of high-quality research. Strictnterpretation of statistical significance would imply that high qual-ty research is all that matters, however the size of the coefficientnd its high standard error suggests some caution, and opens theoor to the possibility that there is something more, not capturedy standard measures of university output.
. Conclusions
In this paper we focus on the economic effects of universi-
ies, and in particular on their effects on innovation. It is widelyelieved that the presence of a university in a region is benefi-ial for industrial innovation activity. We have taken advantagef certain unusual features of university expansion in Italy dur-
ing the 1980s and 1990s in an attempt to identify the effect ofuniversity presence on regional innovation. According to ex postevaluation of the expansion programmes, university schools werecreated “like rain”, independently of underlying economic featuresof the regions. This experiment permits a nice way out of standardendogeneity problems.
Our first result indicates that there is indeed a significant effectof the creation of new university schools on regional innovationactivity. Industrial patenting activity in the region increases quitesignificantly even within five years of a new school opening. Thusthe general impression seems to be correct: university activity ispositively correlated with local innovation activity, and a policytool to increase the latter is indeed to increase the former.
The effect of new schools depends on the economic character-istics of the area. Poor regions with low levels of R&D and humancapital investment are the ones that benefit most from an increasein university presence. This suggests that one role of universities isto fill gaps in missing R&D infrastructure. If this is the case, and ifthere are positive feedbacks in innovation dynamics, then openinga university in an innovation-poor region can be an effective part ofa development strategy for that region. Our results suggest that inan initial period when the region is poorly endowed with “innova-tion assets”, the university presence can compensate. This, underthe proviso that other necessary assets are present, could help topush a region onto a higher innovation path.
How are these benefits created? Marshall might suggest that itarises simply from the agglomeration of agents pursuing relatedactivities; Mike Lazaridis22 asserts that it arises through the pro-duction of highly trained graduates; supporters of the Bayh-Doleact assert that it comes from controlled technology transfer throughacademic patenting. Given the time frame we examine, namelyeffects within 5 years of a school opening, we exclude from theanalysis any effects of universities driven by the new graduateswho enter regional labor markets. To address the question of otherchannels through which this influence flows, we have performed anaccounting exercise estimating how the gross effect of increasingthe number of universities is affected when we add to the modelproxies of factors that might be intermediary in the process. Thefactors that we focus on include professors’ ability to produce scien-tific research and their ability to produce patentable inventions. Wemeasure the former by the number of ISI publications and the latterby the number of patent applications done with the participationof academic inventors.
their citations, has a strong effect on innovation, whereas academic
22 Founder and former CEO of Research in Motion, maker of the Blackberry.
search
ptdrgf
eaetgrAotmtt
A
KtahoHSeaCa
TM
Ns
TT
Ns
R. Cowan, N. Zinovyeva / Re
atents cannot explain observed university effects. At the sameime, we do not exclude the possibility that other types of aca-emic human capital, apart from the ability to produce frontieresearch, could be relevant as well. Such competences as collecting,eneralizing and classifying existent knowledge might be relevantor industrial inventors.
Universities perform many activities, and contain many vari-ties of skill and knowledge. Almost certainly it is a mixture of skillsnd knowledge that serves the interests of the local economy. How-ver, the results presented in this paper suggest that it is possibleo tilt the skill set in directions that are not in fact useful. They sug-est that forcing universities to change tack towards more appliedesearch may not, in fact, provide for the needs of local industry.t the very least, introducing patenting activity as a new measuref university performance may not capture what matters, and, inhe face of Goodheart’s law, as a policy tool to make universities
ore relevant to industry may be self-defeating. The traditionalool that values high-quality scientific research may remain moreo the point.
cknowledgements
We would like to acknowledge the inputs of members of theEINS project, and particularly Francesco Lissoni and Bulat Sandi-
ov, for their gracious openness and valuable help with the data. Were also grateful to Francesco Quatraro who provided us with theistoric data on Italian regional R&D collected from various issuesf ISTAT. We also acknowledge the helpful comments of Bronwynall, Jacques Mairesse, Joel Baum and all the participants of XXXIIymposium of Economic Analysis in Granada and DIME Confer-
nce “Knowledge Based Entrepreneurship: Innovation, Networks,nd System” in Milan. This research was supported by the DYREChaire d’ Excellence of Robin Cowan, funded by the French ANR,nd grants from ESF COST and APE-INV projects.
able A.1easurement of patent characteristics.
Extrapolated citations
a) Citations per patentLog Schools −0.11
(0.32)
R-squared 0.109
Number of observations 220
Absolute number
b) NPL citations per patentLog Schools −0.27
(0.41)
R-squared 0.174
Number of observations 220
otes: First-difference model estimates. Regional characteristics and year dummies are intandard errors clustered by region. *p-Value <0.100, **p-Value <0.050, ***p-Value <0.010
able A.2ime clustering of new schools’ opening.
All observations No openings for at least 3before considered period
Log Schools 0.68*** 1.13***
(0.21) (0.44)
R-squared 0.144 0.229
Number of observations 220 94
otes: First-difference model estimates. Regional characteristics and year dummies are intandard errors clustered by region. *p-Value <0.100, **p-Value <0.050.*** p-Value <0.010.
Policy 42 (2013) 788– 800 799
Appendix A. Robustness and specification checks
A.1. Measurement of patent characteristics
In our dataset we observe information on patents only until2004. There exist several ways to tackle the truncation problemin citations for recent patents. Our preferred method is to extrapo-late citations using the method proposed by Jaffe and Trajtenberg(1996). Alternatively, one can impose the same truncation bias onall observations and count only those citations that were receivedwithin a certain time window after patent filing. Another alter-native would be to normalize citations for patents submitted inthe same year. In order to account for different propensities to citeacross fields of science and technology, one could also normalizecitations within the same field.
Table A.1 summarizes estimations for (i) extrapolated patentcitations, (ii) patent citations received within 5 years after publi-cation, and (iii) patent citations normalized by the year of patentpublication and patent class group. None of these definitionsreveals any effect of new schools on patent quality.
Given that potentially there might be differences across patentclasses in the propensity to refer to scientific literature, we also redoestimations for non-patent literature citations normalizing themfor patents within the same type of patent class and year. Onceagain, we do not observe any significant effect of new schools onthe propensity to refer to scientific literature.
A.2. Time clustering of new schools’ opening
In the paper we hypothesize that the short time span (5-years)
considered after the opening of new schools allows us to excludethe knowledge transfers occurring through graduates. Still, if thereexists time clustering in the opening of new schools within a region,this might affect the interpretation of our results. To avoid this
5-Year window after publication Normalized for patents of thesame year and patent class
−0.16 −0.09(0.26) (0.17)0.147 0.107220 220
Normalized for patents of thesame year and patent class
−0.10(0.23)0.216220
cluded in all regressions. Independent variables are 5-year lagged. In parentheses.
years Rest of observations Excluding Lombardy, Lazio andEmilia Romagna
0.80 0.65***
(0.59) (0.21)0.272 0.148126 187
cluded in all regressions. Independent variables are 5-year lagged. In parentheses
8 search
ptwsoT
e–whr
R
A
A
A
A
A
A
A
B
B
B
B
B
B
B
B
C
C
C
C
D
00 R. Cowan, N. Zinovyeva / Re
roblem we have always clustered standard errors for observa-ions belonging to the same region, allowing them to be correlatedithin region across time. We also perform estimations only on the
ubsample of observations, for which there were no prior schoolpenings during at least 3 years. Results are presented in Table A.2.he estimates, if anything, are large on this subsample.
We also perform estimations on the subsample of regionsxcluding outliers in terms of university expansion and innovation
Piedmont, Lombardy and Emilia-Romagna. 22 of 65 new schoolsere opened in these three regions and the average R&D stockere is four times bigger than the country average. Once again, ouresults are robust to the exclusion of these outliers.
eferences
bramovsky, L., Harrison, R., Simpson, H., 2007. University research and the locationof business R&D. The Economic Journal 117, C114–C141.
cs, Z.J., Audretsch, D.B., Feldman, M.P., 1992. Real effects of academic research:comment. The American Economic Review 92 (1), 363–367.
grawal, A., Henderson, R., 2002. Putting patents in context: exploring knowledgetransfer from MIT. Management Science 48, 44–60.
kers, N., 2000. The referencing of prior art documents in European patents andapplications. World Patent Information 22, 309–315.
ndersson, R., Quigley, J.M., Wilhelmsson, M., 2009. Urbanization, productivity, andinnovation: evidence from investment in higher education. Journal of UrbanEconomics 66, 2–15.
rundel, A., Geuna, A., 2004. Proximity and the use of public science by innovativeEuropean firms. Economics of Innovation and New Technology 13, 559–580.
udretsch, D., Stephan, P., 1996. Company-scientist locational links: the case ofbiotechnology. The American Economic Review 86 (3), 641–652.
agues, M., Sylos Labini, M., Zinovyeva, N., 2008. Differential grading standards anduniversity funding: evidence from Italy. CESifo Economic Studies 54, 149–176.
alconi, M., Breschi, S., Lissoni, F., 2004. Networks of inventors and the role ofacademia: an exploration of Italian patent data. Research Policy 33, 127–145.
aldini, N., Grimaldi, R., Sobrero, M., 2006. Institutional changes and the commer-cialization of academic knowledge: a study of Italian universities’ patentingactivities between 1965 and 2002. Research Policy 35, 518–532.
ekkers, R., Bodas Freitas, I.M., 2008. Analysing knowledge transfer channelsbetween universities and industry: to what degree do sectors also matter?Research Policy 37, 1837–1853.
lundell, Richard, Griffith, Rachel, Windmeijer, Frank, 2002. Individual effects anddynamics in count data models. Journal of Econometrics 108, 113–131.
ottazzi, L., Peri, G., 2003. Innovation and spillovers in regions: evidence from Euro-pean patent data. European Economic Review 47, 610–687.
ratti, M., Checchi, D., de Blasio, G., 2008. Does the expansion of higher educationincrease the equality of educational opportunities? Evidence from Italy. Bank ofItaly Working Paper 679.
reschi, M.S., Lissoni, F., 2009. Mobility of skilled workers and co-inventionnetworks: an anatomy of localized knowledge flows. Journal of Economic Geog-raphy 9, 439–468.
arayol, N., Matt, M., 2004. Does research organization influence academic pro-duction? Laboratory level evidence from a large European university. ResearchPolicy 33, 1081–1102.
assiman, B., Veugelers, R., Zuniga, P., 2008. In search of performance effectsof (in)direct industry science links. Industrial and Corporate Change 17,611–646.
ohen, W.M., Nelson, R.R., Walsh, J.P., 2002. Links and impacts: the influence of
public research on industrial R&D. Management Science 48, 1–23.
owan, R., David, P.A., Foray, D., 2000. The explicit economics of knowledge codifi-cation and tacitness. Industrial and Corporate Change 9, 211–253.
asgupta, P., David, P.A., 1994. Toward a new economics of science. Research Policy23, 487–521.
Policy 42 (2013) 788– 800
Dorfman, N., 1983. Route 128: the development of a regional high technology econ-omy. Research Policy 12, 299–316.
D’Este, P., Patel, P., 2007. University–industry linkages in the UK: what are the fac-tors underlying the variety of interactions with industry? Research Policy 36,1295–1313.
Evangelista, R., Iammarino, S., Mastrostefano, V., Silvani, A., 2001. Measuring theregional dimension of innovation. Lessons from the Italian Innovation Survey.Technovation 21, 733–745.
Feldman, M.P., 1994a. The university and economic development: the case of JohnsHopkins University and Baltimore. Economic Development Quarterly 8, 67–76.
Feldman, M.P., 1994b. Knowledge complementarity and innovation. Small BusinessEconomics 6, 363–372.
Feldman, M.P., Florida, R., 1994. The geographic sources of innovation: technolog-ical infrastructure and product innovation in the United States. Annals of theAssociation of American Geographers 84, 210–229.
Gordon, R., 1990. Measurement of Durable Goods Prices. University of Chicago Press(for NBER), Chicago.
Grimpe, C., Hussinger, K., 2008. Formal and Informal Technology Transfer fromAcademia to Industry: Complementarity Effects and Innovation Performance.ZEW discussion paper 08-080.
Hall, B.H., Jaffe, A., Trajtenberg, M., 2001. The NBER Patent Citations Data File:Lessons, Insights And Methodological Tools. NBER Working Paper Series 8498.
Hall, B.H., Link, A.N., Scott, J.T., 2003. Universities as research partners. The Reviewof Economics and Statistics 85, 485–491.
Henderson, R., Jaffe, A.B., Trajtenberg, M., 1998. Universities as a source of com-mercial technology: a detailed analysis of university patenting 1965–1988.Econometrica 52, 909–938.
Jaffe, A.B., 1989. Real effects of academic research. The American Economic Review79 (5), 957–970.
Jaffe, A.B., Trajtenberg, M., 1996. Flows of knowledge from universities and federallabs: modeling the flow of patent citations over time and across institutionaland geographic boundaries. Proceedings of the National Academy of Sciences ofthe United States of America 93, 12671–12677.
Jaffe, A.B., Trajtenberg, M., Fogarty, M., 2000. The Meaning of Patent Citations: Reporton the NBER/Case-Western Reserve Survey of Patentees. NBER Working Paper,7631.
Jaffe, A.B., Trajtenberg, M., Henderson, R., 1993. Geographic localization of knowl-edge spillovers as evidenced by patent citations. Quarterly Journal of Economics108, 577–598.
Link, A.N., Siegel, D.S., Bozeman, B., 2007. An empirical analysis of the propensity ofacademic to engage in informal university technology transfer. Industrial andCorporate Change 16, 641–655.
Lissoni, F., Sanditov, B., Tarasconi, G., 2006. Keins database on academic inventors:methodology and contents. CESPRI Working Paper 181.
Loof, H., Brostrom, A., 2008. Does knowledge diffusion between university andindustry increase innovativeness? Journal of Technology Transfer 33, 73–90.
Lundvall, B., 1992. National Systems of Innovation. Pinter Publishers, London.Mairesse, J., Mohnen, P., 2010. Using innovation surveys for econometric analysis.
In: Hall, B.H., Rosenberg, N. (Eds.), Handbook of the Economics of Innovation.Elsevier, Amsterdam, pp. 1130–1155.
Ponds, R., van Oort, F., Frenken, K., 2010. Innovation, spillovers anduniversity–industry collaboration: an extended knowledge productionfunction approach. Journal of Economic Geography 10, 231–255.
Rogers, E., Larsen, J., 1984. Silicon Valley Fever. Basic Books, New York.Saxenian, A., 1985. Silicon Valley and Route 128: regional prototypes or historic
exceptions? In: Castells, M. (Ed.), High Technology, Space and Society. SagePublications, Beverly Hills and London, pp. 81–105.
Stephan, P.E., 1996. The economics of science. Journal of Economic Literature 34,1199–1235.
Stephan, P.E., Gurmu, S., Sumell, A.J., Black, G., 2007. Who’s patenting in the univer-sity? Evidence from the survey of doctorate recipients. Economics of Innovation
and New Technology 16, 71–99.
Wooldridge, J.M., 2002. Econometric Analysis of Cross Section and Panel Data. MITPress, Cambridge and London.
