D. Felsensteinare not fully exploited. This is because innovationis <strong>related</strong> to the seedbed environment, which ishypothesized to be an unknown quantity at theearly stages <strong>of</strong> firm development.3. Science <strong>parks</strong> as seedbedsEvidence exists suggesting that, in spatial termsat least, universities do have a seedbed effect ontheir local economies. A series <strong>of</strong> aggregateanalyses on the effects <strong>of</strong> universities on metropolitanareas <strong>or</strong> regions in the United States hasshown this effect to be wide ranging. ThusJaffe [18] has found a relationship between firminnovation rates (measured by patents) and thelevel <strong>of</strong> local university research. This suggests theexistence <strong>of</strong> technological ‘spillovers’ that benefitfirms in proximity to universities. Further evidence<strong>of</strong> this aggregate spillover effect comes from Bania,Eberts and Fogarty [19], who attribute higherlevels <strong>of</strong> .new firm f<strong>or</strong>mation rates to those placeswith concentrations <strong>of</strong> highly skilled universitylabour, and from Beeson and Montgomery [20],who suggest that this spillover effect can alsoaffect occupational composition. They show thatthe level <strong>of</strong> R&D funding at a local universityincreases the odds <strong>of</strong> being employed locally as ascientist <strong>or</strong> engineer <strong>or</strong> being employed in a localhigh-technology industry. This seedbed functionalso serves to entrench future rounds <strong>of</strong> growthas evidenced by universities’ ability to attractscientific infrastructure. F<strong>or</strong> metropolitan areas, ithas been shown that industrial R&D labs tend toconcentrate in those areas where levels <strong>of</strong> universityresearch are highest [21]. This relationship hasalso been found on the basis <strong>of</strong> case-study analysis[22].1It should be noted, however, that evidence fromoutside the US is m<strong>or</strong>e equivocal. In the case <strong>of</strong>Japan, f<strong>or</strong> example, Eto and Fujita [28] reject thehypothesis that universities are instrumental ingenerating high-tech firm growth. They find strongevidence <strong>of</strong> the self-entrenching effects <strong>of</strong> hightechgrowth and that scientific-industrial agglomerationswill tend to reproduce themselves. This can<strong>of</strong>ten take place in proximity to leading universities,although they find no real causality in this process.Similarly, Fl<strong>or</strong>ax and Folmer [29] show that, inthe case <strong>of</strong> Holland, the diffusion <strong>of</strong> knowledgeis not necessarily a function <strong>of</strong> spatial clusteringaround universities. This seems to imply littlespillover effect, in small countries at least.Overall, however, the claims f<strong>or</strong> a significantuniversity seeding effect on the local economy,with respect to innovation level, new firm startuprates, occupational composition and so on,seem to be well founded. Evidence attributing asimilar role to <strong>science</strong> <strong>parks</strong>, however, is ratherm<strong>or</strong>e mixed. In terms <strong>of</strong> new firm f<strong>or</strong>mation,Massey et al. [6] find mixed evidence f<strong>or</strong> British<strong>science</strong> <strong>parks</strong>. On the one hand, new start-upsf<strong>or</strong>m a clear min<strong>or</strong>ity (less than 30%) <strong>of</strong> firms on<strong>science</strong> <strong>parks</strong> and in much celebrated seedbeds,such as Cambridge, this figure is less than 10%.On the other hand, new start-ups on <strong>science</strong> <strong>parks</strong>have a much lower m<strong>or</strong>tality rate than that <strong>of</strong> newfirms in general (less than 2.5% a year). Thiscould however reflect a screening process f<strong>or</strong><strong>science</strong> park entry that selects only firms withgood survival chances. In the US, case studyevidence shows great variation. On the ResearchTriangle Park, f<strong>or</strong> example, over 70% <strong>of</strong> firms arepart <strong>of</strong> multi-plant <strong>or</strong>ganizations whose existencecannot be attributed to the park. The <strong>University</strong><strong>of</strong> Utah Research Park, on the other hand, haslocal, single-plant <strong>or</strong>ganizations comprising m<strong>or</strong>ethan half the park population [ll].Science park-based spin-<strong>of</strong>fs that trace their rootsto the university are another seedbed characteristic.While much attention traditionally has beenfocused on existing firms spawning new firms [30],spin-<strong>of</strong>fs emanating from a university environmenthave received much less attention [31]. Nevertheless,accounts <strong>of</strong> locally based company genealogiesnearly always put the local university <strong>or</strong> <strong>science</strong>park at the apex <strong>of</strong> any ‘family tree’ account <strong>of</strong>seedbed growth.Survey evidence from firms located on Dutchand Belgian <strong>science</strong> <strong>parks</strong> indicates that only 37%<strong>of</strong> firms in the f<strong>or</strong>mer and 16% in the latterattribute their <strong>or</strong>igins to universities [32]. In the96 Technovation Vol. 14 No. 2
Science <strong>parks</strong> - seedbeds <strong>or</strong> enclaves <strong>of</strong> innovation?US, again the variation across individual <strong>parks</strong> isvery great. Thus, Luger and Goldstein [II] havefound that over 120 spin-<strong>of</strong>f companies in thevicinity <strong>of</strong> the Stanf<strong>or</strong>d Research Park haveuniversity antecedents whereas f<strong>or</strong> the ResearchTriangle Park the number <strong>of</strong> university spin-<strong>of</strong>fsis virtually nil. Massey et al. [6] show that, f<strong>or</strong>UK <strong>science</strong> <strong>parks</strong>, on aggregate 25% <strong>of</strong> firms mayhave their beginnings in academia although some<strong>of</strong> these are likely to be analytic service units <strong>or</strong>technical troubleshooting operations that, pri<strong>or</strong> tothe establishment <strong>of</strong> the <strong>science</strong> park, were locatedwithin the university. Even at the level <strong>of</strong> theindividual park unit, while a location such as theCambridge Science Park is much heralded as aspin-<strong>of</strong>f incubat<strong>or</strong> and (in 1987) nearly 400 localfirms owed their ultimate <strong>or</strong>igin to the localuniversity in one way <strong>or</strong> another, direct universityinvolvement in their establishment was much m<strong>or</strong>ecircumspect. Segal Quince Wicksteed [33] rep<strong>or</strong>tsthat less than 20% <strong>of</strong> new companies were f<strong>or</strong>medby direct university entrants.If the <strong>science</strong> park functions as an effectiveseedbed, it is assumed that the level <strong>of</strong> R&Dconducted by firms located on <strong>science</strong> <strong>parks</strong> isgenerally higher than that <strong>of</strong> <strong>of</strong>f-park firms andthat this reflects the level <strong>of</strong> (close) interactionpresumed to exist between <strong>science</strong> park firms andthe local university. The <strong>science</strong> park is thusperceived as an imp<strong>or</strong>tant conduit in technologytransfer out <strong>of</strong> the university and into the localeconomy.Van Dierdonck et al. [32] present a ratherdifferent picture based on <strong>science</strong> <strong>parks</strong> in Belgiumand Holland. They show rather low levels <strong>of</strong>interaction perf<strong>or</strong>mance between park tenants andlocal universities, with the overall level <strong>of</strong> R&Dactivity perf<strong>or</strong>med on the park being lower thanpopularly anticipated. Only 32% <strong>of</strong> Dutch firmsand 57% <strong>of</strong> Belgian firms surveyed rep<strong>or</strong>ted inhouseR&D, and little R&D interaction wasrep<strong>or</strong>ted with either local (<strong>science</strong> park) firms <strong>or</strong>the local university. In fact, f<strong>or</strong> many <strong>science</strong> parkfirms, external research linkages were not locallydefined and were conducted on an internationalscale, pointing to the existence <strong>of</strong> research netw<strong>or</strong>ksunconstrained by national boundaries. Similarly,Massey et al. [6] find <strong>science</strong> park tenant firms inthe UK less ‘leading edge’ than generally imagined.On the inputs side, over 40% <strong>of</strong> employees on<strong>parks</strong> were scientists and engineers2 and R&Dexpenditures in relation to sales were very high.On the outputs side, despite a greater propensityto patent amongst on-park than <strong>of</strong>f-park firms,most firms were found to be engaged in modification<strong>of</strong> existing technologies rather than thedevelopment and production <strong>of</strong> totally new innovations.Complementing this picture is the evidencepointing to the generally low level <strong>of</strong> links between<strong>science</strong> park firms and local universities. Mostaccessing <strong>of</strong> academic resources relates to lowlevelcontacts based on recruiting university graduates,<strong>or</strong> inf<strong>or</strong>mal contacts [6]. Joint research <strong>or</strong>subcontracting relations are much less pr<strong>of</strong>use.F<strong>or</strong> UK <strong>science</strong> <strong>parks</strong> only 14% <strong>of</strong> companiesrep<strong>or</strong>ted such links, and a survey <strong>of</strong> two maj<strong>or</strong><strong>parks</strong> in Israel rep<strong>or</strong>ted a similar figure [34]. Firmson US <strong>science</strong> <strong>parks</strong> have also indicated thatrecruitment <strong>of</strong> graduates and use <strong>of</strong> universityfacilities f<strong>or</strong>m their main points <strong>of</strong> contact withuniversities. Thus, f<strong>or</strong> both the Research TrianglePark and the <strong>University</strong> <strong>of</strong> Utah Research Park,technology transfer to local firms (both on- and<strong>of</strong>f-park) was low [ll]. F<strong>or</strong> these firms the localuniversity was not a particularly imp<strong>or</strong>tant source<strong>of</strong> R&D inputs <strong>or</strong> innovations. Even f<strong>or</strong> theStanf<strong>or</strong>d Research Park, generally accredited withbeing an archetypal ‘seedbed’, local firms did notrep<strong>or</strong>t a significant seeding effect. Thus, while80% <strong>of</strong> on-park firms rep<strong>or</strong>ted links with Stanf<strong>or</strong>d,over 70% <strong>of</strong> <strong>of</strong>f-park firms said they had noconnection.These observed levels <strong>of</strong> university-<strong>science</strong> parkinteraction have led to suggestions that hightechnologyfirms are m<strong>or</strong>e dependent on linkagesand inf<strong>or</strong>mation flows from other similar firmsthan on interaction with universities [35]. If thisis the case, there should be evidence <strong>of</strong> <strong>science</strong>park firms seeding the local economy throughtheir material input-output patterns. However,examining the external purchase and sales linkagesTechnovation Vol. 14 No. 2