Industry-university S&T transfers: what can we learn from Belgian CIS-2 data? Henri Capron and Michele Cincera (ULB-DULBEA) WORKSHOP ON THE PROCESS OF REFORM OF THE UNIVERSITY ACROSS EUROPE. SIENA, ITALY, MAY 24-26, 2004
OBJECTIVES OF THE PAPER: Role played by universities as a source of innovation information for firms Main determinants of industryuniversity cooperative agreements in innovation CIS-2 data - Sample of 1205 Belgian manufacturing firms in 1994-96
OUTLINE OF THIS TALK: Stylised facts Data Econometric framework Empirical findings Conclusions
STYLISED FACTS: interactions between firms & universities: Focus of R&D policy in Belgium towards a closer collaboration between enterprises and universities Large debate about the real effects of universities to permanent interaction with industry Important barriers to industry-university collaborations Industry and universities are not natural partners
STYLISED FACTS: interactions between firms & universities Functioning of European universities is not based on the same rules than American ones. Rosenberg (2001): American universities are more market-driven than their European counterpart. In the EU, over the last years, universities have been increasingly encouraged either directly or indirectly to co-operate with industry in order to alleviate the burdens of public expenditure.
STYLISED FACTS: interactions between firms & universities University-industry interactions can take several forms: industry sponsored research; collaborative research that can be encouraged by public funding; research consortia that put together some companies and universities engaged in various research efforts of common group interest; technology licensing from universities to companies for commercialisation; start-up companies involving universities and having licensing agreements to access university technologies; exchange of research materials to expedite the performance of research; university consultancy and services; graduate and continuing education.
STYLISED FACTS: interactions between firms & universities Industry-university collaborations can be beneficial for both partners: Hicks (2001): universities can enhance their scientific impact by reinforcing collaborations with the industrial sector, e.g. counting of the number of papers among the most cited 1,000 papers from 1981 to 1992 puts forward that 3.3 of every 1,000 university-industry collaborative papers were among the most cited papers against 2.2 for universityuniversity collaborative papers and 1.7 for single university papers. Mansfield (1991, 1992 and 1998): over 10% of the new products and processes introduced by firms could not have been developed without substantial delay in the absence of academic research. The importance of recent research was rated highest by the pharmaceutical industry.
STYLISED FACTS: interactions between firms & universities CIS-2: Number of innovators with very important sources of information for innovation (% and rank), 1996: BE EEA- % rk % Rk Internal information Within enterprise 44 2 47 1 Enterprise group 23 3 25 3 Market information Competitors 23 4 16 6 Clients or customers 54 1 42 2 Consultancy enterprises 3 10 5 8 Suppliers 15 6 19 5 Publicly available information Patent disclosures 2 12 3 11 Professional conferences 5 8 8 7 Computer based information networks 2 11 4 9 Fairs, exhibitions 20 5 22 4 Other Government/PNP research institutes 5 9 3 12 Universities 7 7 4 10 Source: EC, 2001, own calculations
STYLISED FACTS: interactions between firms & universities CIS-2: Number of innovators by type of partners as a share of innovation co-operators, 1996 enterprises within the group universities suppliers clients government or RTO's competitors consultancy enterprises 0 10 20 30 40 50 60 70 EEA BE Source: EC, 2001, Belgian CIS-2, own calculations
DATA Belgian CIS-2: period 1994-96, 2170 surveyed firms, 1378 answers. OSTC (1998): no selection bias. trimming procedure: all observations with DL, DX or D(L/Y) < 50% or >100%, with R&D/Y >50% and log(l/y) below the lower centile or beyond the upper centile have been excluded. final sample 1205 manufacturing firms. 290 firms reported >0 R&D expenditures in 1996. these firms are representative of 40.4% of Belgian total Business Expenditures on Research and Development in 1996.
DATA NACE Industry definition Code (Rev.1) 15-16 Manufacture of food products and beverages Manufacture of tobacco products 17-19 Manufacture of textiles Manufacture of wearing apparel; dressing and dyeing of fur Tanning and dressing of leather; manufacture of luggage, 20-22 Manufacture of wood and of products of wood and cork, Manufacture of paper and paper products Publishing, printing and reproduction of recorded media 23-24 Manufacture of coke, refined petroleum products and nuclear fuel Manufacture of chemicals and chemical products Industry abbrev. # of firms in the sample Avera-ge # of emplo-yees % of innovators IND1 157 167.0 21.0 IND2 141 177.3 19.1 IND3 134 142.2 21.6 IND4 120 713.4 45.8 25 Manufacture of rubber and plastic products IND5 68 139.9 47.1 26 Manufacture of other non-metallic mineral products IND6 82 186.8 22.0 27-28 Manufacture of basic metals IND7 188 231.7 32.4 Manufacture of fabricated metal products, except machinery 29 Manufacture of machinery and equipment, not elsewhere classified IND8 92 217.5 42.4 30-33 Manufacture of office, accounting and computing machinery Manufacture of electrical machinery and apparatus, not elsewhere classified Manufacture of radio, television and communication equipment Manufacture of medical, precision and optical instruments, watches IND9 90 385.7 53.3 34-35 Manufacture of motor vehicles, trailers and semi-trailers IND10 62 707.2 50.0 Manufacture of other transport equipment 36 Manufacture of furniture; manufacturing, not elsewhere classified IND11 71 83.1 16.9
DATA Belgian region Region abbr. Brussels Capital REG1 126 666.2 31.7 Flanders REG2 799 234.7 30.9 W allonia REG3 280 204.4 35.0 Size class (number of employees) 1 < 50 SM ALL 423 25.9 20.1 2 50-150 M EDIUM 391 92.4 30.4 3 > 150 LARGE 391 720.4 46.3 TOTAL 1205 272.8 31.9
DATA Variable All firms in the sample Innovating firms All firms Firms indicating universities as a source of information Collaborating firms All firms Firms that collaborate with universities LEMP mean 1,9464 2,1816 2,3078 2,4477 2,6528 TURNC mean 0,0837 0,1190 0,1157 0,1070 0,1176 EXPC mean 0,2807 0,4224 0,2716 0,2344 0,1979 RDI mean % 0,0060 0,0186 0,0235 0,0219 0,0273 IEI mean % 0,0443 0,1386 0,0388 0,0292 0,0265 FGP % 0,3214 0,4519 0,5287 0,5978 0,6170 GMTSUP % 0,0905 0,2805 0,3648 0,4134 0,5851 PAT % 0,0997 0,3013 0,3689 0,4525 0,5745 INMAR % 0,1487 0,4468 0,4713 0,5363 0,5745
ECONOMETRIC FRWK: Sources of information for innovation: outcomes of the dependent variable takes values 0 (sources not used), 1 (slightly important), 2 (moderately important), 3 (very important). Ordered probit model: natural candidate to account for the ordinal nature of this type of variable. But, only innovating firms answer to the sources of information variable, the importance of information sourcing is not observed for the non innovators: selection bias? Control for selection into the sample: estimate probit model discriminating between innovators and noninnovators. Error term jointly distributed with that from the ordered probit model.
ECONOMETRIC FRWK: Outcomes of the CIS-2 questions as regards collaborative agreements are asked in a sequential way: 1) are you innovative or not? 2) If yes, have you any co-operation arrangements with any type of partners? 3) If yes, with which type of partner, e.g. universities? Trivariate probit model with censoring (Mohnen and Hoareau, 2002) Simulated maximum likelihood method using the Geweke- Hajivassiliou-Keane (GHK, 1994) smooth recursive simulator.
ECONOMETRIC FRWK: Variable Definition Expected impact Continuous log (L) Size (log of number of employees) + log(y) Growth of sales + log(x) Growth of exports + R&D int. R&D intensity + inno. int. Non-R&D innovation intensity +/- Binary product object. process object. other object. Objectives of innovation activity = replace products being phased out, improving product quality, extend product range or open up new markets or increase market share Objectives of innovation activity = improve internal business process flexibility, reduce labour costs, materials consumption or energy consumption Objectives of innovation activity = fulfilling regulations or standards or reduce environmental damage foreign group Part of a foreign group +/- public support Firms that received public funds (reimbursable loans, subsidies) to support their + innovation activities patent Firms that applied for at least one patent over 1994-96 + drastic inno Introduction on the market of a product that is new for the market and non only for + the firm merger During 1994-1996, firm s turnover increased due to merger with another enterprise or + part of it sale/closure During 1994-1996, firm s turnover decreased due to sale or closure of part of the - enterprise IND1-IND11 Industry dummies +/- REG1-REG3 Regional dummies????
RESULTS: Universities as sources of information for innovation Innovation equation Universities as source of information equation est. s.e. est. s.e. C -1.040* 0.180 log (L) 0.628* 0.085 0.517* 0.193 log(y) 0.479* 0.179 0.408 0.275 log(x) 0.060** 0.033-0.012 0.062 foreign group 0.161** 0.096 0.358* 0.128 merger 0.190 0.190 0.274 0.237 sale/closure -0.466** 0.277-0.031 0.472 public support 0.463* 0.182 patent 0.172 0.136 R&D int. 0.014 0.016 inno. int. 0.000 0.002 drastic inno -0.003 0.118 product object. 0.077 0.089 process object. 0.117 0.085 other object. 0.156* 0.079 MU1 1.408** 0.782 MU2 2.322* 0.542 MU3 3.333* 0.402 ρ -0.602 0.587 log-likelihood -1079.9 # of obs. 1204 Notes: *(**) statistically significant at the 5(10)% level; regional and industry dummies included.
RESULTS: Universities as sources of information for innovation Innovation equation Type of partner as source of information equation SUNI SGRP SCOM SCLI SCON SSUP SGMT SPUBL SUNI SGRP SCOM SCLI SCON SSUP SGMT SPUBL log (L) + + + + + + + + + + + + + + + log(y) + + + + + + + + + + + log(x) + + + + + foreign group + + + + + + + + + + + + merger + + sale/closure - - - - - - - - public support + - + + patent - - + R&D int. inno. int. + drastic inno product object. + + + process object. + + + + + other object. + - + + + Notes: positive (+) and negative (-) coefficients statistically significant at the 10% (5%) level.
RESULTS: Innovation collaborations between firms and universities Innovation eq. Cooperation eq. Cooperation with universities eq. est. s.e. est. s.e. est. s.e. C -0.666* 0.209-2.612* 0.378-3.647* 0.741 log (L) 0.643* 0.082 0.826* 0.135 0.913* 0.146 log(y) 0.475* 0.184-0.034 0.310 0.562 0.405 log(x) 0.062* 0.031 0.001 0.046-0.122 0.154 foreign group 0.150 0.101 0.415* 0.140 0.175 0.156 merger 0.257 0.259-0.139 0.302 0.522 0.422 sale/closure -0.471** 0.251-0.631 0.443-0.606 0.465 public support 0.687* 0.171 0.972* 0.153 patent 0.495* 0.160 0.484* 0.167 R&D int. -0.002 0.018-0.011 0.020 inno. int. 0.000 0.001 0.000 0.001 drastic inno 0.230 0.142 0.071 0.173 product object. 0.437* 0.120 0.374* 0.114 process object. -0.022 0.103 0.210** 0.110 other object. 0.002 0.090-0.065 0.102 ρ12 0.447* 0.173 ρ13 0.139 0.137 ρ23 0.901* 0.042 LR test ρ12 = ρ13 114.761* log-likel. -983.9 # of obs. 1204 Notes: *(**) statistically significant at the 5(10)% level; regional and industry dummies included.
RESULTS: Innovation collaborations between firms and universities Innovation eq. Cooperation eq. Cooperation with partner eq. CUNI CGRP CCOM CCLI CCON CSUP CGMT CUNI CGRP CCOM CCLI CCON CSUP CGMT CUNI CGRP CCOM CCLI CCON CSUP CGMT C - - - - - - - - - - - - - - - - - - - - - log (L) + + + + + + + + + + + + + + + + + + + + + log(y) + + + + + + + + + log(x) + + + + + + + foreign group + + + + + + + + + + + + + + merger - Sale/closure - - - - - - - - - - Public support + + + + + + + + + + + patent + + + + + + + + + + R&D int. - + inno. int. + + drastic inno + + + + + + product object. + + + + + + + + + + + + process object. + + other object. Notes: positive (+) and negative (-) coefficients statistically significant at the 10% (5%) level.
CONCLUSION: At European level, universities do not appear to be the most important source of information for firms innovative activities. The most important sources are internal, i.e. within the enterprise or with other firms of the group and with clients or customers. Information from universities appears however to be more important in Belgium as compared to the European average. A different picture emerges when we look at universityindustry collaborations: In Belgium, universities are the second most important types of collaborators (after the firms within the group) and 53% of Belgian companies claim having co-operative agreements with universities, which is significantly above the European average of 38%.
CONCLUSION: factors explaining the use of a particular source of information for innovation not the same according to the type of source. Access to university-based information positively influenced by size of firms, public support to innovation and membership to foreign group. Among the objectives to innovate, only the fulfilling of regulations and standards and the reduction of environmental damage positively affect firms to access to university-based information. In terms of industry-universities S&T collaborations, size, government support, patent applications as well as product and process related objectives tend to increase the probability to innovate.