Science and Technology Infrastructure and Small Countries Ability to attract Foreign R&D A Comparison between Finland and the Netherlands

Transcription

1 Paper to be presented at the workshop Public Research Institutions, International Business and Technological and Economic Catch Up in Developing Regions March 19 th 20 th, Catania, Italy Session 5: Connections (or lack of them) between universities and public laboratories, and multinationals Title Science and Technology Infrastructure and Small Countries Ability to attract Foreign R&D A Comparison between Finland and the Netherlands Cees van Beers Elina Berghäll Tom Poot This paper is provisional. Please do not quote! Keywords: Multinational Enterprises, innovation, R&D collaboration, Science and technology institutions. JEL-code: O32 March 2004 Corresponding author. Cees van Beers and Tom Poot are in the section economics of innovation, Delft University of Technology, Jaffalaan 5, 2628 BX Delft, the Netherlands. c.p.vanbeers@tbm.tudelft.nl; a.p.poot@tbm.tudelft.nl. Elina Berghäll is in the Government Institute for Economic Research (VATT), Helsinki, Finland, . elina.berghall@vatt.fi. The authors are grateful to Statistics Netherlands for using the Dutch Community Innovation Survey data and to the Finnish Statistical Office for the Finnish Community Innovation Survey. 1

2 Abstract This paper addresses the question whether affiliates of foreign firms in Finland cooperate significantly more with domestic universities and public R&D institutions than in the Netherlands. These two countries have implemented different innovation policies in the last fifteen years. Finland released many resources for the improvement of the domestic S&T infrastructure s quality, while the Netherlands did not. We present a detailed empirical analysis in which we use Community Innovation Survey data to estimate the determinants of R&D-co-operation of foreign and domestic firms with the S&T system in Finland and the Netherlands. The results show that Finnish domestic firms co-operate more often with (domestic) S&T institutions than Dutch domestic firms. Finally, the probit estimates reveal that foreign firms in the Netherlands co-operate significantly less with institutions in the (domestic) S&T infrastructure than domestic firms. In the Finnish case this negative effect is not present, which means that MNE s R&D collaboration schemes with the Finnish S&T infrastructure is comparable with that of the domestic firms. This suggests that R&D co-operation of the foreign MNEs in the domestic S&T infrastructure is stronger in Finland than in the Netherlands, which can be considered as a result of different innovation policies. 2

3 1. Introduction Foreign direct investment (FDI) through multinational enterprises (MNEs) affects the domestic economy of small countries much more than large economies in terms of trade flows, employment and knowledge generation. The decision to close a subsidiary in a host country generally has more impact in a small than in a large economy. Therefore it is of great importance for small countries to attract firms that are not footloose. At the same time it appears to be hard for small host countries to attract non-footloose FDI because of limited market size, which is considered as one of the most important economic fundamentals to attract MNEs (see Dunning (1993), Globerman and Shapiro (1999)). FDI investment policies of small countries should therefore focus on other important determinants like skill levels and a good physical infrastructure. Blomström and Kokko (2003) point out that in recent years the view on the importance of investment incentives have started to change mainly due to the internationalization of the world economy. Increased trade liberalization at a global level has increased both foreign trade and foreign direct investment flows. At a regional level, increased market integration (EU, NAFTA) has reduced the importance of market size as an important asset a host country can offer foreign investors. Therefore small countries are now in a better position to compete for foreign investments that would have been directed to large markets many years ago. The importance for small countries to attract MNEs that are not footloose requires the engagement of a relationship with foreign firms from which both parties benefit. R&D-co-operation between foreign firms and elements of a countries science and technology system (S&T-system) like universities and public R&D institutes is such a relationship for two reasons. First, universities and public R&D institutes resources and skills are attractive for (foreign) firms because these are complementary to their own skills and knowledge (Bower, 1993; Santoro and Gopalakrishnan, 2000). S&T institutions have access to basic research at a global level while firms have practical knowledge of market opportunities for new products and financial resources to finance R&D. The complementarity characteristic of R&D co-operations between foreign firms and S&T institutions is most important for the (foreign) firms because it gives them access to knowledge they are not able to produce themselves economically. The incentive for universities and public R&D institutes arises of 3

4 increasing pressure of public budget cuts. R&D-co-operation with firms allows them to increase their financial means. Second, internationalization of research & development (R&D) through multinational enterprises (MNEs) has increased substantially in the last decade. Global value chains break up the production processes in parts that are located in different countries. This is important for small countries because their domestic markets are too small to enable domestic enterprises to recoup high R&D costs especially in a time of shortening product life cycles (Fors and Svensson, 2002). A high quality S&T infrastructure is an asset that attracts R&Dintensive FDI and hence provides a means to gain access to new technology and knowledge by tapping in the knowledge driven part of the global value chains (UNIDO, 2002). An additional factor of importance is that universities and public R&D institutes produce substantial R&D spillovers and therefore contribute to the goal of maximizing the social return of innovations. S&T-institutions are involved in basic R&D, which is estimated to have a high rate of return (Mansfield (1980); Adams (1990)). Also university patents are more frequently and generally cited than average patents (Henderson et.al, 1998). In addition to attracting FDI these spillovers promote growth and increase the development of local industry (Blomström and Kokko, 2003). This paper examines whether affiliates of foreign firms in Finland co-operate significantly more with domestic universities and public R&D institutions than in the Netherlands. We present a detailed empirical analysis in which we estimate the determinants of R&D-co-operation of foreign and domestic firms with S&T institutions in two small economies in the European Union, Finland and the Netherlands. 1 The choice for these two countries is twofold. First, both countries are small and hence MNE investments are important for them (UNCTAD, 2002). Second, both countries are member of the EU but differ in polices to attract foreign investments. The Dutch favorable geographical position towards the large German market made the lack of a large domestic market a less pressing incentive to focus on other determinants of attracting FDI. Before 1989 the Finnish geographical position was favorable towards the Sovjet Union. The collapse of the Sovjet Union in 1989 provoked a deep economic crisis in Finland and forces the country to a radical policy 1 Although the Netherlands has twice as much inhabitants as Finland, it is still considered a small country in the European Union. 4

5 change by starting to promote an active innovation policy in which the central element was to strengthen the domestic S&T infrastructure (Castells and Himanen, 2002). The well functioning S&T infrastructure of Finland in the 1990s made it an attractive place for R&D intensive investments of MNEs. It is expected that Finland attracts more FDI with R&D co-operations in the domestic public S&T structure than in the Netherlands. The paper is organized in five sections. In the next section we explain shortly the important differences between Dutch and Finnish innovation policies in the last decade in more detail. Section 3 describes a model in which we model the determinants of S&T co-operation. Section 4 describes the data used. Section 5 presents the empirical results and the final section reports the conclusions. 2. The impact of innovation policies on the quality of FDI. Several researchers have tried to classify R&D activities (e.g. Kuemmerle, 1997 and Chiesa 1996). Von Zedtwitz and Gassmann (2002) distinguish several R&D strategies on the basis of whether the research or development activities are centralised or dispersed internationally. Research is defined as the process of discovering new scientific knowledge that has the potential to act as a platform for the subsequent development of commercially viable products and manufacturing processes. Development is the process of creating new products or processes that have commercial value by the application of currently available platforms of scientific knowledge. A distinction is made in national treasure R&D, market driven R&D, technology driven R&D and global R&D. This distinction is of importance for a judgement on the role of MNEs in generating technology transfers in the domestic economy of small countries. National treasure R&D is kept in the MNE s home country and is therefore of no direct use for host countries. Small host countries should aim at market-, technologyand global R&D-driven MNEs. Market driven R&D keeps research at home but development activities are dispersed. Business development is dominated by customer demand and not by scientific exploration. As small countries have a small market this kind of R&D can only be relevant in case of a favourite geographical location towards large adjacent countries. For Finland and Sweden this is of less importance. Location drivers for the 5

6 development part of R&D, which is important in case of market-driven research, are not easy to influence by explicit technological policies because they lie within the sphere of market access, customer proximity and local market requirements and within the cost structures of the firms like country specific cost advantages and scaling-up in manufacturing. These can partly be addressed by financial policies (see Table 3). The research part of technology driven and global R&D are attracted to countries that offer favourite location drivers strongly related to the technology infrastructure. Attracting foreign MNEs with this kind of R&D and keeping home own MNEs R&D requires continuous attention for investments in high quality innovation centres, research parks and universities. A highly educated population and good regional and national innovation systems stimulate the specialization in high technology niche markets. Financial government policies like tax reductions cannot play a role here (see also Blomström and Kokko, 2003). INSERT TABLE 1 A distinction exists between 1) technology transfer (in or between firms across national borders), and 2) dissemination (externalizing the ownership of the technology). Foreign MNE investments can speed up the process of economic development in small countries by providing technology, entrepreneurship and organizational skills. It is an important way to upgrade local technological capability by technology transfer. In order to utilize efficiently the accumulated technological expertise of foreign MNEs, host governments need to pursue a well-defined technological strategy, directed to upgrading human skills and innovatory capacity, and provide the scientific infrastructure that high technology MNEs regard as important for their R&D investments in the host countries. For dissemination of technology it is important to strengthen the technological advantages of local firms and enable them to absorb successfully R&D from foreign MNEs. Technological policies that stimulate the absorptive capacities of local firms for technology dissemination from foreign MNEs also affect the motive of home MNEs to invest in the small home country and lead to a technological upgrading of outward foreign direct investments. 6

7 The quality of inward foreign investments is affected by differences in innovation policies. Inward investments are considered of higher quality if they have a strong ability to bring technological knowledge into the host country, which, through knowledge spillovers, can be transferred to domestic firms and hence improves the knowledge and innovation base of the country. Innovation policies refer to innovation subsidies, investments in education, and the efficiency of the national innovation system. In Finland high public investments in education were aimed at increasing the supply of high skilled personnel. As a result the cost of high-skilled professionals is competitive at the international scale. With regard to innovation subsidies and the efficiency of national innovation systems Finland is most dedicated towards these policies. The main focus of growth-oriented policies in Finland has been on improving Science and Technology Infrastructure (STI) since the 1990s. Foreign firms have generally been welcomed to contribute to the national knowledge and productive base but were not specifically aimed at with policies. National STI policies have promoted the creation of centres of innovation and platforms for technology firms. Success in generating indigenous technological capacities has attracted foreign-owned firms even in the absence of any specific measures beyond general advertisement of opportunities. The National Technology Agency of Finland (Tekes) directs its R&D funding to companies based in Finland and Finnish Research Institutes and universities. Criteria considered include (i) expected impact on company s competitiveness and growth, (ii) the technology targeted, (iii) the company s resources and networking capacities, and (iv) the expected impact of the financial support to the project. The ceiling on grants is 50 % of salary, raw materials, machinery, subcontracting, travel and patent costs. Credits can rise up to 60 % of costs. Since networking and co-operation across borders are encouraged, foreign firms may find themselves at an advantage. Venture capital is also available from semi-public organisations such as Sitra (the National Fund for Research and Development) and Teollisuussijoitus (Finnish Industry Investment Ltd). In the Netherlands a shortage of technically skilled workers exists because of a policy of twenty years of budget cuts in the education system. The tendency of many students to choose for nontechnical studies in the last years may reduce high technology-oriented FDI in the near future. 7

8 With regard to innovation policies in the Netherlands the WBSO (Wet Bevordering voor Speur en Ontwikkelingswerk) allows reduction of income tax liabilities of employers so that wage costs for R&D personnel are reduced. While the Finnish approach is directed toward substantial investments in (technical) education and in improving the organization of the STI in order to increase its effectiveness, the Dutch approach has been concentrated on general innovation policies through financial instruments that address production costs like tax credits. According to the analysis of location drivers described above this suggests that Finland will be an attractive host country for foreign investments in the research part of R&D and the Netherlands policies are friendlier for foreign investors that receive their research part from the headquarters abroad but marginally change the R&D processes to local requirements. 3. The Model As the main aim of the present study is whether innovative foreign firms in Finland co-operate more intensely in their R&D efforts with S&T institutions than in the Netherlands, we specify the following model in which the determinants of several R&D collaboration schemes will be investigated. If firms innovate, did they co-operate in their R&D efforts that led to the reported innovations, with external partners like other firms and - of particular importance for the problem addressed in the present paper partners in the S&T infrastructure of the host country. Equation (2) has been formulated as: Pr( part + β bscrd 7 13 t ) = β + β ln S t 1 + β science t β scale t 1 + β orgimp t t 1 + β for + β rdsal β spectoel t + β finsalunc t 3 t 1 16 t 1 + β va + β rdfreq t β cstimp + β 17 t 1 t 1 pespec t + β brnt β avbrnt + ε t t 1 t 1 + β brnf 6 12 t 1 + β avbrnf t 1 The dependent variable is the probability to co-operate on R&D-efforts with an external partner (Pr(part t )). These can be other firms but also organizations that belong to the Science and Technology (S&T) infrastructure like universities and/or R&D-institutes. The first determinant of innovation is the size of the firm as described by the natural logarithm of the firm s turnover in the previous period (lns t-1 ). The direction of the 8

9 effect is not a priori clear in this case. On the one hand, larger firms have more room to innovate internally and a negative effect for β 1 can be expected. However, the increasing internationalization of R&D in the last decade also means that more R&D co-operation with external partners is likely as speed and complexity of R&D make it hard to do everything alone (UNIDO, 2002). As argued in the section 1, this argument is one of the basic arguments for the present study and therefore a positive effect is expected for β 1. The size variable is lagged with 1 period in order to deal with causality problems. Are large innovating firms more involved in R&D collaboration or are innovating firms involved in R&D collaboration larger than non-collaborating innovating firms. As the R&D collaboration of the firm in the present period (t) cannot determine its size in the past (t-1) the causality must go from size to collaboration. The second effect (for t ) describes the impact of a foreign affiliate on the probability to co-operate in R&D efforts. If the dependent variable is measured as S&Tinstitutions of the host country, a significant positive (negative) coefficient shows that foreign affiliates co-operate more (less) intensively with S&T institutions than domestic firms. The variables rdsal t-1 and rdfreq t-1 refer to R&D expenditures as a share of sales and a dummy variable revealing whether the firm performs R&D frequently (value = 1) or incidentally (value = 0), respectively. Positive coefficients of β 3 and β 4 are expected. The origin of the idea that led to the reported innovation is described by the information source variables. If the information source comes from inside the firm (brnt t-1 ), a positive effect on the probability to co-operate with external partners can be expected. The causality is from the firm in which the idea originates to the partner choice. Knowledge spillovers are proxied by brnf t-1, the information source is from outside the firm. Here a causality problem arises because if the origin of the idea is outside the firm, it probably comes from an already existing R&D partnership otherwise the firm would not be able to receive the information about the idea. An exception may be knowledge from S&T institutions as these are often in fundamental knowledge and therefore more easily available than knowledge from non S&T institutions like private firms. This provides an incentive for firms to co-operate with S&T institutions. The lag of this variable is an attempt to deal with the causality and it leads us to the conclusion that a significant positive effect can be expected if the 9

10 dependent variable is restricted to R&D partnerships with S&T institutions and no effect in case of other R&D partners. The ratio of fundamental to applied innovations (bscrd t-1 ) assumed to work out positively on R&D collaboration with S&T-partners. The barriers-to-innovation variables (orgimp t-1, finsalunc t-1, costimp t-1 ) are considered as an encouragement to co-operate with external partners in order to overcome these barriers. The variables avbrnt t-1 and avbrnf t-1 have been included in order to deal with heterogeneity in the sample. The variable pavitt t,i consists of three dummy-variables science, scale and spsup. These measure the relationship between different industry characteristics and innovation performance. As described by Pavitt (1984), science-based industries (denoted by science) such as chemicals or electronics are heavily dependent on knowledge, skills and techniques from academic research. In these industries, the ability to continuously produce a stream of innovation, mostly measured in patenting activity, has been vital for overall corporate success. In scale-intensive industries (scale) such as extraction and processing of bulk materials or automobiles, technological progress consists mainly of incremental technological improvements in complex production processes and complex products. The design, building and operation of complex production processes and/or products are considered as the main source of technological accumulation. As in science-based industries, innovation performance is equally important for the survival of companies. In specialized supplier industries (spsup) such as machine tools or software, technological progress has been fast but based on incremental improvements. The majority of companies have been small and provided high performance inputs into complex production systems. 2 The last two explanatory variable va t and pespec t are only relevant for the services sector. These are no Pavitt-dummies and they refer to financial, ICT and engineering services (va t ) and to utility and trading services (pespec t ). The sector dummies of the Pavitt (1984) classification are expected to be positive for the scientific-oriented firms and R&D co-operation with S&T institutions. The large scale orientation of firms cannot be considered of importance but is taken into 2 Pavitt (1984) defines as a fourth category supplier-dominated industries with textiles as a typical example. In these industries, suppliers drive technological change. For econometric reasons, this category is intentionally left out. However its relative importance is captured by the constant term. 10

11 account as it belongs to the Pavitt classification. With regard to the service sector dummies a negative sign can be expected as services do generally not require fundamental knowledge as provided by S&T institutions. However, if other partners are involved the sign might be positive as also in services sectors co-operation with non S&T institutions may be of importance. 4. Data The data used in the present study are from the Community Innovation Surveys (CIS) of Finland and the Netherlands. In the Finnish case we had the availability of data for the years 1996 and 2000 and in the Dutch situation the years 1996 and 1998 were available. The total number of firms in the Finnish CIS data (1,797 respectively 1,641) is substantially lower than in the Dutch data (9,455 respectively 10,152). The same remark is valid for the number of foreign firms, 250 and 185 in the Finnish case, and 2,078 and 2,041 in the Dutch case. In order to get an indication of how innovative foreign firms are relative to domestic firms, and how important the innovative foreign firms R&D collaboration with domestic S&T institutions are, we first provide some data in Table 2 INSERT TABLE 2 Table 2 should be interpreted with care because different sampling techniques may affect the results of the permanent innovators as a percentage of the total number of firms. Three conclusions can be drawn. First, the part of foreign firms that innovate permanently is higher than the share of domestic firms that report to innovate permanently for all years in both countries. This is a general pattern that is often explained by the size of foreign firms (see also De Backer and Sleuwaegen, 2003). Second, in both countries more than 60 % of the permanent innovators are involved in some kind of R&D co-operation. Third, in Finland between 80 % and 90 % of the R&D collaboration schemes take place with the (domestic) S&T sector. In the Netherlands this is limited to 30 % till 45 %. In Finland foreign firms R&D cooperations with the S&T sector take only place with the domestic S&T sector while in the Netherlands some R&D co-operation of foreign firms with foreign S&T sectors exists. This suggests that in Finland affiliates of foreign firms co-operate significantly 11

12 more with domestic universities and public R&D institutions than with other partners in the small domestic economy while in the Netherlands it is the other way around. 5. Estimation results In Table 3 and 4 the probit estimates of equation 1 have been reported. A panel dataset has been constructed with the year 2000 as the actual year for Finland and 1998 for the Netherlands. The lagged variables refer to 1996 for both countries. The sample has been restricted to innovating firms only. 3 INSERT TABLE 3 Regressions 1 and 2 explain whether a firm has a partner with which it co-operates in R&D. These include any partner consisting of S&T- and non-s&t institutions. Regressions 3 4 show the firms determinants of co-operating with a partner in the S&T sector and in regressions 5-6 it is the domestic S&T sector that is the dependent variable. The variable of main interest is For t, which shows whether foreign affiliates are more or less intensively involved in R&D co-operation than domestic firms. This is a very strong and biased test as it is quite likely that domestic firms co-operate more with domestic institutions than foreign firms. Hence, a negative bias of the For t coefficient may be built in a priori by testing in this way. However, we compare between Finland and the Netherlands. It may be possible that the negative bias is less dominant in one of these countries. However, that would reveal that foreign firms in one country is more involved with S&T institutions as compared with the domestic firms than foreign firms on the other country. Table 3 shows that foreign affiliates of MNEs do not affect significantly co-operate more or less in R&D than domestic firms. Table 2 already pointed in this direction as it revealed that in the year 2000 there is hardly any difference in the percentages for R&D co-operation irrespective of the collaborator and for collaboration schemes with (domestic) S&T institutions. In the Dutch regressions (2, 4 and 6) the insignificant effect of foreign firms on R&D-collaboration irrespective of the partner becomes 3 Correction for selectivity bias by means of a Heckman procedure is not necessary here because firms that do not innovate also do not have any R&D collaboration scheme. 12

13 significantly negative if the dependent variable is (domestic) S&T partners. This implies that if firms innovate, affiliations of foreign MNEs co-operate less with Dutch (domestic) S&T-institutions than domestic firms. In Table 4 this effect disappears or gets weaker, which implies that in the Netherlands particularly foreign firms in the non-industrial sector (mainly services) are responsible for the negative effect. INSERT TABLE 4 The size variable is significantly positive in all regressions of Tables 3 and 4 as expected. R&D-input (rdsal t-1 ) is significantly positive in Finland s industrial firms, which implies that R&D-input of firms in Finland is an important drive to co-operate with all but also specific (S&T-) partners than in the Netherlands. This is consistent with Nieminen and Kaukonen (2001: p. 104)) who argue that in Finland higher R&D investments go together with more co-operation with universities. Rdfreq t-1 is insignificant in nearly all regressions, which can be attributed to possible multicollinearity with the R&D-input variable. The significant positive effect of the information source from inside the firm (brnt t-1 ) in all the Finnish regressions reveal that Finnish firms actively look for all kinds of external R&D-partnership. In the Dutch cases firms do not actively look for partnerships with (domestic) S&Tinstitutions. A positive significant effect for brnf t-1, knowledge spillovers described by information source is from outside the firm can be found in the Dutch regressions in case of S&T co-operation. 4 The basicness of research variable is not significant in the Finnish regressions but do reveal the expected positive significant effect on R&D collaboration in the Dutch regressions. This might suggest that Finnish firms co-operate with external (domestic S&T) partners also in cases of applied research, while in the Netherlands co-operation with these partners is only of relevance if fundamental research is at stake. The 4 This can be attributed to a causality problem. Information sources from outside the firm might be the result of an existing R&D co-operation, particularly with S&T institutions. Although we try to deal with the causality problem by estimating with a lagged variable, in case of continuous R&D cooperation with S&T institutions in 1996 and 1998 and 2000 this might mean that an idea coming from outside the firm in an existing S&T-collaboration scheme might lead to continuous collaboration with the S&T institutions. As a result Dutch firms still do not produce ideas themselves but get them from S&T institutions. It might also be possible that Dutch S&T-institutions develop ideas on their own that are relevant for firms while Finnish universities are waiting for ideas from private firms and then start working on it. This needs to be investigated further! 13

14 positive coefficients in the Dutch regressions increase in case of R&D co-operation with (domestic) S&T institutions, which is consistent with the observation that basic research is of more importance in S&T institutions. Barriers to innovations due to lack of finance or sales are a motive of Finland s firms to start R&D co-operation with S&T institutions. The lack of significance for the variables avbrnt t-1 and avbrnf t-1 reveal that no heterogeneity of the sample could be found. With regard to the Pavitt dummies, as expected, Science-dummy affects all partnerships positively. The scale dummy is less important and the specialized suppliers are important in the industry only. The service sector dummies are negative as expected though mostly not significant except with regard to S&T collaboration in case of the Dutch sales and utilities services sector (Pespec t ). Firms in these sectors do co-operate systematically less with external partners than firms in other sectors. 6. Conclusions This paper addresses the question whether affiliates of foreign firms in Finland cooperate significantly more with domestic universities and public R&D institutions than in the Netherlands. These two countries have implemented different innovation policies in the last fifteen years. Finland released many resources for the improvement of the domestic S&T infrastructure s quality, while the Netherlands did not. We present a detailed empirical analysis in which we estimate the determinants of R&D-co-operation of foreign and domestic firms with the S&T system in two small economies in the European Union, Finland and the Netherlands. The data come from three waves of Community Innovation Survey (CIS-) data of Finland and the Netherlands. For both countries we estimate for the period what the determinants of R&D-co-operation of domestic and foreign firms with the domestic S&T infrastructure in both countries is. We focus on differences between the two countries in the R&D collaboration intensity of all innovating firms and industrial firms with domestic S&T institutions. The innovation policies of Finland suggest a priori that Finland is a more attractive place for foreign R&D investments than the Netherlands. The results point out that Finnish domestic firms co-operate more often with domestic S&T institutions than Dutch domestic firms. Finally, the Heckman corrected 14

15 probit estimates reveal that foreign firms in the Netherlands co-operate significantly less with institutions in the domestic S&T infrastructure than domestic firms. In the Finnish case this negative effect is not present, which means that MNE s R&D collaboration schemes with the Finnish S&T infrastructure is comparable with that of the domestic firms. This suggests that R&D co-operation of the foreign MNEs in the domestic S&T infrastructure is stronger in Finland than in the Netherlands, which can be considered as a result of different innovation policies. 15

16 References Adams, J.D., 1990, Fundamental stocks of knowledge and productivity growth, Journal of Political Economy, 98, 4, p Blomstöm, M. and A. Kokko, 2003, The Economics of Foreign Direct Investment Incentives, NBER Working Paper 9489, NBER, Cambridge. Bower, J.D., 1993 Successful joint ventures in science parks, Long Range Planning, 26, 6, p Castells, M. and P. Himanen, 2002, The Information Society and the Welfare State. The Finnish Model, Oxford University Press, Oxford, UK. De Backer, K. and L. Sleuwaegen, 2003, Why are foreigns more productive than domestic firms? Paper presented at EARIE 2003, Helsinki. Dunning, J., 1993, Multinational Enterprises and the Global Economy, Reading: Addison-Wesley Publishing Company. Fors, G. and R. Svensson, 2002, R&D and foreign sales in Swedish multinationals: a simultaneous relationship?, Research Policy, 31, p Henderson, R., A.D. Jaffe and M. Trajtenberg, Universities as a Source of Commercial Technology: A Detailed Analysis of University Patenting, , Review of Economics and Statistics, 80, 1, p Kuemmerle, W., 1997, Building effective R&D capabilities abroad, Harvard Business Review, 3/4, p Mansfield, E., 1980, Basic research and productivity increase in manufacturing, American Economic Review, 70, p

17 Nieminen, M. and E. Kaukonen, 2001, Universities and R&D Networking in a Knowledge-Based Economy. A glance at Finnish developments, SITRA Reports Series 11, Helsinki. Pavitt, K., 1984, Sectoral Patterns of Technical Change: Towards a Taxonomy and a Theory, Research Policy, 13, p Santoro, M.D. and S. Gopalakrishnan, 2000, The institutionalization of knowledge transfer activities within industry-university collaborative ventures, Journal of Engineering and Technology Management, 17, p UNIDO, 2002, Industrial Development Report 2002/2003. Competing through innovation and learning, Vienna. Van Beers, C., 2003, The Role of Foreign Direct Investments on small Countries Competitive and Technological Position, VATT Research Reports 100, Helsinki Von Zedtwitz, M. and O. Gassmann, 2002, Market versus technology drive in R&D internationalization: four different patterns of managing research and development, Research Policy, 31, pp

18 Table 1. Location drivers of research and development Reasons to locate research in a particular location Proximity to local universities and research parks Tapping informal networks Proximity to centres of innovation Limited domestic science base Access to local specialists Dissipating risks among several research units Support of local development projects Adhering to local regulations Local patenting issues Subsidies Low acceptance of research in home country Source: Von Zedtwitz & Gassmann (2002) Reasons to locate development in a particular location Local market requirements Global customers request local support Customer proximity and lead users Cooperation with local partners Market access Local citizen image Simultaneous product launching Use of different time zones Country specific cost advantages Facilitating scale-up in manufacturing Process innovation to local production National protection (intellectual property rights) 18

19 Table 2. Innovativeness and R&D collaboration of industrial firms with domestic S&T institutions: percentages for Finland (1996 and 2000) and the Netherlands (1996 and 1998). Percentage for the industrial sector only in italics Finland Netherlands D F D F D F D F % of total permanent innovators % perm. innovators R&D co-operation % R&D co-operation Public S&T institutions Dom. S&T institutions Source: Community Innovation Survey databases of Finland and the Netherlands: years 1996, 1998 and 2000 D = domestic firms; F = foreign firms 19

20 Table 3 Heckman corrected probit estimates of co-operation with partners in the Dutch and Finnish S&T-infrastructure: all sectors in 1998 (NL) and 2000 (F) partner S&T partner domestic S&T partner Finland Netherlands Finland Netherlands Finland Netherlands Constant *** (1.680) *** (0.372) *** (1.472) *** (0.469) *** (1.472) *** (0.470) LnS t *** (0.087) For t (0.347) Rdsal t *** (3.259) Rdfrq t (0.292) Brnt t * (0.556) Brnf t (0.702) Bscrd t (0.179) Orgimp t * (1.222) Finsalunc t (0.480) Costimp t (0.949) Avbrnt t (1.537) Avbrnf t (2.543) Science t 0.828** (0.405) Scale t (0.354) Spectoel t (0.321) Va t (1.060) Pespec t (0.673) *** (0.023) (0.073) ** (0.801) (0.067) ** (0.140) (0.182) *** (0.035) *** (0.209) (0.134) (0.277) (0.634) (0.818) *** (0.167) (0.095) (0.115) (0.132) (0.093) 0.426*** (0.085) (0.282) (2.247) (0.275) 1.386** (0.522) (0.649) (0.183) (0.770) (0.435) (0.934) (1.372) (2.182) ** (0.387) (0.361) (0.309) (0.767) (0.604) *** (0.027) ** (0.898) ** (0.083) ** (0.083) (0.166) *** (0.211) *** (0.042) (0.232) (0.142) ** (0.306) (0.812) (0.948) ** (0.200) (0.115) (0.137) (0.168) ** (0.119) 0.426*** (0.085) (0.282) (2.247) (0.275) 1.386** (0.522) (0.649) (0.183) (0.770) (0.435) (0.934) (1.372) (2.182) ** (0.387) (0.361) (0.309) (0.767) (0.604) *** (0.027) ** (0.091) (0.794) ** (0.083) (0.169) ** (0.211) *** (0.042) (0.238) (0.144) *** (0.308) (0.792) (0.932) ** (0.204) (0.117) (0.139) (0.168) ** (0.121) n Wald Chi 61.56*** *** 80.88** ** (15 df) Pseudo R

21 Table 4 Heckman corrected probit estimates of co-operation with partners in the Dutch and Finnish S&T-infrastructure: industrial sectors in 1998 (NL) and 2000 (F) partner S&T partner domestic S&T partner Finland Netherlands Finland Netherlands Finland Netherlands Constant *** (2.343) LnS t *** (0.111) For t (0.392) Rdsal t *** (3.601) Rdfrq t (0.364) Brnt t ** (0.825) Brnf t (0.875) Bscrd t (0.2431) Orgimp t (1.634) Finsalunc t (0.484) Costimp t (1.051) Avbrnt t (1.803) Avbrnf t (4.419) Science t 1.378** (0.496) Scale t (0.419) Spectoel t ** (0.376) * (0.447) *** (0.032) (0.095) (1.043) (0.092) * (0.192) (0.241) *** (0.047) * (0.271) (0.161) (0.361) (0.851) (1.178) *** (0.205) (0.114) (0.130) *** (2.218) 0.474*** (0.116) (0.323) 8.846*** (3.120) (0.352) 1.675** (0.726) (0.809) (0.210) (1.272) ** (0.484) (1.115) (1.777) (4.250) 1.087** (0.463) (0.392) ** (0.345) *** (0.538) *** (0.037) (0.110) (1.056) (0.110) (0.222) ** (0.268) *** (0.055) (0.312) (0.171) * (0.389) (1.043) (1.347) *** (0.231) (0.135) ** (0.153) *** (2.218) 0.474*** (0.116) (0.323) 8.846*** (3.120) (0.352) 1.675** (0.726) (0.809) (0.210) (1.272) ** (0.484) (1.115) (1.777) (4.250) 1.087** (0.463) (0.392) ** (0.345) *** (0.447) *** (0.037) * (0.112) (1.070) (0.112) (0.226) * (0.271) *** (0.055) (0.330) (0.174) ** (0.391) (1.033) (1.374) ** (0.237) (0.137) * (0.152) n Wald Chi 54.96*** *** 56.26*** *** 56.26*** *** (15 df) Pseudo R

22 Annex. Explanation of variables in equations 1 and 2. Variable St t LnS t-1 Explanation Firm co-operates with other domestic or foreign partners in the S&T system in period t the natural logarithm of the size of the firm, measured by its turnover in period t-1. For t 1 if the firm is a dependent or independent part of a firm and 0 otherwise in period t. Rdsal t-1 R&D expenditures as fraction of the turnover in period t-1 Rdfrq t-1 Firm innovates permanently in period t-1 Brnt t-1 Information source for the innovation is from inside the firm in period t-1 Brnf t-1 Information source for the innovation is from outside the firm in period t-1 (proxy for knowledge spillovers) Bscrd t-1 natural logarithm of the ratio of fundamental to applied innovations in period t-1 (basicness of research) Orgimp t-1 Score variable: a higher value means more barriers to innovation in its organization in period t-1 Finsalunc t-1 Score variable: a higher value means more barriers to innovations due to uncertainty with regard to sales and finance in period t-1 Costimp t-1 Score variable: a higher value means more cost barriers to innovations in period t-1 Avbrnt t-1 Average of inside the firm innovation source variable in period t-1 Avbrnf t-1 Average of outside the firm innovation source variable in period t-1 Science t Pavitt dummy for Science Based Industrial firms in period t Scale t Pavitt dummy for Scale Intensive Industrial firms in period t Spectoel t Pavitt dummy for Specialised Equipment Suppliers in Industrial firms in period t Va t-1 Dummy for valued added services in period t-1 Pespec t-1 Dummy for pre-specified services in period t-1 22