Industrial clusters have attracted considerable attention worldwide for their expected contribution to regional innovation. Recently, policymakers in various countries have developed specific cluster policies. However, there exist few empirical studies on cluster policies. In this study, we use original questionnaire data to empirically evaluate the Industrial Cluster Project (ICP) initiated by the Ministry of Economy, Trade and Industry in 2001. We address two research questions on the effect of the ICP: are the project participants who exploit various support programs more successful in alliance/network formation within the cluster than others and which kind of ICP support program contributes to firm performance.In a departure from previous such projects, the ICP aims at the autonomous development of regional industries and includes both direct R&D support and indirect networking/coordination support. The focus of public support for local firms has clearly shifted toward enabling networking and coordination for those who can help themselves. Thus, we pay special attention to the differences between direct R&D support and indirect networking/coordination support, which indicate the conditions necessary for the effective organization of cluster policies to improve firm performance.Our empirical evaluation is based on a recent original survey of a sample of 511 firms. We first employ the propensity score and the difference-in-differences estimation to analyze the degree of alliance/network formation before and after participating in the ICP. Thereafter, we use Heckman's two-step procedure and the negative binomial model to examine the effects of support programs on firm performance.The estimation results suggest that cluster participants who exploit support programs (especially indirect support measures) expand the industry-university-government network after participating in the ICP. Moreover, we find that not every support program contributes to firm performance; firms should therefore select the program that is most aligned with their aims. Indirect support programs have an extensive and strong impact on output, especially innovation outcomes, whereas direct R&D support has only a weak effect.Keywords: cluster policy, industrial cluster, R&D support, subsidy, networking JEL codes: O25; O38; R11 2 1. Introduction Industrial clusters have been attracting considerable attention worldwide for their expected contribution to regional innovation and development, especially through enhanced collaboration and knowledge spillover among local firms and research organizations. Ever since the 1990s, policymakers in various countries have launched specific cluster policies (see Appendix 1).However, to the best of our knowledge, there exist few empirical studies on the effects of cluster policies on the performance of local firms, whereas agglomeration theories derive a major implication that regional differences in agglomeration explain differences in regional industry structure and performance (Porter, 1998;Glaeser...

Industrial clusters have attracted increasing attention as important locations of innovation. Therefore, several countries have started promotion policies for industrial clusters. However, there are few empirical studies on cluster policies. This paper examines the effects of the "Industrial Cluster Project" (ICP) in Japan on the R&D productivity of participants, using a unique dataset of 229 small firms, and discusses the conditions necessary for the effective organization of cluster policies. Different from former policy approaches, the ICP aims at building a collaborative network between universities and industries and supports the autonomous development of existing regional industries without direct intervention in the clustering process. Thus far, the ICP is similar to indirect support systems adopted by successful European clusters. Our estimation results suggest that participation in the cluster project alone does not affect R&D productivity. Moreover, research collaboration with a partner in the same cluster region decreases R&D productivity both in terms of the quantity and quality of patents. However, cluster participants apply for more patents than others without reducing patent quality when they collaborate with national universities in the same cluster region. These results imply the effectiveness of indirect support systems that remove obstacles and relax constraints in the clusters. In order to improve the R&D efficiency of local firms, it is also important to construct a wide-range collaborative network within and beyond the clusters, although most clusters focus on the network at a narrowly defined local level. These characteristics may be important factors for the effective organization of cluster policies.

Despite various expected advantages, university-industry research collaboration (UIC), a relationship between two different worlds, often faces serious difficulties. Thus, the performance of UIC depends on the research partners' strategies to bridge the gaps between them according to the institutional environment. In Japan, UIC has developed rapidly since the late 1990s based on drastic institutional changes regarding universities. We pay special attention to the role of the university intellectual property (IP) policy introduced after 2003 and empirically examine its impact on the performance of UIC projects. A clear and equitable IP policy that can be applied flexibly to the needs of partners would be optimal for a UIC to be efficiently managed. Otherwise, the project might face serious conflicts of interests and low incentive for cooperation. Using a sample of Japanese firms from our original survey, we find that the IP policy of partner universities indeed has a positive and significant impact on various performances of UIC projects, controlling for firm and project characteristics and considering potential selection bias from UIC participation.

R&D consortia (collaborative R&D projects among private firms, universities, and public research institutes) have been attracting increasing attention as an effective means of promoting innovation. Especially for SMEs, such collaboration provides important opportunities to access and obtain advanced scientific knowledge generated by universities and public research institutes. It is expected that not only the participants in R&D consortia will enhance their performance, through direct knowledge spillovers, but also that the business partners of consortia members may enjoy indirect effects (rent spillovers), through their business transactions. This paper empirically examines the spillover effects through government-sponsored R&D consortia using firm-level data and the propensity score method.Focusing on a major support program for R&D consortia in Japan, the "Consortium R&D Project for Regional Revitalization" by METI, we confirm that there are both direct (knowledge) spillover effects from firms' participation in this program and indirect (rent) spillover effects on the customer firms of the consortia members. Moreover, by comparing SMEs and large firms, we find that only SMEs obtain knowledge spillovers in R&D consortia, whereas, among their customers, only large firms enjoy rent spillovers.

No abstract