Web & Social Media Analytics Previous Year Question Paper.pdf
Juan Alcacer: Local R&D Strategies and Multi-location Firms: The Role of Internal Linkages
1. Local R&D Strategies and Multi-location Firms: The Role of Internal Linkages Juan Alcacer, HBS Minyuan Zhao, Univ. of Michigan Dec 12 2010
2. Benefits and costs from collocating Firms co-locate to reap benefits of agglomeration in a given location (Marshall,1922) Localized knowledge spillovers is force behind co-location in high-tech industries & R&D function . . .But co-location also have costs – and these are likely to be most pronounced for leading firms (Shaver & Flyer, 2000) Knowledge flows in both directions Leading firms may lose valuable knowledge if they co-locate with competitors How can leading global firms reduce their cost from outward spillovers to competitors when geographic distance is not an option?
3. Our contribution We explore alternative strategies that MNEs can follow to reduce outward knowledge flows when they co-locate with competitors Internal linkages We find that innovation generated in clusters with high competitor density are more likely to be associated to cross-cluster teams are more self-cited by other locations (more internalized) are less cited locally by competitors (less knowledge outflows) We bring together the local competitive environment and the global element of MNEs to better explain how firms can benefit from location-specific advantages without compromising their ability to profit from innovation
4. How do firms appropriate knowledge? Raising barriers for imitation Geographic distance Rules, routines that encourage secrecy Patents & trade-marks Legal reputation Reducing incentives for imitation Complementary assets Physical assets Intangible assets (marketing skills, managerial skills) Internal linkages However, knowledge flow is not location free; the actual spillovers – hence the actions to prevent it – mostly happen at specific locations
5. Internal linkages… Field work Interviews with IP managers, R&D managers in 7 semiconductor firms Numerous ways to raise barriers for imitation Use of complementary assets Fabs: most process innovation not useful to competitors Managerial skills: time is money Internal linkages: Cross-cluster teams Rotation of engineers Enhanced communication channels
6. Internal linkages dual use… For knowledge appropriation but also… For knowledge sourcing… Strong internal linkages lead to external knowledge absorption (Lahiri, 2010) Knowledge flows within MNEs (Frost and Zhou, 2005)
7. Internal linkages for knowledge appropriation… Strong internal linkages increase control (Nobel and Birkinshaw, 1998; Edstrom and Galbraith 1977) and coordination within MNEs To the extent that MNEs are able to integrate and internalize knowledge on a global basis they can build on new technologies faster than imitators (Buckley & Casson, 1976; Bartlett & Ghoshal, 1990 ) Strong internal linkages increase internal interdependence (Liebeskind, 1996; Zhao, 2006)
8. Propositions We expect to observe more cross-cluster teams at locations with higher appropriation risks, e.g. in clusters with a large number of direct competitors more intensive intra-firm knowledge flows across clusters with the presence of cross-cluster teams less knowledge outflow to local competitors with the presence of cross-cluster teams
9. Data and Sample The global semiconductor industry, from 1998 to 2001 Agglomeration benefits are well documented Innovations easily traced through patents Technology used across different industries Derwent as main source for innovation Characterize clusters with comprehensive profile: Basic science: 50,387 publications – ISI Web of Knowledge Innovation: 60,880 patents (28,334 US patents) belonging to 2,217 organizations Development 974 plants – World Fab Watch Production: 549 fabless firms – Directory of Fabless Semiconductor Companies …but 16 leading innovators of the industry as focal firms
11. Empirical approach Test whether innovations generated in clusters with high density of competitors differ as predicted in terms of Presence of cross-cluster teams Cross-cluster self-citations (internalization) Citations by local competitors
12. Empirical approach Patent families instead of patents Test whether innovations generated in clusters with high density of competitors differ as predicted in terms of Presence of cross-cluster teams Cross-cluster self-citations (internalization) Citations by local competitors
13.
14. Empirical approach Organic algorithm instead of administrative units Test whether innovations generated in clusters with high density of competitors differ as predicted in terms of Presence of cross-cluster teams Cross-cluster self-citations (internalization) Citations by local competitors
20. Data for clusters Obtain latitude and longitude for all locations International: Geonet Names Server (GNS) by the National Geospatial Intelligence Agency Dataset with 5.5 million names of locations worldwide Alternative spelling USA: Geographic Names Information System (GNIS) by the US Geological Survey (USGS) Initial clustering based on inventor locations, then add plants, fabless, publications
22. Empirical approach Test whether innovations generated in clusters with high density of competitors differ as predicted in terms of Presence of cross-cluster teams Cross-cluster self-citations (internalization) Citations by local competitors Alternative definitions
23. Identifying actors in cluster Small Firms Based on USPTO Large Firms Industry Profit No industry Based on Hoovers Segment Organizations in cluster (2,217) No segment Based on names, USPTO, internet Competitor No competitor Universities Non-Profit Gov’t entities Based on names Other non-profit
24. Empirical approach Test whether innovations generated in clusters with high density of competitors differ as predicted in terms of Presence of cross-cluster teams Cross-cluster self-citations (internalization) Citations by local competitors Based on patent family data
25. Dependent Variables Cross-cluster teams Binary level indicating of family is associated to a cross-cluster team Internalized value Cross-cluster self-citations (Trajtenberg et al., 1997; Hall et al., 2005) Cross-cluster self-citations per patent family =∑cross-cluster self-citations of US patents Main analysis using only assignee citations Citations by local competitors Competitors according characterization of competition in cluster: all assignees, for profit assignees, firms in industry, firms in segment and competitors.
29. + significant at 10%; * significant at 5%; ** significant at 1% cross_clusterfict= Cict+ Xf+ Yict + Zct + ζt +υi + τctry +γtech+ εfict
30. + significant at 10%; * significant at 5%; ** significant at 1% cross_cluster_self_citationfict= Cict+ Xf+ Yict + Zct + ζt +υi + τctry +γtech+ εfict
31. + significant at 10%; * significant at 5%; ** significant at 1% local_citations_by_ENTITYfct= Cict+ Xf+ Yict + Zct + ζt +υi + τctry +γtech+ εfict
32. Summarizing results Innovations generated by MNEs in clusters with high competitor density are different More likely to be associated to a cross-cluster team More cross-cluster self-cited (beyond effect of cross-cluster teams) Innovations associated to cross-cluster teams are different More cross-cluster self-cited (cross-cluster teams increase internalization) Less cited by local competitors (cross-cluster teams as appropriability)
33. Robustness checks Hierarchical clusters vs. organic clusters Percentages instead of counts for self-citations and citations by local competitors Adding examiner self-citations and citations: weaker results Analysis for top 5% of multi-location companies in semiconductors Analysis for more than 25 clusters Collinearity: Orthogonalized variables 1 variable to characterize clusters Ratio variables All locations Including 1 variable at a time
34. Contributions Explain how MNEs can still benefit from innovating in clusters where competitors are also present Offer a richer picture of clusters combining two elements Local environment: organizations innovating across technological space Global actors: deciding what to allocate to specific clusters depending on global competition Bring location to the appropriability literature Bring internal organization to cluster literature
35. Contributions Definition of clusters A dataset with latitude and longitude data for all patents since 1969 Identifying results by differentiating technology space vs. product market space
36. Measuring control Number of family patents with inventors across-clusters Differentiating between core and peripheral Peripheral cluster Core cluster
50. Location & Strategy Knowledge seeking & agglomeration International expansion & competition Patents & MNEs Knowledge Seeking and FDI Location, Mgt Sci ‘02 Strategic Interaction in International Strategies, under review AMR Patent Citations as a Measure of Knowledge Flows, REStat ’06 Location Strategies & Knowledge Flows, forthcoming Mgt Sci Location Choices Across the Value Chain, Mgt Sci 06 Patent Quality, R&R Research Policy Location Strategies and Agglomeration Economies Competition and Dynamic International Strategies International Patenting Strategies The Impact of Firm Rivalry on Location Choices Transferring intangibles across border Global Competitors as Next-Door Neighbors: Competition and Geographic Co-location in the Semiconductor Industry Global value chain fragmentation Globalization The geography of drug development and clinical trials, NBER The Intergovernmental Network and the Governance of FDI Fragmentation in the Value Chain in the Wireless Telecommunication Industry
51. How firms can use their location decisions to… …acquire new capabilities? …compete across geographic markets? Knowledge Seeking & FDI Location, Mgt Sci ‘02 Location Choices Across the Value Chain, Mgt Sci ‘06 Location Strategies & Knowledge Flows, Mgt Sci forthcoming Strategic Interaction in International Strategies, under review AMR Location Strategies & Agglomeration Economies Competition and Dynamic International Strategies Location in drug industry, NBER chapter What do patent data mean? Global Competitors as Next-Door Neighbors: Competition & Geographic Co-location Patent Citations as a Measure of Knowledge Flows, REStat ’06 Patent Quality, R&R Research Policy Transferring intangibles overseas
52. Technological distance to minimize outward flows MNEs can differentiate across multiple dimensions For an outward knowledge spillover to occur, the receiving firm must be able to identify and absorb knowledge Technological distance affects competitors’ scope of search (Stuart & Podolny 1996) and their capacity to absorb knowledge (Cohen & Levinthal 1990) Leading firms are more likely to generate innovations that are technologically distant from competitors’ innovations in clusters with high competitor density
53. Does technological distance mean no learning? Jacobs-Porter vs. Marshall-Arrow-Romer clusters Different actors across value chain in same industry Same technology used in other industries Automobiles Aerospace Defense Electronics Telecommunications Groups within the cluster with semiconductor MNEs as centers or small groups Still plenty of opportunities for knowledge seeking
54. Propositions In clusters with high competitor density, MNEs will generate innovations that are technologically distant from innovations generated by competitors
55. + significant at 10%; * significant at 5%; ** significant at 1% average_technological_distanceict = Cict +Xict + Yct + ζt + υi + τctry+εict