The structural transformations of internationalized R&D activities: An analysis of patents data Lucio Picci and Luca Savorelli University of Bologna III Workshop The Output of R&D Activities: Harnessing the Power of Patent Statistics 13 14 June 2011 JRC IPTS, Seville, Spain
The Eschirichia Coli epidemics The Eschirichia Coli epidemics killed in little time 17 people and infected 1600 across 13 European countries only 3 hours for researchers to produce a preliminary genetic sequence identifying this particularly virulent E. coli bacterium using the machine in China and Germany This very fast result could happen thanks to the international collaboration between German and Chinese researchers: BIG research center Shenzhen, China the University Medical Centre Hamburg Eppendorf, Germany
Structure of the presentation 1) Research questions and summary of preliminary results 2) The data: the global picture and its structural evolution 3) Determinants of internationalization in R&D: a gravity model 4) Conclusions and future research
The internationalization of inventive activity Patel and Pavit, 1991: internationalization of production vs. localized inventive activity The times they are a changing: Patel and Vega, 1999; Le Bas and Sierra, 2002 Anecdotal evidence: The Economist, 2010 Companies in the Fortune 500 list have 98 R&D facilities in China and 63 in India. [ ] General Electric s health-care arm has spent more than $50m in the past few years to build a vast R&D centre in India s Bangalore,[ ] Cisco is splashing out more than $1 billion on a second global headquarters Cisco East in Bangalore, [ ] Microsoft s R&D centre in Beijing is its largest outside its American headquarters in Redmond. Industry case studies Pharmaceuticals; Bennato and Magazzini, 2009 Biotech; Shan and Song, 1997 Semiconductors; Almeida, 1996
Research questions We use worldwide patents data (Patstat) following an innovative approach (de Rassenfosse et al.,2011). 1.What s the global view on international collaborations in patenting? 2.What are the differences across technologies and across countries? in the intensity in the nature 3. What are the determinants of international collaborations and how do they evolve across time?
Contribution to the literature /1 Structural changes in internationalized patenting activity Guellec and van Pottelsberghe de la Potterie,2001; OECD, 2008; Picci, 2010: take global view regardless of technological sectors We break down data by WIPO s International Patent Classification (IPC) taxonomy. Specific contributions: 1.Differences in intensity and nature by tech sector; informative contrast between different metrics; home base augmenting vs home base exploiting ; Kummerle, 1997 2.Disentangle pure growth effects from compositional effects
Contribution to the literature /2 Determinants of international collaborations Gravity model at aggregate level and at sectorial level. Role of: 1.Physical distance; 2.Cultural distance; Determinants of evolution over time.
Preliminary results Important structural differences in the process of internationalization of R&D appear both across countries and across technologies 1.Increase of the electrical engineering sector and prominent role in internationalization. 2.Considerable levels of internationalization in some countries and specific sectors: e.g. UK: more than 30% of patents in the electrical engineering sector are int. 3.Qualitative aspects: pre eminent role of US applicants abroad; more tangled figures for some European countries (e.g. Germany) 4.Positive but modest role of compositional effects 5.Geographic distance negatively affects collaborations increasingly over time sensibly more in some sectors (e.g. mechanical engineering) 6. Cultural proximity plays a positive role
Fractional counting 1 Patent 4 inventors from: US, US, IT, ES 2 applicants: US, ES
Fractional counting 1 Patent 0,25 0,25 0,25 0,25 0,50 4 inventors from: US, US, IT, ES 2 applicants: US, ES
InvApp absolute measure of internationalization 1 Patent 0,25 0,25 0,25 0,25 0,50 4 inventors from: US, US, IT, ES 2 applicants: US, ES InvApp US, ES =Inv US x App ES = 0.5 x 0.5 = 0.25 InvApp ES, US =Inv x App ES US = 0.25 x 0.5 = 0.125
InvApp Inv relative measure of internationalization InvApp US, ES =Inv US x App ES = 0.5 x 0.5 = 0.25 1. This is the absolute measure for 1 patent. 2. Sum across patents and get the country score for internationalization. 3. Weight over the total fractional counting according to inventors. What you get is the InvApp Inv measure, % of international patents over the total of a given country: InvApp US, ES Inv US
Overall degree of internationalization, 1990 2006 InvApp Inv metric, the United Kingdom, France, Germany, US, and Japan
Macro technological sectors United Kingdom Other fields 33 Furniture, games 34 Other consumer goods 35 Civil engineering Electrical engineering 1 Electrical machinery, apparatus, energy 2 Audio visual technology 3 Telecommunications 4 Digital communication 5 Basic communication processes 6 Computer technology 7 IT methods for management 8 Semiconductors
Patents shares of technological sectors, 1990 2006 Inv. from the USA, Germany, and Japan United States Germany Japan
Shares of international patents by tech sector, 1990 2006 InvApp Inv metric for the USA, France, and the United Kingdom 6% United States 20% France 30% United Kingdom
Comparison between measures of internationalization / 1 InvApp Inv and InvApp App metrics for the USA, France, and Germany, 1990 2006 InvApp App = InvApp ES, US App US United States France Germany Figura 5
Comparison between measures of int. by tech sector / 2 InvApp Inv and InvApp App in Germany, 1990 2006 Germany Chem Electr Mech Figura 5
Breakdown in compositional and pure growth effects Growth rate of int., InvApp Inv metric for selected countries Figura 5
A gravity model of sectorial inventive activity ln( InvApp s ijt ) s s = 0 + β1 ln( Ait ) + β2 ln( Ajt ) + β3 ln( distij ) + λlijt + β4dit + β5 β D + ε jt ijt macro sector Inv.mass country i (pat count) Inv.mass country j (pat count) Distance between i & j capital cities Vars vector: Com. borders Com. language Similar language Religion EU members Eu. Monetary Un. Tech(s) 35 sect Tech 5 sect IP protection Country spec. Fixedfxs
OLS estimation (1) All collaborations (2) Non-zero (3) (4) (5) (6) (7) VARIABLES linvapp 0 All technologies linvapp 0 All technologies linvapp 1 Electrical linvapp 2 Instruments linvapp 3 Chemistry linvapp 4 Mechanical linvapp 5 Other linvselfx_0-5 -0.0707 0.558*** 0.989*** 0.301 0.551** 0.900*** 0.546 (0.246) (0.141) (0.226) (0.206) (0.219) (0.240) (0.423) lappselfy_0-5 0.151 0.507*** 0.710*** 0.455** 0.364 0.678*** 0.419 (0.238) (0.140) (0.219) (0.209) (0.223) (0.237) (0.394) ln(distance) -0.323*** -0.281*** -0.282*** -0.260*** -0.391*** -0.407*** -0.242*** (0.0263) (0.0392) (0.0686) (0.0667) (0.0613) (0.0621) (0.0715) Border 0.525*** 0.268*** 0.370*** 0.266** 0.295*** 0.558*** 0.643*** (0.0540) (0.0673) (0.118) (0.114) (0.105) (0.108) (0.123) Language 0.171** 0.386*** 0.637*** 0.131 0.368** 0.417*** 0.730*** (0.0708) (0.0915) (0.164) (0.155) (0.143) (0.144) (0.167) EU Union -0.0476-0.384*** -0.253-0.370* -0.934*** -0.682*** -0.0493 (0.0695) (0.125) (0.220) (0.212) (0.196) (0.198) (0.230) Euro Zone 0.0409 0.130 0.181 0.383** -0.0320-0.0365 0.276 (0.0793) (0.112) (0.196) (0.189) (0.175) (0.177) (0.207) Tech prox 0 1.397*** 1.106*** 1.436*** 1.205*** 1.065*** 0.792*** 0.562* (0.0979) (0.165) (0.342) (0.287) (0.293) (0.269) (0.312) Tech prox 1-5 -0.564* -0.846*** 0.0193 0.272-0.447*** (0.312) (0.244) (0.179) (0.240) (0.161) ipr_x -0.378-0.0796 0.507-0.471-0.0900-0.176 0.294 (0.262) (0.268) (0.462) (0.442) (0.412) (0.428) (0.359) ipr_y 0.0247-0.0327 0.205 0.325-0.485 0.284-0.483 (0.272) (0.527) (0.930) (0.889) (0.821) (0.830) (0.987) Language sim 1.010*** 0.694*** 1.099*** 1.032*** 0.429** 0.685*** 0.115 (0.0872) (0.111) (0.194) (0.190) (0.173) (0.177) (0.207) Religion sim 0.497*** -0.557** -0.607-0.558-0.864** 0.141-0.178 (0.0943) (0.223) (0.390) (0.377) (0.350) (0.351) (0.406) Observations 6439 1658 1658 1658 1658 1658 1658 R-squared 0.754 0.857 0.800 0.759 0.752 0.769 0.653
Elasticity of distance over time (1) (2) (3) (4) (5) (6) (7) VAR. linvapp 0 All technologies linvapp 0 All technologies linvapp 1 Electrical linvapp 2 Instruments linvapp 3 Chemistry linvapp 4 Mechanical linvapp 5 Other a. 1990-2006 ldist -0.323*** -0.281*** -0.282*** -0.260*** -0.391*** -0.407*** -0.242*** (0.0263) (0.0392) (0.0686) (0.0667) (0.0613) (0.0621) (0.0715) b. 1990-1998 ldist -0.280*** -0.228*** -0.175* -0.283*** -0.245*** -0.333*** -0.308*** (0.0362) (0.0520) (0.101) (0.0999) (0.0859) (0.0944) (0.0988) c. 1999-2006 ldist -0.367*** -0.327*** -0.366*** -0.210** -0.524*** -0.474*** -0.199* (0.0383) (0.0582) (0.0954) (0.0909) (0.0869) (0.0836) (0.104) Obs. (a,b,c) 6439, 3016, 3423 1658, 746, 912 1658, 746, 912 1658, 746, 912 1658, 746, 912 1658, 746, 912 1658, 746, 912
Summary of results & future research 1. Increase of the electrical engineering sector and prominent role in internationalization. 2. Considerable levels of internationalization in some countries and specific sectors: e.g. UK: more than 30% of patents in the electrical engineering sector are int. 3. Geographic distance negatively affects collaborations increasingly over time (but not in all sectors) sensibly more in some sectors (e.g. mechanical engineering) Next step: detecting the sources of change in elasticities across time and across sectors.
Thanks! Contacts Luca.savorelli@unibo.it Lucio.picci@unibo.it