Université Libre de Bruxelles (ULB) Solvay Brussels School of Economics and Management (SBS-EM) European Center for Advanced Research in Economics and Statistics (ECARES) The R&D-patent relationship: An industry perspective Jérôme Danguy, Gaétan de Rassenfosse and Bruno van Pottelsberghe The Output of R&D activities: Harnessing the Power of Patents Data II IPTS, Sevilla May 2010
Outline Motivation: global patent warming : R&D Patent The results basic model full model Dynamic propensities The R&D-patent relationship: An industry perspective 2
Evolution of patent filings Most widely used indicator of technology output 400000 350000 Global Patent Warming 300000 250000 200000 150000 100000 50000 0 NPFCORR EPO TRIADIC USPTO What does this trend reflect? Increase in research inputs? Increase in research productivity? Increase in propensity to patent? The R&D-patent relationship: An industry perspective 3
Starting point: G. de Rassenfosse & B. van Pottelsberghe (2009) Both productivity and propensity matter Education S&T IP system Research Invention Input Productivity Propensity Patent(s) Test different patent indicators Industry and time dimensions The R&D-patent relationship: An industry perspective 4
Industry : a missing link in the litterature Research effort: key determinant of patenting NPFCORR Patents per country (2004) 120.000 100.000 80.000 60.000 40.000 20.000 0 $ 0 $ 100 NPFCORR Patents per industry (2004) 90.000 80.000 70.000 60.000 50.000 40.000 30.000 20.000 10.000 0 $ 0 $ 50 R&D Billions R&D Billions Strong country variation, stronger industry variation: Alternative mechanisms of appropriation Strategic role of patents The paper aims at better understanding the R&D-patent relationship at industry level The R&D-patent relationship: An industry perspective 5
Two propensities to patent A high «appropriability» does not necessarily imply a high propensity to patent Appropriability Share of inventions that are patented* 80 70 60 50 40 30 20 10 0 PHAR INST CHEM COMP MACH ELEC COMM RUBB FABM TRAN MINE AUTO PETR META FOOD 0 0,5 1 1,5 2 2,5 3 3,5 Patents per R&D expenditures (2000) *Source: Arundel and Kabla (1998) The R&D-patent relationship: An industry perspective 6
A new R&D-patent relationship Research Input Research productivity Invention Appropriability propensity Patent Y/N Strategic propensity How many? The share of inventions which are patented The number of patents filed to protect a given invention Many potential determinants in addition to R&D Appropriability mechanisms Arundel and Kabla (1998) Human capital Cohen et al. (2000) Academic research International competition Strategic filings Design of patent system The R&D-patent relationship: An industry perspective 7
The Error Correction Model Short-run impacts Long-run impacts p r z x ( p c r z x + v ij, t = ψ i + ψ j + ψ t + γ s ij, t + βs ij, t + α s ij, t λ ij, t 1 γ ij, t 1 β ij, t 1 α ij, t 1) ij, t Where each individual is an industry (i) in a country (j) p is the log of patent counts (5 alternative indicators) r is the log of R&D expenditures x is the vector of variables affecting the research productivity Share of basic research Share of higher education International competition z is the vector of variables affecting the propensity to patent Appropriability IP Index ψ, ψ, ψ i j t are the industry, country and time fixed effects Panel data composed of 18 industries, 19 countries and 19 years (1987 to 2005) The R&D-patent relationship: An industry perspective 8
The dependent variable: 5 patent indicators There exist many patent indicators each carrying its own meaning 1. Corrected count of national priority filings (NPFCORR) 2. Patent applications filed at the EPO 3. USPTO 4. REGIONAL: EPO for EU and USPTO for RoW 5. TRIADIC patent families Protection in local market Extension in the Triad VALUE VALUE The R&D-patent relationship: An industry perspective 9
The basic R&D-patent relationship log(#patents) log(r&d) log(#patents) (t-1) NPFCORR TRIADIC EPO USPTO REGIONAL (1) (2) (3) (4) (5) 0.009 0.013 0.009-0.013 0.014 (0.007) (0.015) (0.009) (0.010) (0.009) -0.119*** -0.290*** -0.155*** -0.145*** -0.149*** (0.007) (0.010) (0.008) (0.008) (0.008) log(r&d) (t-1) 0.014*** 0.032*** 0.018*** 0.017*** 0.019*** (0.002) (0.005) (0.003) (0.003) (0.003) Countries dummies Yes *** Yes *** Yes *** Yes *** Yes *** Industry dummies Yes *** Yes *** Yes *** Yes *** Yes *** Time dummies Yes *** Yes *** Yes *** Yes *** Yes *** Number of observations 4943 4943 4943 4943 4943 Adjusted R-Squared 0.197 0.187 0.156 0.171 0.129 Long-run impact of R&D 0.118*** 0.110*** 0.116*** 0.123*** 0.128*** Notes: Standard errors in parentheses, ***, **, * denote significance at the 1, 5 and 10-percent levels, respectively. The R&D-patent relationship: An industry perspective 10
Including the productivity and propensities variables log(#patents) NPFCORR TRIADIC REGIONAL (1) (2) (3) APPROPRIABILITY 0.004*** 0.012*** 0.005*** (0.000) (0.001) (0.000) IP INDEX 0.031** 0.053** 0.073*** (0.012) (0.023) (0.015) log(r&d) -0.003-0.010-0.008 (0.008) (0.016) (0.010) INTL COMP -0.002 0.098*** 0.052*** (0.016) (0.030) (0.019) SHARE HIGHER EDU -0.010*** -0.002-0.008*** (0.002) (0.004) (0.002) log(#patents) (t-1) -0.142*** -0.279*** -0.137*** (0.008) (0.012) (0.009) log(r&d) (t-1) 0.014*** 0.013** 0.007* (0.003) (0.006) (0.004) INTL COMP (t-1) 0.028*** 0.100*** 0.056*** (0.009) (0.017) (0.011) SHARE HIGHER EDU (t-1) 0.0002-0.002 0.005*** (0.001) (0.002) (0.001) Countries dummies Yes *** Yes *** Yes *** Industry dummies Yes *** Yes *** Yes *** Time dummies Yes *** Yes *** Yes *** Number of observations 3696 3696 3696 Adjusted R-Squared 0.237 0.190 0.140 Long-run impact of R&D 0.099*** 0.047** 0.051* Long-run impact of IC 0.197*** 0.358*** 0.409*** Long-run impact of SHE 0.001 0.007 0.036*** Notes: Standard errors in parentheses, ***, **, * denote significance at the 1, 5 and 10-percent levels, respectively. The R&D-patent relationship: An industry perspective 11
The dynamic propensities country and industry Dummies captures the remaining propensity to patent the «dynamic propensity» net of the impact of all other variables Comparative analysis: normalized coefficients Country Dummies Industry Dummies 1 0,9 TRIADIC 1 0,9 TRIADIC 0,8 0,8 0,7 0,6 0,5 0,4 0,7 0,6 0,5 0,4 0,3 0,3 0,2 0,1 0 KR PL NO ES IE DK FI NL DE JP FR CA IT BE GB US SE 0,2 0,1 0 COMM COMP INST ELEC PHAR AUTO TEXT MINE CHEM TRAN FOOD MACH META RUBB FABM WPAP PETR Catching-up effect ICT effect The R&D-patent relationship: An industry perspective 12
The dynamic propensities time 300 250 200 150 USPTO EPO TRIADIC NPFCORR 100 50 0 y1987 y1990 y1993 y1996 y1999 y2002 y2005 Globalisation effect: Companies do not file particularly more national patents but have a higher willingness to extend them abroad The R&D-patent relationship: An industry perspective 13
R&D is a significant determinant of patent filings at industry level Both productivity and propensity matters in the R&D-patent relationship International competition IP-index: design of patent systems Academic research Appropriability propensity Two components of the propensity to patent: appropriability propensity and strategic propensity but lack of data The dynamic propensity to patent differs across countries and industries: catch-up and ICT The propensity has strongly increased over time for international applications: globalization phenomenon The R&D-patent relationship: An industry perspective 14
Thank you Questions? 15
Back-up slides 16
Detailed description of the model Research efforts (R) lead to inventions (I) which, in turn, may lead to patent applications (P) I = ΩR γ P = ΦI (1) Where Ω and Φ capturing, respectively, the productivity and the propensity (appropriability and strategic) effects Taking logs and defining x and z as the variables affecting productivity and propensity (2) 2 Expanding the patent production function gives for industry (i), country (j) and year (t): (3) With c= c 1 + c 2 (4) p i = c 1 + α x + γ r p = c + β z + i = c Taking first differences, assuming that the remaining productivity effect c 1 is constant over time p = c2, + γ r + β z + α x + ε + γ r + β z + α x + υ (5) Estimating the patent production function by means of an error correction model (ECM) p ψ ψ ψ γ β α λ ( γ β α + v ij, t = i + j + t + s rij, t + s zij, t + s xij, t pij, t 1 c rij, t 1 zij, t 1 xij, t 1) ij, t Dummies Short-term impacts Long-term impacts The R&D-patent relationship: An industry perspective 17
Descriptive facts on our 5 patent indicators Research effort and patenting activity(2004) 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 374,106 104,230 218,673 231,927 50,504 R&D NPFCORR EPO USPTO REGIONAL TRIADIC EU US-CA JP-KR 0,30 Share of triadic in NPFCORR by industry 0,25 0,20 0,15 0,10 0,05 PHAR CHEM PETR Average MACH FABM 0,00 1987 1992 1997 2002 The R&D-patent relationship: An industry perspective 18
: the impact of basic results log(#patents) NPFCORR TRIADIC REGIONAL (1) (2) (3) log(r&d) 0.019-0.004 0.020 (0.013) (0.029) (0.018) SHARE BASIC 0.016*** -0.0002-0.005 (0.003) (0.007) (0.004) log(#patents) (t-1) -0.114*** -0.365*** -0.192*** (0.011) (0.019) (0.014) log(r&d) (t-1) 0.016*** 0.041*** 0.022*** (0.005) (0.011) (0.007) SHARE BASIC (t-1) 0.019*** 0.029*** 0.023*** (0.003) (0.006) (0.004) Countries dummies Yes *** Yes *** Yes *** Industry dummies Yes *** Yes *** Yes *** Time dummies Yes *** Yes *** Yes *** Number of observations 1811 1811 1811 Adjusted R-Square 0.331 0.241 0.170 Long-run impact of R&D 0.140*** 0.112*** 0.115*** Long-run impact of SB 0.167*** 0.079*** 0.120*** Notes: Standard errors in parentheses, ***, **, * denote significance at the 1, 5 and 10-percent levels, respectively. The R&D-patent relationship: An industry perspective 19
: Strategic propensity measured by Complexity log(#patents) NPFCORR TRIADIC REGIONAL (1) (2) (3) APPROPRIABILITY 0.004*** 0.012*** 0.005*** (0.000) (0.001) (0.000) IP INDEX 0.031*** 0.053** 0.073*** (0.012) (0.023) (0.015) COMPLEXITY* 0.003*** 0.005*** 0.003*** (0.000) (0.000) (0.000) log(r&d) -0.003-0.010-0.008 (0.008) (0.016) (0.010) INTL COMP -0.002 0.098*** 0.052*** (0.016) (0.030) (0.019) SHARE HIGHER EDU -0.010*** -0.002-0.008*** (0.002) (0.004) (0.002) log(#patents) (t-1) -0.142*** -0.279*** -0.137*** (0.008) (0.012) (0.009) log(r&d) (t-1) 0.014*** 0.013** 0.007* (0.003) (0.006) (0.004) INTL COMP (t-1) 0.028*** 0.100*** 0.056*** (0.009) (0.017) (0.011) SHARE HIGHER EDU (t-1) 0.0001-0.002 0.005*** (0.001) (0.002) (0.001) Countries dummies Yes *** Yes *** Yes *** Industry dummies Yes *** Yes *** Yes *** Time dummies Yes *** Yes *** Yes *** Number of observations 3696 3696 3696 Adjusted R-Squared 0.236 0.190 0.140 Long-run impact of R&D 0.099*** 0.047** 0.051* Long-run impact of IC 0.197*** 0.358*** 0.409*** Long-run impact of SHE 0.001 0.007 0.036*** * source: von Graevenitz et al. 2008 The R&D-patent relationship: An industry perspective 20