Measuring patent quality and radicalness of patents filed by regions: example of the ICT sector Hélène Dernis OECD Directorate for Science Technology and Industry, Economic Analysis and Statistics Patent statistics, innovation management and IPR, Paris June 5, 2012, 5 June 2012
Measuring patent quality The patent based indicators proposed: Are based on existing literature; Try to capture the technological importance of the invention, its economic value, and the possible impact on subsequent technological developments. Rely on information contained in the patent documents: can be constructed for all patents, rely on a homogeneous set of information, comparable across countries and over time. Are compiled for each EPO patent document from EPO s Worldwide Statistical Database (PATSTAT). Indicators are normalised with respect to maximum values obtained in cohorts - year & technology field (WIPO/Schmoch, 2010) ICT patents are identified using OECD definition.
Measuring patent quality The patent based indicators proposed: Are based on existing literature; Try to capture the technological importance of the invention, its economic value, and the possible impact on subsequent technological developments. Rely on information contained in the patent documents: -> can be constructed for all patents, NOTES -> rely on a homogeneous set of information, The assessment of the patent quality relies on the sole content of patent document -> comparable as available across in patent countries databases. and over time. Are compiled for each EPO patent document from EPO s Worldwide Statistical Database (PATSTAT). No additional information about e.g. market transactions or real use of the (patented) technology available are exploited. Indicators are normalised with respect to maximum values obtained in cohorts - year & technology field (WIPO/Schmoch, 2010) Using different data sources may lead to different results. ICT patents are identified using OECD definition.
Patent scope Is associated with the technological and economic value of inventions: Patent scope relates to the valuation of a firm; broad patents are more valuable (Lerner, 1994). Patents scope to be used to foster early disclosure of fundamental innovations (Matutes et al., 1996). Definition: (follows Lerner, 1994). Number of 4-digit subclasses of the International Patent Classification (IPC) the invention is allocated to. Larger number broader scope. 0.5 0.4 0.3 0.2 0.1 0.0 Normalised index, 2005-08 (average) All patents ICT All (median) ICT (Median) Source: OECD calculations based on PATSTAT, EPO, October 2011.
Patent family size Families are patents filed in several countries and related to each other by one or several common priorities. Patent family size: Is associated to the economic value of patents (Lanjouw et al., 1998). Large international patent families have been found to be particularly valuables (Harhoff et al., 2003). Definition: number of patent offices at which an invention has been protected by a patent. Larger number more valuable patent. 0.5 0.4 0.3 0.2 0.1 0.0 Normalised index, 2005-08 (average) All patents ICT All (median) ICT (Median) Source: OECD, calculations based on PATSTAT, EPO, October 2011.
Number of claims Is associated with the technological and economic value of inventions: Determines the technology and subject matter protected by law. Reflects the expected economic value of a patent (Tong & Davidson, 1994; Lanjouw & Schankerman, 2001, 2004). Definition: number of claims per patent. Larger number more valuable patent. 0.5 0.4 0.3 0.2 0.1 0.0 Normalised index, 2005-08 (average) All patents ICT All (median) ICT (Median)
Backward citations Patents cited in the patent document. Are used to assess patentability: Are positively related to value of patents (Harhoff et al., 2003). BUT May signal inventions of incremental nature (Lanjouw & Schankerman, 2001). Definition: Number of patents cited in the patent document. Includes self-citations. More citations more valuable patent. 0.4 Normalised index, 2005-08 (average) All patents ICT All (median) ICT (Median) 0.3 0.2 0.1 0.0
References to Non-Patent Literature (NPL) Part of backward citations. Help assess patentability. Reflect closeness to scientific knowledge (Callaert et al., 2006). Patents with NPL contain more complex and fundamental knowledge (Cassiman et al., 2008). Patents with NPL are of significant higher quality (Branstetter, 2005) Definition: Share of NPL citations in a patent document. Higher NPL share more valuable patent 0.3 Shar e of NPL in total backward citations, 2005-08 (average) All patents ICT All (median) ICT (Median) 0.2 0.1 0.0
Forward citations Citations received by subsequent patents. Mirror technological importance for subsequent developments (e.g. Trajtenberg et al., 1990). Include self-citations, as these may be more valuable than external cites (Hall et al., 2005). Definition: Number of citations received in 7-year time after publication. Corrected for patent equivalents. More citations received more valuable patent. 0.15 Normalised index, 2000-04 (average) All patents ICT All (median) ICT (Median) 0.10 0.05 0.00
Breakthrough inventions Capture the extent to which inventions serve as basis for future technological developments. Are associated with entrepreneurial strategies. Patenting grows much more in cities and technologies where breakthrough inventions occur (Kerr, 2010). Definition (follows Ahuja & Lampert, 2001) Top 1% cited patents in each cohort (technology field and year). Forward citations counted up to 7 years after publication. Counts corrected for equivalents. Breakthroughs more valuable. 1.00 0.80 0.60 0.40 0.20 0.00 Share of breakthrough patents in total, 2000-04 (average) All patents ICT
Radicalness difficult to measure. A radical invention is novel, unique, and impacts on future technology (Dahlin & Behrens, 2005). Radicalness linked to firm formation and entrepreneurship. Radicalness Definition (originality based on Hall, 2005 ): Herfindahl-type concentration index N 2 Radicalness p =Σ k j N, where Nk is the fractional count of backward citations in different IPC classes k in total backward citations N out of n p 4-digits IPC tech classes. More radicalness more valuable patent. 0.15 Normalised index, 2005-08 (average) All patents ICT All (median) ICT (Median) 0.10 0.05 0.00
Grant lag The time elapsed between application and grant dates reveals applicants belief about value of the patent: Well-documented patents are approved faster (Harhoff & Wagner, 2009). Time to grant depend on effort made by filing party (Régibeau & Rockett, 2010). Definition: Grant pi = 1 t Max ti, where: t is the grant lag (days); Max ti the max lag of cohort i Shorter grant lag more valuable patent. 0.6 Normalised index, 2000-04 (average) All patents ICT All (median) ICT (Median) 0.4 0.2 0.0
Mirrors number and distribution of forward citations and IPC classes cites belong to. Captures importance of patents for later developments, and number of fields where they happen (Bresnahan & Trajtenberg, 1995; Hall et al., 2001, Hall & Trajtenberg, 2004). Corrections using weights to reflect the technological proximity of the IPC classes. 0.8 0.6 0.4 0.2 0.0 Generality Definition: (adapted from Hall & Trajtenberg, 2004). account for technological distance between IPC classes (weighted measure) GW p = 1 α ij c 2 j j i p i, C p where p i is the share of IPC class i in the patent; α ij is the propensity of IPC class i to be cited by IPC class j; c j the share of IPC class i in the patent c citing p and Cp total forward citations of p. Higher generality more valuable patent. Normalised index, 2000-04 (average) All patents ICT All (median) ICT (Median)
Patent quality Patent quality (PQ) synthesises economic and technological value of patents. Accounts for 4 to 6 dimensions: PQ(4): Forward citations; family size; number of claims; generality index. PQ(6) = PQ(4) + Backward citations; grant lag. Definition (derived from Lanjouw & Shankerman, 2004): Unweighted average of 4 (or 6) normalised components. Patents cohorts stratified by year & tech field. Definition differs from Lanjouw & Shankerman (sector specific weighted average of claims, bwd & fwd cits and family size) Higher quality more valuable patent. Normalised index, 2000-04 (average) All patents (PQ4) ICT (PQ4) All (PQ6) ICT (PQ6) 0.5 0.4 0.3 0.2 0.1 0.0
All indicators have been constructed at the level of the patent document. Using the OECD REGPAT database makes it possible to assess the performance of selected regions (OECD s Territorial Level 3 TL3 regions) Benchmarking regions having contributed to more than 250 ICT-related EPO patents in 2005-08 (90 regions). Insights at the regional level Share of ICT in total EPO patents Top 30 regions in ICT-related patents Waterloo - ON (CA3530) Shenzhen - Guangdong (CN191) Seattle ( ) - WA (US152) Gyeonggi-do (KR013) Uusimaa (FI181) Kanagawa (JPC14) San Jose-San Francisco-Oakland - CA (US146) Noord-Brabant (NL41) Kyoto (JPF26) San Diego ( ) - CA (US145) Rochester ( ) - NY (US139) Stockholms län (SE110) Seoul (KR011) Hauts-de-Seine (FR105) Tokyo (JPC13) Taiwan (CN34) Paris (FR101) Osaka (JPF27) Unterer Neckar (DE68) Los Angeles ( ) CA (US097) Houston ( ) - TX (US075) New York ( ) - NY-NJ-CT-PA (US118) Inner London - West (UKI11) Boston ( )- MA-NH (US022) München (DE93) World total Chicago ( ) - IL-IN-WI (US032) Aichi (JPE23) Philadelphia ( ) - PA-NJ-DE-MD (US127) Stuttgart (DE72) Minneapolis ( ) - MN-WI (US109) 0 20 40 60 80 100 1 2 3 2000-04 2005-08
Insights at the regional level (top 20 regions) Patent Scope (normalised), 2005-08 average All patents ICT Toronto metropolitan - ON (CA3520) Montréal - QC (CA2466) Basel-Stadt (CH031) Saitama (JPC11) Västra Götalands län (SE232) Osaka (JPF27) Ottawa-Carleton - ON (CA3506) Düsseldorf (DE42) Starkenburg (DE52) Vaud (CH011) Kyoto (JPF26) Tokyo (JPC13) Uusimaa (FI181) Rhein-Main (DE51) Wien (AT130) Hyogo (JPF28) Berlin (DE30) Shizuoka (JPC22) Albany ( ) - NY (US004) Stockholms län (SE110) San Diego ( ) - CA (US145) Boise City-Nampa - ID (US021) Miami ( ) - FL (US106) Regensburg (DE90) Seattle ( ) - WA (US152) Isère (FR714) Philadelphia ( ) - PA-NJ-DE-MD Cambridgeshire CC (UKH12) Seoul (KR011) Raleigh-Durham-Cary - NC (US133) Family size (normalised), 2005-08 average All patents ICT Minneapolis ( ) - MN-WI (US109) Noord-Brabant (NL41) Sydney - NSW (AU105) Noord-Holland (NL32) Phoenix ( ) - AZ (US128) San Jose ( ) - CA (US146) Orlando ( ) - FL (US121) Boston ( ) - MA-NH (US022) Basel-Stadt (CH031) Ottawa-Carleton - ON (CA3506) 0.0 0.2 0.4 0.6 0.0 0.2 0.4 0.6
Insights at the regional level (top 20) Generality index (weighted), 2000-04 average All patents ICT Albany ( ) - NY (US004) Phoenix ( ) - AZ (US128) Rhein-Main (DE51) Miami ( ) - FL (US106) Saitama (JPC11) Düsseldorf (DE42) Boston ( ) - MA-NH (US022) Västra Götalands län (SE232) Cleveland ( ) - OH (US035) Washington ( ) - DC-MD-VA-WV Köln (DE44) Toronto metropolitan - ON (CA3520) Hannover (DE19) San Diego ( ) - CA (US145) Denver ( ) - CO (US045) Minneapolis ( ) - MN-WI (US109) Berlin (DE30) Stuttgart (DE72) Detroit ( ) - MI (US047) Basel-Stadt (CH031) 0.0 0.2 0.4 0.6 0.8 Radicalness / originality, 2005-08 average All patents ICT Tel Aviv District (IL051) Petah Tiqwa Sub-District (IL042) Washington ( ) - DC-MD-VA-WV Daejeon (KR051) Salt Lake City ( ) - UT (US142) Denver ( ) - CO (US045) Chicago ( ) - IL-IN-WI (US032) Atlanta ( ) - GA-AL (US011) Bouches-du-Rhône (FR824) Dallas ( ) - TX (US042) Seattle ( ) - WA (US152) Orlando ( ) - FL (US121) Cleveland ( ) - OH (US035) Bonn (DE46) Portland ( ) - OR-WA (US131) San Diego ( ) - CA (US145) Phoenix ( ) - AZ (US128) Miami ( ) - FL (US106) Dublin (IE021) Los Angeles ( ) - CA (US097) 0.0 0.1 0.1 0.2
Insights at the regional level (top 20) Forward citations (normalised), 2000-04 average All patents ICT Yamanashi (JPC19) Aichi (JPE23) Hyogo (JPF28) Rochester ( ) - NY (US139) Breakthrough (share in total patents), 2000-04 All patents ICT Nordsjælland (DK013) Basel-Stadt (CH031) Waterloo - ON (CA3530) Uusimaa (FI181) Gyeonggi-do (KR013) Industrieregion Mittelfranken (DE86) Kyoto (JPF26) Tokyo (JPC13) Osaka (JPF27) Kanagawa (JPC14) Schwarzwald-Baar-Heuberg (DE76) Basel-Stadt (CH031) Zürich (CH040) Seattle ( ) - WA (US152) Nagano (JPC20) Waterloo - ON (CA3530) Chiba (JPC12) Shizuoka (JPC22) Surrey (UKJ23) Südlicher Oberrhein (DE77) Schwarzwald-Baar-Heuberg (DE76) Starkenburg (DE52) Prov. Antwerpen (BE21) Miami ( ) - FL (US106) Yamanashi (JPC19) Gyeonggi-do (KR013) Rochester ( ) - NY (US139) Seattle ( ) - WA (US152) Osaka (JPF27) Surrey (UKJ23) Københavns omegn (DK012) Vaud (CH011) Regensburg (DE90) Seoul (KR011) Industrieregion Mittelfranken (DE86) Nagano (JPC20) 0.0 0.1 0.2 0.3 0.0 0.5 1.0 1.5
Insights at the regional level (top 20) Patent quality index (PQ4), 2000-04 average All patents ICT Seattle ( ) - WA (US152) Denver ( ) - CO (US045) San Diego ( ) - CA (US145) Phoenix ( ) - AZ (US128) Cleveland ( ) - OH (US035) Cambridgeshire CC (UKH12) Boston ( ) - MA-NH (US022) Vaud (CH011) Sydney - NSW (AU105) Washington ( ) - DC-MD-VA-WV Minneapolis ( ) - MN-WI (US109) Philadelphia ( ) - PA-NJ-DE-MD Uusimaa (FI181) Raleigh-Durham-Cary - NC (US133) Västra Götalands län (SE232) Basel-Stadt (CH031) Salt Lake City ( ) - UT (US142) Toronto metropolitan - ON (CA3520) Waterloo - ON (CA3530) Albany ( ) - NY (US004) Patent quality index (PQ6), 2000-04 average All patents ICT Cambridgeshire CC (UKH12) Phoenix ( ) - AZ (US128) Cleveland ( ) - OH (US035) San Diego ( ) - CA (US145) Minneapolis ( ) - MN-WI (US109) Wien (AT130) Düsseldorf (DE42) Vaud (CH011) Dublin (IE021) Philadelphia ( ) - PA-NJ-DE-MD Seattle ( ) - WA (US152) Uusimaa (FI181) Västra Götalands län (SE232) Schwarzwald-Baar-Heuberg (DE76) Tel Aviv District (IL051) Sydney - NSW (AU105) Detroit ( ) - MI (US047) Waterloo - ON (CA3530) Rhein-Main (DE51) Bouches-du-Rhône (FR824) 0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5
Future developments Ongoing OECD work: Refine and validate indicators (further econometric analysis) Extend the compilation to other IP offices Analyse patent portfolio of firms (e.g. determinants of entrepreneurship, high growth firms, etc.) Thank you! helene.dernis@oecd.org
Announcement 2012 conference organised by the EPO and OECD Further information at: www.oecd.org/sti/patents Submissions up to : 30 June 2012