In the 2000s China moved from modest contributor to global patents to become the number one patent producing country in the world. What is quality of Chinese patents compared to those of US/other countries? To what extent is China’s patent growth frontier inventions vs catch-up of products new to China but not the world? What is the relation of patents with economic outcomes? This talk will answer these questions with statistical and case evidence.
Find out more at Iems.ust.hk/china-patent
8447779800, Low rate Call girls in Shivaji Enclave Delhi NCR
Richard Freeman: China's Patent Explosion
1. China's Patent Explosion
1. The Phenomena and Questions for Study
2. Data Sets: SIPO via Google linked to
Firms and USPTO
3. Analysis and Findings
4.Conclusion: future work
Richard B. Freeman, Harvard University, NBER
Lintong Li, Peking University
HKUST, Feb 22, 2017
Preliminary Analysis of New Data; Comments appreciated.
4. The study of patents
Primary economic motivation is to explain growth. First major modern
researcher was Jacob Schmookler, Invention and economic growth
(Harvard University Press, 1962). Zvi Griliches developed econometric
analysis with firm level data.
“Patent and patent statistics have fascinated economists for a long
time. Questions about sources of economics growth, the rate of
technological change , the competitive position of different firms and
countries……all tend to revolve around notions of
inventiveness……In this desert of data, patent statistics loom up as a
mirage of wonderful plentitude and objectivity. They are available;
they are by definition related to inventiveness” (Griliches, 1990)
Patents are not innovation in the Schumpeter/Oslo convention sense
since most patents do not lead to new products or processes, but they are
valuable indicator of new ideas intended to produce innovation and thus
provide insight into China's move from developing economy to
innovative knowledge-based economy.
5. Debate over innovation in China
May 2014, VP Joe Biden at Air Force Graduation: “Name to me
one innovative project, one innovative change, one innovative
product that has come out of China”
March 2014 Harvard Business Review, experts debated whether
China's government structure is compatible with “true spirit of
entrepreneurship”
February 2015, Economist debate “Is China a global innovative
powerhouse?” with debaters focused on how much government
domination of economy discourages innovation
Winter 2017 “ Chinese firms have ... a capacity to become more
innovative in response to wage pressure and global
opportunities…we should not be pessimistic about ... a
successful transition to a more innovation-based growth
model.” Wei, Xie and Zhang, JEP,
6. Goals for Innovation in 2016-2020 5 Year Plan
Rankings in National Innovation Capacity 18 15
Contribution of Science and Technology to Production (%) 55 60
R&D intensity (%) 2.1 2.5
High-tech Firm's Revenue (Trillion RMB) 22.2 34
Share of value added from knowledge intensive service industry in total GDP (%) 15.6 20
Share of R&D expenditure in Revenue for above-scale Industrial Firms (%) 0.9 1.1
Rank in International Science Paper's Citations 4th 2nd
Count of PCT applications 3.05 6.1
Patents per 10k people 6.3 12
Revenue of technology contract (billion RMB) 984 2000
Fraction of citizens with science literacy (%) 6.2 10
7. Five Big Questions
1 – What is quality of Chinese patents and impact of
quality on number and growth of patents compared to
US/other countries?
2 – How much patent growth is catch-up (inventions new
to China) vs frontier (inventions new to world)?
3 – What is impact of “ innovation” associated with
patents on economic outcomes?
4 – What are driving forces behind the patent explosion?
5-- Can patent-related innovation help ameliorate China's
increased inequality and the pollution/ environmental
costs of rapid growth?
8. 2. SIPO-Firm-USPTO data set
State Intellectual Property Office (SIPO): bibliometric data
1985-2012, by application year; Google patent search: web
scraping (using Java Jsoup package) with additional data. Data
set includes references (backward citations) for SIPO patents
granted in 2009-2015s/forward citations for earlier patents put in
by examiners/applicants; technology classes of patents; addresses
for assignees and inventors; names of firms matched with 1998-
2007 data in Annual Survey of Industrial Firms.
United States Patent and Trademark Office (USPTO): 1976-
2015 (grant year)-- using public data but parsed for latest years
(Used Python Element tree package)
USPTO-SIPO Matched Patent subset for same patents in
USPTO as in SIPO
9. Data Set includes patent references/citations
Patent references are on the front page of a granted patent.
They identify “prior art” upon which the current invention builds.
The greater the number of backward citations, the more a patent
relies on previous work
Using references/backward citations, we construct forward
citation measure for every patent. Since we only have backward
citation data 2009-2015. the forward citations is downward
biased for patents granted before 2009.
The number of forward citations shows the impact of a patent
on other inventions and is an indicator of the value of the patent.
(Trajtenberg, 1990; B. H. Hall, 2000; Bloom and Reenen, 2002)
and of the geography of knowledge spillovers (Jaffe, Trajtenberg
and Fogarty, 2000; Thompson and Fox-Kean, 2005)
10. Identifying Same Patents in SIPO and USPTO as bridge between
patent offices from 35,989 candidate matches
National patents provide IP protection in country so companies will seek
protection for more important inventions in more countries
11. Incentives for more patents notwithstanding, Chinese firms
do not break one USPTO patent into several SIPO patents.
More likely they take the same patent and translate it.MatchingbetweenSIPOandUSPTOforthe“MatchingPart”
USPTO
1 2 3 4+ Total
SIPO
1 11,472(83%) 1,050 687 109 12,871
2 330 76 42 8 456
3 9 30 9 33 81
4+ 33 26 24 324 407
Total 11,844 1,182 315 474 13,815
Number of matched SIPO-USPTO patents
12. 1 – What is quality of Chinese patents?
Patent quality is lower in China than in US and declined during
explosive growth but upward trend in Chinese patents raised its
share of world patent citations as well as its share of world patents.
1.1 Chinese patents make lower claims than US patents
1.2 SIPO patents make smaller number of citations than
USPTO patents, but part of difference is due to greater
propensity for US firms to reference older patents in USPTO.
1.3 Trend decline in citations per SIPO patent
1.4 China explosion of patents occurs in all major patent
offices. This means China had most rapid growth of top
patents among major countries. (Has distribution by citations
gotten more unequal?).
3. Analysis and Findings: answers to questions
13. 1.1 US firms make more patent claims than Chinese firms
Patent claims state the subject of inventions and are principle factor in determining whether a
proposed invention infringes on existing patents. Dependent claims refer to other patents and
claims; independent claims have no references.
● Chinese firm SIPO patents have fewer claims than Chinese firm SIPO-SUPTO patents. 2005
decline in USPTO claims increase of $18 to $50 per claim in excess of 20. (Harhoff, 2016)
14. 1.2 More Patent citations in USPTO than SIPO:
quality or patenting norm? same China patent cites less in SIPO
15. 1.3 Decline in citations per China patent Relative to US
addressed patent in USPTO (from annual regressions)
16. 1.4 China explosion of patents not just in SIPO
2010-2015 growth rates of 27% in SIPO; 25% in USPTO; 27%
EPO; and fastest of 43% in JPO
17. ● Alternative Exchange Rates Between China Patents
in USPTO AND SIPO and US USPTO patents
18. Q2 How much patent growth is catch-up (inventions new
to China) vs frontier (inventions new to world)? Most
likely catch-up but rapid growth at frontier.
2.1 China origin patents grow rapidly in USPTO,
country becomes #5 foreign source of patents
19. 2.2 But Citations to China in USPTO fall from
above to normal for to other USPTO Patents
20. 2.3 China Share of USPTO Citations Rises,
especially for newest cohorts of patents
21. 2.4 Cosine Similarity Distribution of Chinese Patents
Converges toward US Patents in Technology Space
Patent offices use numerical technology codes to categorize technologies of
patents; USPTO has USPC 150,000 codes; SIPO, IPC 70,000 codes. Number
of classes of patents increases in SIPO (improved patent office?); country
patterns similar, save for China variability with small numbers in early USPTO
22. q3 –What is impact of “ innovation” associated
with patents on economic outcomes?
China patents are associated with higher productivity, higher
profit margin and higher growth rate, within industry and
weaker effects for same firm over time.
Estimated effects on outcomes in production function
regressions as comparable to those for US patents
Relation associated largely with higher citation patents.
23. (1) (2) (3) (4) (5) (6)
Ln(output) Margin Employment
Growth
Ln(output) Margin Employment
Growth
Ln (patent stock+1) 0.067*** 0.026*** 0.031*** 0.013*** 0.001 0.005
(0.002) (0.001) (0.002) (0.003) (0.001) (0.003)
Industry FE Yes Yes Yes No No No
Firm FE No No No Yes Yes Yes
Observations 1911990 1911990 1351307 1788105 1788105 1258923
R2 0.947 0.079 0.019 0.971 0.574 0.216
(1) (2) (3) (4) (5) (6)
Ln(output) Margin Employment
Growth
Ln(output) Margin Employment
Growth
Ln (patent stock) 0.027*** 0.014*** 0.021*** 0.011*** 0.001 0.006
(0.003) (0.001) (0.002) (0.004) (0.001) (0.004)
Industry FE Yes Yes Yes No No No
Firm FE No No No Yes Yes Yes
Observations 36644 36644 31590 33682 33682 28964
R2 0.971 0.090 0.047 0.987 0.631 0.261
Note: In column 1&4, ln(capital), ln(employment) and ln(materials) are omitted. We control for years, ages and
ownership types. Cluster standard errors at firm level in parentheses. Output and input have been deflated by 2-digit
industry deflators. * p < 0.10, ** p < 0.05, *** p < 0.01
Substantial positive patent effects on productivity, profitability and
employment growth within a industry. and on productivity and
weakly on employment growth for same firm
24. (1) (2) (3) (4) (5) (6)
Ln(output) Margin Employment
Growth
Ln(output) Margin Employment Growth
Ln (high quality
patent stock+1)
0.063*** 0.025*** 0.031*** 0.015*** 0.002 0.008
(0.003) (0.002) (0.003) (0.004) (0.002) (0.005)
Ln (low quality
patent stock+1)
0.027*** 0.010*** 0.009*** 0.001 -0.001 -0.003
(0.004) (0.002) (0.003) (0.005) (0.002) (0.005)
Industry FE Yes Yes Yes No No No
Firm FE No No No Yes Yes Yes
Observations 1911990 1911990 1351307 1788105 1788105 1258923
R2 0.947 0.079 0.019 0.971 0.574 0.216
(1) (2) (3) (4) (5) (6)
Ln(output) Margin Employment
Growth
Ln(output) Margin Employment
Growth
Ln (high quality
patent stock)
0.021*** 0.009*** 0.009** 0.012 0.007** 0.001
(0.005) (0.002) (0.004) (0.009) (0.003) (0.010)
Ln (low quality patent
stock)
0.008* 0.003 0.006 0.001 0.001 0.000
(0.005) (0.003) (0.004) (0.008) (0.003) (0.008)
Industry FE Yes Yes Yes No No No
Firm FE No No No Yes Yes Yes
Observations 10127 10127 9033 9125 9125 8151
R2 0.977 0.120 0.072 0.989 0.655 0.260
Relation Stronger for Highly Cited Patents and Outcomes
25. 4—Conclusion and Future Work
4.1 Magnitude of China's explosion in patents overwhelms lower
and falling quality so that China's share of world patent citations and
of world claims has increased.
4.2 Despite quality issues, China's patents have similar impact on
production measures as patents in advanced countries.
4.3 China's patents converging in technology space toward US.
Data on technical classes provides a route to study whether patent-
related innovation could ameliorate China's increased inequality and the
pollution/environmental problem and its movement toward higher-tech
industries. Expect to see rapid growth in all classes and industries.
Likely driving forces behind the patent explosion include: expanded
number of university graduates, government incentives for patents,
foreign direct investment and transfer of knowledge.
26. Finally, our data set can be used to examine other questions
regarding the patent explosion in China.
Using data on the location of patents, researchers can
examined local knowledge spillovers effect – 25% citation pairs are
located within a province.
Using backward citations to USPTO patents; EPO patents and
to JPO patents. Researchers can uncover the knowledge diffusion
path from developed countries to developing countries and assess
the effect of FDI on technology transfer
With information on particular products in different industries,
and of the share of new products and processes in firm sales the
patent data can also tighten the link between patent-driven
innovation and economic success.
28. Odd Patterns in USPTO Data
JUSPTO shows increased references by applicants in USPTO,
especially for US origin applicants while examiners references
remain stable. No such pattern in SIPO
32. Comparison in USPTO (1998-2015)
Note: Robust standard errors in parentheses. * p < 0.10, ** p < 0.05, *** p < 0.01. We control assignee’s type, application year dummies, grant year dummies, 7 digit IPC
dummies and 7 digit IPC dummies interact time trend. The benchmark for the set of dummies are a combination of countries unlisted here.
(1) (2) (3) (4) (5) (6)
Citations Citations
(excluding
self-citations)
Citations
(excluding
own country
ones)
Citations(appl
icant)
Citations(exa
miner)
Claims
US assignee 3.549*** 2.962*** -4.841*** 2.846*** 0.405*** 2.873***
(0.033) (0.032) (0.026) (0.028) (0.005) (0.019)
CN assignee 0.543*** 0.405*** -2.393*** 0.357*** 0.067*** -2.532***
(0.034) (0.033) (0.029) (0.029) (0.009) (0.045)
DE assignee -1.441*** -1.503*** -2.707*** -0.970*** -0.383*** -0.985***
(0.036) (0.036) (0.032) (0.031) (0.007) (0.027)
JP assignee -1.766*** -1.929*** -3.423*** -1.646*** -0.077*** -4.131***
(0.033) (0.032) (0.028) (0.028) (0.006) (0.020)
KR assignee -1.363*** -1.369*** -3.107*** -1.175*** -0.138*** -2.373***
(0.040) (0.040) (0.035) (0.034) (0.008) (0.029)
TW assignee -1.163*** -1.153*** -2.888*** -1.156*** -0.069*** -2.546***
(0.042) (0.041) (0.035) (0.035) (0.009) (0.031)
IN assignee -0.101 -0.065 -2.768*** -0.252*** -0.047** -0.814***
(0.090) (0.089) (0.078) (0.078) (0.019) (0.141)
N 3495698 3495698 3495698 3495698 3495698 3495698
r2 0.236 0.239 0.171 0.198 0.259 0.0964
Y mean 8.665 8.165 3.815 5.738 2.042 17.07
33. Correlation of citations, claims, technologies
China SIPO patents
(1) (2) (3) (4) (5)
Citations Citations Citations Citations Citations
Firm (Dummy) 0.139***
0.129***
0.136***
(0.005) (0.005) (0.005)
Institution (Dummy) 0.323***
0.326***
0.330***
(0.006) (0.006) (0.006)
Number of Claims 0.010***
0.009***
(0.001) (0.001)
Number of Technologies 0.050***
(0.001)
Application & Grant Year FE Yes Yes Yes Yes Yes
Technology FE No Yes Yes Yes Yes
N 807,202 806,689 806,689 806,689 806,689
Adjusted R2 0.078 0.135 0.139 0.139 0.142
Y mean 0.907 0.907 0.907 0.907 0.907
US USPTO patents
(1) (2) (3) (4) (5)
Citations Citations Citations Citations Citations
Number of Claims 0.291***
0.282***
(0.003) (0.003)
Number of Technologies 0.566***
(0.008)
Application & Grant Year FE Yes Yes Yes Yes Yes
Technology FE No Yes Yes Yes Yes
Assignee Type FE No No Yes Yes Yes
N 2,254,907 2,247,091 2,247,090 2,247,090 2,247,090
Adjusted R2 0.083 0.208 0.208 0.219 0.221
Y mean 16.104 16.097 16.097 16.097 16.097