2. Introduction
• R&D and Economic Growth
• Capitalization of R&D investment
• R&D Spillover Effects
• Missing Link in capturing Economic impacts of R&D
2
3. R&D (≠ Knowledge ≠ Innovation)
• R&D {Knowledge Innovation} Economic Growth
• R&D activity : Knowledge creation from materials, labor and
capital inputs
• A part of knowledge published as patents, academic papers, or
other format
• Success/Failure of R&D projects
• Innovation : New product, new process, new organizational
and marketing method
• Success/Failure of Innovation
• Increase in Productivity
• Increase in Productive Capacity (and/or Demand)
• Economic growth
3
5. R&D-Knowledge Linkage
• Patents as an indicator for knowledge production
• R&D (Knowledge) Patent
• R&D performer ≒ Assignee/Applicant of Patent
• Matching patent data to R&D survey/Economic Census
• e.g., NISTEP-RIETI Joint Project (Leader: Prof. Motohashi)
• Cover not only firms but also university/public research institutes.
• !Caveat
• Not all new knowledge are published as patents.
• E.g., trade secrets, human embodied tacit knowledge.
5
6. Innovation-Economic Growth Linkage
• Measuring Innovations
• Subjective measure
• Oslo Manual (OECD/Eurostat)
• Japanese National Innovation Survey (JNIS): 2003/2009/2012/2015
• Conducted by the National Institute of Science and Technology Policy
• Questionnaire to firms
• Objective measure
• (Indirect) TFP growth, Patent
• (Direct) Literature-Based Innovation Output (LBIO):
• Count new products on trade journals, press releases*, news papers, …
*Project of the NISTEP-GRIPS/SciREX Center using Nikkei Telecom
Database (2003-2014)
6
7. Questionnaire of JNIS
• Technological Innovation
• Product Innovation(5), Process Innovation (6)
• Non-Technological Innovation
• Organizational Innovation (a-c), Marketing Innovation (d-g)
7
Source: Japanese National Innovation Survey 2012 (NISTEP)
8. History and the next wave of J-NIS
8
J-NIS 2003 J-NIS 2009 J-NIS 2012
J-NIS 2015
(TBD)
Oslo manual Rev. 2 Rev. 3 Rev. 3 Rev. 3
Corresponding CIS
CIS 3
(2000/2001)
CIS 2008 CIS 2010 CIS 2014
Period 1999-2001 2006-2008 2009-2011 2012-2014
Industries covered
Agriculture,
mining,
manufacturing,
some service
industries
Service
industries
expanded
More
expansion in
service
industries
(Same as in J-
NIS 2012)
Population size 216,585 firms 331,037 412,753 380,226
Sample size 43,174 firms 15,137 20,191 24,825
Response rate 21.4% 30.3% 35.2% (TBD)
Product innovative
firms %
20% 29% 20% (TBD)
International
comparison
OECD (2003)
Innovation in
Firms
(NISTEP DP 68)
OECD STI
Scoreboard
2013
(TBD)
9. Percentages of firms with innovations (2009-11)
9
14
7
12
22
24
12
6
10
20
23
19
8
17
29
2525
14
25
43
32
0
5
10
15
20
25
30
35
40
45
Product
innovation
New to market
product
Process
innovation
Organizational
innovation
Marketing
innovation
%
All firms Small firms
(10-49 emp.)
Medium-size
(50-249 emp.)
Large firm
(250+ emp.)
Source: Japanese National Innovation Survey 2012 (NISTEP)
10. Example of output indicator:
Innovative firms in OECD member countries
10
Manufacturing
Services
Source: OECD Science, Technology and Industry Scoreboard 2013
Notes:
• The rate depends on
composition of
industry in each
country.
• Japanese respondents
tend to perceive
“innovation” in
narrower sense than
western countries.
11. Example of a micro-data analysis using J-NIS (2009)
• Public financial support enhances R&D intensity.
• R&D and research cooperation with business partners and universities lead to innovations.
• Product and process innovation accelerate labor productivity growth.
11
R&D per employee
(million yen, 2008)
Probability of product innovation
(2006-2008)
Probability of product innovation
(2006-2008)
Growth rate in labor productivity
(2006-2008)
Source: Ikeuchi and Okamuro (2013) “R&D, innovation, and business performance of Japanese start-ups: A comparison with established firms” NISTEP
DISCUSSION PAPER No.104
Productivity
R&D
Innovation
Collaboration
with university
Public financial
support
0.00
0.05
0.10
0.15
0.20
0.25
No Yes
Public financial support
Start-up firm
Established firm
0.00
0.20
0.40
0.60
0.80
1.00
-3 0 3
R&D intensity in log.
Start-up firm
Established firm
0.00
0.20
0.40
0.60
0.80
1.00
No Yes
University cooperations
Start-up firm
Established firm
-0.40
-0.20
0.00
0.20
0.40
No Yes
Product or process innovation
Start-up firm
Established firm
12. Literature-Based Innovation Output Indicators (LBIO)
12
Literature-based Innovation Output Indicators
(LBIO)
Innovation inputs Innovation outputs Innovation impacts
R&D activity
Design activity
Marketing activity
Training
Acquisition of machinery,
equipment and software
Patent
Trademark
Design
New/significantly improved:
Products
Processes
Organizational methods
Marketing methods
Productivity
Profitability
Growth
Enterprise value
Intellectual
Property Rights
(IPR)
Press release
Advertisement
Public Relations
23. Productivity Effects of R&D Spillovers
• Ikeuchi, Belderbos, Fukao, Kwon and Kim, (2013) NIPTEP DP#93
• “Sources of Private and Public R&D Spillovers: Technological, Geographic
and Relational Proximity”
• Data:
• Census for Manufactures (plant-level microdata)
• R&D Survey (firm-level microdata)
• Specification:
𝑄𝑖𝑡 = 𝑓 𝐿𝑖𝑡, 𝐾𝑖𝑡, 𝑀𝑖𝑡 𝑔 𝑅𝑖𝑡−1, 𝑆𝑖𝑡−1, 𝑃𝑖𝑡−1, X 𝑖𝑡 𝑈𝑖𝑡
• Where:
• 𝑄𝑖𝑡: Gross output of the plant
• 𝐿𝑖𝑡, 𝐾𝑖𝑡, 𝑀𝑖𝑡: Inputs of plant 𝑖 in year 𝑡
• 𝑅𝑖𝑡−1: Firm-level R&D stock
• 𝑆𝑖𝑡−1: Private R&D stock
• 𝑃𝑖𝑡−1: Public R&D stock
• X 𝑖𝑡: a vector of other observable factors (control variables) affecting
plant productivity
• 𝑈𝑖𝑡: plant-year specific unobserved efficiency.
23
24. Spillovers: Private R&D Stocks
• Firms’ R&D distinguished by 30 fields: mapped into 25 (2-
digit) industries
• Other firms’ R&D by field weighted by technological
proximities between industries, allowing for geographic
decay in the effectiveness of spillovers
𝑆𝑖𝑓𝑠𝑡 =
𝑓′≠𝑓 𝑠′
𝐾 𝑓′ 𝑠′ 𝑡 𝑇 𝑠𝑠′ 𝑒
𝜏𝑑 𝑖𝑓′ 𝑠′ 𝑡
• where:
• 𝑑𝑖𝑓′ 𝑠′ 𝑡: Minimum geographic distance between plant 𝑖 and the
plant of firm 𝑓′ in the field 𝑠′ in year 𝑡;
• 𝑇 𝑠𝑠′: the technological proximity weight;
• 𝑒
𝜏𝑑 𝑖𝑓′ 𝑠′ 𝑡: Weight for geographic proximity of plant 𝑖 to R&D stock
firm 𝑓′
for field 𝑠′
;
• 𝜏: an exponential decay parameter, with 𝜏 < 0 (e.g.
Lykachin et al. , 2010)
24
25. Technological Proximity between Industry
• Technological relatedness between firms are derived from
patent citation data and based on Leten et al. (2007).
• Patent citation data are available at the 4-digit IPC level.
• (IPC: International Patent Classification)
• The IPC codes can subsequently be mapped onto industries using
the industry-technology concordance table developed by Schmoch
et al. (2003) in which each technology field is uniquely linked to its
corresponding NACE two-digit industry.
• Weights for the own industry normalized at 1.
25
27. Public R&D Spillover Pools
• R&D expenditures by Universities from R&D survey (~100% resp. rate)
• Calculate R&D stocks with 15 percent depreciation rate
• Allocated to science fields based on # of researchers per science field
• Public R&D spillovers:
• R&D per science field weighted by its relevance for specific industries
• Allow for decaying effects due to geographic distance (plant-institution)
𝑃𝑖𝑡𝑠 =
ℎ 𝑚
𝐴ℎ𝑚𝑡 𝑇𝑠𝑚 𝑒 𝜃 𝑑 𝑖ℎ
• 𝐴ℎ𝑚𝑡: R&D stock of public institutes in location ℎ for academic field 𝑚 in year 𝑡;
• 𝑇𝑠𝑚: Knowledge flow matrix between industry/R&D field 𝑠 and science field 𝑚;
• 𝑑𝑖ℎ: geographic distance between plant 𝑖 and institution location ℎ;
• 𝜃: the geographic decay parameter, 𝜃 < 0.
• Technological proximity:
• Citations in patents to scientific literature by academic field ( Van Looy et al,
2004): gives concordance between science fields and IPC/technology classes
• Technology class to industry: IPC/technology class to industry concordance (Smoch et al.
2003)
27
29. Relational proximity
• R&D stocks of supplier and customer industries
• Identifying the importance of supplier and customer transactions
from Input-Output tables for 58 JIP industries.
• Supplier industry proximity weights 𝑆𝑈𝑃 𝑠𝑠′ and customer
proximity weights 𝐶𝑈𝑆 𝑠𝑠′, with:
• Supplier industry weights: 𝑆𝑈𝑃 𝑠𝑠′ 𝑡 =
𝑄 𝑠′ 𝑠𝑡
𝑗 𝑄 𝑗𝑠𝑡
• Share of industry 𝑠′sales to industry s in the sum of sales by all
industries j to industry s (input share).
• Customer industry weights: 𝐶𝑈𝑆 𝑠𝑠′ 𝑡 =
𝑄 𝑠𝑠′ 𝑡
𝐸𝑋𝑠𝑡+𝑄 𝑠𝑡
• Weights for customer R&D stocks are the shares of sales by
industry s to industry 𝑠′
in total sales (domestic sales + exports)
industry s (output share).
29
31. Decomposition of the inter-firm R&D spillovers effects:
technological vs. relational proximity
31
0.306
0.140
0.069 0.059
0.287
0.132
0.099 0.071
0.276
0.108
0.078
0.044
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
1987-1992 1992-1997 1997-2002 2002-2007
Customer industry R&D spillovers
Supplier industry R&D spillovers
Tech-related industry R&D spillovers
(Balanced panel)
Source: Ikeuchi et. al (2013)
32. Decomposition of the inter-firm private R&D spillovers
effects: entry and exit
32
(Balanced panel)
-0.60
-0.40
-0.20
0.00
0.20
0.40
0.60
0.80
1.00
Entries of plants with R&D
Surviving plants R&D effects
Exits of plants with R&D
Total
Source: Ikeuchi et. al (2013)
34. Comprehensive Framework for Public R&D Spillovers
(On going RIETI-NISTEP joint projects)
34
Higher Education
Institutes & Public
Research Institutes
Industry
(Firm)
Paper
publication
Patent
application
Patent application
Direct
collaboration
1. Non-Patent Literature Citation
+2. Academic
Patent Citation
+3. Joint Patent invention
“Science intensity by industry by linking dataset of science, technology and industry” (with
Kazuyuki Ikeuchi, Ryuichi Tamura and Naoshi Tsukada), presented at OECD BlueSky
Conference Sept. 2016
35. Database Construction
35
IIP-PD
• Inventor Part (1995-2013 application)
Disambiguation of inventor (Li et. al,
2014)
• using information of name, address,
applicant, IPC4, co-inventor
• Telephone directory rare name as
training set
• KAKEN info as correct reference
info to decide threshold value
English inventor name (PATSTAT link)
and inventor affiliate (by single applicant
patent)
• Applicant Part (all IIP patent)
Standardized applicant name
Standardized address
Scopus
• Author ID information
(1999-2012 publication)
Author name
Author affiliateNISTEP
Institution ID
Economic Census
• Firm and institution ID
information (2001, 2004,
2006, 2009 and 2012)
Name
Address
Match by name and
location (using
establishment level info)
37. Scientific intensity within academic inventors
37
Original Data : # of research papers by science field “s” (Rs)
# of patents by technology field “t” (Pt) for researcher “i”
Step 1: Construct W matrix, 𝑊𝑠𝑡
Paper
=
𝑖
𝑅𝑖𝑠 ×
𝑃𝑖𝑡
𝑡 𝑃𝑖𝑡
Step 2: Normalize by patent counts of “i” 𝐿 𝑡
Paper
=
𝑠 𝑊𝑠𝑡
𝑃𝑎𝑝𝑒𝑟
𝑖𝑡 𝑃𝑖𝑡
Step 3: S-T Linkage by technology class “t” is calculated by
𝑆𝐼𝑡 = 𝐿 𝑡
Paper
∗ 𝐴𝑐𝑎𝑑𝑒𝑚𝑖𝑐_𝑆ℎ𝑎𝑟𝑒𝑡
38. Science Intensity by Industry, 2000-2003 (papers per employees*1000)
38Source: Motohashi et. al (2016)
39. Science Intensity by Industry, 2008-2011 (papers per employees*1000)
39Source: Motohashi et. al (2016)
40. Conclusions
• R&D and Economic Growth
• Knowledge and Innovation
• R&D-Knowledge Linkages
• Innovation-Economic Growth Linkages
• Capitalization of R&D investment
• Effects on TFP measures and Growth Accounting
• R&D Spillover Effects
• Pecuniary spillovers (Static/Dynamic Unit TFP)
• Technological Spillovers and Knowledge Flow
• Academic or Science Activities in Economy
• Missing Link
• Knowledge-Innovation linkages Knowledge flow matrix
• Liking with IO-table?
40
41. Thank you for your
attentions.
Kenta Ikeuchi (RIETI/GRIPS/NISTEP)
ks.ikuc@gmail.com
41
Editor's Notes
The last slide in page 6 shows the history of our survey and the new next research.
We start this survey from 2003 and we have conducted the survey three times by now.
We plan to conduct next survey in this year.
Response rate have been gradually improved but not so much high.
Also, you can see the percentage of the firms with a product innovation has not been in a steady trend. That rate fallen between JNIS 2009 and 2012 from 29% to 20%. Between 2009 and 2011, Japan economy had been influenced by two economic shocks. One is the Global financial crisis occurred in the fall of 2008 and the another is the big earthquake in the east side of Japan. These shocks seems to negatively influence on innovation activities in firms.
This is all of my presentation. Thank you very much for your attention.
Next, I’d like to show several results of our survey.
First, the figures in page 2 are derived from OECD’s Science, Technology and Industry Scoreboard 2013. We provided the results of Japanese Innovation Survey to OECD and OECD collect the data and compare the results among countries.
The figure shows the percentage of the firms which introduce each type of innovations in each countries. If we focus on product and process innovation, you can see that the share of innovative firms in Japan is relatively low.
We also have conducted data analysis using the micro data of Japanese National Innovation Survey. Figure in page 4 is an example. In this research, we use the micro data of J-NIS 2012 and analyze and compare the determinants of R&D, innovation and productivity between start-up firms and established firms.
As the results of this analysis, we found public financial support significantly enhances R&D intensity of the firms and R&D expenditure and research cooperation with university increase the probability of firms to introduce an innovation. Also, we found the introduction of product and process innovation accelerate labor productivity growth.
Next, I’d like to explain about empirical model and data.
First, we adopt the standard knowledge stock augmented production function framework.
Qit denotes the gross output of the plant
L denotes labor inputs, K denotes capital and M denotes intermediate inputs.
Rit-1 is the one-year lagged firm-level R&D stock
Sit-1 is the private R&D stock
P is the public R&D stock
X is a vector of other observable factors affecting plant productivity
Finally, Uit denotes the plant-year specific unobserved efficiency.
Using the firm/field-level R&D stock, next, we construct the private R&D spillover variables.
We define the technologically relevant private R&D stock as a spillover pool:
as the sum total of other firms’ R&D, K, assigned to their plants on the basis of the industry,
weighted by the technological relatedness, T, with the field of the plant
and geographic decay in the effectiveness of spillovers.
We model an exponential decay function in the effectiveness of spillovers with the distance d between plants and the parameter, tau, to be estimated.
Our technological relatedness measure between fields is derived from patent citation data and based on a previous study b Laten et al.
The relatedness between technologies will be reflected in the intensity with which technologies in a field build on prior art in a different field.
Weights for the own industry normalized at 1.
This table shows the technological relatedness index between industries used for the construction of technologically relevant R&D stocks.
I’d like to skip the detailed explanation for time saving.
We also calculate the public R&D stock as another source of R&D spillovers
Public R&D expenditure data is also derived from the R&D surveys as well.
We differentiated public R&D by location and academic field.
We apply fifteen percent (15%) depreciation rate.
Same as the private R&D stock, we consider the technological relatedness, T tilde, between plant industry and each science fields of public R&D stock and the decay in the geographic distance, d tilde.
Applied technological relatedness measure is also based on the data on patent citation to science literature.
We also measure relationally proximate R&D stocks by the R&D stocks of supplier and customer industries.
To calculate the supplier and customer R&D stocks, we use supplier industry weights and customer industry weights instead of the technological proximity between industries.
We identify the importance of supplier and customer transactions from Input-Output tables.
Supplier weights are measured as the input share.
Customer industry weights are measured as the output share.
Given these regression results,
we can decompose the actual TFP growth in Japan into several factors:
Firm internal R&D effects, private R&D spillovers effects, and public R&D spillovers effects.
We can see that; declining R&D spillovers, in particular private R&D spillovers, play an important role in the decline in TFP in Japan.
Moreover, we can decompose changes in private R&D spillovers;
To the tech-related industry R&D, and customer and supplier industry R&D spillovers.
Results show that; declining spillovers is generally due to the observed declining growth of private R&D stocks.
In addition, we can see that; supplier spillovers are relatively persistent compared to technological proximate R&D spillovers.
We can further decompose the changing private R&D spillovers into effects;
due to exit of plants, entry of plants, and surviving plants.
This exercise shows that;
in particular, the effects of exit of plants operated by R&D intensive parent firms has a mostly important role in the spillovers decline.
Finally, we can also decompose these effects into the regional level.
Results show the most of these exit effects appear in the major industrial agglomerations in Japan;
Such as Tokyo area including Kanagawa and Saitama prefectrue
And Aichi prefecture which is the home town of Toyota automobile.