Prof. of Economics, Seoul Nat’l University Schumpeterian Analysis of Economic Catch-up: Firm, Sector and country Levels Keun Lee 李根 Prof. of Economics, Seoul Nat’l University Director, Center for Economic Catch-up www.keunlee.com
Winner of 2014 Schumpeter Prize Innovation system at 3 levels: Winner of 2014 Schumpeter Prize
The book endorsed by: Richard Nelson; Justin Lin Franco Malerba’ John Cantwell Ha-Joon Chang; John Mathews “The story Keun Lee tells is fascinating and thought provoking. This book lays out a convincing new perspective on the conditions behind the remarkable development of manufacturing in South Korea and Taiwan. The well documented argument is that in several of the industries where firms in these countries were very successful, technology at the frontier was going through a transition, with the technology coming in requiring a different set of skills and capabilities than the technology becoming obsolete. This diminished the advantage of the old industrial leaders and provided a window of opportunity for effective entry in these two economies. The story Keun Lee tells is fascinating and thought provoking.” Richard Nelson, Columbia University
What is catch-up? “Catching up, forging ahead, and falling behind” (Abramovitz, 1986 JEH) 1) national level: per capita income, share in world GDP, 2) firm-level: market share, sales growth, productivity => rise and decline of nations and firms (and sectors)
Record of Catching-up / Falling behind: As % of the US per capita Income ( in 2005 Constant PPP)
Trend of the Income Levels as Percentage of that of Japan: => Korea, Taiwan: No catching up in 60s, 70s:-> only from 1980s
Many countries are in the MIT (Middle income trap) Source: World Bank 2012,
Flat R&D/GDP as source of MIT (middle income trap)
What is MIT? MIT = a slowdown of growth in middle-income countries (MICs) -- caught between low-wage manufacturers and high-wage innovators. (Lin 2012; Williamson 2012; Yusuf and Nabeshima 2009; World Bank 2010 and 2012; S pence 2011); Why important? - 70% of World Poor in MICs; - It is future of the LICs (low income C’s) 9
Lee, Keun & B. Kim (2009, World Development) Confirms importance of Innovation (patent count) for middle and higher income countries This study: beyond just patent counts; => more details of the NIS Eg) cycle time of tech. localization of knowledge creation,
Schumpeterian Perspective: National Innovation systems Lundvall (1992): defines NIS = elements and relationships 1) which interact in the production, diffusion and use of knowledge 2) rooted inside the borders of a nation state.’ It is about efficiency in acquisition, creation, diffusion, and utilization of knowledge. -> The differences in NIS determines different levels of competitiveness of nations.
->b/c extractive vs inclusive institution Acemoglu and Robinson, Why Nations Fail ->b/c extractive vs inclusive institution Bill Gates’ book review “Never explain how to move to more “inclusive” institutions” Inclusive vs. extractive : -> relevant more in low income or pre-modern economy b/c less difference among middle income countries => Why Nations Fail at Middle Income Stage: due to not-Innovative systems
( constraints on executives) Quality of Political Institutions ( constraints on executives) 1965 1980 2000 Korea 3 1 6 Taiwan 2 Philippines 5 Thailand 7 Malaysia 4 China India Brazil Argentina Chile Mexico Source: Polity IV Dataset; from Lee and Kim 2009 table 1
Who are Neo-Schumpeterians? Adam Smith: birth of new world Karl Marx: law of motion of capitalism John Keynes: demand side Joseph Schumpeter: supply side Tech. change as key engine of economic change Neo-Schumpeterian Tech. change can be endogenous , predictable ; Law of motion in innovation The study by Keun Lee and other scholars: Innovation also key engine of catching-up ; catching-up can be explained and predictable
The 3 Failures as Justification for Gov’t Activism Market failure System failure Capability failure Focus Market institutions Interaction among actors Actors (firms) Source Knowledge as public good Cognition failure from tacitness of knowledge historically given; No learning opportunity Example problem Sub-optimal R&D R&D impact: low No R&D Solutions R&D subsidies Reducing cognitive distance Access to knowledge and help in learning School Analogy Tuition support Making more friends Targeting student learning Relevance Developing and advanced countries More unique to developing countries From Keun Lee, a chapter in Stiglitz & Lin eds, 2014
Analysis at 3 levels => different question Country: What determines catching-up growth: -> per capita income growth Sector: Why easy to catch up in some sectors; why not in others ? -> Country’s US Patent share in sectors 3) Which the CIS (corporate innovation system) a good fit for catching up; sales growth, profitability, firm value, productivity => different question -> same answers = knowledge variables
Key Variable = cycle time of technologies Cycle time = speed of change in the knowledge base of a technology = mean citation lag in patent citations “To catch up, specialize in Short cycle technology-based sectors“ because 1) old knowledge quickly obsolete/useless -> less need to bother about them: less disadvantageous for the latecomers +2) new knowledge tend to emerge more often -> greater opportunity for growth => technological sectors with less reliance on the old technologies You got to be different from the North, than trying to be similar from the beginning (opposite to Hausmann )
Overall: Short cycle technology matter
Four Key Variables and Hypotheses at three Levels of country, sector & Firms 4 Hypotheses : Growth strategies Technological specialization 1 (short vs. long cycle) Technological. Specialization 2 (high vs. low originality ) Localization of knowledge creation & diffusion (vs. reliance on foreign sources) Balanced vs. Concentration of knowledge creation
(1) Patent number; (2) Grant year (3) Grant date Data: NBER USPTO patent data Patents registered from 1963 to 1999; later update to 2006 Patent citation data made from 1975 to 1999; updates to 2006 1975-1999: at (http://www.nber.org/patents/) 1976-2006: NBER website (https://sites.google.com/site/patentdataproject/Home). Information in the database (1) Patent number; (2) Grant year (3) Grant date (4) Application year (starting in 1967) (5) Country of first inventor ; (7) Assignee identifier, if the patent was assigned (starting in 1969) (8) Assignee type (individual, corporate, or government; foreign or domestic) (9) Main U.S. patent class (UPC); (11) citing patent number and cited patent number (patent citation pair data) Retrieved at the country, tech. classes (sectors) and firm levels (Korean, US firms) Regression analysis
1st Question: Recent Debates on Long Run Economic Growth Institution vs. Policies vs. Geography 1) Not policies but Institutions matter: (Acemoglu, et al 2001, 2002; Rodrik et al 2004) Institution as a savior of the Washington Consensus 2) Geography matter: A destiny? ( J. Sachs) 3) Does institution matter? -> Human capital ( Glaeser, La Porta, Lopez-de-Silanes & Shleifer, 2004) => Innovation still missing; different factor matter at different stage of development
From just patent count to other details of knowledge base: the Key NIS variables by country groups High Income Countries Middle Income Countries Korea and Taiwan Mean(μ0) Mean(μ1) Mean(μ2) Herfindhal index 0.105 0.374 0.206 Intra-national knowledge diffusion index 0.092 0.036 0.041 Technology cycle (years) 9.341 9.946 8.512 Originality index 0.284 0.265 0.217 Number of Patent 4965.2 52.4 920.7 Quality of Patent 17519.1 73.0 1583.2 Technology capability 9.00E+08 8.74E+03 3.09E+06
Intra-national Citation in Patents (~self-citation)
Middle Income Countries Basic Descriptive Data of the country groups High Income Countries Middle Income Countries Korea and Taiwan Mean(μ0) Mean(μ1) Mean(μ2) Real GDP per capita growth rate (four-year average %) 0.102 0.067 0.260 Initial GDP per capita, each period 15823 2883 5170 Population growth rate (four-year average %) 0.032 0.077 0.054 Fixed investment per GDP (%) 23.5 23.0 27.9 Enrollment rate of secondary education (%) 95.6 58.6 80.7 Number of Patent 4965.2 52.4 920.7 Quality of Patent 17519.1 73.0 1583.2 Technology capability (share) 9.00E+08 8.74E+03 3.09E+06 Notes: High Income Countries are those whose GDP per capita in 2000 constant prices exceed US$ 10000. In both groups, only those countries that have more than 10 U.S patents in every period are included.
Fixed Effect Panel Model System-GMM Model per capita income growth High Income Middle Income Dummy Model Log of initial GDP per capita for each period -0.1575 -0.3460 -0.2433 -0.1399 -0.1160 -0.1799 (-2.57)** (-4.02)*** (-4.78)*** (-2.81)*** (-3.07)*** (-4.87)*** Growth rate of population for each four-year-period -0.7749 -1.1817 -1.1293 -0.0737 -0.1480 -0.2823 (-1.71)* (-1.70)* (-2.90)*** (-0.09) (-0.21) (-0.39) Fixed capital investment per GDP 0.0030 0.0146 0.0093 0.0089 0.0242 (1.08) (4.12)*** (4.24)*** (2.58)*** (3.43)*** (2.68)*** Enrollment rate of secondary education 0.0006 0.0009 0.0001 -0.0032 -0.0026 (0.99) (0.69) (1.65) (0.03) (-1.40) (-1.38) Herfindhal index -0.2254 -0.1494 -0.3355 -0.2189 -0.1867 -0.5009 (-2.40)** (-2.14)** (-3.17)*** (-1.94)* (-2.85)*** (-2.70)*** Intra-national knowledge diffusion 0.5128 0.3296 0.6397 0.3581 1.0351 0.8361 (1.87)* (0.90) (1.96)** (1.74)* (1.07) (3.32)*** Technology cycle 0.0249 0.0145 0.0319 0.0284 0.0274 0.0636 (3.13)*** (2.30)** (3.50)*** (2.42)** (2.01)** Originality index -0.1042 0.1152 0.0239 -0.1778 0.4870 -0.3502 (-0.52) (0.73) (0.10) (-0.67) (-0.90) Dummy (middle ) (dropped) -0.3999** Dummy*Herfindhal index 0.2059 0.4534 (2.19)** Dummy*Intra index -0.2784 0.1565 (-0.63) (0.49) Dummy*Technology cycle -0.0183 -0.0337 (-1.79)* (-2.04)** Dummy*Originality index 0.0674 0.8946 (0.26) (1.73)*
Results with a dummy for 4 Asian : fixed effect models High Income group (1) Middle Income Group (2) Whole World (3) Log of initial GDP per capita for each period -0.3364 -0.3493 -0.2880 (-4.57)*** (-4.32)*** (-5.48)*** Growth rate of population for each four-year-period -1.1684 -1.1337 -1.0742 (-2.26)** (-1.76)* (-2.74)*** Fixed investment per GDP 0.0070 0.0140 0.0102 (2.72)*** (4.36)*** (5.05)*** Enrollment rate of secondary education 0.0008 0.0012 (1.70)* (0.62) (2.43)** Herfindhal index -0.1952 -0.1579 -0.1442 (-2.42)** (-3.04)*** Intra-national knowledge diffusion index 0.5939 0.3411 0.4261 (2.52)** (0.98) (2.09)** Technology cycle 0.0381 0.0147 0.0195 (5.21)*** (2.45)** (4.49)*** Originality index 0.0200 0.1236 0.0741 (0.10) (0.83) (0.70) Dummy (Asian 4) 1.2881 1.1123 1.1134 (2.37)** (2.91)*** Dummy*Herfindhal index -0.4630 -0.6132 -0.4820 (-2.11)** (-1.87)* (-1.88)* Dummy*Intra index -0.2715 -0.5187 -0.7163 (-0.31) (-0.39) (-0.68) Dummy*Technology cycle -0.0873 -0.0589 -0.0616 (-3.16)*** (-1.39) (-1.78)* Dummy*Originality index -0.8980 -0.9542 -1.0547 (-1.53) (-1.09) (-1.48) Constant 2.6933 2.3479 2.1035 (4.16)*** (4.11)*** (4.97)*** Number of obs 112 95 191 Number of groups 28 27 51 R2 0.282 0.204 0.118
Regressing growth onto National Innovation systems: Asian 4 as benchmark High Income middle Inc. World Tech cycle time (-)* (+)* Localization of knowledge + Originality HH: inventor concentration Asian 4 Dummy (+) * Controls: Initial income, Population, Investment, secondary enrollment Shorter cycle leading to growth in Asian 4
Summary: country-levels High income countries are those with more even distribution of inventors, higher rate of intra-national knowledge diffusion, by higher rate of originality shorter time of technology cycle (than other MICs) but longer than Asian 4s. 2) These variables are all significantly related to growth, except originality variable. 3) Among two suspects of catching-up growth: intra-national knowledge diffusion; short cycle technologies
* Difference: Specialization into short cycle time technologies Longer cycle time is positively related to economic growth in both of high and middle income countries, but negatively (thus shorter time is positively) related to economic growth in the Asian 4 -> detour from long cycle to shorter cycles over the catching up period -> others, eg LA, MIC are still in the long cycle fields
Top 10 Classes of G5 vs Korea-Taiwan ->no overlap Class Name Patent count 1 514 Drug, Bio-Affecting and Body Treating Compositions 10349 2 428 Stock Material or Miscellaneous Articles 3883 3 73 Measuring and Testing 3789 4 123 Internal-Combustion Engines 3479 5 424 3389 6 210 Liquid Purification or Separation 2853 7 435 Chemistry: Molecular Biology and Microbiology 2852 8 250 Radiant Energy 2639 9 264 Plastic & Nonmetallic Article Shaping or Treating 2349 10 324 Electricity: Measuring and Testing 2325 Korea-Taiwan Class Class Name Patent count 1 438 Semiconductor Device Manufacturing: Process 1189 2 348 Television 712 3 439 Electrical Connectors 408 4 257 Active Solid-State Devices ( Transistors, Solid-State Diodes) 374 5 362 Illumination 6 280 Land Vehicles 355 7 365 Static Information Storage and Retrieval 346 8 70 Locks 340 9 360 Dynamic Magnetic Information Storage or Retrieval 313 10 482 Exercise Devices 311
Top 10 Classes of G5 vs 8 middle-income countries -> similar Class Name Patent count 1 514 Drug, Bio-Affecting and Body Treating Compositions 10349 2 428 Stock Material or Miscellaneous Articles 3883 3 73 Measuring and Testing 3789 4 123 Internal-Combustion Engines 3479 5 424 3389 6 210 Liquid Purification or Separation 2853 7 435 Chemistry: Molecular Biology and Microbiology 2852 8 250 Radiant Energy 2639 9 264 Plastic & Nonmetallic Article Shaping or Treating 2349 10 324 Electricity: Measuring and Testing 2325 8 mid income's Class Class Name Patent count 1 514 Drug, Bio-Affecting and Body Treating Compositions 120 2 424 76 3 435 Chemistry: Molecular Biology and Microbiology 54 4 75 Metallurgical Compositions, Metal Mixtures 52 5 65 Glass Manufacturing 44 6 604 Surgery 7 210 Liquid Purification or Separation 40 8 423 Chemistry of Inorganic Compounds 9 502 Catalyst, Solid Sorbent or Product 10 123 Internal-Combustion Engines 38
2nd Question: Sectoral difference in catch-up Answer: 2nd Question: Sectoral difference in catch-up Answer: . (Sectoral Systems of Innovation; Technological Regimes of Sectors) Lee & Lim (2001, Research Policy) Park and Lee (2006: Industrial & Corporate Change) Jung and Lee (2010: Industrial & corp. change)
TFP Catch-up : Korea vs Japan Rapid catch-up (about 30%) Sustain Gap (about 10%) Source: Jung and Lee (2010: Industrial & corp. change) 36
Convergence of Productivity in IT: korea vs Japan Sam. Elect.: OVER While Industry : JUST 37
“Still Gap in Autors: Hyundai vs. Toyota H.M. : Under like industry 38
First tier 2 :Korea and Taiwan Extension of Park & Lee, 2006, (Industrial and Corporate Change) To: 2) Second Tier: 8 2-a) Asia 4: China, India, Malaysia, Thailand 2-b) LA 4: Brazil, Mexico, Argentina, Chile
1st step: Occurrence of Catch-up : (1 or 0 in positive change in paten shares in 417 classes, 80-95) No differ. bt. Asia 4 and LA Mean G5 0.348 Korea_Taiwan 0.814 Asia 4 + LA4 0.151 Asia4 0.140 Latin America4 0.162 China 0.226 India 0.170 Malaysia 0.096 Thailand 0.066 Argentina 0.130 Brazil 0.276 Chile 0.059 Mexico 0.181
2nd step: Speed of Catch-up (%P. change, 80-95): Big difference bt 2nd step: Speed of Catch-up (%P. change, 80-95): Big difference bt. Asia 4 and LA Mean G5 -2.246 Korea_Taiwan 3.677 Asia 4 + LA4 0.030 Asia 4 0.059 Latin America4 0.000 China 0.106 India 0.066 Malaysia 0.035 Thailand 0.028 Argentina -0.015 Brazil -0.008 Chile 0.023 Mexico 0.002
3rd step: Level of technological capability: (average share for the 1980-95) Mean % G5 8.804 Korea_Taiwan 2.164 Asia 4 + LA 0.080 Asia 4 0.041 Latin America 0.120 China 0.061 India 0.059 Malaysia 0.024 Thailand 0.019 Argentina 0.078 Brazil 0.211 Chile 0.020 Mexico 0.170
8 Independent variables measuring tech regimes 1) Technological opportunities : average growth rate 2) Cumulativeness of technical advances (persistence) : share of persistent registrant (with more than one at every year) 3) Appropriability of innovations : share of self-citation received 4) Property of the knowledge base : originality (broad base of knowledge) *5) Fluidity (Uncertainty) of technological trajectory : Fluid2 = (Maximum count-Minimum)/average count of patent *6) Initial stock of accumulative knowledge : initial share *7) Relative technological cycle time (speed of change) : relative citation lag *8) Accessibility to external knowledge flows (spillover/ACCESS) : citation from non-G7 to G7
Regression models 1) Occurrence of technological catch-up = F (technological regimes) whether or not there is positive change in the US patent share of a country (probit regressions) 2) Speed of technological catch-up degree of positive change in share (regressions for the sectors with occurrence) 3) Level of technological Capability (share by a country)
Sector level: In which sector more patents? (Korea +Taiwan) vs. G5 => Korea+Taiwan: more share in short cycle; cf) G5 more share in long cycle sectors
8 Catch-up (Asia 4 + LA), Asia 4, LA : Catch-up and capability Variable Occurrence of catch-up Speed of catch-up Level of tech. capability 8 catch 4 Asia 4L A 4L.A 4LA OPPOR -0.989* -0.576 -1.560** 1.160** 0.40 2.90** -0.078 -0.00 -0.15 CUMUL1 -0.599** -0.454 -0.727* 0.078 -0.06 0.17 -0.134** -0.02 -0.25** APPRO 0.25 0.198 0.323 1.204** 2.93*** 0.14 0.210*** 0.18*** 0.24* NATURE 0.340 0.164 0.874* 1.21** 0.58 0.034 0.09* FLUID2 0.12 0.102 0.149 0.079 -0.04 0.07 -0.001 -0.01 0.01 INITIAL 1.09*** 1.02*** 1.15*** -0.38*** -0.29*** -0.41*** 0.00 0.02 CYCLE -0.489** -0.76** -0.19 1.52*** 0.28 2.32*** ACCESS -1.12 -2.18 -0.31 2.07 0.80 3.77 0.31 0.04 Adjusted R2 275.5 125.9 155.578 0.189 0.22 0.045 F-statistic 0.118 0.10 0.117 8.010 6.48 7.1 10.5 3.86 6.6 No. obs. 3008 1504 453 210 243 For next tier 8: cycle time-> positive and insignificant Cf) Korea+Taiwan: negative and signficant
Levels of technological capabilities First 2, higher in short cycle time sectors, but Next 8, lower in short cycle sectors Two faces of leapfrogging argument (Perez & Soete): new technological paradigm may permits leapfrogging or act as additional barriers with truncation of learning (Lall 1992; 2000) Changing paradigm = either a window of opportunity or a barrier to catch-up How to ride well this changing paradigm determines the chance for sustained catch-up (vs. short-lived)
Summary: Sectoral patterns of catching up: -> catch-up happened in such sectors as: Short cycle technology sectors High appropriability sectors Plus (Jung & Lee 2010) Explicit /codified (less tacit) knowledge sectors Oligopolistic sectors with firms with export-orientation (domestic rents + external discipline); -> Schumpeter’s insight
3rd Question: Catch-up at the firm level: CIS (corporate Innovation systems);
in the advanced countries and the late-comer countries Our departing point Firms in the advanced countries and the late-comer countries are different in many aspects, including their levels of capabilities and behavior. *There should be more differences => this study -> knowledge bases of the firms cf) Schumpeterian theory of the firms (Winter, Nelson): accounting inform. not enough to show heterogeneity of firms
Comparing Korean firms with the American firms the former representing firms from late-comer countries ; the latter representing the firms from the advanced countries. ; focuses on innovation systems of firms, -> such as cycle times of firms’ technology, self-citation (self-production of knowledge) technological diversity of firms, Originality of firms’ knowledge base,
Specializing in Short cycle technologies and Leapfrogging: Hypo 1 A shorter cycle => latercomers do not have to master older knowledge and patents. -> less disadvantageous in such sectors; -> So, catching-up firms better specialize in short-cycle technologies. -- consistent with the leapfrogging hypothesis (Perez & Soete 1988): So far no research that confirms this view at the firm level, cf) Park and Lee (2006:ICC) confirm this at the sector level using patent data.
Catching-up (Korean) vs. Mature (US) firms: The former in short cycle technologies
Consolidation of self-production of knowledge: high in US vs Consolidation of self-production of knowledge: high in US vs. low in Korea : Hypo 2 Will compare the self-citation ratio in the patent citations between catching-up vs. advanced firms. self-citation: = degree to which one’s innovation builds upon its own knowledge pool accumulated over the past. Hypo. : the more advanced a firm is, the higher self-citation ratio. Hypo: some correlation between self-citations and performance
Catching-up (Korean) vs. Mature (US) firms: The former in low self-citation (localization) USA Firms
Self-citation Ratio of Samsung Electronics and Sony: (Joo and Lee 2010)
2 Additional Aspects: Originality and Diversity Impact of the originality of the knowledge base of a firm -- U.S. firms listed on the stock market would be on the frontier of technologies >their knowledge base/pool is full of more original ideas Hypo: originality , higher in U.S. firms compared to Korean firms. Cf) no study has confirmed the significance of this variable; the literature tends to emphasize the importance of “being creative or original” in today’s world of competition in the era of knowledge-based economy. 2) Diversity into more technological fields vs. concentration into a few fields: Hypo: more advanced (mature) = more diversity than catching-up firms
Measurement : originality Def) If a patent cites patents from more diverse fields, more original -> The broader is the technological root of the underlying knowledge or research related to the patents, the higher originality of a patent. --Synthesis of divergent ideas is probably characteristic of research that has high originality, and hence is basic in that sense (Trajtenberg et al., 1997). speed. Formula: for each firm i. ORIGINALITY = 1- where k is the technological sector (more precisely patent class k), NCTINGik = No of citations made by the patent held by firm i to patents that belong to patent class k, and NCITINGi is the total number of citations made by the patent held by firm i.
Measurement : Diversity (concentration) Diversity measured in two ways. simply the number of technological sectors (classes) that each firm has applied its patents. one minus the Herfindahl–Hirschman Index (HHI) of concentration, and HHI can be measured for firm i at time t as follows: HHI = summation over k classes of share of each class’s patent in total patents of each firm at time t
Comparison of the Knowledge Variable: means Part A: Sample means Variable US Korea US-KOR Gap t-value Patent Count 18.5 9.56 8.94 4.592** Patent quality 1.14 0.73 0.41 6.821** Number of sectors with patents 6.9 4.15 2.75 5.044** HH index 0.51 0.71 -0.2 -8.808** (degree of sector concentration) Originality 0.42 0.3 0.12 7.662** Technology cycle (years) 14.05 11.91 2.15 4.39** Intra-firm diffusion (self-citation) 0.03 0.09 18.001** Catching-up or Korean firms , inferior to the American firms in every aspects of knowledge variables: patent counts, quality, originality, and its diversity. * But note that : Korean firms to have patents with shorter cycle times than the American firms.
Catching up firms tend to pursue sales growth Comparing the catching-up (Korean) firms with the advanced (US) firms : behavior Variables US Korea Difference (US-KOR) t-value Number of employees 13719.57 6857.53 6862.04 7.634** (unit : person) Sales per employee 187.11 294.47 -107.35 -7.445** (unit : thousand dollars) ROA: Return on assets(%) 9.3 8.2 1.1 3.232** ROS: Return on sales(%) 4.7 9.9 -5.2 -5.228** TOBIN-Q 1.76 1.01 0.74 34.171** Sales growth rate(%) 8.8 12.1 -3.3 -2.788** Investment Propensity(%) 1.0 2.6 -1.6 -3.356** Debt to equity ratio(%) 266.1 302.7 -36.6 -0.342 Capital Labor ratio ( unit : thousand dollar) 60.24 153.91 -93.67 -7.251** Catching up firms tend to pursue sales growth by borrowing and investing more, while the advanced firms, to purse profitability and firm values.
Knowledge and firm performance I: benchmark with patent count only -> not much new and not much difference of US and Korea ( based on Hausman test: random vs fixed US firms Dependent variables: GROWTH ROA ROS SALES/EMP TOBINQ Patent Count (-) (+)* (+) (+)** No. of workers (-)** (-)* Investment Propensity Debt to Equity Ratio Capital Labor Ratio Obs 3475 3479 3478 3362 Korea firms (+)+ (-)+ 239 240 127
Investment propensity Knowledge and firm performance II: US; one variable in each model: no significance of cycle time US firms Dependent GROWTH ROA ROS SALES/EMP TOBINQ HH Index (+)* (-) (+) ** Originality (+)** Tech. Cycle Self-citation (+)+ No. of Emp. + Investment propensity Debt to Equity Ratio Capital Labor Ratio (-)** (- )**
Investment Propensity Knowledge and firm performance II: Korean; no significance of self-citations Korea firms Dependent GROWTH ROA ROS SALES/EMP TOBINQ Independent HH Index (-) (+) (-)+ Originality (+)+ (+)* Tech. Cycle (-)** (-)* Self-citation No. of Emp. Investment Propensity Debt to Equity Ratio Capital Labor Ratio Obs 239 231 217 240 232 218 127 122 114
Knowledge and firm performance III: with 3 relevant variables only US firms Dependent GROWTH ROA ROS SALES/EMP TOBINQ H-H Index (+)* (-) (+) (-)** Originality (+)** Self-citation (+)+ No. of workers (-)+ Investment Propensity Debt to Equity Ratio Capital Labor Ratio Obs 3468 3472 3471 3355 Korea firms H-H Index Tech. Cycle (-)* Investment propensity 231 232 122
Summary: basic profiles Behavior: => catching up firms tend to pursue sales growth by borrowing and investing more, while the advanced country firms, to purse profitability and firm values in stock markets. knowledge bases: => catching-up or Korean firms are inferior to the American firms in every aspects of the patent counts, their quality, their originality, and its diversity. --- except Korean firms tend to have patents with shorter cycle times than the American firms.
From Knowledge to Performance 1 Self-citation: significant in US firms, not in Korean firms Cycle time: not important in the US firms; quite significant in Korean firms originality does not matte much in both US and Korea firms For Korea: Short cycle leading to higher profitability (but not to high growth or value): -> catching-firm pursued growth by physical investment but pursued profitability by moving into emerging/new industries, eg. From footwear to automobile / IT. or, in cross-sectional terms, short cycle sectors ( eg, IT) give them higher profitability than longer cycle fields (footwear)
From Knowledge to Performance 2 2) self-citation pattern at firm level: -> insignificance in the Korean firms indicates still weak level of the mechanism of self-production of knowledge,; self-citation is only 3% in the Korean firms, compared to four time higher level of 12% in the US firms; medians: zero for the Korean firms, 9% US firms. The next task for the Korean firm is to consolidate more their intra-firm knowledge creation mechanism so that it may translate into higher productivity and firm values.
at the 3 Levels of catching-up Summary of the Study at the 3 Levels of catching-up
Overall: Short cycle technology matter
in academic research, too Short cycle matters in academic research, too Jones and Weinberg (2012) on the age-achievement relationship in the natural sciences Young scientists (late entrants) to make more contributions at a younger age in the fields of abstract /deductive knowledge than in the more inductive fields that draw on accumulated knowledge, and in which existing knowledge is slow to reach obsolescence
Japan caught up with Germany by the mid 1970s
Japan also specializing in shorter cycle sectors than Western incumbents ->path creation
shorter cycle technology sectors Diversification by moving into shorter cycle technology sectors Tech. turning point
China from late 90’s but India not yet? India still not shorter than 9 years
China and India Average Cycle Time of China’s top 30 class US patents = 8.1 years (2000-2005 yrs) Cf) Korea and Taiwan = 7.7 yrs (avg of 1980-95) Brazil & Argentina = 9.3 yrs (avg. 1980-95) China more similar to Korea & Taiwan than to Brazil and Argentina Cf) India: applying IT (short-cycle tech) to Services Average weighed Cycle time = 8.7 yrs (2000-05)
Not yet over the Turning points In Latin American countries
Synthesis 3 variables out of 5 more closely related to catch-up = those showed a different pattern between more and less successful economies -> cycle time, localization, & diversification One Transition variables = cycle time (how to catch up) and Two End-state variables = localization (who is leading catching-up) and diversification (what is catch-up)
Tech. Diversification = No of sector with patents / 417 cf) 417 = No of 3 digit classes in USPTO; Diversification matter but where to diversify? => into short cycle
Other criteria of targeting 1) high opportunity sectors -- but: how about competition/entry barriers 2) Justin Lin: GIF framework: target mature (left-over) sectors of country above you’ -- Makes sense; b/c to find a niche (lower entry barrier); but might not be enough as you get close to Frontier (to be a HIC ; to take over) - > Lee (2013) : short cycle time: more theoretical & more for upper middle C’s 82
How about trade-based diversification than short cycle sectors? ** Hausman et al. (2007 JEG); Imbs and Wacziarg’s (2003 AER) : - Diversification (of export structure) as a necessary condition for growth-> a key challenge for developing countries -> A measure of the sophistication of tradable products using income level as the weighting factor. -> but, tautological, as it say “you have to move into goods currently being produced by richer countries if you want to be rich” -> not an effective criterion for specialization; where to diversify first? ** We state the opposite: “To specialize in sectors with short-cycle technologies instead of those already being dominated by rich countries (long-cycle technologies)”
Can take a Detour if you have a high driving skill , when the straight road is jammed Straight Road: but traffic jam (adding-up problem) 1980 Detour: No jam but rough & winding road -> need skill (tech. capability)
Take-home message of the book You cannot catch up by directly emulating or replicating the practices of the forerunning economies. Catch-up comes only if you take a different path (to find a niche) (b/c catch-up is chasing a moving target)
Should allow ‘detour’ for latecomers Should allow ‘detour’ for latecomers!!! cf) direct replication of the developed 1) Immediate trade liberalization vs. asymmetric/selective liberalization => Korea, Taiwan used to be more protective; but now most open 2) High vs. gradual protection of IPRs => used be low in Korea & Taiwan but now very high 3) Big Bang vs. Gradualism in system transition => Washington Consensus vs. BeST (Beijing-Seoul-Tokyo) Consensus
To From Middle to High Income Countries From Trade Specialization to Technology Specialization Stages Low or low middle income Upper middle income To high income Type of specialization Trade specialization Technology specialization Source of specialization Comparative advantages from resource endowment Absorption/design capability from learning/R&D effort Type of sector Labor intensive/resource industries Short cycle/emerging technologies End goal competitive export industries Indigenous knowledge creation & diffusion Background theory Product life cycle (inheriting) Catch-up cycle (leapfrogging)
Now, How to drive the Detour: Implementation Strategies The detour is not just smooth and easy; -> requires certain level of technology capacity, not only firm-level but also at the national-level
3 Steps along the Detour (to move beyond OEM/assembly) Acquiring Design Capability (to move beyond OEM/assembly) 2) Targeting/Entering the mature /low-end segment of short cycle Sectors 3) Leapfrogging into New/Emerging Technologies in the Short-cycle Sectors
So, Industrial Policy is 1) Not protection but earning time to learn 2) Not picking winner but picking good students and matching them with good teachers 3) to facilitate to break into higher segment of GVC and new emerging sectors 4) Sacrificing short/static gains for long-term /dynamic term gains 91
Space for Industrial Policy under WTO regime (along the stage of econ dev’t) 1) Production subsidies or SOEs for initial promotion (instead of local contents requirement) 2) Under-valued currency for export promotion (b/c export subsidies prohibited) 3) R&D subsidies for innovation capa. buidling 4) Public-private Joint R&D for leapfrogging Ind Policy = not zero sum but positive sum for a globe (global public goods) ( also, allow IP for entry into near-monopoly market
2 Kinds of Latecomers’ Advantage = Low cost inheriting and Leapfrogging 1) Advantage in Mature Industries (Gerschekron) -> possibility of low cost-based entries without bothering to bear the burden of R&D costs -> you can adopt most up-to-data facility or products 2) Advantages in Emerging Industries -> entry at an earlier stage with the same entry costs but without the sunk costs or being locked into old technologies. -> Leapfrogging
Leapfrogging vs. Incumbent Trap Leapfrogging: Perez and Soete (1988) ->Emerging technological paradigms = a window of opportunity -> not being locked into the old technological system and thus being able to grab new opportunities in the emerging industries. for the catching up country, Cf) Forerunners’ Trap: locked into exiting technologies due to the sunk costs of their investment.
Risk and Potentials of Leapfrogging: Path-following = entry with oldest generation tech (1G) 2) Stage-skipping = entry by up-to-date generation 2 Tech 3) Path-creating /leapfrogging = entry by generation 3 technology
Risk of leapfrogging without Ind. Policy: Solyndra in US; Solar panel cost in 2 generation techologies 1st G: amorphous silicon cells China enters 2nd G: thin-film solar cell: solyndra Solydra entered with 2 generation tech-> failed: Source: BNEF Bazilian et al (2012), Fig. 1
Now Another Paradigm Shift 1) New Energy Revolution (Renewable Energies) to replace fossil-fuel 2) Fusion of Technologies (IT, BT, NT,) in search for new solutions -> Best Time for Leapfrogging and already happening; former latecomers are no more latecomers 97
Stage-skipping and Leapfrogging in the Environmental Kuznets Curve Source: adapted from Assefa (2011)
More Examples Late Entry: Entering Mature Segment of Short cycle Sectors: eg) High speed Train in China, India’s IT service, Middle sized Jets by Brazil Entry into notebooks by P-P in Taiwan Suggestion: Nigeria can build oil refinery, rather than keep exporting crude oils Leapfrogging into New/Emerging Segments of Shorter-cycle Sectors: eg) Solar PV and Wind power in China and India, Electric Vehicles by China Ethanol or Biofuels in Brazil
Every country for IT’s? Another Adding-up? Analogous to the adding-up problem or risk of specialization in labor-intensives by all low-income countries. Big difference: 1) specialization based on factor endowments : fixed with few opportunities for change, 2) specialization in short-cycle technologies : no fixed list of technologies but rather specializing in a field or sector where new technologies always emerge to replace old ones, as existing technologies become obsolete soon.
Last, putting China‘s NIS into Comparison vs. Germany, Japan vs Last, putting China‘s NIS into Comparison vs. Germany, Japan vs. Korea, Taiwan, India
Is China innovative enough? Cf) Korea, India
Korea also now shorter than Japan, and China too in the future?
Technological Diversification: No of sectors with patents/417 Japan , Germany Korea, Taiwan
Originality (3 year moving average)
<Inventor concentration (HH) Index: 3 year moving average>
<Intra-National creation and Diffusion of knowledge: localization> Taiwan
China lower than India: Intra National Diffusion>
NIS in China 5 patterns and a puzzle? Overall Assessment: China seems to have passed the technological turning point and to be moving toward shorter cycle technologies Steady record of diversification into diverse fields already some level of balances in assignees concentration and of originality but a puzzle: somewhat lower or slow progress in localization of knowledge creation & diffusion; even lower than India or Korea in late 80s. not so much indigenous actor oriented Overall Assessment: Satisfactory in terms of establishing the engines of catch-up by moving into shorter cycle technologies. But yet to consolidate local knowledge basis by promoting more indigenous firms R&D and their interaction
ありがとう! Gracias! Meu Amigo! Obrigado! Thank you! 謝謝大家 감사합니다
References (www.keunlee.com) Lee, Keun, & BY Kim,”Both Institutions & Policies matter but differently at differnent income levels: long run economic growth,: World Development (2009). Lee, K. "Capability Failure & industrial policy to move beyond the middle income trap“ in Stiglitz and J. Lin (eds.), Industrial Policy Revolution I. 2013. Lee, Keun, & C. Lim (2001), “Technological Regimes, Catching-up & Leapfrogging: the Findings from the Korean Industries”, Research Policy, 459-483. Lee, Keun, Chaisung Lim, and Wichin Song (2005), "Digital Technology as a Window of Opportunity and Technological Leapfrogging: Catch-up in Digital TV by the Korean Firms”, Inter.J. of Tech. Management, Vol. 29, 1/2, pp. 40-64. Lee, Keun, “Making a technological Catchup.” Asian J.of Tech. Innovation, 2005. Park, K., and Keun Lee (2006), “Linking the Technological Regime to Technological Catch-up: An Empirical Analysis Using the US Patent Data,” Industrial and Corporate Change, July 2006 Jung, M & K. Lee, (2010), “Sectoral systems of Innovation and Productivity Catch-up: between the Korean and Japanese firms,” Industrial & Corporate Change.
Keun’s more recent topics: A Theory of Catch-Up Cycle: Changes in Industrial Leadership and Catch-Up by Latecomers
* More than 2 times Changes in industry leadership Steel: U.S. Japan Korea (partly) China Automobile: Germany U.S. Japan partly Korea ; China Shipbuilding: U.S. Britain Japan Korea partly China Semi-conductor (memory chips): US-> Japan -> Korea Mobile phones: motorola -> Nokia -> Samsung ( Apple) Camera: Germany -> Japan 1 -> Japan 2 IT service: US – Ireland(partly) –India Mid size jet: Europe -> Canada (Bombadier) – Brazil (Embraer) Wine: France -> US, Australia -> Italy Why do this often occur in many sectors? Why the leader cannot persist but decline?
Our Theory: Industry Catch-Up Cycle 3 Windows of Opportunity, 3 strategies, 3 Cycles Each cycle is that of a leading firm or a collection of firms in a nation; -> a new cycle replacing an old cycle
Secrets of Catch-up Cycles = windows of opportunity + Incumbents Responses (incumbents’ trap) and Latecomer’s Advantages and Disadvantages
Why Incumbent declines? -> Incumbent Trap/Lock-in Winners tend to falling into trap: (of ignoring new technologies) -> be complacent with the current success (with the current/dominant technologies). not necessarily by their mistakes but by rational choice; -- given uncertainty of new tech, and given fixed investment whose life cycle has not finished -> emergence of new paradigm/generations of technologies + incumbent trap leadership changes
3 catchup strategies: Leapfrogging with Gov’t support Path-following strategy = start from generation 1 technoloiges stage-skipping = entry with generation 3 tech (most productive and stable) Path-creating/ leapfrogging = jump to generation 4 (emerging) technology)
‘History-friendly’ calibration: radical change in leadership after technology shocks Evolution of total market shares: -> immediate leadership changes Text with No Subtitle
-> no leadership change: = Aborted catch-up (failure to catch-up) Experiment 1 – no Lock-in effects -> reduce the intensity of lock-in (incumbent trap) effects -> no leadership change: = Aborted catch-up (failure to catch-up) Text with No Subtitle