• Share
  • Email
  • Embed
  • Like
  • Save
  • Private Content
Lecture 5 - Technology diffusion and technology transfer
 

Lecture 5 - Technology diffusion and technology transfer

on

  • 25,011 views

 

Statistics

Views

Total Views
25,011
Views on SlideShare
24,754
Embed Views
257

Actions

Likes
6
Downloads
685
Comments
1

7 Embeds 257

http://ocw.unu.edu 191
http://www.slideshare.net 58
http://www.ocw.unu.edu 4
http://translate.googleusercontent.com 1
http://static.slideshare.net 1
http://webcache.googleusercontent.com 1
http://localhost 1
More...

Accessibility

Categories

Upload Details

Uploaded via as Adobe PDF

Usage Rights

© All Rights Reserved

Report content

Flagged as inappropriate Flag as inappropriate
Flag as inappropriate

Select your reason for flagging this presentation as inappropriate.

Cancel

11 of 1 previous next

  • Full Name Full Name Comment goes here.
    Are you sure you want to
    Your message goes here
    Processing…
  • tt
    Are you sure you want to
    Your message goes here
    Processing…
Post Comment
Edit your comment

    Lecture 5 - Technology diffusion and technology transfer Lecture 5 - Technology diffusion and technology transfer Presentation Transcript

    • Lecture 5: Technology diffusion and technology transfer
    • Introduction • Diffusion=adoption of innovation(s) • Technology transfer=much broader [Diffusion plus] • Accumulation of knowledge takes place with the interaction among actors.[this process builds on investment decisions and allocation of resources –compare that to the technological capabilities approach] • Useful line of research for better understanding of the supply side of innovation dynamics and the structure of R&D systems. [increasingly relevant to industrializing countries with indigenous R&D systems] • Availability of data for empirical research and well established methodologies [sectoral linkages, stocks of knowledge, trade flows, data on innovation dynamics]. • An emerging field of research in Developing Countries with dynamic export oriented, high-tech sectors • Relevant to several policy issues: FDI, IPR, Public policies supporting R&D.
    • Theories of Diffusion • Up to this point, we have focused on the development of new knowledge. However, technology can only have a wider impact on the economy if it is used by a large population of firms/comsumers. Thus, the picture is incomplete. Now, we look at the adoption of new innovation. This is the study of diffusion. • Today we look at models of diffusion, to see how economists model the decision to adopt technology. We’ll also look at some empirical results. Next, we' discuss policies that encourage diffusion within a ll country. Finally, we’ll look at international technology diffusion. Key questions: – What is the rate of adoption and innovation? – What variables affect this rate? – How does policy affect diffusion?
    • Early studies of diffusion • Early studies of diffusion focused on agricultural technologies. The first studies were done by anthropologists and sociologists, not economists. • The diffusion traditions of various disciplines began to merge in the 1960s with the Ryan-Gross study of hybrid corn. This was a sociology study. • Hybrid corn, introduced by Iowa State in 1928, had yields 15-20% higher. It had been adopted by most Iowa farmers by 1940. • Ryan and Gross studied what factors influenced its adoption. They found that communication between previous and potential adopters was important. They found an S-shaped rate of adoption. This result is still typical today. • Sociologists focus on the following reasons to explain the S-shaped curves: – What is the relative advantage of the innovation over the existing technology? – Is the technology compatible with potential adopters current way of doing things, and with current social norms? – How complex is the technology? – Can the technology be tested? – How easy is it to evaluate the innovation after it has been tried? • They also define five categories of adopters: – Innovators – Early adopters – Early majority – Late majority – Laggards
    • Griliches’ (1957) hybrid corn study • Economists have added to the diffusion literature by examining individual differences (and decisions) in diffusion. • Griliches looked at the diffusion of hybrid corn both within regions (as did Ryan/Gross) and across regions. Goal: to explain why individual adoption decisions vary: – The “acceptance problem” – what explains variation in adoption rates within a region. – The “availability problem” – what explained the timing of the development of hybrid corn for specific areas. • Griliches did this by fitting an S-shaped logistic trend diffusion curve to data on the percentage of corn area planted with hybrid seed. • Key features of the curve: – Origin: the starting point. Griliches defines as date at which 10% of a region’s corn was hybrid. This is meant to indicate commercial availability of hybrid corn. – Average lag between technical availability and commercial availability was 2 years. – Agricultural stations did more work on hybrid corn in regions where corn was important (e.g. Iowa, Wisconsin). The date of origin is earlier here. – Slope: indicates the rate of acceptance – Ceiling: measures the percentage of acceptance when usage stabilizes • Interpretation: differences in rate of acceptance (slope) and level of acceptance (ceiling) can be explained by differences in how profitable it is to shift to hybrid corn. • .
    • Diffusion of industrial technology • The standard S-shaped diffusion curve has also been found in studies of industrial technology diffusion. Mansfield (1968) looked at factors influencing inter-firm and intra-firm diffusion. He examined the diffusion of 12 different technologies in 4 different industries. • Mansfield found that key variable was Profitability Example: Mansfield used regression techniques to explain differences in diffusion patterns Significant variables included: – Profitability of investing in diesel locomotive – Inter-firm differences in size and liquidity – Differences in the starting date
    • Environmental technologies • Recent work in environmental economics shows that environmental regulations do lead to faster diffusion of environmental technologies. • However, rates of adoption are slow. Cost-effective technologies diffuse very slowly. Why might that be? • Potential market failures include: – Inadequate information – Agency problems The person installing the technology might not be rewarded for doing so (e.g. landlord/tenant relationship) – Consumers have high discount rates. Thus, they place little weight on potential benefits. – Lack of access to credit markets • Subsidies have been found to be more effective increasing diffusion than higher prices, suggesting that access to credit is important -- people will adopt when they can afford to.
    • Recent work on diffusion has focused on trying to explain the prevalence of the S-shaped diffusion curve - The epidemic model • The epidemic model considers information to be the key to diffusion. As more people adopt the technology, information of it spreads quickly, leading to a period of rapid adoption. • The epidemic model models technology as a “contagious disease.” Adoption occurs as potential adopters learn about the new technology. Adoption is slow at first, as few people (or firms) know about the technology. The more people “infected” (that is, those that have adopted), the more likely others will also be “infected.” Thus, as information spreads, a period of rapid adoption follows. • Shortcoming – This model assumes that, once potential adopters learn of a technology, they will use it. – This model assumes the quality of the technology is the same over time. • Implications – Adoption includes a positive externality. The decision to adopt makes it more likely that others will also learn about the innovation. This suggests that gradual diffusion is the result of a market failure. It also suggests that, until market saturation is reached, the economy is in disequilibrium. • Recent modifications focus on equilibrium. These models assume there is perfect information on the technology, so that the epidemic model is not relevant. Rather, there are differences among users that explain gradual diffusion. Firms must pay a cost, ct, to adopt the technology at any time t. This price changes over time. Each firm weighs the benefits of adoption at time t against the cost of adoption at time t. As the costs or benefits of adoption change, the number of adopters changes. • Implication: – Gradual diffusion is rational. It is the result of profit-maximizing behavior, rather than a market failure.
    • Other diffusion models • Stock models • As the number of users of the new technology increases, the gross benefits from adoption decline. That can be due to the effect of technology adoption on prices or due to prices in factor markets (supply effects). • Order models • The firm’s position in the adoption order determines its gross return from adoption. Early adopters typically get a higher gross return. This suggests the first-mover advantage dominates the advantage of waiting for better technologies. However, costs are important to get net return from adoption. • Very recent work combines equilibrium and epidemic models. These models are hazard models. Hazard models combine a baseline epidemic diffusion curve with firm-specific variables to capture the effects above. • Firms adopt when the Net Present Value of adoption is: – Positive, so that adoption is profitable, and – Higher than it would be if the firm waited until a later date to adopt – Thus, unlike the other models, there may be a benefit to waiting. • This approach allows the researcher to capture the magnitude of each effect. Indeed, these recent results suggest the firm-specific effects are more important.
    • Types of equilibrium models: Rank (or probit) models • Potential users differ in some important characteristic. Thus, some firms benefit from adoption more than others do. The net benefits can be ranked across firms. Those with the highest ranks go first. Examples of rank effects found to be important: – Firm size (generally a positive effect) – R&D expenditure – Market share – Market structure (ambiguous effect) – Input prices – Characteristics of the technology – Government regulations
    • 11 Neoclassical Equilibrium VS Evolutionary Disequilibrium Neoclassical Equilibrium Evolutionary Disequilibrium Assumptions • Full info/limited info • Necessarily limited info • Infinite rationality • Bounded rationality • Equilibrium mechanism • Disequilibrium mechanism • Exogenous/endogenous • Necessarily endogenous • Continuous & quantitative • Continuous & quantitative, or • Discontinuous & qualitative Characteristics of the diffusion process • Predictable • Unpredictable • A-historical • Path-dependent (historicity) • Efficient • Efficient, or • (possibly) Inefficient
    • U.S. diffusion of major inventions 100 90 80 Telephone 70 Refrigerator 60 Share (%) Washing machine 50 40 VCR 30 20 10 Electric Service 0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 Year
    • Economic determinants The adoption decision will depend on the factors that usually affect investment decisions: – Benefits – Costs – Risk and uncertainty/information – Environment and institutions – market and/or regulatory
    • Benefits depend on • Closeness of substitute technologies – Mobile and landline telephones • Networks and standards – ATM adoption by banks – VHS/Beta • Experience and Learning – Investment in adopter skills
    • Costs depend on • Price of new technology • Complementary investments – Infrastructure and other capital equipment – Training/skills • Scale – Due to fixed cost nature of adoption in many cases • Cost of finance – Large body of literature on innovation investment where there is uncertainty
    • Uncertainty/information • New technology – less understood, more uncertainty about how well it works • Uncertainty about whether it will be successful (standards) • Benefits are a flow, costs are upfront => benefits may be more uncertain • The option to delay decision in order to acquire more information may cause delay in adoption
    • Environment – market structure Size and/or market power of adopters: • Accelerates diffusion – Scale economies • Delays diffusion – Slower and less flexible Size and/or market power of suppliers: Accelerates diffusion [Sponsoring a standard (e.g., IBM and – the personal computer) Delays diffusion [Higher prices, Less fear of market share • loss to entry (see ATT in 1960s)]
    • Environment - regulatory • Accelerates adoption – mandates pollution or safety standards – solves coordination problems in network industries • Delays adoption – Safety regulation, e.g., new pharmaceuticals and medical instruments. – Standard-setting process - telecommunications lighting innovation?
    • Some empirical examples (B. Hall) Date Authors Technology Observations Factors 1957 Griliches hybrid corn Midwest farms profitability; need to specialize product 1975 David mechanical reaper US,UK farms minimum efficient scale regulation; concentration;firmsize;holding co. (risk);cost of 1984 Hannan/MacDowell ATMs US banks substitutes 1995 Saloner/Shepard ATMs US banks network size;customer deposits (size) US auto prod worker wage;tech complexity;size;stable customer 1995 Helper CNC machine tools component firms relationshp 1997 Kennickell/kwast electronic banking US consumers education; assets; learning (older services versus newer) Majumdar/Vankatara 1998 man elec switching tech US telecomms network effect and scale (weaker over time) 1998 Gray/Shadbegian new tech in paper US paper plants environmental regulation on-time benefits;stable customer relationship - helps 1998 Hubbard on-board IT US trucking monitoring firms cross- 2000 Stoneman/Toivanen robot technology country real options; volatility in uncertain investments? education level of workers;openness;overall investment 2001 Caselli/Coleman computers OECD countries rate European 2001 Gruber and Verboven mobile telephones consumers concentration of providers;tech improvements
    • Technology Transfer - Types of technology transfer: • Cooperative research and development • Licensing or sale of intellectual property • Either across firms or to a start-up spin-off firm • Technical assistance • Public exchange of information [e.g. conferences, publications, networking]
    • Three competing paradigms for government support • First, Market failures: The government intervenes to correct market failures. Examples used include externalities, high transactions costs, and imperfect information. However, transactions costs need not be a market failure. Transactions costs are legitimate costs paid by individuals. However, the government can still take advantage of opportunities to lower transactions costs. • The evolution of universities as the source of basic research after WWII follows from this. • Examples of other policies: – Intellectual property rights – R&D subsidies – R&D tax credits
    • Three competing paradigms for government support • Second, Mission technology paradigm: – Government should perform R&D in service of well-specified missions for the national interest that aren’t well-served by private R&D. Has been used to justify agricultural research for a long time. – More prominent after WWII, in areas such as defense, energy, and health. In some ways, this is related to market failure, as that may be why private R&D doesn’t serve the need. Other times, private R&D won'serve the need because t there is no market -- it is the government who is purchasing the good [missing markets]. • Third, Cooperative technology paradigm: – Government plays an active role in technology transfer. Has received more attention recently [think of Taiwan in lecture 4]. Example: patent citations from government labs. Jaffe, Banks, and Fogerty (1998) look at patent citations received by NASA patents. NASA patents are more general (that is, cited by patents in more patent classes) and more important (that is, cited more often) than other patents
    • Institutions for Technology Transfer- US, Research joint ventures (RJV) • They are a type of strategic research partnership (SRP). • Types of actors: – 88% for-profit firms – 9% non-profits (including universities) – 3% government • However, more RJVs have some participation from government or universities. – 15% of RJVs have a university member • Varies by industry: from 1985-2001, 30% of RJVs in electronics had at least one university member. • 12% of RJVs have a government laboratory member. – Two-thirds of partnerships in electric equipment, computers, chemicals, or transportation. • Size of RJVs: – Largest: Oil and gas RJVs had a median of 8 members. – Chemicals: median of 5 – Electronics: median of 6 • Motivation: – Alleviate the spillover problem by internalizing leakage of R&D – Improve coordination of R&D efforts to avoid wasteful duplication – Spread the risks associated with large-scale projects – Assure access to complementary knowledge – Take advantage of scale economies • However, unintended transfer of technologies is a concern of industry • When are they likely to be effective? – When spillovers are only moderately high; If spillovers are too high, there is no incentive to join. Being a free rider makes more sense; If they are too low, there is no reason to internalize them ; When overall IP protection is weak; If IP protection is strong, the costs of sharing information are greater (you give up more of your private benefits by sharing).
    • Government Laboratories • In the US, of the $93.3 billion government funded R&D in 2004, 26% was performed by the government. • Types of research done at government laboratories – Research in support of agency activities. This contributes to technologies purchased by the government. – Data collection (e.g. Department of Commerce, NSF) – Basic and applied science in areas with a public interest, such as: • Biomedical research at NIH • Basic physics research at DOE • Meteorology research at NIST • Agricultural research at Agricultural Research Stations – Supporting the commercial activities of firms • Unlike the above three (“mission research”), this type of research focuses on technology transfer. The rationale for Government laboratories – Scale – some research projects rely on large capital expenditures (e.g. medical institutes, large telescopes) – Security – some projects require direct government supervision [DOD is in charge of a large share R&D]. – Mission and regulatory requirements – agencies such as FDA are required to perform a certain amount of R&D – Knowledge management – Long-term R&D kept in house to preserve control of projects and keep close connections with sponsors
    • Technology diffusion across countries • An important consideration is the ability of a country to absorb new knowledge. Trade can provide the elements of technology, but these other elements must also be in place for technology transfer to be successful. – Production capabilities – Investment capabilities: does the country have the ability to expand production facilities and build new ones. – Invention capabilities: the ability of the local efforts to adapt, improve, and develop technology. Very similar ideas to technological Capabilities. However, they elaborate further with concepts yielding empirical results
    • The first basic mechanisms of technological diffusion - direct learning (active spillovers) • Direct learning about foreign technological knowledge. This means learning about the blueprint of the technology, so that the recipient country can reproduce the technology. • If the cost of obtaining the knowledge is less than the cost of invention, a spillover has occurred. Such spillovers should increase the productivity of domestic inventive activity. It becomes part of the domestic knowledge stock off which inventors build. • No purchase is necessary for active spillovers. They can simply be transferred via blueprints. Such transfer can be low-cost. • However, without licensing agreements, the inventor may choose to keep the blueprints secret. • Although no direct purchase is required, activities such as trade or FDI can help create and maintain communication channels.
    • Direct learning-2 • Active technology transfer refers to tacit knowledge also. • Tacit knowledge can be transferred by: – Demonstrations – Personal instruction – Provision of expert services – Hiring workers away from the innovating firm.
    • The second basic mechanisms of technological diffusion: passive spillovers • Technology embodied in specialized and advanced intermediate products that have been invented abroad. • Its use should make production more efficient. However, no new knowledge is passed on directly to domestic inventors. Thus, the productivity of domestic R&D does not increase. If the product is purchased for less than the opportunity cost of producing the product, including R&D costs, a spillover occurs. • Then we have a passive spillover. Griliches calls this a pecuniary externality. • For passive spillovers (embodied technological change), a purchase must occur. Passive spillovers are transferred via: – International trade – Foreign direct investment (FDI)
    • Economic transaction Country j Intermediate inputs Rent Inputs Spillovers Country i Capital Investment goods Output Patent flows R&D Knowledge Knowledge Knowledge Technological proximity spillovers Four different channels through which economic transactions can result in spillovers: input-related spillovers (P); investment-related spillovers (P); Knowledge spillovers; and patent-related spillovers (A)
    • Some empirical evidence • Role of trade • Coe & Helpman (1995) estimate the effect of both domestic and foreign R&D on TFP growth for 22 countries. – Domestic R&D more important for larger countries, foreign R&D more important for smaller countries in the sample. – Competition may play a role here. There are two competing possibilities. Foreign knowledge: • Increases productivity through spillovers. • Competes with domestic products. – For a technology leader, competition is likely to be more important. – Domestic R&D elasticity. 0.08 for smaller countries, 0.23 for G- 7 countries. – Foreign R&D elasticity: 0.12 for smaller countries, 0.06 for G-7 countries. – Blyde (2001) finds OECD imports have a stronger effect on TFP than Latin American imports. Explanation: more R&D is embodied in OECD imports.
    • Alternative trade weighting methodologies Coe and Helpman (1995) measure the foreign R&D capital stock for a country as the sum of every other country’s R&D capital stock weighted by bilateral import shares. The underlying assumption is that trade is a mechanism which disseminates knowledge between countries. m ijt = •K F K “[variable]_tdquot; (1) it jt m it j≠i F where K is the foreign knowledge stock of country i at period t, mijt is the imports of it m ijt country i from country j, mit is country i’s total imports, is the import share of m t country j in country i’s imports, and Kjt is the knowledge stock of country j. The weights sum to one. Their preferred specification scales equation (1) by country i’s import intensity on the assumption that a country’s level of imports relative to its GDP affects the benefits a country receives from foreign R&D: m m it ijt = •K F K “[variable]_tch” (2) it jt y it m j≠i it m it where is country i’s total imports over its GDP ratio. y it Lichtenberg and van Pottelsberghe (1998) identify two problems with equations (1) and (2): the specifications are not invariant to the level of data aggregation — combining any • two country’s stocks in (1) would always increase the stock of foreign R&D; and m it the addition of the term , combined with transforming the level of R&D stocks • y it into indexes and taking logs, results in equation (F2) being misspecified. They propose the alternative formulation: m ijt = •K F K “[variable]_ti” (3) it jt y j jt where yjt is country j’s GDP. In this formulation, the stock of R&D that country i receives from country j is country j’s R&D stock times the fraction of country j’s output that is exported to country i. Source: Coe and Helpman (1995); Lichtenberg and van Pottelsberghe (1998).
    • Firm-level Technology Transfer • Firms that wish to serve foreign markets must choose between producing the good at home and exporting it or setting up production abroad. If they set up production abroad, they can use FDI through a subsidiary, or they can license their technology to another firm. • Two key choices: – Export vs. Production Abroad [Empirical evidence suggests these are complementary processes. Likely because intermediate goods must be exported to subsidiaries] – FDI vs. Licensing
    • Empirical findings – FDI does not necessarily lead to strong positive spillovers. • Firms choose to operate through a fully-owned subsidiary (FDI) rather than through joint ventures or technology licensing because FDI helps keep the private returns of technology internal to the firm. • Overseas R&D activity is still the least globalized of MNEs’ activities, largely concentrated in developed countries – 91% of overseas R&D activity of US MNEs is located in other developed countries. The rest is concentrated in a handful of relatively advanced developing countries
    • Spillover channels of FDI Driver Source of the productivity gain Example or Description Imitation or Adoption of new production methods. The local firms learn by imitating an demonstration MNE. Adoption of new management effect practices. Competition Reduction in X-inefficiency. The local firms have to improve their performance because of competition effect Faster adoption of new technology. from more efficient/productive MNEs. Training Increased productivity of MNE trained workers move to local (Human capital) complementary labour. firms or set up their own firms. Tacit knowledge. Exports Scale economies. MNE may generate export information externalities and may Exposure to technological frontier. provide a demonstration effect to local firms. Sources: Blomstrom and Kokko (1998); Görg and Greenaway (2001); Görg and Strobl (2002); Görg and Greenaway (2003).
    • What makes technology transfer successful? • Newly acquired technologies are rarely used at their peak productivity. – Both the initial productivity and the time it takes to master the technology depend on the starting level of ability. – Three key factors: • Labor cannot be effectively trained without experience with the activity. • Technologies usually need to be integrated into larger systems. Such integration takes experience. • Technologies are sensitive to local circumstances. • An important factor is the ability of the country to absorb new knowledge. This can be affected by: – Human capital [This may explain why rich countries benefit more from FDI.] – R&D [R&D may be necessary to adapt technology to local conditions.] • Geography matters – Features that affect technology transfer include: • Physical constraints, such as soil & climate conditions • Economic constraints • Social factors such as legal systems, transaction costs, etc.
    • Readimgs… Read Hall, Cort, Saggi Branstetter replaces Keller Forbes [on Moday] Rogers (general reading)