The Effects of Green Energy Policies on Innovation
by
Ryan Prescott
BSc, University of Victoria, 2006
A thesis submitted in partial fulfillment of the requirements for the degree of
Master of Science
in
The Faculty of Graduate Studies (Agricultural Economics)
THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)
April 2009
© Ryan Prescott, 2009
ABSTRACT
Over the last twenty years, climate change and energy stability have increasingly played a role in government policy. Innovation is a key ingredient to increased growth while achieving climatic goals. The policy paths taken by different countries have been extremely diverse. What effects do these policies have on innovation and what policies are best at encouraging the market? This thesis uses a theoretical model to examine the effect that different policies have on innovation and empirically tests a series of hypotheses obtained from the model. Both the theoretical and empirical results support the hypothesis of a “home bias”, whereby innovation in domestic markets is impacted more than innovation in world markets when a domestic policy is initiated. A second result is that mandatory renewable energy minimum levels for power companies have Ғ◌݆
relatively strong impacts on installed wind capacity. In general, a country’s wind capacity is determined by both economic and non-economic considerations.
ii
TABLE OF CO TE TS
Abstract ............................................................................................................................... ii Table of Contents............................................................................................................... iii List of Tables ..................................................................................................................... iv List of Figures ..................................................................................................................... v 1
Introduction................................................................................................................. 1 1.1 Background ..................................................................................................... 1 1.2 Problem Statement .......................................................................................... 3 1.3 Research Objectives........................................................................................ 5 1.4 Thesis Description .......................................................................................... 6 2 An Overview of Green Energy Markets ..................................................................... 7 2.1 Technology Background ................................................................................. 7 2.2 Policy Background........................................................................................ 12 2.3 Literature Review of R&D Policy ................................................................ 17 3 Theoretical Model..................................................................................................... 23 3.1 Introduction................................................................................................... 23 3.2 Stage Two Cournot Competition˖.................................................................. 26 3.3 Stage One R&D Race ................................................................................... 33 4 Empirical................................................................................................................... 41 4.1 Empirical Model ........................................................................................... 41 4.2 Patent Data .................................................................................................... 47 4.3 Policy Variables ............................................................................................ 50 4.4 Other Variables ............................................................................................. 51 5 Results....................................................................................................................... 53 5.1 Patenting – Regression One .......................................................................... 53 5.2 Market Size – Regression Two ..................................................................... 56 6 Brief Country Review ............................................................................................... 60 6.1 Germany........................................................................................................ 60 6.2 France............................................................................................................ 61 6.3 Denmark........................................................................................................ 62 7 Conclusion ................................................................................................................ 63 References......................................................................................................................... 65 Appendix........................................................................................................................... 68
iii
LIST OF TABLES Table 4.1: IPC codes for Wind ......................................................................................... 49 Table 5.1: Results of Patent Regression ........................................................................... 54 Table 5.2: Fixed Effects Market Size Results................................................................... 57 Table 5.3: Random Effects Market Size Results .............................................................. 58 Table 8.1: Change in Research as Feed-in-Tariff Changes .............................................. 68 Table 8.2: Data.................................................................................................................. 68 Table 8.3: Descriptive Statistics ....................................................................................... 78
뿰ۓ
iv
LIST OF FIGURES Figure 4.1: Patents from 1990-2006 for Germany, Denmark, Great Britain and Austria in Wind Energy ..................................................................................................................... 50 Figure 4.2: Electricity Output (GWh) for Germany, Denmark, Great Britain, Austria and OECD................................................................................................................................ 52
뿰ۓ
v
1
I TRODUCTIO 1.1 Background With ongoing worldwide concern about the effects of climate change and large
fluctuations in oil prices, governments are increasingly looking for policies and incentives to help curb carbon emissions and allow continued economic growth. Europe is leading the way with a mandated decrease in greenhouse gas levels by 20% compared to 1990 levels and a mandated 20% increase in renewable energy by 2020 (BBC News, 2009).
As developing countries such as China and India continue to grow, world energy demand is expected to increase substantially. Renewable energy will undoubtedly play a key role in meeting both the world’s climatic goals and allowing increased energy price 뿰ۓ
stability. Energy price instability is due to increased worldwide demand, political instability in large oil producing countries, and decreases in easily accessible oil stocks. There has also been speculation that current oil stocks may have peaked already and that worldwide production may decline in the near future (Bentley, 2002). Currently most renewable energy sources (excluding large scale hydro) are in their infancy of development and are not economically feasible without government incentives. Therefore if governments want to meet their future growth and climate objectives without large subsidies, innovation will have to play a key role.
The effects of government policies on the growth of renewable energy sources can be seen in annual growth rates of renewable energy sources of countries such as Denmark, Spain and Germany (OECD/IEA, 2004). Denmark was one of the first to
1
implement a carbon tax in the early 80’s, while Germany and Spain introduced several policies including feed-in tariffs in the early 90’s. Knowing how different types of government policies affect incentives for innovation by different sources of renewable energy will better allow government officials to design policies which leverage their individual country’s competitive advantage at a comparatively low cost.
There are several fundamental economic reasons why policies should be introduced to encourage research and development, and innovation. Economists generally agree that research and development (R&D) should be encouraged due to the spillover effects associated with R&D and innovation. The spillovers are a result of the innovating firm not being able to capture all the potential benefits of the innovation because of imperfect property rights, which results in a social valuation greater than the private valuation. This ˖
can be the result of such things as the innovation not being patentable or the inability of the innovating firm to protect its innovation from copying. However, if through a patent or some other form of intellectual property rights an innovator is protected these spillovers would be eliminated. This would remove the need for a government policy intervention.
It is generally agreed that private industry on its own does not supply a sufficient amount of R&D because private valuation of a product or innovation is less than the social valuation of the product or innovation because innovators cannot capture all of the gains from the innovation due to imperfect property rights. When looking at the issue of green technology, the pollution externality also plays a significant role in social benefits calculations. From a social benefit perspective there is an added external benefit in
2
encouraging the growth and innovation in green technology in the form of less pollution. Therefore it is in a government’s interest to examine ways to encourage innovation and R&D in renewable energy to both address spillover effects and reduce the social cost of pollution.
A variety of policies have been used throughout the world to try to raise the level of R&D and growth in alternative energy to achieve a more socially optimal outcome. The most basic policy is the provision of an R&D subsidy to increase R&D to a more socially optimal level. Other policies, which are more directly related to the environmental externality, are designed to increase reliance on renewable energy or to decrease reliance on fossil fuel based energy. Permits and taxes have been used in the area of energy supply in much the same manner as any other form of environmental regulation. Price
guarantees or price tariffs have been used to raise the price paid for renewable energy above the price paid for traditional energy, to allow for continued growth and innovation in renewable energy.
1.2
Problem Statement
The market structure for renewable (“green”) energy is unique in many ways. Green energy firms compete in a race with each other to develop new green technologies. The winner of the race then uses their newly developed technology to compete with traditional forms of energy in the energy markets, often relying on government policies to make them more cost competitive. Given the broad array of innovation and green energy policies which currently exist in global energy markets, it is imperative that governments
3
understand the comparative impacts of their policies on domestic versus foreign firms and understand which policies are the most effective in absolute terms. Governments may want to implement policies that encourage domestic innovation more than international innovation in order to grow their domestic firms and give them a comparative advantage in the future. The reduction of the pollution externality and the encouragement of domestic firms may in practice have conflicting results. Domestic firms may not be the most efficient at producing green energy resulting in less pollution abated than if an international firm was supplying the green energy.
Underinvestment in private R&D and innovation due to spillovers is widely acknowledged; however, for green energy, additional environmental benefits are associated with innovations which result in less polluting forms of energy production.
These social benefits, which extend beyond private benefits, are the main reasons governments have been increasingly designing policies which encourage innovation in the area of green energy. With increased globalization, governments also need to look at whether it is domestic green energy firms or foreign green energy firms that are being encouraged by their regulations. This could have a direct impact on what type of policy should be implemented, depending on the goal of the regulation.
Due to the imperfect competition nature of green energy and the domestic and international components there are few questions which should be asked: How are the R&D incentives of domestic and international firms impacted differently by the introduction of a particular R&D policy? Which policies have been the most effective in stimulating the growth in green energy? Knowing the answers to these questions will
4
better enable governments to implement policies which encourage domestic innovation and growth in green energy.
1.3
Research Objectives
The aim of this thesis is to examine the comparative impact of different policies on the innovation incentives facing domestic and international firms. The specific research objectives are to:
• •
review current literature on innovation and the environment develop a theoretical model of a R&D race between innovation firms, the winner of which goes on to compete in a non-competitive energy market • use the theoretical model to generate a set of testable hypotheses • use empirical analysis to test the various hypotheses ˖ to continued growth and addressing Innovation in renewable energy is the key
climatic objectives. Unfortunately, innovation is inherently difficult to measure. Most research on innovation resort to use patent counts as a measure of innovation activity. Patents allow innovators temporary monopoly power and thus there is a major incentive to patent any new innovation. For innovations to be patentable they must be novel, involve inventive steps and be capable of industrial application (Dernis & Guellec, 2001).
Patent counts from European countries will be used in the empirical analysis. Specifically, results from patent count empirical analysis will be used to assess the absolute effectiveness of policies which are designed to encourage growth of renewable energy sources. Patent count data will also be analyzed to examine the comparative impact of an R&D policy on innovation activity in domestic and foreign countries. Later
5
in this thesis I conduct a case study on wind power, the most advanced of the developing renewable green energy sources.
1.4
Thesis Description
The thesis will be divided into seven different chapters. Following this introductory chapter, Chapter 2 provides an overview of policy initiatives and technology changes in the green energy markets, and also reviews the relevant literature. The development of the theoretical model is contained in Chapter 3. Chapter 4 presents the data and empirical model. The empirical results are presented in Chapter 5. A brief discussion of specific green energy policies countries is contained in Chapter 6. Chapter 7, the final Chapter, concludes the thesis.
6
2
A OVERVIEW OF GREE E ERGY MARKETS This overview chapter will be broken up into four sections. Section 2.1 will focus
on the current technology of the four most developed renewable energy sources: wind power, ocean power, geothermal power, and biomass. Section 2.2 will look at the various policies that have been used throughout the world to encourage growth and innovation in alternative energy sources. The final section, 2.3, will be a literature review relating to R&D policy and the economic rational proposed to introducing government policy to encourage growth and innovation in alternative energy.
2.1
Technology Background
Renewable energy sources can be grouped into three main categories based on 썈ۓ
maturity levels: first generation energy sources, which emerged from the industrial evolution, second generation energy sources, which are now becoming commercially viable, and third generation sources, which are those sources still under development (International Energy Agency, 2006). Excluding hydropower, which is widely used throughout the world, the remaining renewable energy sources fall into the following categories: biomass and geothermal (first generation); wind, solar hot water, solar photovoltaics and advanced bioenergy (second generation); and third generation, which consists of ocean energy, advanced geothermal, biorefinery technologies and advanced biomass. These technologies, which are still under development, have the potential to be important sources of energy in the future.
7
From the beginning, and currently in many developing countries, combustion of wood or biowaste combustion was used for trivial tasks such as cooking food and heating living quarters. However advances in technology have produced an array of new options for producing bioenergy. Bioenergy includes biogas production, combustion, gasification and proylsis.
Biogas: The major source of biogas is through anaerobic digestion which involves the breaking down of biodegradable material using microorganisms in the absence of oxygen to produce methane and carbon dioxide rich biogas. The methane is then used in the same manner as natural gas. The most common source of fuel for anaerobic digestion is agricultural products and waste such as landfill waste or sewage. This technology is often used for the treatment of waste water with high organic content.
Biomass combustion: The combustion of biomass is used in several different ways to produce bioenergy. The combustion of biomass such as wood or wood pellets has been used for heating of residences and with automation and standardization has lead to increased efficiency. The combustion of biomass has also been used in combined heat and power (CHP) plants throughout Scandinavia for district heating. Biomass, co-fired with coal, has also been used in large scale combustion plants for the production of electricity.
Biomass fermentation: Fermentation involves the conversion of sugars from such sources as corn, sugarcane or sugar beet into ethanol and another byproduct such as methane using microorganisms. The fermentation of anything that contains sugar can
8
result in ethanol; however, sources like sugar cane, sugar beets, molasses and fruits are most efficient because of their ability to produce ethanol directly from the sugars. Sources such as grains or cellulose using wood waste need an intermediate step to convert the primary source to simple sugars.
Prylosis: Prylosis is process in which biomass is converted to liquid (bio-oil), gaseous and solid (charcoal) through the super heating of the biomass in the absence of oxygen. There however has been little market implementation of this technique at the present time.
Geothermal: Geothermal energy comes from the harnessing of super heated underground aquifers or reservoirs. These reservoirs are heated via an outward transfer of
heat from the earth’s core. Rainwater is usually used as the medium to transfer heat to the surface. When the rock characteristics are appropriate (i.e., permeable) the water will pool forming underground reservoirs of high pressure and temperature, essentially geothermal fields. These fields can be grouped into two types: either water-dominated or vapour-dominated. In the simplest case, steam from the geothermal well is passed through a steam turbine to produce electricity. If the well produces hot water instead of steam, another working fluid is used with a low-boiling point. Water is used instead to vaporize the working fluid which is then passed through a turbine, condensed in water and reused.
Solar: Solar energy can be broken up into two streams, photovoltaics (PV) and solar thermal power plants. Photovoltaics involves the use of sunlight to split positive and
9
negative charge carriers which produces an electrical voltage. The DC voltage is then converted to AC and distributed to the grid. The use of PV has increased substantially with installed capacity increasing from 110 MW in 1992 to 1809 MW in 2003 (International Energy Agency, 2006). The second stream of solar power is the solar thermal stream which involves the concentration of the sun’s energy to produce either a super heated liquid or gas. Small scale solar thermal is used to heat water and heat an individual residence. On the large scale there have been several proposed designs for large scale solar thermal plants such as the parabolic trough and tower plants. However the main components are all the same. There is a concentrator to concentrate the solar radiation. That concentrated solar energy is then used to heat a medium like water, resulting in steam. That steam is then used to produce energy like any other thermal plant. A major advantage of the solar thermal is that in times of bad weather a fossil fuel
burner can drive the water steam cycle guaranteeing continuous supply of energy.
Wind: Wind is by far one of the most established renewable energy sources. The design of wind turning a horizontal or vertical turbine has remained constant for decades; however, since the 1980’s major advances in aerodynamics, structural design and resource assessment have resulted in 5% annual increases in energy yield of turbines (Herbert, Iniyan, Sreevalan, & Rajapandian, 2007). This has resulted in wind turbines increasing in size from less than 100kW in 1980’s to in excess of 3.5 MW presently.
Ocean: Ocean energy can be broken down into three areas: tidal energy, wave energy, and ocean thermal energy. Tidal, wave and ocean geothermal energy are all still
10
in their infancy and in the research, design and development stage. Therefore there is no leader at the moment.
Many companies have come up with very different ideas to harness the power of the tides. The most basic is a turbine in the water, whereby the tide rotates the turbine producing power. Currently the main source of tidal power is the result of barrage tidal power, where a barrage or dam is placed across an estuary. The barrage has openings for tidal water to flow into on its way to the estuary; the water turns a turbine resulting in power.
Wave power also has several competing technologies with no clear winner depending on whether it is offshore, near shore or shoreline waves. Presently onshore
devices include the oscillating water column which consists of a partially submerged structure where air is trapped above the water free surface. Waves force water in from below forcing the trapped air outwards spinning a turbine. Another onshore device uses a convergent channel which water rushes into it until it overflows into a reservoir. That reservoir is then used as a conventional low head dam. Offshore devices include the Palemis, a floating cylindrical snake like device. Its hydraulic joints extract energy with the movement of the waves.
Ocean thermal uses warm water from the surface to produce steam in a working medium with a low boiling point to produce steam and run a generator with cold water from the depths to condense the working medium. This is, however, extremely inefficient and there are presently only demonstrations plants producing electricity of this nature.
11
With so many competing technologies and with ocean technology still in its infancy, it is likely that the best technology has yet to be discovered.
2.2
Policy Background
Countries throughout the world have taken vastly different approaches to encouraging renewable energy growth and innovation. Europe has a long history of encouraging alternative energy through not only practical demonstrations, but also policy intervention. This started in the early 1970’s and before, coinciding with the fear over fossil fuel supplies (OECD/IEA, 2004). In the 1970’s the main policy interventions throughout Europe, Japan and United States involved the funding of research and development and implementation of demonstration plants. ˖
The pace of policy interventions throughout Europe and the United States continued slowly throughout the late 70’s and early 80’s as prices of oil receded following the oil shocks of 1973-1974. This changed in the late 80’s at which time the European commission, which previously had minimal effect on establishing a place in European energy policy making, started to establish a position (McGowan, 2000).
Many different policies have been used over the last thirty years to encourage the growth and subsequently the innovation in renewable energy sources. From country to country, the type and target of policies vary substantially depending on their needs. For example, Denmark has targeted growth in wind energy with capital grants incentives for wind turbines implemented in 1979 followed by guaranteed prices in 1981 (OECD/IEA,
12
2004). Canada did little to encourage any form of growth until the early 1990’s most likely due in part to substantial cheap hydro electricity.
The theoretical and empirical literature on the effects on these policies can be taken from not only environmental theory, but also trade theory. The environmental literature looks at the effects of policies such as taxes and pollution permits on the pollution. Trade theory looks at the effect of protectionist policies to protect a domestic industry. This is similar to renewable energy policy when a government is trying to protect the renewable industry and increase its output while competing with the traditional energy source. The most recent literature is from the biofuels debate which uses both branches of literature. Some approaches used by countries to encourage green energy are producer grants, individual or company grants, obligations, carbon taxes, green certificates, targets and
feed-in-tariffs.
Tax incentives such a carbon taxes, charge carbon polluters in an attempt to internalize the effect of the externality resulting in less pollution. The optimal level of pollution can be achieved if the marginal damage of the pollution is known. Baumal and Oates (1971) discuss the difficulties of implementing a tax system to properly capture the damage of the pollution externality. Firms decrease the level of pollution until the marginal cost of abatement of another unit is equal to the marginal benefit. This does not directly subsidize non polluting firms but rather increases the cost of polluting firms resulting in non polluting firms being more cost competitive. The carbon tax increases the price of electricity resulting in a user pay system whereby the more you consume the more you pay in taxes. Empirical literature on pollution taxation is widespread. For
13
example Dahl (1979) looked at the effects of a gasoline taxation and finding the elasticity of demand to be -0.442.
Producer grants are given to producers to produce renewable energy. These grants are subsidies to increase the amount of renewable energy used. The subsidy decreases the cost of producing renewable green energy resulting in an increased share of the market. It does not, however, give the traditional sources of energy incentive to abate emissions. The resulting market price of electricity decreases because of the subsidy. This kind of subsidy results in everyone paying an equal portion of the subsidy. There is also a subsequent decrease in price of electricity.
Obligations require that a certain amount of renewable energy must be used in a Ҳ
country’s energy supply. This is usually done through a bid system whereby firms bid to supply some much renewable energy at a given price per kilowatt hour. The lowest cost bids are then selected until the amount equal to the obligation is filled. Each bidder receives a price equal to their bid price of supplying energy. This implies that the subsidy to each generator may be different as the price paid to each generator should be equal to its marginal cost (Menanteau, Finon, & Lamy, 2003). The additional price is passed along to consumers in the form of higher electricity bills implying a user pay system of financing. Obligations are equivalent to a subsidy on renewable output and a tax on traditional energy output (Fischer & Newell, 2004).
Recent research in the area of obligations has focused on ethanol mandates. Joust and de Gorter (2009) look at the impact of mandates combined with other incentives such
14
as biofuel tax credits and find they may work the same as consumption subsidies with no effect on ethanol consumption. Different models have been used to look at the effects of the bio-fuel mandate on reaching national pollution targets and the overall effect on the economy (Dixon & Osborne, 2007).
Another example of obligations is domestic content protection. In this case there is a requirement to use a certain amount of domestic content in the production of a good. If the domestic good and the international good are perfect substitutes similar market reactions that would occur as a result of the introduction of renewable obligations. Krishna and Itoh (1988) develop a theoretical model to look at content protection in oligopolistic markets and find that domestic firms benefit from content protection if the two inputs are perfect substitutes. Other literature looks at other market structures such as ᓠҳ
perfect competition and monopolies and the effects of goods which are compliments or not perfect substitutes. Empirical literature from Beghin and Lovell (1993) looked at the effect of domestic content protection on the Australian cigarette and tobacco industry, finding that increased domestic content protection led to increased use of domestic Australian tobacco.
Feed-in-tariffs or price guarantees, guarantee a price for renewable energy above the price of traditional energy sources. Electricity is generated from the renewable energy source until its marginal cost equals the price given by the feed-in-tariff (Menanteau et al., 2003). Since the feed-in-tariff is greater than renewable energy suppliers would receive under normal conditions there is more renewable output. The benefit to renewable firms varies with their cost of production. Firms with low cost of production
15
collect large economic rents. The increased price given to the renewable energy source is funded through an increase in the price of electricity to consumers resulting in a user pay system.
Consumer/individual grants are grants to individuals or companies to install renewable energy sources in their homes or businesses. The result is an increase in personal sources of renewable energy which might result in a small increase in renewable energy.
Targets are normally unbinding renewable energy targets set by government and depend on other policies to reach to the targeted goal. If targets were binding they would work in the same manner as obligations. 奠ҳ
Green certificates function by allotting a certain amount of renewable energy that must be produced by each producer. This producer can either produce that output themselves or purchase it from another producer or from a specialized renewable energy firm in the form of green certificates (Menanteau et al., 2003). The renewable energy firm receives both the price received for the electricity and the price of the green certificate. The increased price is passed along to consumers in the form of higher electricity prices.
When all necessary information is available, price approaches such as taxes and quantity approaches such as obligations or green certificates can have similar effects on reducing emissions and increasing renewable energy output. The key difference is that
16
that with a tax or feed-in-tariff the exact amount of pollution abated or new renewable energy supplied won’t be known until ex post, while in the case of obligations or green certificates it will be exactly known how much renewable energy will be added or pollution abated.
2.3
Literature Review of R&D Policy
The literature on innovation relates to two main areas of renewable energy. The broadest area involves public good aspects of research and development (R&D) and the spillovers resulting from R&D. Economists have long argued for R&D tax subsidies or tax credits in order to achieve a more socially optimal level of R&D. The second area is the literature on innovation and the environment, which falls under the broader umbrella of induced innovation. Policies are implemented leading to a change in relative prices. Ҵ
This change in relative prices leads to the environmentally friendly technology becoming comparatively cheaper, resulting in innovation.
Increased research and innovation in alternative energy sources will lead to increased growth in alternative energy sources resulting in decreased use of traditional polluting sources of energy. This increased growth may come as a result of the creative destruction of innovations, a notion pioneered by Schumpeter (1942). The idea is that new innovations come along and make old technology obsolete. In the case of renewable energy innovation this could be a new way harnessing wind energy or turbine that produces more energy. Either will result in increased growth in output of renewable energy.
17
The idea of spillovers associated with R&D has been pointed out by many economists starting with Arrow (1962) and Nelson (1959). Nelson noted that private firms have little incentive to undertake basic research because some basic research is not patentable and because of the narrow technology base and therefore limited applications of technology. In many cases the significance of the basic research will not be realized for some time. With innovation being the result of R&D, policies aimed at increasing R&D will eventually increase innovation.
Theoretical research into R&D spillovers starts with Spencer (1984), who proposed a model with varying amounts of spillover, ranging from no spillovers in the case where patents are 100 percent effective to complete spillovers, where all firms share equally in the benefits flowing from the innovation. He finds that when R&D is costly and can lead 䠰Ҵ
to significant cost reduction, social welfare can be raised by providing firms with R&D subsidies. If firms know spillovers will occur there is less incentive for them to invest in R&D and as the number of firms increases the spillovers will increase, thereby decreasing an individual firm’s incentive to innovate. Spencer suggests that the most direct way to deal with low incentives to innovate is to subsidize research.
Grilches (1992) reviews empirical evidence showing that private rates of return on R&D have been significantly less than the social rate of return. The studies reviewed by Grilches use essentially two techniques: the simple measurement of the improvement of an innovation, such as the yield of farm product; or the regression based estimates of spillover benefits to outside sources from R&D expenditures. Much of the recent work on spillovers has used patent citations. Patent data facilitates an examination of cross
18
citations to look at how patents in one sector are cited by patents in another sector. These results have lead many countries to implement market based solutions such as subsidizing R&D or giving tax credits designed to increase R&D to more socially efficient levels.
A large body of empirical evidence documents the effects of R&D policy on R&D expenditures. Hall and Van Reenen (2000) review previous studies of this market based approach. They review 12 US studies and 10 non-US studies, published between 1983 and 1998, looking at the effects of different tax treatments on R&D spending. A variety of methods from survey to econometric methods were used to determine the price elasticity of R&D and a cost-benefit ratio. The econometric methods were usually based on one of two methods. The first method is the quasi experimental approach regressing R&D investment on explanatory variables and a policy dummy, allowing one to examine ۗ
how R&D changes as policies change. The second is very similar to the first except instead of a policy dummy there exists a user cost of R&D. For survey results, the researchers ask the senior managers of leading firms how things have changed. The estimated elasticities range from insignificant or negative to close to 2, with the majority being less than 1.
It is widely agreed that induced innovation was first described by Hicks (1966):
“The real reason for the predominance more of labor saving inventions is surely that which was hinted at in our discussion of substitution. A change in the relative prices of the factors of production is itself a spur to invention, and to invention of a particular kind - directed to economising the use of factor which has become relatively expensive.”
19
The extension of induced innovation to environmental policy is only a recent revival of what Hicks first pointed out. If the environment is included as an input to producing a good and that input changes value, there is incentive to innovate. Environmental policy stimulates innovation by creating these changes in input prices and as a result several empirical papers look at the effects of environmental policy on innovation.
Induced innovation as a result of energy price changes has been looked at by several authors. Newell, Jaffe and Stavins (1999) look at the effect energy prices have on room air conditioners, central air conditioners and gas water heaters.
Popp (2002) uses patents from the US patent and trademark office from 1970 to 1994 to look at the effect of energy prices on energy efficiency innovations. He limits his ۗ
research to American inventors. He regresses the log of the ratio of the number of patents in a particular field and year over the total number patents filed that year against energy prices, stock of knowledge, federal energy R&D and other explanatory variables. Instrumental variables are used for federal energy R&D because of its correlation to energy prices. A knowledge variable is created using a second regression looking at the number of citations a patent receives. The results show the expected positive effect that energy prices have on innovation.
Popp (2006) used patent data on patents filed by the US, Germany and Japan in their home countries to look at innovation spillovers. He looks at whether latecomers of NOx and SO2 regulation adopt innovation of early regulators or continue to innovate on their own. Although most of the scrubbers used to reduce NOx and SO2 are produced
20
domestically for the three countries, the author goes a step further, looking at whether the innovations from the country that first regulate the emissions influences the innovations from the countries which regulate at a later date. This is done using citations counts, specifically by looking at whether innovations resulting from early regulation get more citations. He finds that foreign innovations are not being directly adopted and therefore domestic R&D will continue to play a substantial role in innovation.
The work most closely related to the model developed in this thesis is Johnston et. al (2008) who were among the first to look at policy changes over countries in a panel data setting. They looked at the effect of policy changes within most industrialized nations on innovation using country patent counts. Using a fixed effect binomial maximum likelihood estimate to regressing patents per country per time period against ۗƘ
explanatory variables such as public R&D expenditure, electricity price, electricity consumption and policy dummies, they look at the effect policy changes have on the number of patents. Their results vary significantly from policy to policy and from renewable energy source to renewable energy, with obligations and tradeable permits having the greatest positive effect on patenting.
In a race situation like a patent race where multiple firms compete to discover the same or very similar ideas, there may actually be too much R&D and firms may duplicate R&D especially if there is a limited amount of spillover. Loury (1979) and Lee and Wilde (1980), for example, create a theoretical model that produces several interesting results, one of which says that in the short-run with a fixed market structure each firm may be researching more than the socially optimal due to duplication.
21
Dasgupta and Stiglitz (1981) developed a theoretical model that included another important factor, the size of the innovation. They use their model to examine oligopoly and monopoly market structures and the frequency and size of market innovations. In the oligopoly case, they determine that there may be more innovation than is socially optimal. In the monopoly case there may be not enough innovation and the innovation that does occur is larger than that which is socially optimal.
Nordhaus (1972) looked at the optimal length of patents. He suggests that different patent lengths should be designed for different types of patents in order to maximize social welfare. He suggested that society may want to error on the side of a longer patent length rather than a patent length that is too short.
ۗ
Along with subsidies, other approaches such as public R&D, prize competitions or public private partnerships have been examined for more socially optimal results. But determining how these policies address inefficiently low levels of research is complicated, with no clear cut best response.
22
3
THEORETICAL MODEL This theoretical chapter section will be divided into four sections. Section 3.1 will
describe the basic model including variables, structure and timing of the relevant events. Section 3.2 will focus on solving the second stage of the model, which involves a Cournot-Nash game between the renewable energy firm and the traditional energy firm. Section 3.3 will examine stage one of the game, where two renewable energy firms from different countries compete for a patent which allows them to be the sole provider of renewable energy in stage two. The final section 3.4 presents comparative statics results on how different government policies or changes in demand for energy will increase or decrease a firm’s propensity to patent.
3.1
Introduction
ۗ
The main way to protect a new innovation is through the filing of a patent. A patent allows filers protection against other individuals or companies using their idea. For simplicity we will assume that patents give permanent and total protection to the filer of the patent. In reality this is typically not true as patents can quite often be innovated around.
Different technologies typically have different marginal costs (MC) of producing a MW/h of energy. Innovation lowers a firm’s MC and normally dominates the old technology. A pair of competing innovating firms are assumed to take part in a two stage game. In stage one, the two firms from two different countries (countries m and n) compete in an R&D race knowing the type of intervention that the governments of m and
23
n have chosen. In the final stage, the winner of the patent race, which has the lowest marginal cost amongst all renewable energy sources and therefore completely dominates the market for renewable energy, takes its new technology and competes in a Cournot competition with a traditional supplier of electricity (coal, natural gas, hydro).
The winner has two markets which it has an opportunity to service: a domestic market and an international market. International markets and domestic markets have different levels of accessibility. Accessibility is the ease at which a firm can enter a particular market. It is reasonable to assume that the cost of entering an international market exceeds the cost of entering a domestic market. For this analysis, accessibility is exogenously determined and is a result of government policies and regulations. The traditional (fossil fuel) supplier is expected to have a lower cost than the renewable ۗŏ
energy supplier and therefore will supply most of the market, much like what is currently observed.
The renewable energy firm supplies q ri MW/h of electricity to the market in country i while the traditional energy firm supplies q ti MW/h of electricity to the market in country i, where i is either country m or n. Since a MW/h of electricity from a renewable source is equivalent to a MW/h of traditional source in the minds of consumers, the two goods are homogenous. We will assume that both traditional and renewable sources of energy have constant marginal cost curves cr and ct; however the cost of renewable energy will be higher in the international market because of the added accessibility cost δ .
24
The demand for MW/h of power is determined by the utility companies. For simplicity we will assume a linear demand curve, where P(Q) is the price per MW/h of electricity:
(3.1)
P i (Q ) = α i − β i Q
Q = qti + q ri
i = m, n
Assume that a firm’s ability to service one market does not affect its ability to service another market. There are several policy interventions that government can use, one of which is guaranteed prices or feed-in-tariffs which usually involve the government guaranteeing a price above the current market value. Therefore renewable energy sources become more cost competitive.
To establish a relationship between theۗprices of renewable and traditional energy sources, assume that:
(3.2)
Pri (Q ) = sPt i (Q ) If s>1 then the government is using feed-in-tariffs or price guarantees and the price
received by the renewable energy source is greater than the price received by the traditional energy source. If s=1 then there is no government intervention. Alternatively s = (1 + j ) where j is the percentage increase in price of renewable energy sources over traditional energy sources.
25
3.2
Stage Two Cournot Competition
There are two subgames that must be solved in stage two, depending on whether it is the domestic firm or the international firm which innovates. If a firm from country m wins the patent race, it will compete in its domestic market m and an international market n. Whereas if a firm from country n wins it is a mirror case with its domestic market being n and international market being m. Therefore only one of the cases will be solved.
Assumption: Firm in country m wins and competes with traditional energy source.
In this case the domestic firm competes in its domestic market and receives profits
π rm and it also competes in country n’s market and receives a profit π rn . Therefore the profit for the firm from country m should it innovate is: ۗ
(3.3)
π m = π rm + π rn Within country m the firms supplying the traditional and renewable sources of
energy solve the maximization problems:
(3.4a) max π rm (q rm , qtm ) m qr (3.4b) max π tm (q rm , qtm ) m qt where: (3.5a) π rm = Prm (Q )q rm − c r q rm = s m (α m − β m (q rm + qtm ))q rm − c r q rm (3.5b) π tm = Pt m (Q )q tm − ctm qtm = (α m − β m (q rm + qtm ))q tm − ctm q tm
26
Recall from equations (3.1) and (3.2) that P i (Q ) = α i − β i Q and Pri (Q ) = sPt i (Q ) . The domestic market and international market can be solved separately due to their constant marginal cost structure. If the renewable energy firm had increasing marginal costs then the maximization of profits from the domestic market and international market would have to be solved jointly.
In the international market, it is assumed that the marginal cost for the firm from country m is equal to c rn = c r + δ , where δ is an international accessibility parameter. If
δ = 0 there is no additional cost of supplying the international market relative to the domestic market. If δ > 0 there is an additional cost of doing business in the international market as a result of such things as increased transportation costs or non tariff barriers such as a complex regulatory environment. The renewable energy and traditional energy ۙڮ
firms solve the maximization problems:
(3.6a) max π rn (q rn , qtn ) n qr (3.6b) max π tn (q rn , qtn ) n qt where: (3.7a) π rn = Prn (Q )q rn − c r q rn = s n (α n − β n (q rn + qtn ))q rn − (c r + δ )q rn (3.7b) π tn = Pt n (Q )qtn − ctn qtn = (α n − β n (q rn + qtn ))qtn − ctn qtn
Equilibrium quantities: Solving the first order conditions for the various profit maximization problems, we get the reaction functions for the renewable energy firm and the traditional firm in the domestic market of country m:
27
(3.8a) q rm =
α m − β m qtm −
cr sm
2β m
α m − β m q rm − ctm (3.8b) q = 2β m m t
In firm m’s international market the reaction functions are:
(3.9a) q rn = (3.9b) q tn =
α n − β n qtn −
cr + δ sn
2β n
α n − β n q rn − ctn 2β n
The reaction function allows one to determine the optimal production response of MW/h of electricity supplied by one firm in ۙ response to the other firm’s supply. In solving the reaction functions we are looking for the Cournot-Nash equilibrium quantity on MW/h produced by each firm. The Cournot Nash equilibrium production by the renewable energy and traditional firms in the domestic market is given by:
(3.10a) q rm * =
α m + ctm −
2c r sm
3β m c α m + mr − 2ctm s (3.10b) q tm * = 3β m The corresponding quantities in the international market are:
28
(3.11a) q rn * =
α n + ctn −
2( c r + δ ) sn
3β n (c + δ ) α n + r n − 2ctn s (3.11b) q tn * = 3β n Substituting equations (3.10a) and (3.10b) into equation (3.5a) and equations (3.11a) and (3.11b) into equation (3.7a) we can determine the equilibrium level of profit from the renewable energy source in each market. In the domestic market, firm m earns:
m cr α + m + c tm s (3.12) π rm * = s 3
m 2c α + c tm − mr s m 3β
m 2c r m α c + − m m m m t s α + c r + s ct s ۙm − c r = 3 3β m
m 2c α + c tm − mr s −c m r 3β
m 2c + ctm − mr m α s = s βm 3
2
In the international market, firm m earns:
(3.13)
n cr + δ n 2(cr + δ n ) 2(cr + δ n ) n n n n α + n + ct α + ct − α + ct − n n n s s s πr * = s − (cr + δ ) n n 3 3β 3β n n 2( c r + δ n ) 2( c r + δ n ) n n α + ct − n n n s nα n + (c r + δ ) + s n ctn α + ct − s n s s − (c r + δ ) = = n n 3 3 β 3β
29
2
Therefore the combined equilibrium profit for the firm from country m is:
(3.14) π m* = π rm* + π rn* Policy impacts: The issue of interest is the differential impacts of a domestic versus an international feed-in tariff policy on the profits of firm m. The impact of the policy on profits is important because, as shown below, the size of the profits earned in stage two determines a firm’s incentives to innovate in stage one.
The following two results can be established.
An increase in domestic feed-in-tariff or price guarantee increases profits π m* . ۙӖ
2c m m α + ct − mr ∂π m * 1 s (3.15) = m m 3 ∂s β
2
m + 2s βm
2c m m α + ct − mr s 3
2c r 3( s m ) 2
> 0
An increase in international feed-in-tariff or price guarantee increases profits π m* .
2
n n 2(cr + δ n ) n n 2(cr + δ n ) α + ct − α + ct − n n n 2(cr + δ n ) ∂π m * 1 2s s s >0 (3.16) = m + n 3(s n ) 2 3 3 ∂s n β β As we can see from equations (3.10a) and (3.11a), an increase in feed-in tariff or price guarantee shifts the reaction curve of the renewable energy source outwards leading
30
to an increased market share for the renewable energy firm. This increased market share results in increased profit for the renewable firm.
Another issue is what happens to the price faced by the traditional energy provider and the renewable energy provider when a feed-in tariff or price guarantee is implemented. Using equation (3.1) the equilibrium price faced by traditional energy firms is:
Pt m* = α m − β m (q tm* + q rm* ) In the domestic market and
Pt n* = α n − β n (q tn* + q rn* )
˖
in the international market.
If there is an increase in the domestic feed-in tariff or a price guarantee there is a decrease in traditional energy prices Pt m* .
∂Pt m* ∂s
m
∂q m* ∂q m* cr 2c r = − β m tm + rm = − β m − + m m 2 m ∂s 3β ( s m ) 2 3β ( s ) ∂s
< 0
An increase in the international feed-in tariff or price guarantee results in a decrease in traditional energy prices Pt n* in the international market.
31
n* ∂Pt n* c +δ ∂qrn* 2( c + δ ) m ∂qt = − β m − mr m 2 + mr m 2 < 0 = − + β n n n ∂s ∂s 3β ( s ) 3β ( s ) ∂s
Therefore an increase in price guarantee or feed-in tariff decreases the price that traditional energy sources receive. This would likely be small because the share of the energy market that renewable energy sources hold is very small compared to traditional energy sources. The renewable energy firm has two forces affecting it, the decrease in the price of the traditional source which it is tied to and the increase in price as a result of the feed-in tariff or price guarantee. Using equation (3.2) the equilibrium price faced by the renewable energy firm is:
Prm* (Q ) = s m Pt m* (Q ) ۙ
In the domestic market and
Prn* (Q ) = s n Pt n* (Q )
in the international market.
If there in an increase in domestic feed-in tariff or price guarantee there is a an increase in the renewable energy price Prm* .
∂P ∂s
m* r m
∂Pt ∂s m
m*
= Pt m* + s
m cr α + m + ctm s = 3
− s m c r 3( s m ) 2
> 0
32
An increase in international feed-in tariff or price guarantee results in an increase in the renewable energy price Prn* in the international market.
∂Pt ∂P = Pt n* + s n t n n ∂s ∂s n*
n*
n cr + δ α + + ctn n s = 3
− s n − c r + δ > 0 3( s n ) 2
Note that an increase in the feed-in tariff decreases the price received by the traditional energy source but this is counteracted by the feed-in tariff, resulting in an increase in the price renewable energy.
3.3
Stage One R&D Race ۙ
For stage one the two renewable firms from two separate countries (m and n) compete to try and discover the new technology. This technology takes the form of a drastic innovation that obsoletes other innovations and allows the innovator to be the sole supplier of renewable energy to the energy markets. Assume Firm i’s probability of discovering the new technology P(xi) is determined only by xi ,, which is the amount of money spent on research. Specifically, assume that:
0 ≤ P(xi) ≤ 1 P′(xi) = 1/k > 0 P′′(xi) = 0
i=m,n
This assumption implies an increase in xi increases the probability of a successful discovery at a constant rate 1/k. Finally if both firms discover the new technology then
33
the winner will be determined by a coin flip. Consequently, the probability of Firm i winning is:
(3.17) Pr(firm i is the winner) = Pr(firm i discovers)Pr(firm j doesn’t discover) + 0.5 Pr(firm i discovers)Pr(firm j discovers) In the coin flip situation there is a negative effect associated with both firms innovating at the same time resulting in one firm’s research investment being completely wasted. In reality their effort may not be completely wasted as they may have the opportunity to file a patent in a smaller less lucrative market which would allow them to capture some of their lost research investment.
1−ϕ
Both firms have the same quadratic R&D cost function
2
i ( x ) 2 , where 0 ≤ φ ≤ 1 i i
is a government subsidy of research. If φi = 0ۙ then there is no government research subsidy. There may be a desire to subsidize research on the part of the government due to the traditional spillovers associated with imperfect property rights or in a desire to increase the speed in innovation.
Recall from equation (3.14) that π i* is the stage two profit that the winner will receive with the patent for supplying renewable energy to the domestic energy markets and the international market. Consequently, expected stage one profits for the pair of firms can be expressed as:
1−ϕ
m (x )2 2 m 1−ϕ (3.18b) E ( profit ) = [ P ( x )(1 − P ( x )) + 0.5 P ( x ) P ( x )]π * − 2 n ( x ) 2 n n m n m n n
(3.18a) E ( profit ) = [ P ( x )(1 − P ( x )) + 0.5 P ( x ) P ( x )]π m
m
n
m
n
m
*−
34
By assumption, the loser of the patent race will receive zero profits.
The two firms choose the amount of money to dedicate to R&D simultaneously. Therefore we can determine a Nash equilibrium level of expenditure on research. Solving the maximization profit we get the first order conditions, where π ri* is the exogenous profit earned in stage three for being the sole supplier of MW/h of renewable energy:
(3.19a) P ' ( x m )(1 − 0.5 P ( x n ))π m* − (1 − ϕ m ) x m = 0 (3.19b) P ' ( x n )(1 − 0.5 P ( x m ))π n* − (1 − ϕ n ) x n = 0 For simplicity the left hand side of the first order equations (3.19a) and (3.19b) will be described as a function of their variables. This allows for easier notation when taking the total derivatives.
ۙ
(3.20a) P ' ( x m )(1 − 0.5 P ( x n ))π m* − (1 − ϕ m ) x m = f m ( x m , x n , π m* , ϕ m ) = 0 (3.20b) P ' ( x n )(1 − 0.5 P ( x m ))π n* − (1 − ϕ n ) x n = f n ( x m , x n , π n* , ϕ n ) = 0
Taking the total differential of equations (3.20a) and (3.20b) with respect to xm, xn and sm and dividing by dsm we get:
df m dx m df m dx n df m dπ m* (3.21a) + + =0 dx m ds m dx n ds m dπ m* ds m df n dx m df n dx n df n dπ n* (3.21b) + + =0 dx m ds m dx n ds m dπ n* ds m Solving the system of equations the resulting equations are:
35
(3.22a)
(3.22b)
dx m = ds m
dx n = ds m
−
−
df m dπ m* df n df m df n dπ n* + dπ m ds m dx n dx n dπ n ds m df m df n df m df n − dx m dx n dx n dx m df n dπ n* df m df n df m dπ m* + dπ n ds m dx m dx m dπ m ds m df m df n df m df n − dx m dx n dx n dx m
where:
(3.23) (3.24) (3.25) (3.26) (3.27) (3.28)
df m = −(1 − ϕ m ) < 0 dx m df n = −(1 − ϕ n ) < 0 dx n df m 1 = −0.5 P ′( x m ) P ′( x n )π m = −0.5 2 π m < 0 dx n k ۙ df n 1 ′ ′ = −0.5 P ( x m ) P ( x n )π n = −0.5 2 π n < 0 dx m k df m 1 = P ′( x m )(1 − 0.5 P( x n )) = (1 − 0.5 P( x n )) > 0 * k dπ m df n 1 = P ′( x n )(1 − 0.5 P ( x m )) = (1 − 0.5 P( x m )) > 0 * k dπ n Substituting equations (3.23) to (3.28) into (3.22a) and (3.22b) the result is:
(3.29a)
dx m = ds m
−
dπ * dπ * 1 1 1 (1 − 0.5 P ( x n )) m (ϕ n − 1) + (−0.5 2 π m ) (1 − 0.5 P ( x m )) n k ds m k ds m k 1 1 (ϕ m − 1)(ϕ n − 1) − (−0.5 2 π n )(−0.5 2 π m ) k k
36
(3.29b)
dx n = ds m
−
dπ * dπ * 1 1 1 (1 − 0.5 P ( x m )) n (ϕ m − 1) + (−0.5 2 π n ) (1 − 0.5 P ( x n )) m k ds m k ds m k 1 1 (ϕ m − 1)(ϕ n − 1) − (−0.5 2 π n )(−0.5 2 π m ) k k
Result 1: If both countries are completely accessible (δm=δn=0) and have equal research subsidies then an introduction of a feed-in-tariff by one country will result in an equal increase in both firms’ equilibrium level of research.
Proof: Suppose a feed-in-tariff is introduced by country m.
From equations (3.12) and (3.13) it can be seen that if δ = 0 then π m* = π n* = π * dπ n* dπ m* dπ * and from (3.15) and (3.16) if δ = 0 then = = > 0 . If both research ds mۙ ds m ds m subsidies are equal then φm = φm = φ. Since equations (3.20a) and (3.20b) are symmetric xn = xm =x. Substituting equations into (3.29a) and (3.29b) we get:
1 dπ * 1 (1 − 0.5 P ( x)) (−(ϕ − 1) − 0.5 2 π ) dx m k ds m k = >0 1 ds m 2 2 (ϕ − 1) − (0.5 2 π ) k
1 dπ * 1 (1 − 0.5 P ( x)) (−(ϕ − 1) − 0.5 2 π ) dx n k ds m k = 1 ds m (ϕ − 1) 2 − (0.5 2 π ) 2 k
37
Therefore if (ϕ − 1) 2 > (0.5
dx n dx m 1 2 π ) then = >0 ds m ds m k2
Q.E.D
With two completely accessible economies a feed-in-tariff will result in both firms increasing their research equal amounts. Since both firms can capture all of the potential profits from both markets, if the potential profits increase in one market they should increase their research to attempt to capture the increased profits.
Taking the total differential of equations (3.20a) and (3.20b) with respect to xm, xn and φm and dividing by dφm:
df m dx m df (3.30b) n dx m (3.30a)
dx m df m dx n df m + + =0 dϕ m dx n dφ m dφ m dx m df n dx n + =0 dφ m dx n dφ m
ۙ
Solving the system of equations:
df m df n dx dφ m dx n (3.31a) m = df m df n df m dφ m − dx m dx n dx n df n df m dx dx m dφ m (3.31b) n = df m df n df m dφ m − dx m dx n dx n −
df n dx m
df n dx m
38
(3.32)
df m = xm > 0 dϕ m
Result 2: If one country’s government implements a research subsidy then the firm from that country will increase its equilibrium level of money spent on research while the firm from the other country will decrease its equilibrium amount of money spent on research.
Proof: Suppose φm decreases as a result of a research subsidy.
Substituting in equations (3.23) to (3.26) and (3.32) into (3.31a) and (3.31b).
dx m = dφ m
− ( x m )(ϕ n − 1) ۙ 1 1 (ϕ m − 1)(ϕ n − 1) − (−0.5 2 π n )(−0.5 2 π m ) k k
1 πn) 2 dx n k = 1 1 dφ m (ϕ m − 1)(ϕ n − 1) − (−0.5 2 π n )(−0.5 2 π m ) k k ( x m )(−0.5
Therefore if ϕ i − 1 > (0.5
dx m 1 dxn π ) then > 0 and <0 i 2 dφ m dφ m k
Q.E.D
An increase in research subsidy will increase the equilibrium level of money spent on research by the firm in the country which implemented the research subsidy and decrease the equilibrium level on money spent by the firm whose country didn’t
39
implement the research subsidy. The research subsidy lowers the cost of research to the domestic firm and therefore the domestic firm will do more research. The international firm’s best response to the domestic firm’s increase in research without any additional profits from stage two is to decrease their research. This is a unique result to this model which results in the domestic firm acting like a first mover and the international firm as a follower.
The following results were developed using artificial data to attempt to determine the effects on research in asymmetric situations. All solutions must satisfy the expected profit first order conditions equations (3.19a) and (3.19b). Several different parameter values were used in order to try and determine sign and relative size of several derivatives in an asymmetric environment (See appendix). ۙ
Result 3: If a country is not perfectly accessible, a feed-in-tariff will increase both domestic and international research; however, the increase to domestic firms will be greater.
Proof: From table 1 in the appendix it is clear that in all simulated situations
dx m dx n dx dx , >0 and m − n > 0. ds m ds m ds m ds m
If a country is not completely accessible, policies which should affect all firms equally result in an unequal level of research.
40
4
EMPIRICAL This empirical Chapter is divided into four subsections. Section 4.1 describes the
basic empirical model. Sections 4.2-4.4 describe the data which will be used. Section 4.2 is a description of the patent data, followed by section 4.3 which describes the policy data and section 4.4 which provides a description of the other variables.
4.1
Empirical Model
This section looks at the effects that domestic output and world output have on innovation. To do this the effect that domestic output of wind energy (GWh) has on patents is compared to the world wind energy output (GWh). If there is a home bias domestic output (GWh) should have a greater effect on patenting than world output 摠ۛ one possible explanation for a home bias (GWh). Following from the theoretical section,
would be that policies implemented to encourage the increased output of wind energy encourage domestic firms to innovate more than international firms. However, the policy effects on innovation may not be the only explanation captured by this regression. Alternatively there may be a government bias in the procurement process towards domestic firms resulting in the growth of domestic firms. This would result in large wind producing countries having large domestic firms which would have incentive to invest in R&D and innovate. This bias and its effects on domestic growth of firms has been pointed many (Branco, 1994; Miyagiwa, 1991) and may still be prevalent throughout Europe (Nielsen & Hansen, 2001).
41
The literature on panel patent count data started from Hausman et al (1984) which looked at the effects R&D expenditures had on patent counts of firms and was the first to develop random and fixed effects models for Poisson and negative binomial distributions. This was followed up by Cameron and Trivedi (1998) who pulled together all the applicable research on count data into one source. To test if there is a “home bias”, a simple model is constructed:
~
(4.1)
λi ,t = exp( β1 DS i ,t + β 2WS t + u i )
~
Where λ i,t is the exponential mean function and is a function of the domestic size of market (DSi,t) in country i in time t, world size of market (WSt) and a fixed effect or ۙ random effect (ui). The domestic size and world size will be measured in terms of
thousands of gigawatt hours of electricity (GWh). For simplicity it will now be written in matrix form:
~
(4.2)
λi ,t = exp(u i ) exp( β1 DS i ,t + β 2WS t ) = vi exp( β1 DS i ,t + β 2WS t ) = vi λi ,t = E[ pat i ,t | xi ,t ] In order to test for “home bias” both Poisson and negative binomial distributions
will be tested due to the significant amount of zeros. If there is a “home bias” we would expect to see either the coefficient on domestic size being larger than that of world size if both are significant or the coefficient on domestic size being positive and the coefficient on world size being insignificant. The major difference between the Poisson and the negative binomial model is that the Poisson model forces the mean and variance to be 42
equal not allowing for overdispersion of the data. Overdispersion is often seen in cases where there are a lot of zero counts resulting in a variance larger than the mean. The negative binomial distribution allows for overdispersion. Fixed effects and random effects models will be tested for each distribution using the random and fixed effects developed by Cameron and Trivedi (1998).
Fixed Effects Models: Both a negative binomial and Poisson fixed effects models will be used to look for “home bias”. The fixed effect model allows one to attempt to capture unobserved heterogeneity. In the fixed effects model u i is allowed to be correlated with x'i ,t and is assumed to be fixed. The fixed effects model results in different intercepts for each country. The coefficients are estimated using a conditional maximum likelihood estimates. For the Poisson distribution it is assumed pat i ,t are ~
independently identically distributed with Poisson distribution and mean λi ,t , the conditional joint density for the ith observation is:
(4.3)
T (∑t pat i ,t )! λi ,t Pr pat i ,1 , K , pat i ,T | ∑ pat i ,t = ∏t ∑ λi , s ∏ t pat i ,t ! t =1 s
pati ,t
For the negative binomial model it is assumed that the assume pat i ,t is independently identically distributed with a negative binomial distribution and mean λi,t and variance λi ,t (vi + vi2 ) , the conditional joint density for the ith country is:
43
(4.4)
(
)(
)
T Γ(λi ,t + pat i ,t ) Γ ∑t λi ,t Γ ∑t pat i ,t + 1 Pr pat i ,1 , K , pat i ,T | ∑ pat i ,t = ∏t Γ ∑t λi ,t + ∑t pat i ,t Γ(λi ,t )Γ( pat i ,t + 1) t =1
(
)
Random Effects Models: In the random effects model, vi is a country specific random variable. With the random effects model, u i must be independent of x'i ,t . For the Poisson model, vi is assumed to be distributed as a gamma random variable with parameters (δ , δ ) normalized so that E [vi ] = 1 and Var [vi ] = 1 / δ . This leads to the joint density for country i of:
(4.5)
Pr ( pat i ,1 , K , pat i ,T
(
)
Γ ∑t pat i ,t + δ δ | xi ,1 , K xi ,T ) = Γ(δ ) δ + ∑t λi ,t
δ
λi ,t pati ,t − ∑ pati , t t ∏ t ∑t λ i , t + δ pat i ,t !
(
)
Ҵ
For the negative binomial model,
1 is assumed to distributed as beta (a,b) 1 + vi
random variable. This leads to the joint density for country i of:
(4.6)
( (
)(
Γ(λi ,t + pat i ,t ) Γ(a + b )Γ a + ∑t λi ,t Γ b + ∑t pat i ,t Pr ( pat i ,1 , K , pat i ,T | xi ,1 , K xi ,T ) = ∏ t Γ(a )Γ(b )Γ a + b + ( ) ( ) Γ λ ! Γ pat + 1 ∑t λi,t + ∑t pat i,t i ,t i ,t
Equations (4.2) to (4.4) are maximized to obtain the coefficient estimates.
Another interesting question that arises is what determines the domestic output (GWh) of a market? Some small countries have a disproportionately large domestic
44
) )
output while some large countries have a disproportionately small domestic market. What will encourage the growth in renewable energy? There are several factors that could determine the output of a domestic market. One of the major factors that will determine the domestic supply of output will be the policies within that country. Ideally one would like to look at the effect that policies have on the supply of renewable wind energy. However, the supply of wind power (GWh) has a random component to it in that the supply of wind is determined in part by the weather. Therefore instead of using the supply of wind power the change in capacity over the year is used. Change in capacity is nonrandom and policies should have the same effect on the change in capacity as the change in supply.
The policy data available through OECD/IEA (OECD/IEA, 2004) poses a major ۞
problem in that it doesn’t provide a measure of the size of the policy. This could lead to misleading or incorrect results because it treats all policies as being of equal size. Another major issue is that in many cases multiple policies are implemented at the same time resulting in multicollinearity between policies. This will result in large standard errors between policies most likely resulting in many policies being insignificant in the results. The policies are broken up into seven subfields: tax incentives, feed-in-tariffs, green certificates, producer grants, consumer/individual grants, obligations and targets. From the theoretical literature all policies, except targets which tend be non-binding, should result in positive changes in supply of wind energy. This should result in their coefficients being positive implying that these policies increase the change in wind capacity. We would predict the expected price of oil to have an impact on the growth of renewable energy sources. As the expected price of oil rises, there should be more 45
incentive to invest in green energy as there is an expected relative drop in price. If this is true there should be a positive coefficient on the price of oil. For the expected price of oil, a simple change in oil price from the current year to the previous year is used. This is used as an approximation to what the expected change in oil prices will be in the next year.
Other factors that could affect the size of the domestic green energy market would be the size of the economy (GDP) and the size per capita (GDP per person). In general one would expect a large economy with a large demand for electricity would have an increased use of all forms of energy including green energy sources and traditional energy sources resulting in a positive coefficient on GDP. It has also been speculated amongst economists that as individuals’ wealth rises they tend to put more value on a ۙ
clean environment. Since green energy isn’t polluting it would seem likely that as per capita income increases so too should the demand for renewable energy. This would be seen in a positive coefficient on GDP per capita. Due to multicolinearity with GDP an average of the past three years is used for GDP per capita. This eliminates the multicolinearity issue which arises from using GDP and GDP per capita from the same year and smoothes GDP per capita from business cycles. With this in mind a second model is created to look at the determination of domestic markets, Where ∆capi,t is the change in capacity.
46
(4.7) gdpcap i ,t − 2 + gdpcap i ,t −3 + gdpcap i ,t − 4 ∆cap i ,t = β 1 pol i ,t −1 + β 2 ln gdp i ,t −1 + β 3 ln 3 + β 4 (ln oil i ,t −1 − ln oil i ,t − 2 ) + α i + δ t + ei ,t
The model includes either fixed or random country effects αi and time effects δt. If
α i is a fixed effect, it is assumed that it is nonrandom and can be estimated using dummy variables. If α i is random it is assumed that it comes from a normal distribution which must be accounted for in the error covariance matrix. If α i is considered to be a random effect, two additional variables are included: a coastline dummy and an area variable. One might expect a country with a larger area to have a large change in wind capacity due to the potential of better locations for wind turbines. Additionally with coastline providing some of the best wind potential, countries with coastline may have larger ۙ changes in wind capacity. These are not included in the fixed effects model because they are constant and therefore captured by the fixed effect. Time fixed effects are used to capture effects that are different across time periods but not countries.
4.2
Patent Data
Country count data on patents filed with the European Patent office seeking protection within the European Region will be used to measure the effects of policy changes on innovation. The European Patent Office maintains a searchable patent database (European Patent Office, 2008) of European, worldwide and US patents from 1978 to the present. It allows a search of the database using 10 fields such as keyword(s)
47
in title or abstract, application number, publication number, priority number, inventor, applicant, European classification (ECLA) and International Patent Classification (IPC).
There are several issues associated with using patent data as a way of measuring innovation. One is that all patents are equal regardless of how minor or major the innovation is. One way to alleviate this effect is to only look at patents filed to the European Patent Office rather than individual country patent offices. Since all patent applications require expenditures of time and money one could assume that only the truly worthwhile inventions would take the time and the money to file with the broader European Patent Office. Another major issue is determining the date of the invention, which will be crucial to determine the effect of policy changes. Although this information isn’t expressly given, the priority date does allow one to know the first date that the Ҵ
patent was filed with any patent office and thus is the closest date to the invention date. One would also assume that a patent would be filed soon after invention to allow for protection from competitors. This is more useful than the publication number or application number because many inventors will first file in their home office or the office of one of the larger European countries (Great Britain, Germany or France) before filing for a European patent.
Patents related to the use in producing wind energy can be determined using International Patent Classification (IPC) codes (World International Patent Organization, 2008). The hierarchal coding system allows the isolation of relevant classes or subclasses suitable for each form of alternative energy. For wind power the relevant classification are displayed in table 4.1.
48
Table 4.1: IPC codes for Wind Wind Wind motors with rotation axis substantially in wind direction Wind motors with rotation axis substantially at right angle to wind direction Controlling wind motors Adaptations of wind motors for special use; Combinations of wind motors with apparatus driven thereby Details, component parts, or accessories not provided for in, or of interest apart from, the other groups of this subclass ie Transmission of power, e.g. using hollow exhausting blades or Mounting structures
Class F03D F03D F03D
Sub-Classes 1/00 3/00 7/00
F03D
9/00
F03D
11/00
Once the categories are determined, the database is searched using the IPC classification and the results searched to determine that the patent is relevant. A patent result is relevant if its intended use is in producing wind energy. The description section of each patent allows the innovator to discuss the intended use of their innovation. This minimizes the risk of counting patents which should not be included. The risk of not counting a patent is small because most patents are filed under numerous IPC codes. ۙ
Patent counts were determined for all major patenting countries along with patent priority dates and country of inventor. A panel data set is constructed for 1990 to 2006. From the twenty countries of the European Union, sixteen were found to have patented during that period of time. As one can see from figure 4.1 each country has significantly different rate of patenting and there has been a substantial increase in overall patenting in the last several years.
49
Figure 4.1: Patents from 1990-2006 for Germany, Denmark, Great Britain and Austria in Wind Energy 140
120
100
Germany Denmark Great Britain Austria Europe
Patents
80
60
40
20
˖
0
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Year
The drop off in patenting in the final years is a result of patent counts being based on priority date, the first date that a patent is filed at any patent office. As a result there will be many patents that have a priority date from 2005-2006 that have not yet been filed with the European patent office or have not yet made it through the European patent office system.
4.3
Policy Variables
Policy variables are determined using the International Energy Agency Global Renewable Energy Policies and Measures Database (OECD/IEA, 2008) and their
50
published policy reviews (OECD/IEA, 2004). The database allows the searching of policies based on type of policy, country and policy target. Policy target is the type of renewable energy that the policy is supposed to target since different policies have been used to target specific renewable energy sources. Policy type will allow the grouping of policies into the following categories: grants to producers, grants to companies/individuals to invest in wind energy, tradeable green certificates, tax incentives, feed-in-tariffs, obligations (producers are required to use dedicated amounts of certain renewables) and targets. For each policy type, dummy variables are created. The policy dummy is 1 if it is implemented in a given country and remains a 1 until it is phased out. The policy dummy will only be connected with the renewable energy source that they are supposed to be targeting, since usually policies are designed to target only certain renewable energy sources. The IEA policy database starts with policies
implemented as early as 1974 and has been updated up to July 2008. It would be desirable to know the strength of these policies, such as the size of the feed-in-tariff but this information is not readily available.
4.4
Other Variables
Wind capacity and supply is available from 1990-2006 through the IEA database (International Energy Agency, 2009). Like the trend in patenting there has been a steady increase in the use of wind energy. Some countries like Germany have seen a significant growth in output from wind energy over the 16 year period from 1990 to 2006 (see Figure 4.2).
51
Figure 4.2: Electricity Output (GWh) for Germany, Denmark, Great Britain, Austria and OECD 140000
120000
Electricty Output (GWh)
100000
Germany Denmark Great Britain Austria OECD
80000
60000
40000
20000
ۙ 0 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 Years
The IEA database also provides the cost of crude oil imports to Europe. GDP and GDP per capita are available from the International Monetary Fund (2009). Coast dummies and land area are available through the Central Intelligence Agency (2008). For complete descriptive statistics see tables 8.2 and 8.3.
52
5
RESULTS The results chapter is broken up into two sections. Section 5.1 examines the results
of the first regression to determine whether there is a “home bias” and section 5.2 examines the second regression, which looks at what determines the size of a market.
5.1
Patenting – Regression One
The results to both the Poisson and negative binomial regressions are supplied in table 5.1. In both cases domestic output and total output have a positive impact on the amount of patenting done by a given country. A dummy variable is included for 2006 to capture the decrease in patents due to the priority lag. In all but one case the coefficient on domestic output is greater than the coefficient on total OECD output. ۞ֹ
53
Table 5.1: Results of Patent Regression Poisson Output in thousands of GWh Domestic Output World Output Dummy 2006 Constant Observations Number of Countries
Output in thousands of GWh Domestic Output World Output Dummy 2006 Constant
Fixed Effects Negative Binomial
0.025 0.015 (3.87)*** (2.28)** 0.01 0.019 (7.63)*** (11.62)*** -1.366 (9.74)***
0.023 (1.69)* 0.012 (5.90)***
0.025 (1.96)* 0.019 (8.18)*** -1.372 (5.00)*** -0.185 (0.89) 272 16
-0.265 (1.41) 272 272 272 16 16 16 Random Effects Poisson Negative Binomial 0.026 0.018 (4.18)*** (2.66)*** 0.01 0.018 (7.64)*** (11.64)*** -1.368 (9.78)*** ֹۙ
0.027 (2.14)** 0.011 (5.69)***
-0.255 (1.37) 289 16
0.030 (2.41)** 0.018 (8.02)*** -1.353 (5.03)*** -0.181 (0.88) 289 16
Observations 289 289 Number of Countries 16 16 Absolute value of z-statistics in parentheses * significant at 10% level;** significant at 5% level; *** significant at 1% level
When looking at the two results one must also remember the major difference in the Poisson and negative binomial models. The Poisson model assumes that the mean is equal to the variance. This may not be true when it comes to patents which tend to display overdispersion. Overdispersion is caused by a large number of zero entries resulting in the variance being larger than the mean. In all regressions domestic size of market and world size of market are significant at 1%, 5% or 10% significance level and jointly significant. An F-test was run to check whether the coefficients for domestic size and world size were significantly different. In all the Poisson models the null hypothesis
54
that the coefficients on domestic size and world size were equal could not be rejected at 5%. Comparatively in all the negative binomial models the null hypothesis that the coefficients on domestic size and world size were equal could always be rejected in favor of the alternative that they are statistically different.
The Haussman test is run with the negative binomial fixed and random effect models with 2006 dummies to attempt to determine which model specification is most appropriate. The null hypothesis that the coefficients estimated by the efficient random effects estimator are the same as the ones estimated by the consistent fixed effects estimator are rejected (p-value 0.0006); therefore the fixed effects model should be used.
In order to interpret the coefficients the conditional mean is differentiated with Ҵ
respect to domestic market size (DSi,t) and world market size (WSt) to look at the effects of marginal changes on the conditional mean.
∂E[ pat i ,t | xi ,t ] ∂WS i ,t ∂E[ pat i ,t | xi ,t ] ∂DS i ,t
= β 2 λ i ,t = β 1 λ i ,t
In all but one case a change in the size of the domestic market has a larger effect on the expected patent count than a change in the size of the international market. In some cases it has almost double the effect; however in only the negative binomial specifications are the coefficients significantly different. Another significant issue is that
55
the maximum likelihood negative binomial estimators are inconsistent under misspecification.
5.2
Market Size – Regression Two
The results of the market size regression can be seen in tables 5.2 and 5.3. The results show inconsistencies. GDP and GDP per capita which one would assume to be both be positive and significant in fact change signs and in some cases are significant and in other cases are not significant. The results as presented in tables 5.2 and 5.3 are also highly influenced by outliers. If Spain and Germany are excluded from the fixed effects regressions, GDP and GDP per capita become insignificant in regression 1 and in regression 2 GDP per capita remains significant. ۙъ
56
Table 5.2: Fixed Effects Market Size Results Oil ((ln oil i ,t −1 − ln oil i ,t − 2 ) ) GDP per capita gdpcap i ,t − 2 + gdpcap i ,t −3 + gdpcap i ,t − 4 ln 3
GDP (ln gdpi ,t −1 ) Green Certificates Tax Incentives Targets Obligations Feed-in-tariffs ۙъ
Grants to Companies or Individuals Grants to Producers of Electricity Time Dummies Constant
(1)
(2)
Dropped
117.69 (1.46)
477.10 (2.14)**
375.92 (2.45)**
-379.40 (1.88)* -277.51 (3.62)*** -26.29 (0.34) -89.76 (1.04) 451.27 (6.44)*** -104.67 (1.58) -43.71 (0.55) 18.20 (0.29) Yes -2464.51 (1.48) 272 17 0.4
33.86 (0.27) -188.79 (2.58)*** 24.40 (0.33) -81.18 (0.94) 578.90 (9.20)*** -45.96 (0.71) 16.46 (0.21) 51.62 (0.83) No -3825.24 (3.29)*** 272 17 0.34
Observations Number of Countries R-squared Absolute value of t-statistics in parentheses * significant at 10% level;** significant at 5% level; *** significant at 1% level
57
Table 5.3: Random Effects Market Size Results Oil ((ln oil i ,t −1 − ln oil i ,t − 2 ) ) GDP per capita gdpcap i ,t − 2 + gdpcap i ,t −3 + gdpcap i ,t − 4 ln 3
GDP (ln gdpi ,t −1 ) Green Certificates Tax Incentives Targets Obligations Feed-in-Tariffs ۞
Grants to Companies and Individuals Grant to Producers Time Dummies Area ln(area i ) Coastline Dummy Constant
(1)
(2)
Dropped
115.75 (1.50)
106.58 (0.86)
216.61 (1.98)**
20.58 (0.27) -254.90 (3.58)*** -65.94 (0.96) -130.42 (1.66) 468.12 (7.12)*** -69.56 (1.14) 14.98 (0.21) 35.60 (0.62) Yes
52.68 (0.75) -155.19 (2.23)** -9.23 (0.13) -73.40 (0.92) 583.29 (9.66)*** -12.78 (0.21) 73.40 (1.00) 68.98 (1.16) No
46.59 (0.55) 96.65 (0.51) -1754.483 (1.08) 272 17
45.11 (0.55) -22.86 (0.12) -2896.45 (2.47)** 272 17
Observations Number of Countries Absolute value of z-statistics in parentheses * significant at 10% level; ** significant at 5% level; *** significant at 1% level The only significant results that remain constant are that obligations result in an increase in the change in wind capacity and green certificates result in a decrease in the
58
change in wind capacity. It is not surprising that countries that set obligations to the amount of wind energy or renewable energy that must be used show increases in the change in wind capacity. This result holds even as different countries are removed from the regression to check if an individual country is driving the result. The result that green certificates are negative throughout all the regressions may be of some surprise; however if Italy is removed from the regression, green certificates are insignificant. Italy uses green certificates but has been very unsuccessful at encouraging growth in renewable energy sources. The lack of significance amongst other policies could be a reflection of how policies are introduced, with several policy types being introduced at the same time. The result is high correlation amongst policies resulting in large standard errors, which may cause none of the other policies to be significant. ۞г
The lack of significance of many of the economic variables that one may expect to be a factor in the increase in wind capacity, leads one to look for other social, political or infrastructure constraints that may have an impact on the change in demand for wind energy. In the following section, several countries will be examined to determine the other factors that may have lead to increases in demand of wind energy.
59
6
BRIEF COU TRY REVIEW If changes in wind power capacity can’t be explained by any of the typical
economic variables, what does determine changes in country’s wind power capacity? In the next sections we will look at a few countries which have been both leaders in the development of wind capacity and followers. In Europe the environment has become an issue when it comes to energy policy; however a country’s response varies significantly. In many cases a country’s response to the use of nuclear power plays a significant role in a country’s growth in renewables. Nuclear power is a cheap and reliable way of producing electricity without the emissions of greenhouse gases. Section 6.1 will look at Germany a world leader in wind capacity. Section 6.2 will look at France a country in many ways similar to Germany, but with little development of wind potential and section 6.3 will look at Denmark a relatively small country with large wind development. ۙг
6.1
Germany
Germany is the largest economy in Europe, the fifth largest consumer of the electricity in the world and second leading producer of wind energy in 2008 behind only the United States (NBC News, 2008). In 2006 5% of its total generation was from wind resulting in 30710 GWh of electricity (Energy Information Administration, 2009). Wind output has grown substantially from 71 GWh in 1990 to 30710 GWh in 2006. So what has prompted this rapid growth in wind energy? Most European countries started to take an increasingly serious look at their energy makeup following the oil crisis of the 80’s. This was followed in the early 90’s by a push for stronger environmental regulation. In the case of Germany there seems to be a political part to the story not captured in the
60
economic data. In 1998 the Social Democrats-Green coalition came into power with one of their prime objectives to phase out the use of nuclear power in Germany. Since 1997 the amount of nuclear power generated has, as a percentage of total generation, decreased from approximately 31% to approximately 27% in 2006 resulting in a 30000 GWh decrease in the use of nuclear. Germany has taken a very aggressive role in implementing and researching green energy sources which can be seen in the increase in share of all renewables in the past 15 years. The strength which the renewables are encouraged may not accurately be represented in the binary nature of the policy variables.
6.2
France
France is in many cases very similar to Germany. France and Germany are of similar size with France being slightly smaller in terms of GDP, but having a slightly ۙ
larger per capita GDP. Both seem very similar politically with left leaning and right leaning political parties being in power over similar amounts of time; however the results have been very different when it comes to wind power. The French by comparison have been the largest European net exporter of electricity for the last 15 years (Energy Information Administration, 2009), the result of an overcapacity. Starting in the 1980’s the French diversified their energy portfolio by investing in nuclear power. By the 90’s 90% of their domestic electricity needs and 75% of there total generation were met by nuclear power resulting in substantial exporting of electricity. The reliance on nuclear has also resulted in a power grid designed for large consistent centralized power plants (Reiche & Bechberger, 2004), which may be poorly designed to handle the strains of wind energy. This combined with what seems to be a general acceptance of nuclear
61
power by the French people has resulted in the slow growth of wind power and has probably resulted in extremely modest targets for renewable energy sources such as wind power.
6.3
Denmark
Denmark is in a position quite different from France and Germany in that it is relatively small. Denmark consumed around 1% of Europe’s electricity in 2006 of which 18% of its electricity was produced from wind. In 1985 Denmark abandoned any idea of the use of nuclear power as part of their power portfolio (Climate Change and Power Economic Instruments for the European Electricity, 2002). Denmark has taken an even stronger taxation approach than many European countries. As imported fuel prices decreased in the late 80’s taxation increased, resulting in a neutralization in the drop of ۙН
energy prices. This coupled with the moratorium on nuclear power resulted in the perfect environment for increased use in wind power. This can be seen not only in the use of wind power and other renewables but also the total energy intensity (consumption per GDP) which has not increased since 1972 (Energy Information Administration, 2009). Denmark also has a distinct advantage compared to other countries in that it is connected to two large grids, the Nordic grid and the German grid and it has relatively small output. The structure of the Danish wind market is vastly different than most as well. Whereas as many countries are moving towards privatization there are over 3000 co-op wind farms in Denmark leading to 150,000 owners (Reiche & Bechberger, 2004). This has the effect of leading to increased acceptance amongst the Danish people.
62
7
CO CLUSIO Based on the theoretical model developed, government regulation or other factors
which increase the cost of doing business in an international market may lead to domestic firms innovating more than international firms. Empirical evidence shows that domestic wind energy output might increase the expected number of patents in a country more than international wind energy output. The results of the regressions are inconclusive.
The answer to the second question, which policies have the most impact on the domestic change in wind capacity, is unclear. The only variable that is robust to the inclusion and exclusion of different countries is obligations which remains significant and positive throughout. This implies the implementation of a policy involving an obligation to use a certain amount of renewable energy results in a positive increase in ҴН
the change in capacity of wind energy. This is not surprising due to the strength and ease of implementation of this policy. Other policies were generally insignificant for most model specifications, which could be due to the multicolinearity. When looking at several different countries one can see that the choice of whether to implement policies to encourage wind power and the strength of those policies is very individualistic. Exactly what contributes to the size of the market is difficult to determine because factors other than economic factors, such as social attitudes towards wind or nuclear power and constraints of the grid and domestic potential of other sources of reliable energy, all play a part.
63
A critical missing element to this research is the strength of the policy implemented; however, this is extremely hard to determine because within a country there are numerous regional governments implementing many different policies. As the data set gets longer and more green energy sources are widely used throughout the world, the factors leading to increases in capacity may become clearer. Future work should include some level of the stringency of the policies; however, at this time there is not enough information to produce any strong results on different policy effects on the change in wind capacity.
ۙ
64
REFERE CES Arrow, K. J. (1962). Economic Welfare and the Allocation of Resources for Invention. The RA=D Corporation. Baumol, W. J., & Oates, W. E. O. (1971). The Use of Standards and Prices for the Protection of the Enivonment. The Swedish Journal of Economics, 73(1), 42-54. BBC News. (2009). EU climate package explained Retrieved March 16, 2009 Beghin, J. C., & Lovell, C. A. K. (1993). Trade and Efficiency Effects of Domestic Content Protection: The Australian Tobacco and Cigarette Industries. The Review of Economics and Statistics, 75(4), 623-631. Bentley, R. W. (2002). Global oil & gas depletion:an overview. Energy Policy, 30, 189205. Branco, F. (1994). Favoring domestic firms in procurement contracts. Journal of International Economics, 37, 65-80. Cameron, A. C., & Trivedi, P. K. (1998). Regression Analysis of Count Data. New York: Cambridge University Press. Central Intelligence Agency. (2008). CIA world fact book. Climate Change and Power Economic Instruments for the European Electricity. (2002). London: Earthscan. Dahl, C. A. (1979). Consumer Adjustment to a Gasoline Tax. The Review of Economics and Statistics, 61(3), 427-432. Dasgupta, P., & Stiglitz, J. (1981). Entery, Innovation, Exit Towards a Dynamic Theory І of Oligopolistic Industrial Structure. ۙEuropean Economic Review, 15, 137-158. de Gorter, H., & Just, D. R. (2009). The Economics of a Blend Mandate for Biofuels. American Journal of Agricultural Economics. Dernis, H., & Guellec, D. (2001). Using Patent Counts for Cross-Country Comparisons of Technology Output: OECD. Dixon, P. B., & Osborne, M. T. R. (2007). The Economy-Wide Effects in the United States of Replacing Crude Petroleum with Biomass. Paper presented at the GTAP Conference. Energy Information Administration. (2009). Official Statistics from the US Government. European Patent Office. (2008). esp@cenet. Retrieved September 19, 2008, from http://ep.espacenet.com/ Fischer, C., & Newell, R. G. (2004). Environmental and Technology Policies for Climate Change and Renewable Energy. Griliches, Z. (1992). The Search for R&D Spillovers. Scandinavian Journal of Economics, 94, 29-47. Hall, B. H., & Van Reenen, J. (2000). How effective are fisical incentives for R&D? a review of evidence. Research Policy, 29, 449-469. Hausman, J., Hall, B. H., & Griliches, Z. (1984). Econometric Models for Count Data with An Application to the Patents-R&D Relationship. Econometrica, 52(4), 909937. Herbert, G. M. J., Iniyan, S., Sreevalan, E., & Rajapandian, S. (2007). A review of wind energy technologies. Renewable and Sustainable Energy Reviews, 11(6), 11171145.
65
Hicks, J. R. (1966). The Theory of Wages. New York: St Martin's Press. International Energy Agency. (2006). Renewable Energy: RD&D Priorities Insight from Technology Programmes. Paris: International Energy Agency. International Energy Agency. (2009). International Energy Agency Data Services. Retrieved January 5, 2009, from http://wds.iea.org/wds/default.aspx International Monetary Fund. (2009). World Economic Outlook Database. Retrieved January 5, 2009, from http://www.imf.org/external/pubs/ft/weo/2008/02/weodata/index.aspx Johnstone, N., Hascic, I., & Popp, D. (2008). Renewable Energy Policies and Technological Innovation: Evidence Based on Patent Counts. Krishna, K., & Itoh, M. (1988). Content Protection and Oligopolistic Interactions. The Review of Economic Studies, 55(1), 107-125. Lee, T., & Wilde, L. L. (1980). Market Structure and Innovation: A Reformulation. The Quarterly Journal of Economics, 94, 429-436. Loury, G. C. (1979). Market Structure and Innovation. The Quarterly Journal of Economics, 93, 395-410. McGowan, F. (2000). Reconciling EU Energy and Environmental Policy. In P. Jasinski & W. Pfaffeberger (Eds.), Energy and Environment: Multiregulation in Europe (pp. 1-21). Burlington: Ashgate Publishing Company. Menanteau, P., Finon, D., & Lamy, M.-L. (2003). Prices versus quantities: choosing policies for promoting the development of renewable energy. Energy Policy, 31, 799-812. Miyagiwa, K. (1991). Oligopoly and Discriminatory Government Procurement Policy. American Economic Review, 81(5), ҴІ 1320-1328. NBC News. (2008). U.S. Edges Out Germany as World Wind Power Leader [Electronic Version]. Retrieved Febuary 18, 2009 from http://www.nbcnewyork.com/news/green/U_S__Edges_Out_Germany_as_World _Wind_Power_Leader.html. Nelson, R. (1959). The Simple Economics of Basic Research. Journal of Political Economy(67), 297-306. Newell, R. G., Jaffe, A. B., & Stavins, R. N. (1999). The Induced Innovation Hypothesis and Energy-Saving Technological Change. The Quarterly Journal of Economics, 114(3), 941-975. Nielsen, J. U.-M., & Hansen, L. G. (2001). The EU public procurement regime— Does it work? . Intereconomics, 36(5), 255-263. Nordaus, W. (1972). The Optimum Life of a Patent: Reply. American Economic Review, 62(3), 428-431. OECD/IEA. (2004). Renewable Energy Market & Policy Trends in IEA Countries [Electronic Version]. OECD/IEA. (2008). Global Renewable Energy Policies and Measures. Retrieved October 10, 2008, from http://www.iea.org/textbase/pm/?mode=re Popp, D. (2002). Induced Innovation and Energy Prices. The American Economic Review, 92(1), 160-180. Popp, D. (2006). Internatinal innovation and diffusion or air pollution control technologies: the effects of NOx and SO2 regulation in the US, Japan and Germany. Journal of Environmental Economics and Management, 51, 46-71.
66
Reiche, D., & Bechberger, M. (2004). Policy differences in the promotion of renewable energies in the EU member states. Energy Policy(32), 843–849. Schumpeter, J. A. (1942). Capitalism, Socialism and Democracy. New York: Harper and Brothers. Spencer, M. (1984). Cost Reduction, Competition, and Industry Performance. Econometrica, 52(1), 101-121. World International Patent Organization. (2008). International Patent Classification. Retrieved October 18, 2008, 2008, from http://www.wipo.int/classifications/ipc/en/
Ҵ
67
APPE DIX Table 8.1: Change in Research as Feed-in-Tariff Changes deltan 0.00 0.00 0.00
deltam 5.00 5.00 5.00
Cr 20.00 20.00 20.00
Ct 10.00 10.00 10.00
alpham 200.00 200.00 200.00
betam 0.50 0.50 0.50
alphan 300.00 300.00 300.00
betan 2.00 2.00 2.00
sm 1.00 1.20 1.50
Phim 0.00 0.00 0.00
Phin 0.00 0.00 0.00
1/k 300 300 300
profit m 10472 12373 15254
profit n 9739 11606 14454
xm 33.12 38.76 47.08
xn 30.67 36.19 44.40
change xm
change xn
5.63 8.33
5.52 8.21
0.00 0.00 0.00 0.00
10.00 10.00 10.00 10.00
20.00 20.00 20.00 20.00
10.00 10.00 10.00 10.00
200.00 200.00 200.00 200.00
0.50 0.50 0.50 0.50
300.00 300.00 300.00 300.00
2.00 2.00 2.00 2.00
1.00 1.10 1.70 1.90
0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00
300 300 300 300
10472 11420 17186 19124
9050 9957 15581 17491
33.25 36.09 52.76 58.15
28.49 31.19 47.37 52.65
2.84 16.68 5.39
2.70 16.17 5.28
0.00 0.00 0.00 0.00
10.00 10.00 10.00 10.00
30.00 30.00 30.00 30.00
10.00 10.00 10.00 10.00
300.00 300.00 300.00 300.00
0.50 0.50 0.50 0.50
200.00 200.00 200.00 200.00
2.00 2.00 2.00 2.00
1.00 1.30 1.40 1.70
0.00 0.00 0.00 0.00
0.00 0.00 0.00 0.00
200 200 200 200
15139 21361 23453 29758
13006 19084 21141 27369
65.40 86.86 93.52 112.16
54.40 74.70 80.99 98.47
21.46 6.66 18.64
20.30 6.29 17.48
Table 8.2: Data
Austria
Output OECD Patents Change GDP US GDP US per Crude Oil Grant to Grants to Feed-in- Obligationsd, Targetsd,e Tax Land dummy b Gwha total Wind Billionsc Capitac Pricesa Producers companiesd,e tariffsd,e e incentivesd,e Green Areaf coastline a d,e f output Capacity certifacates d,e Gwha 1990 0 3845 0 165.396 21541.98 22.66 0 0 0 0 0 0 0 0 82444
Austria
1991
0
4148
0
0
173.382
22357.74
19.53
0
0
0
0
0
0
Austria
1992
0
4537
0
0
195.107
24883.82
18.54
0
0
0
0
0
0
Austria
1993
0
5473
0
0
189.709
23996.95
16.14
0
0
0
0
0
0
Austria
1994
0
7056
2
0
203.972
25701.69
15.63
0
0
0
0
0
0
Austria
1995
1
7349
1
1
238.55
30012.69
17.06
0
0
0
0
0
0
Austria
1996
5
8380
0
9
234.234
29429.96
20.67
0
0
0
0
0
Austria
1997
20
10721
1
9
207.126
25994.61
18.88
0
0
0
0
0
Country
Time
0 82444 0 82444
0
0 82444 0 82444
0 0
0
0 82444 0 82444
0
0 82444
0
0 0 0
68
Austria
1998
45
14425
0
8
212.439
26632.13
12.3
0
0
0
0
0
0
Austria
1999
51
19343
0
8
211.206
26426.19
17.32
0
0
0
0
0
0
Austria
2000
67
28551
2
19
191.761
23935.5
27.89
0
0
0
0
1
0
Austria
2001
172
34632
4
15
190.319
23662.5
23.92
0
1
1
0
1
0
Austria
2002
203
48335
0
64
206.684
25567.66
24.3
0
1
1
0
1
0
Austria
2003
366
58438
2
210
252.516
31106.68
28.39
0
1
1
0
1
0
Austria
2004
924
76864
1
217
289.419
35403.98
36.53
0
1
1
0
1
0
Austria
2005
1328
93657
2
267
304.529
36987.45
51.74
1
1
1
0
1
0
Austria
2006
1722
116182
0
142
323.071
39098.61
62.77
1
1
1
0
1
0
Belgium
1990
7
3845
1
197.859
19831.6
22.66
0
0
0
0
0
0
Belgium
1991
8
4148
0
0
202.936
20263.23
19.53
0
0
0
0
0
0
Belgium
1992
9
4537
0
0
225.577
22434.17
18.54
0
0
0
0
0
1
Belgium
1993
8
5473
0
0
216.297
21428.12
16.14
0
1
0
0
0
1
Belgium
1994
9
7056
0
0
236.116
23317.53
15.63
0
1
0
0
0
1
Belgium
1995
9
7349
0
0
277.151
27324.34
17.06
0
1
1
0
0
1
Belgium
1996
8
8380
0
0
275.708
27109.89
20.67
0
1
1
0
0
1
Belgium
1997
8
10721
0
0
249.639
24493.6
18.88
0
1
1
0
0
1
Belgium
1998
11
14425
0
1
255.56
25020.6
12.3
0
1
1
0
0
1
Belgium
1999
13
19343
2
4
253.851
24792.54
17.32
0
1
1
0
0
1
Belgium
2000
16
28551
6
4
232.653
22669.1
27.89
0
1
1
0
0
1
Belgium
2001
37
34632
3
12
232.105
22512.64
23.92
0
1
1
0
0
1
Belgium
2002
57
48335
3
5
252.714
24405.05
24.3
0
1
1
1
0
1
Belgium
2003
88
58438
3
36
310.526
29869.79
28.39
0
1
1
1
0
1
Belgium
2004
142
76864
0
29
359.3
34395.94
36.53
0
1
0
1
0
1
Belgium
2005
227
93657
1
71
376.399
35809.98
51.74
0
1
0
1
0
1
Belgium
2006
366
116182
0
45
398.192
37618.56
62.77
0
1
0
1
0
Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic
1990
0
3845
0
52.128
5058.64
22.66
0
0
0
0
0
۞
0 82444 0 82444
0
0 82444 0 82444
0
0 82444 0 82444
0
0 82444 0 82444
0
0 82444 0 30278
0
0 30278 0 30278
1
0 30278 0 30278
1
0 30278 0 30278
1
0 30278 0 30278
1
0 30278 0 30278
1
0 30278 0 30278
1
0 30278 0 30278
1 1
1
0 30278 0 30278
0
77276
0
77276
0
77276
0
77276
0
77276
0
77276
0
0 0 0 0 1 1 1 1 1 1 1 1 1
0 1991
0
4148
0
0
27.18
2636.3
19.53
0
0
0
0
0
0 0
1992
0
4537
0
0
31.898
3090.94
18.54
0
0
0
0
0
0 0
1993
0
5473
0
0
36.651
3551.5
16.14
0
0
0
0
0
0 0
1994
0
7056
0
0
42.538
4117.9
15.63
0
0
0
0
0
0 0
1995
0
7349
0
0
55.256
5348.65
17.06
0
0
0
0
0
0 0
69
Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic Czech Republic Denmark
1996
0
8380
0
0
1990
610
3845
0
Denmark
1991
740
4148
1
70
Denmark
1992
915
4537
0
Denmark
1993
1034
5473
1
Denmark
1994
1137
7056
Denmark
1995
1177
Denmark
1996
Denmark
62.011
6011.54
20.67
0
0
0
0
0
0
77276
0
77276
0
77276
0
77276
0
77276
0
77276
0
77276
0
77276
0
77276
0
77276
0
77276
0
0 42394 0 42394
1
0 42394 0 42394
1
0 42394 0 42394
1
0 42394 0 42394
1
0 42394 0 42394
1
1 42394 1 42394
1
1 42394 1 42394
1
1 42394 1 42394
1
1 42394 0 304473
1
0 1997
0
10721
0
0
57.135
5545.14
18.88
0
0
0
0
0
0 0
1998
0
14425
0
1
61.847
6033.68
12.3
0
0
0
0
0
0 0
1999
0
19343
0
0
60.192
5880.87
17.32
0
0
0
0
0
0 0
2000
0
28551
0
0
56.717
5548.48
27.89
0
0
0
0
0
0 0
2001
0
34632
0
0
61.843
6077.48
23.92
0
0
0
0
0
0 0
2002
2
48335
0
5
75.276
7401.06
24.3
0
0
0
1
0
0 0
2003
4
58438
0
5
91.358
8974.9
28.39
0
0
1
1
0
0 0
2004
10
76864
0
5
109.525
10742.47
36.53
0
0
1
1
1
0 0
2005
21
93657
1
13
124.549
12175.43
51.74
0
0
1
1
1
0 0
2006
49
116182
0
15
142.313
13863.89
62.77
0
0
1
0
1
0 0
136.174
26490.38
22.66
1
0
1
0
0
1
137.189
26618
19.53
1
0
1
0
0
1
45
150.54
29109.62
18.54
1
0
1
0
0
1
33
140.835
27141.09
16.14
1
0
1
0
0
1
5
41
153.901
29559.3
15.63
1
0
1
0
0
1
7349
2
84
182.179
34810.19
17.06
1
0
1
0
0
1
1227
8380
3
226
184.481
35052.38
20.67
1
0
1
0
0
1
1997
1934
10721
2
288
170.642
32287.96
18.88
1
0
1
0
0
1
Denmark
1998
2820
14425
10
313
173.902 ۞ 32783.81
12.3
1
0
1
1
0
1
Denmark
1999
3029
19343
15
316
174.172
32726.84
17.32
1
0
1
1
0
1
Denmark
2000
4241
28551
10
633
160.533
30065.24
27.89
1
0
1
1
0
1
Denmark
2001
4306
34632
8
106
160.583
29967.94
23.92
1
0
1
1
0
1
Denmark
2002
4877
48335
17
394
174.42
32444.25
24.3
0
0
1
1
0
1
Denmark
2003
5561
58438
23
225
212.981
39506.68
28.39
0
0
1
0
0
1
Denmark
2004
6583
76864
23
8
244.983
45329.46
36.53
0
0
1
0
0
1
Denmark
2005
6614
93657
23
4
258.58
47717.25
51.74
0
0
1
0
0
1
Denmark
2006
6108
116182
11
6
276.283
50815.38
62.77
0
0
1
0
0
1
Finland
1990
0
3845
0
139.83
28042.04
22.66
0
0
0
0
0
0
1 1 1 1 1 1 1 1 1
70
Finland
1991
0
4148
0
1
126.422
25215.14
19.53
0
0
0
0
0
0
Finland
1992
2
4537
1
0
110.81
21977.51
18.54
0
0
0
0
0
0
Finland
1993
4
5473
2
4
87.421
17254.83
16.14
0
0
0
0
1
0
Finland
1994
7
7056
0
0
100.714
19793.1
15.63
0
0
0
0
1
0
Finland
1995
11
7349
0
1
130.75
25598.21
17.06
0
0
0
0
1
0
Finland
1996
11
8380
0
1
128.525
25080.17
20.67
0
0
0
0
1
0
Finland
1997
17
10721
1
5
123.428
24014.02
18.88
0
0
0
0
1
1
Finland
1998
23
14425
0
5
130.466
25315.98
12.3
1
0
0
0
1
1
Finland
1999
49
19343
1
21
130.948
25350.7
17.32
1
0
0
0
1
1
Finland
2000
78
28551
0
0
122.222
23612.3
27.89
1
0
0
0
1
1
Finland
2001
70
34632
0
1
125.269
24145.89
23.92
1
0
0
0
1
1
Finland
2002
64
48335
0
4
135.972
26145.42
24.3
1
0
0
0
1
1
Finland
2003
93
58438
1
9
165.031
31657.45
28.39
1
0
0
0
1
1
Finland
2004
120
76864
2
30
189.411
36228.95
36.53
1
0
0
0
1
1
Finland
2005
170
93657
3
0
196.001
37361.28
51.74
1
0
0
0
1
1
Finland
2006
156
116182
4
4
209.745
39827.97
62.77
1
0
0
0
1
1
France
1990
0
3845
1
1248.49
22015.74
22.66
0
0
0
0
0
1
France
1991
0
4148
0
1
1249.15
21924.26
19.53
0
0
0
0
0
1
France
1992
0
4537
0
0
1373.98
24003.98
18.54
0
0
0
0
0
1
France
1993
2
5473
0
2
1292.04
22483.14
16.14
0
0
0
0
0
1
France
1994
5
7056
0
0
1366.08
23692.46
15.63
0
0
0
0
0
1
France
1995
5
7349
1
0
1572.3
27181.59
17.06
1
0
0
0
0
1
France
1996
7
8380
3
3
1574.52
27134.71
20.67
1
0
1
0
0
1
France
1997
11
10721
3
1
1425.86
24496.07
18.88
1
0
1
0
0
1
France
1998
20
14425
0
8
1474.31
25245.93
12.3
1
0
1
0
0
1
France
1999
37
19343
3
3
1458.32
24854.93
17.32
1
0
1
0
0
1
France
2000
77
28551
2
39
1333.2
22577.69
27.89
1
0
1
0
0
1
France
2001
131
34632
5
26
1341.06
22556.05
23.92
1
0
1
0
0
1
France
2002
269
48335
4
50
1463.7
24450.75
24.3
1
0
1
0
0
1
France
2003
391
58438
4
89
1805.18
29954.43
28.39
1
0
1
0
0
1
France
2004
596
76864
1
141
2061.09
33987.16
36.53
1
0
1
0
0
1
France
2005
963
93657
3
360
2147.48
35206.99
51.74
1
0
1
0
0
1
France
2006
2150
116182
1
665
2271.26
37019.84
62.77
1
0
1
0
0
1
Germany
1990
71
3845
1
1547.03
19592.74
22.66
1
1
1
0
0
Germany
1991
215
4148
4
1815.06
22692.8
19.53
1
1
1
0
0
62
۞
0 304473 0 304473
1
0 304473 0 304473
1
0 304473 0 304473
1
0 304473 0 304473
1
0 304473 0 304473
1
0 304473 0 304473
1
0 304473 0 304473
1
0 304473 0 304473
1
0 640053 0 640053
1
0 640053 0 640053
1
0 640053 0 640053
1
0 640053 0 640053
1
0 640053 0 640053
1
0 640053 0 640053
1
0 640053 0 640053
1
0 640053 0 640053
1
0 640053 0 349223
1
0 0
0 349223
1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
71
Germany
1992
291
4537
1
73
2066.73
25523.16
18.54
1
1
1
0
0
0
Germany
1993
674
5473
4
151
2005.56
24657.05
16.14
1
1
1
0
0
0
Germany
1994
1428
7056
6
309
2151.03
26380.45
15.63
1
1
1
0
0
0
Germany
1995
1712
7349
3
494
2524.95
30860.75
17.06
1
1
1
0
0
0
Germany
1996
2078
8380
8
427
2439.35
29743.72
20.67
1
1
1
0
0
0
Germany
1997
3034
10721
13
402
2163.23
26362.45
18.88
1
1
1
0
0
0
Germany
1998
4593
14425
23
706
2187.48
26664.6
12.3
1
1
1
0
0
0
Germany
1999
5528
19343
34
1466
2146.43
26123.92
17.32
1
1
1
0
0
0
Germany
2000
9352
28551
46
1957
1905.8
23168.07
27.89
1
1
1
1
0
0
Germany
2001 10456
34632
76
2659
1892.6
22957.16
23.92
1
1
1
1
0
0
Germany
2002 15856
48335
45
3247
2024.06
24523.16
24.3
1
1
1
1
0
0
Germany
2003 18859
58438
69
2608
2446.89
29647.83
28.39
1
1
1
1
0
0
Germany
2004 25509
76864
62
2020
2748.82
33318.7
36.53
1
1
1
1
0
0
Germany
2005 27229
93657
50
1799
2794.48
33897.93
51.74
1
1
1
1
0
0
Germany
2006 30710
116182
40
2194
2914.99
35422.06
62.77
1
1
1
1
0
0
Greece
1990
2
3845
1
92.195
9073.82
22.66
0
0
0
0
0
0
Greece
1991
2
4148
0
0
99.422
9705.08
19.53
0
0
0
0
0
0
Greece
1992
8
4537
0
15
109.556
10591.03
18.54
0
0
0
0
0
0
Greece
1993
47
5473
1
11
102.608
9816.3
16.14
0
0
0
0
0
0
Greece
1994
37
7056
0
0
109.824
10400.74
15.63
1
0
1
0
0
0
Greece
1995
34
7349
0
0
128.895
12096.48
17.06
1
1
1
0
0
0
Greece
1996
38
8380
0
0
136.273
12688.9
20.67
1
1
1
0
0
0
Greece
1997
36
10721
0
0
133.128
12311.98
18.88
1
1
1
0
0
0
Greece
1998
70
14425
0
11
133.869
12308.92
12.3
1
1
1
0
0
0
Greece
1999
162
19343
0
71
137.829
12610.95
17.32
1
1
1
0
0
0
Greece
2000
451
28551
0
117
127.604
11627.2
27.89
1
1
1
0
0
0
Greece
2001
756
34632
1
44
130.994
11896.61
23.92
1
1
1
0
0
0
Greece
2002
651
48335
0
17
148.827
13482.68
24.3
1
1
1
0
0
0
Greece
2003
1021
58438
2
84
193.663
17510.59
28.39
1
1
1
0
0
0
Greece
2004
1121
76864
0
99
230.291
20785.83
36.53
1
1
1
0
0
0
Greece
2005
1266
93657
0
21
247.418
22290.45
51.74
1
1
1
0
0
0
Greece
2006
1699
116182
1
258
268.69
24157.31
62.77
1
1
1
0
0
0
Ireland
1990
0
3845
0
47.753
13621.19
22.66
0
0
0
0
0
1
Ireland
1991
0
4148
0
0
48.4
13727.88
19.53
0
0
0
0
0
Ireland
1992
5
4537
0
6
54.415
15308.83
18.54
0
0
0
0
0
۞
0 349223 0 349223
1
0 349223 0 349223
1
0 349223 0 349223
1
0 349223 0 349223
1
0 349223 0 349223
1
0 349223 0 349223
1
0 349223 0 349223
1
0 349223 0 130800
1
0 130800 0 130800
1
0 130800 0 130800
1
0 130800 0 130800
1
0 130800 0 130800
1
0 130800 0 130800
1
0 130800 0 130800
1
0 130800 0 130800
1
0 130800 0 130800
1
0 68890 0 68890
1
1 1
0 68890
1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
72
Ireland
1993
15
5473
0
0
50.421
14107.37
16.14
0
0
0
0
0
1
Ireland
1994
19
7056
0
0
55.327
15429.05
15.63
0
0
0
0
0
1
Ireland
1995
16
7349
1
0
67.059
18620.83
17.06
0
0
0
0
0
1
Ireland
1996
14
8380
0
0
74.013
20411.26
20.67
0
0
0
0
0
1
Ireland
1997
50
10721
0
51
81.26
22176.01
18.88
0
0
0
0
0
1
Ireland
1998
169
14425
0
5
88.227
23825.09
12.3
0
0
0
0
0
1
Ireland
1999
187
19343
0
8
96.548
25803.86
17.32
0
0
0
0
1
1
Ireland
2000
244
28551
1
45
96.879
25565.25
27.89
0
0
0
0
1
1
Ireland
2001
334
34632
0
8
104.778
27234.97
23.92
0
0
0
0
1
1
Ireland
2002
388
48335
0
13
122.954
31388.22
24.3
0
0
0
0
1
1
Ireland
2003
454
58438
0
74
157.685
39620.27
28.39
0
0
0
0
1
1
Ireland
2004
655
76864
0
131
185.221
45787.88
36.53
0
0
0
0
1
1
Ireland
2005
1112
93657
0
153
202.022
48870.67
51.74
0
0
1
0
1
1
Ireland
2006
1622
116182
0
252
222.609
52504.7
62.77
0
0
1
0
1
1
Italy
1990
2
3845
1
1135.54
20029.19
22.66
0
0
0
0
0
0
Italy
1991
3
4148
0
1
1198.99
21129.68
19.53
1
0
0
0
0
0
Italy
1992
2
4537
0
3
1271.91
22403.3
18.54
1
0
1
0
0
0
Italy
1993
4
5473
0
11
1022.66
17997.61
16.14
1
0
1
0
0
0
Italy
1994
7
7056
1
3
1054.9
18557.95
15.63
1
0
1
0
0
0
Italy
1995
9
7349
0
1
1126.63
19819.03
17.06
1
0
1
0
0
0
Italy
1996
33
8380
1
12
1259.95
22164.1
20.67
0
0
0
0
0
0
Italy
1997
118
10721
1
85
1193.62
20985.09
18.88
0
0
0
0
0
0
Italy
1998
231
14425
2
45
1218.67
21433.7
12.3
0
0
0
0
0
0
Italy
1999
403
19343
0
68
1202.4
21129.55
17.32
0
0
0
1
0
0
Italy
2000
563
28551
1
131
1100.56
19293.4
27.89
0
0
0
1
0
0
Italy
2001
1179
34632
2
301
1118.32
19541.11
23.92
0
0
0
1
0
0
Italy
2002
1404
48335
8
116
1223.24
21317.51
24.3
0
0
0
1
0
0
Italy
2003
1458
58438
3
94
1510.06
26308.26
28.39
0
0
0
1
0
0
Italy
2004
1847
76864
3
253
1730.1
30119
36.53
0
0
0
1
0
0
Italy
2005
2344
93657
3
508
1779.41
30638.87
51.74
0
0
0
1
0
0
Italy
2006
2971
116182
2
267
1858.34
31801.63
62.77
0
0
0
1
0
0
Netherlands
1990
56
3845
1
295.46
19760.56
22.66
0
0
0
0
0
0
Netherlands
1991
88
4148
0
33
303.462
20136.83
19.53
0
0
0
0
0
0
Netherlands
1992
147
4537
1
18
334.654
22119.79
18.54
0
0
0
0
0
Netherlands
1993
174
5473
1
30
324.39
21286.59
16.14
0
0
0
0
0
۞
0 68890 0 68890
1
0 68890 0 68890
1
0 68890 0 68890
1
1 68890 1 68890
1
1 68890 1 68890
1
1 68890 1 68890
1
1 68890 1 68890
1
0 294020 0 294020
1
0 294020 0 294020
1
0 294020 0 294020
1
0 294020 0 294020
1
0 294020 1 294020
1
1 294020 1 294020
1
1 294020 1 294020
1
1 294020 1 294020
1
1 294020 0 33883
1
0 33883 0 33883
1
0 0
0 33883
1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
73
Netherlands
1994
238
7056
0
21
348.911
22742.86
15.63
0
0
0
0
0
0
Netherlands
1995
317
7349
0
98
419.348
27187.8
17.06
0
0
0
0
0
1
Netherlands
1996
437
8380
2
46
418.106
26985.21
20.67
0
0
0
0
0
1
Netherlands
1997
475
10721
2
28
387.013
24860.92
18.88
0
1
0
0
0
1
Netherlands
1998
640
14425
1
39
403.202
25756.79
12.3
0
1
0
0
0
1
Netherlands
1999
645
19343
4
47
411.997
26141.54
17.32
0
1
0
0
0
1
Netherlands
2000
829
28551
6
37
386.204
24250.65
27.89
0
1
0
0
0
1
Netherlands
2001
825
34632
3
38
400.998
24990.27
23.92
0
1
0
0
0
1
Netherlands
2002
946
48335
4
185
439.357
27206.57
24.3
0
1
0
0
0
1
Netherlands
2003
1318
58438
6
236
539.343
33240.83
28.39
0
0
1
0
0
1
Netherlands
2004
1867
76864
2
167
610.691
37520.24
36.53
0
0
1
0
0
1
Netherlands
2005
2067
93657
1
151
639.579
39210.61
51.74
0
0
1
0
0
1
Netherlands
2006
2733
116182
8
334
677.961
41476.76
62.77
0
0
1
0
0
1
Norway
1990
0
3845
0
117.865
27763.99
22.66
0
0
0
0
0
0
Norway
1991
0
4148
0
0
120.068
28143.58
19.53
0
0
0
0
0
0
Norway
1992
3
4537
0
0
128.591
29965.6
18.54
0
0
0
0
0
0
Norway
1993
3
5473
1
0
118.284
27404.05
16.14
0
0
0
0
0
0
Norway
1994
9
7056
0
0
124.737
28732.29
15.63
0
0
0
0
0
0
Norway
1995
10
7349
0
3
149.007
34149.84
17.06
0
0
0
0
0
0
Norway
1996
9
8380
0
1
160.173
36521.51
20.67
0
0
0
0
0
0
Norway
1997
10
10721
0
0
158.55
35955.94
18.88
0
0
0
0
0
0
Norway
1998
11
14425
1
0
151.156
34075.61
12.3
0
0
0
0
0
0
Norway
1999
25
19343
1
10
159.093
35619.4
17.32
0
0
1
0
0
1
Norway
2000
31
28551
2
-1
168.671
37520.11
27.89
0
0
1
0
0
1
Norway
2001
27
34632
0
0
170.982
37840.31
23.92
1
0
1
0
0
1
Norway
2002
75
48335
2
84
193.175
42525.73
24.3
1
0
1
0
0
1
Norway
2003
218
58438
3
0
225.307
49316.72
28.39
1
0
1
0
0
0
Norway
2004
252
76864
1
55
258.986
56344.18
36.53
1
0
0
0
1
0
Norway
2005
506
93657
5
118
302.175
65604.6
51.74
1
0
0
0
1
0
Norway
2006
673
116182
0
14
337.426
72768.13
62.77
1
0
0
0
1
0
Portugal
1990
1
3845
0
75.967
7599.78
22.66
0
0
0
0
0
0
Portugal
1991
1
4148
0
0
85.975
8623.01
19.53
0
0
0
0
0
0
Portugal
1992
4
4537
0
2
103.394
10375.38
18.54
0
0
0
0
0
0
Portugal
1993
11
5473
0
5
90.982
9121.33
16.14
0
0
0
0
0
Portugal
1994
17
7056
0
0
95.335
9542.48
15.63
0
0
0
0
0
۞
0 33883 1 33883
1
1 33883 1 33883
1
1 33883 1 33883
1
1 33883 1 33883
1
1 33883 1 33883
1
1 33883 1 33883
1
1 33883 0 212460
1
0 212460 0 212460
1
0 212460 0 212460
1
0 212460 0 212460
1
0 212460 0 212460
1
0 212460 0 212460
1
0 212460 0 212460
1
1 212460 1 212460
1
1 212460 1 212460
1
0 91951 0 91951
1
0 91951 0 91951
1
0 0
0 91951
1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
74
Portugal
1995
16
7349
0
0
113.017
11281.89
17.06
0
0
0
0
0
0
Portugal
1996
21
8380
0
10
117.658
11715.22
20.67
0
0
0
0
0
0
Portugal
1997
38
10721
0
11
112.134
11132.65
18.88
0
0
0
0
0
0
Portugal
1998
89
14425
0
19
118.711
11742.31
12.3
0
0
0
0
0
0
Portugal
1999
123
19343
0
9
121.823
12003.54
17.32
0
0
0
0
0
0
Portugal
2000
168
28551
0
26
112.98
11081.91
27.89
0
1
0
0
0
0
Portugal
2001
256
34632
0
42
115.812
11291.37
23.92
0
1
1
0
0
0
Portugal
2002
362
48335
0
65
127.906
12382.79
24.3
0
1
1
0
0
0
Portugal
2003
496
58438
0
78
156.712
15057.65
28.39
0
1
1
0
0
0
Portugal
2004
816
76864
0
285
179.195
17107.4
36.53
0
1
1
0
0
0
Portugal
2005
1773
93657
0
511
185.771
17643.35
51.74
0
1
1
0
0
0
Portugal
2006
2925
116182
0
617
195.186
18466.74
62.77
0
1
1
0
0
0
Spain
1990
14
3845
0
520.709
13407.9
22.66
0
0
1
0
0
0
Spain
1991
15
4148
0
1
560.796
14401.9
19.53
0
0
1
0
0
0
Spain
1992
103
4537
0
30
613.016
15691.09
18.54
0
0
1
0
0
0
Spain
1993
116
5473
0
1
514.949
13140
16.14
0
0
1
0
0
0
Spain
1994
175
7056
1
7
516.718
13149.72
15.63
0
0
1
0
0
0
Spain
1995
270
7349
0
57
597.278
15164.36
17.06
0
0
1
0
0
0
Spain
1996
364
8380
1
129
622.65
15772.03
20.67
0
0
1
0
0
0
Spain
1997
742
10721
0
193
573.376
14485.63
18.88
0
0
1
1
0
0
Spain
1998
1352
14425
0
428
601.625
15146.23
12.3
0
0
1
1
0
0
Spain
1999
2744
19343
7
765
618.691
15495.85
17.32
0
0
1
1
0
0
Spain
2000
4727
28551
1
593
582.377
14464.24
27.89
0
0
1
1
0
0
Spain
2001
6759
34632
5
1191
609.631
14971.13
23.92
0
0
1
1
0
1
Spain
2002
9342
48335
4
1494
688.676
16669.31
24.3
0
0
1
1
0
0
Spain
2003 12075
58438
4
1054
885.358
21077.65
28.39
0
0
1
1
0
0
Spain
2004 15700
76864
16
2372
1045.67
24493.52
36.53
0
0
1
1
0
0
Spain
2005 21176
93657
19
1601
1132.13
26087.08
51.74
0
0
1
1
0
0
Spain
2006 23040
116182
3
1818
1233.43
27989.03
62.77
0
0
1
1
0
0
Sweden
1990
6
3845
1
242.848
28268.9
22.66
0
0
0
0
0
0
Sweden
1991
13
4148
0
4
256.344
29655.26
19.53
0
0
0
0
0
0
Sweden
1992
31
4537
0
8
266.224
30628.56
18.54
0
0
0
0
0
0
Sweden
1993
48
5473
1
9
202.589
23166.01
16.14
0
0
0
0
0
0
Sweden
1994
72
7056
0
11
217.844
24709.03
15.63
0
0
0
0
0
Sweden
1995
99
7349
1
27
254.105
28753.02
17.06
0
0
0
0
0
۞
0 91951 0 91951
1
0 91951 0 91951
1
0 91951 0 91951
1
0 91951 0 91951
1
0 91951 0 91951
1
0 91951 0 91951
1
0 499542 0 499542
1
0 499542 0 499542
1
0 499542 0 499542
1
0 499542 0 499542
1
0 499542 0 499542
1
0 499542 0 499542
1
0 499542 0 499542
1
0 499542 0 499542
1
0 499542 0 410934
1
0 410934 0 410934
1
0 410934 0 410934
1
1 1
0 410934
1
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
75
Sweden
1996
144
8380
0
38
276.23
31231.81
20.67
0
0
0
0
0
1
Sweden
1997
203
10721
2
18
252.614
28551.6
18.88
1
0
0
0
0
1
Sweden
1998
317
14425
0
51
253.154
28590.99
12.3
1
0
0
0
0
1
Sweden
1999
358
19343
6
22
257.225
29027.48
17.32
1
0
0
0
0
1
Sweden
2000
457
28551
0
13
246.372
27735.91
27.89
1
0
0
0
0
1
Sweden
2001
482
34632
7
86
225.546
25316.3
23.92
1
0
0
0
0
1
Sweden
2002
608
48335
2
62
249.378
27892.13
24.3
1
0
0
0
0
1
Sweden
2003
679
58438
0
42
311.763
34734.21
28.39
1
0
0
0
0
1
Sweden
2004
850
76864
1
53
357.721
39696.58
36.53
1
0
0
0
0
1
Sweden
2005
936
93657
2
41
367.162
40580.52
51.74
1
0
0
0
0
1
Sweden
2006
987
116182
0
23
393.606
43190.45
62.77
1
0
0
0
0
1
Switzerland
1990
0
3845
0
239.332
35452.95
22.66
0
0
1
0
0
0
Switzerland
1991
0
4148
0
0
241.971
35361.64
19.53
0
0
1
0
0
0
Switzerland
1992
0
4537
0
0
251.791
36449.46
18.54
0
0
1
0
0
0
Switzerland
1993
0
5473
0
0
244.198
35042.82
16.14
0
0
1
0
0
0
Switzerland
1994
0
7056
0
0
270.913
38597.06
15.63
0
0
1
0
0
0
Switzerland
1995
0
7349
0
0
316.418
44803.52
17.06
0
0
1
0
0
0
Switzerland
1996
1
8380
0
2
305.054
43078.48
20.67
0
0
1
0
0
0
Switzerland
1997
2
10721
0
0
264.703
37300.71
18.88
0
0
1
0
0
0
Switzerland
1998
3
14425
0
1
273.103
38338.08
12.3
0
0
1
0
0
0
Switzerland
1999
3
19343
1
0
268.572
37486.74
17.32
0
0
1
0
0
0
Switzerland
2000
3
28551
0
0
250.195
34802
27.89
0
0
1
0
0
0
Switzerland
2001
4
34632
0
2
255.217
35392.68
23.92
0
0
1
0
0
0
Switzerland
2002
5
48335
2
0
279.515
38659
24.3
0
0
1
0
0
0
Switzerland
2003
5
58438
2
0
325.27
44886.22
28.39
0
0
1
0
0
0
Switzerland
2004
6
76864
4
4
363.428
50051.87
36.53
0
0
1
0
0
0
Switzerland
2005
8
93657
1
3
372.565
51218.67
51.74
0
0
1
0
0
0
Switzerland
2006
15
116182
1
0
388.679
53340.47
62.77
0
0
1
0
0
United Kingdom United Kingdom United Kingdom United Kingdom United
1990
9
3845
2
1017.79
17782.06
22.66
0
0
0
1
0
۞
0 410934 0 410934
1
0 410934 0 410934
1
0 410934 0 410934
1
0 410934 1 410934
1
1 410934 1 410934
1
1 410934 0 39770
1
0 39770 0 39770
0
0 39770 0 39770
0
0 39770 0 39770
0
0 39770 0 39770
0
0 39770 0 39770
0
0 39770 0 39770
0
0 39770 0 39770
0 0
0
0 39770 0 39770
0
241590
1
241590
1
241590
1
241590
1
0 241590
1
1 1 1 1 1 0 0 0 0 0 0 0 0 0
0 1991
11
4148
0
4
1059.26
18441.42
19.53
0
0
0
1
0
0 0
1992
40
4537
0
36
1098.3
19072.63
18.54
0
0
0
1
0
0 0
1993
218
5473
4
81
982.615
17025.6
16.14
0
0
0
1
0
0 0
1994
342
7056
0
22
1061.38
18343.34
15.63
0
0
0
1
0
0
76
Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom United Kingdom
1995
391
7349
2
47
1157.44
19947.19
17.06
0
0
0
1
0
0
241590
1
241590
1
241590
1
241590
1
241590
1
241590
1
241590
1
241590
1
241590
1
241590
1
241590
1
241590
1
0 1996
488
8380
1
38
1220.85
20989.85
20.67
0
0
0
1
0
0 0
1997
667
10721
1
84
1359.44
23312.43
18.88
0
0
0
1
0
0 0
1998
877
14425
0
9
1456.16
24902.18
12.3
0
0
0
1
0
0 0
1999
850
19343
0
26
1502.89
25609.93
17.32
0
0
0
0
0
0 0
2000
947
28551
2
55
1480.53
25142.25
27.89
0
0
0
1
0
0 1
2001
965
34632
2
15
1471.4
24891.24
23.92
0
0
0
1
0
1 1
2002
1256
48335
5
107
1614.7
27219.23
24.3
1
0
0
1
0
1 1
2003
1285
58438
10
208
1862.77
31278.68
28.39
1
0
0
1
0
1 1
2004
1935
76864
5
191
2199.25
36755.86
36.53
1
0
0
1
0
1 1
2005
2904
93657
7
632
2280.06
37863.45
51.74
1
0
0
1
0
1 1
2006
4225
116182
3
Sources: a International Energy Agency, 2009 b European Patent Office, 2008 c International Monetary Fund, 2009 d OECD/IEA, 2004 e OECD/IEA, 2008 f Central Intelligence Agency, 2008
390
2435.7
40237.54
62.77
1
1
0
1
0
1 1
۞
77
Table 8.3: Descriptive Statistics Variable Output (GWh) World Output (GWh) GDP per capita GDP Crude Oil Prices Producer Grants Individual/Company Grants Feed in Tariff Obligation Targets Tax Incentives Green Certificates
Obs 289 289 289 289 289 289 289 289 289 289 289 289
mean 1358.384 31878.59 569.8027 25374.39 25.54529 0.314879 0.217993 0.456747 0.190311 0.121107 0.33218 0.17301
sd 3988.772 34032.4 660.6475 10993.25 13.02466 0.465273 0.413599 0.49899 0.393228 0.326818 0.471812 0.378912
78
