<%BANNER%>

International Transfer of Solar Technology

Permanent Link: http://ufdc.ufl.edu/UFE0045254/00001

Material Information

Title: International Transfer of Solar Technology
Physical Description: 1 online resource (99 p.)
Language: english
Creator: Phalin, Amanda J
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: citations -- diffusion -- environment -- intellectual -- international -- patents -- property -- quality -- solar -- technology -- transfer
Economics -- Dissertations, Academic -- UF
Genre: Economics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: This paper investigates the relationship between patent quality and the international transfer of solar technology. Using data from 84 countries, I also explore whether strengthening a country’s intellectual property rights (IPR) laws increases patent filings in this sector. This paper also examines patent filings as a measure of technology transfer, as well as the structure of the international solar market. I find a generally positive and statistically significant relationship between patent quality and the international transfer of solar technology. The analysis also shows that— contrary to other research —IPR laws alone generally have no effect or a negative effect on technology transfer in this sector when a quality measure is included. Finally, results demonstrate that climate affecting the intensity of sunlight alone does not determine solar technology inflows. Rather, infrastructure, IPR laws, and human capital combined with this indicator are important.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Amanda J Phalin.
Thesis: Thesis (Ph.D.)--University of Florida, 2013.
Local: Adviser: Dinopoulos, Elias.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045254:00001

Permanent Link: http://ufdc.ufl.edu/UFE0045254/00001

Material Information

Title: International Transfer of Solar Technology
Physical Description: 1 online resource (99 p.)
Language: english
Creator: Phalin, Amanda J
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2013

Subjects

Subjects / Keywords: citations -- diffusion -- environment -- intellectual -- international -- patents -- property -- quality -- solar -- technology -- transfer
Economics -- Dissertations, Academic -- UF
Genre: Economics thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: This paper investigates the relationship between patent quality and the international transfer of solar technology. Using data from 84 countries, I also explore whether strengthening a country’s intellectual property rights (IPR) laws increases patent filings in this sector. This paper also examines patent filings as a measure of technology transfer, as well as the structure of the international solar market. I find a generally positive and statistically significant relationship between patent quality and the international transfer of solar technology. The analysis also shows that— contrary to other research —IPR laws alone generally have no effect or a negative effect on technology transfer in this sector when a quality measure is included. Finally, results demonstrate that climate affecting the intensity of sunlight alone does not determine solar technology inflows. Rather, infrastructure, IPR laws, and human capital combined with this indicator are important.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Amanda J Phalin.
Thesis: Thesis (Ph.D.)--University of Florida, 2013.
Local: Adviser: Dinopoulos, Elias.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2013
System ID: UFE0045254:00001


This item has the following downloads:


Full Text

PAGE 1

INTERNATIONAL TRANSFER OF SOLAR TECHNOLOGY By AMANDA J. PHALIN A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2013 1

PAGE 2

2013 Amanda J. Phalin 2

PAGE 3

To my grand parents James Robert Spear and Margaret Spear whose love and support made this work possible 3

PAGE 4

ACKNOWLEDGMENTS I thank my husband, Benj amin Phalin, for his unfailing love and encouragement. I thank my chair, Dr. Elias Dinopoulos, for his guidance and support. I also thank Adam Narkiewicz for his assistance with data management, and the University of Florida's Center for International Busi ness Education & Research for funding that supported the datagathering process. 4

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS .................................................................................................. 4 LIST OF TABLES ............................................................................................................ 7 ABSTRACT ..................................................................................................................... 9 CHAPTER 1 INTRODUCTION .................................................................................................... 10 2 THE STRUCTURE OF THE INTERNATIONAL SOLAR MARKET ......................... 15 Types and Ca pacities of Solar Technologies .......................................................... 15 Degree of Concentration ......................................................................................... 17 Cost Structure ......................................................................................................... 19 Ease of Entry and Barriers to Entry ........................................................................ 20 Availability and Pricing of Substitutes ..................................................................... 21 Input Market ............................................................................................................ 22 Demand .................................................................................................................. 23 Projections .............................................................................................................. 24 3 LITERATURE REVIEW .......................................................................................... 28 4 PATENTS: A RELIABLE MEASURE OF INTERNATIONAL TECHNOLOGY TRANSFER? .......................................................................................................... 34 5 THEORETICAL AND ECONOMETRIC FRAMEWORK .......................................... 43 6 VARIABLES AND DATA ......................................................................................... 46 7 RESULTS: OECD COUNTRIES ............................................................................. 53 Full Dataset ............................................................................................................. 58 Robustness of Results ............................................................................................ 58 8 RESULTS: ALL COUNTRIES ................................................................................. 67 Trimmed Dataset: HighIncome Group ................................................................... 67 Trimmed Dataset: Lower Income Group ................................................................. 70 Full Dataset: High Income Group ........................................................................... 72 Ful l Dataset: Lower Income Group ......................................................................... 74 9 CONCLUSION ........................................................................................................ 83 5

PAGE 6

APPENDIX A COMPUTER PROGRAM USED TO GATHER DATA ............................................ 86 B RESULTS WITH SUN VARIABLE ONLY ............................................................... 88 REFERENCES .............................................................................................................. 90 BIOGRAPHICAL SKETCH ............................................................................................ 99 6

PAGE 7

LIST OF TABLES Table page 2 1 Solar PV operating capacity by country .............................................................. 27 2 2 Market shares of worlds top solar PV manufacturers ........................................ 27 7 1 OECD summary statistics, t rimmed .................................................................... 61 7 2 OECD summary statistics, f ull ............................................................................ 61 7 3 OECD n egative b inomial r esults, t rimmed .......................................................... 62 7 4 OECD n egative b inomial r esults, f ull ................................................................. 63 7 5 OECD b aseline OLS r esults ............................................................................... 64 7 6 OECD l og l inear r esults, t rimmed ....................................................................... 65 7 7 OECD l og l inear r esults, f ull ................................................................................ 66 8 1 Trimmed, h igh i ncome n ations summary statistics ............................................. 77 8 2 Trimmed, l ower i ncome n ations s ummary statistics ........................................... 7 7 8 3 Full, h igh i ncome n ations summary statistics ..................................................... 78 8 4 Full, l ower i ncome n ations summary statistics .................................................... 78 8 5 High i ncome n ations ........................................................................................... 79 8 6 High i ncome n ations w/ sun i nteraction t erms ..................................................... 80 8 7 Lower i ncome n ations ........................................................................................ 81 8 8 Lower i ncome n ations w/ sun i nteraction t erms ................................................... 82 B 1 High i ncome n ations w/ sun o nly ......................................................................... 88 B 2 Lower i ncome n ations w/ sun o nly ....................................................................... 89 7

PAGE 8

LIST OF FIGURES Figure page 7 1 Top 20 destination countries for U.S. solar patents, 1952 2011 ....................... 60 8

PAGE 9

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INTERNATIONAL TRANSFER OF SOLAR TECHNOLOGY By Amanda J. Phalin May 2013 Chair: Elias Dinopoulos Major: Economics This dissertation investigates the relationship between patent quality and the international transfer of solar technology. Using data from 84 countries I also explore whether strengthening a countrys intellectual property rights (IPR) laws increases patent filings in this sector. In addition, t his research examines patent filings as a measure of technology transfer as well as the structure of the international solar m arket. I find a generally positive and statistically significant relationship between patent quality and the international transfer of solar technology The analysis also shows that contrary to other researchIPR laws alone generally have no effect or a negative effect on technology transfer in this sector when a quality measure is included. Finally, results demonstrate that climate affecting the intensity of sunlight alone does not determine solar technology inflows. Rather, infrastructure, IPR laws, and h uman capital combined with this indicator are important. 9

PAGE 10

CHAPTER 1 INTRODUCTION Much discussion has recently been focused on the concept of international environmental technology transfer, and in particular the role of patents in both measuring and facilitating these flows. Key questions emerge: Which patents are likely to be filed abroad, and what characteristics do they share? What do countries with high patent inflows look like? Do patent flows behave differently depending on the sector being examin ed or income level of the country where patenting occurs ? In this dissertation, I investigate the relationship between patent quality and international technology transfer, specifically as it relates to solar technologies. Using patent flows as a measure o f technology diffusion, I hypothesize that solar patents of higher quality, as indicated by a weighted measure of forward citations, are more likely to be filed abroad than their lower quality counterparts. I also hypothesize that the strength of a country s intellectual property rights (IPR) laws will increase patent filings in this sector as well. This research adds to the literature in several key ways. First, I examine data disaggregated at the technology level. While much research has been done on the relationship between intellectual property and technology transfer at the aggregate level, there is an urgent need for increased availability of reliable and objective data on climate technologies, particularly on IPR related aspects (Latif and Maskus, et al., 2011) Other s cholars have previously pointed out the need to examine data in this area that is disaggregated by national income level and sector, including Kumar (1996), Lall (2003), and Basberg (1987). Unlike many previous studies, I conduct a dis aggregated analysis using a dataset of solar patents filed from the United States in 84 countries between 1952 and 2011. The results here will allow me either to confirm that solar 10

PAGE 11

patent flows behave in line with aggregate patent flows and with those in o ther sectors, or to explore why patenting patterns in this sector may be different. Second, I examine the structure of the international solar market. O utlining the buyers, sellers, and major national and business players in the industry can help better ex plain the results of the data analysis. Third, I examine to what extent patent filings represent technology diffusion via a literature review of the topic. To confirm that higher quality patents are more likely to be filed internationally is important beca use it indicates that only more valuable technology will potentially be used internationally. However, a patent filed abroadeven a highquality onedoes not necessarily mean that the technology embodied in that patent will be used in the country where it is filed, or used in a way that will spill over outside the company or research institution that files it Finally, this research offers a new perspective on the relationship between intellectual property rights and the international transfer of solar technologies by considering quality as an additional variable of interest, which has not been studied before to my knowledge. This issue is especially salient now, as world leaders have met recently at several climate c onferences to negotiate steps to curb global warming At the 2011 Durban conference, b oth developed and developing countries committed themselves to formulating a legally binding agreement to reduce climatechangecausing emissions and Kyoto Protocol policies were extended (WRI, 2011). They also made progress on providing financing for poorer nations to access climatechangemitigating technologies, and facilitating technology transfer is important in this context. Thus, solar technology 11

PAGE 12

transfer, an alternative energy likely to see more producti on as carbon emissions are reduced under Kyoto, is an important area of research. Quality is a fundamental aspect of the IPR technology transfer question. Theory predicts that higher quality technologies are more likely to be diffused internationally (Eato n and Kortum, 1994; Eaton and Kortum, 1995, Kortum and Lerner, 1997). How is quality best measured? For many years, beginning especially in the 1980s, scholars have used patent citations for this purpose. Researchers like citations because they provide a c lear path show ing how innovation moves between people, firms, industries, and countries. Citations can be indicative of a products quality because the more often a patent is cited, the higher the probability that it is a useful or valuable development. Quality measures are crucial because they are the only way to determine whether patents are important or of any value. In the past, researchers have used raw patent counts to measure technology transfer, but if those patents are of low quality or do not re present any real innovation, then it is not accurate to say that technology is actually being transferred. Finding a way to measure patent quality allows researchers to capture true flows of technology more accurately. I use a weighted measure of forward patent citations to gauge patent quality. I hypothesize that, per theory, patent quality should be positively correlated with more technology diffusion. I find a positive, statistically significant relationship between the two across a range of specificatio ns To my knowledge, this is the first study examining specifically the relationship between patent quality and the propensity to patent as it relates to solar technology. The results here are the first empirical confirmation that, when it comes to international technology transfer in the solar sector, quality matters. 12

PAGE 13

IPRs are also important for green technologies since this sector requires large initial R&D investments ; innovators need to be able to reap profits from their initial outlays to succeed (Lati f and Maskus, et al., 2011). In addition IPRs importance can vary across technologies ; some sectors that produce easily imitated products (e.g., pharmaceuticals) find IPR strength to be a vital requirement, while other, less imitable, industries (e.g., t raditional manufacturing) may be less concerned with protecting intellectual property I expect that because of the large R&D investment required and the advanced technology often needed to develop and produce solar energy products, a strong IPR system sho uld be positively correlated with patent flows. The only research, to my knowledge, examining the relationship between IPR laws and solar international technology transfer (ITT) (Dechezleprtre, 2011) found that stronger IPR laws have a positive, statistic ally significant effect on solar technology flows. However, I find the opposite result; my results show that when a variable controlling for patent quality is added to the analysis, IPR laws have a negative, statistically significant effect, or no effect a t all. In addition, these effects may differ depending on the level of economic development; in countries with strong IPR laws and high levels of economic development, robust patent protection may actually make it harder for technologies to be patented. This research offers an indication that the effect of IPR laws on technology flows of solar patents may be different for highly developed nations than when examining poor and rich nations together. The dissertation proceeds as follows: Section 2 explores the structure of the international solar market. Section 3 presents a brief review of the literature related to 13

PAGE 14

patents, patent quality and patent citations. Section 4 examines the reliability of patents as measures of technology transfer. Section 5 explains the theoretical underpinnings of the analysis as well as an econometric model drawn from the theoretical model. Secti on 6 details how the variables are constructed an d the sources of data. Section 7 presents and analy z es the empirical results for a subset of 16 OECD nations Section 8 presents and analy z es results for all 84 highand lower income nations .1 Finally, Section 6 offers concluding remarks. 1 I define income according to the World Bank. One exception is the inclusion of Zimbabwe, whose per capita GDP puts it in the low income category, according to the bank. 14

PAGE 15

CHAPTER 2 THE STRUCTURE OF THE INTERNATIONAL SOLAR MARKET The structure of the international solar market may help shed light on global technology flows in this area, and what drivers are likely to be significant in encouraging international diffusion of this type of technology. First, it is important to review the main types of solar technologies and their potential capacities. Then, several facets of the global solar market can be explored, including the degree of concentration in the industry; its cost structure; ease of and barriers to entry; availability and pricing of substitutes; the market for inputs; and the demand side of the market. Finally, the future of solar energy can be analyzed with projections of production and consumption. Types and Capacities of Solar Technologies Solar technology can be broadly divided into four types: collection, conce ntration, photovoltaics, and heating/cooling (Steiner, 2009). Solar collector technologies collect sunlight through such mechanisms as plates, troughs, towers, and dishes. Concentrated solar power, or CSP, uses mirrors or lenses to concentrate sunlight. Solar photovoltaics, or PV, refers to technology that converts sunlight into energy. Solar heating and cooling use either passive design (e.g., reflective roofs) or active design (e.g., solar powered AC units) to reduce the use of fossil fuels in heating and cooling buildings. There are also smaller subsectors in solar powered vehicles, but the technological development there is insignificant compared with the rest of the industry. Finally, nanotechnologies are also beginning to be used in the solar sector, but these developments are truly on the frontier of knowledge; therefore, little information about nanosolar advancements is currently available. The oldest form of solar technology is solar thermal power, a mode of solar collection that peaked in the 1970s. Since then, 15

PAGE 16

solar PV has grown rapidly, particularly in the 1990s (Steiner, 2009). Currently, solar PV and CSP constitute the most important subsectors of the industry. Worldwide, solar capacity has increased by more than 1,500 percent between 1992 and 2003 (WIPO, 2009). Germany, Japan, and the United States account for 85 percent of total capacity. Despite this expansion, solar energy comprises a very small percentage of total energy use globally, 0.02 percent (Sharma, 2011); fossil fuels provide almos t 80 percent of the worlds energy, with nuclear power coming in second at 13 percent (Byrne, 2010). Nonetheless, use of solar technologies is growing. For example, as of 2000, 1.1 million developing world homes used solar PV or lanterns; 10 million homes used solar water heating; and more than 25 countries had policies in place to regulate independent power production (Holm, 2005). By 2010, 70 million homes worldwide used solar water heating (UNEP, 2010), while about 3 million homes used small PV installat ions (Sawin, 2010). Overall, use has doubled every two years (Sharma, 2011). In particular, the PV sector has grown quickly, at 35 percent per year on average since 2000 according to some (Poullikkas, 2010). Other estimates are even more impressive, showi ng 60 percent average annual growth since 2000 (Sawin, 2010). Thanks to this growth, the industry earned almost $40 billion of revenue worldwide (Sharma, 2011). Investors are also noticing the solar industrys potential: 2010 saw $2.3 billion worth of vent ure capital and private equity investment, a compound annual growth rate of almost 60 percent between 2004 and 2010 (U.S. DOE, 2011). Total investments added up to almost $80 billion in 2010, with Germany comprising 45 percent of the total (Hopwood, 2011). (Table 21) 16

PAGE 17

Concentrated solar power (CSP) is also an important subsector, and capacity has been expanding there as well. Spain is the top installer of CSP technologies (just over 55 percent of the global total), followed by the United States (almost 39 percent) and Iran (5 percent) (US DOE, 2011). Degree of Concentration It is important to note that different solar technologies compete in different types of markets. For example, largescale PV and CSP projects may compete with gridconnected utilities, while smaller, off grid, standalone solar power installations more likely compete with diesel and other types of generators (Timilsina, 2011). Likewise, there are markets for both commercial and residential installations, as well as other consumer products ( toys and electronics) and government products (traffic lights, road signs). Markets and their incentives differ widely across the world. European and Asian nations tend to have very centralized policies and incentives, while the United States features some times overlapping federal, state, and local policies (Barker, 2011). As measured by patent filings, Japanese companies dominate solar technology development, including Canon, Sanyo Electric, Sharp, Matsushita Electric, and Kyocera (WIPO, 2009). Specifical ly, in the solar PV sector, 15 firms control 49 percent of the market (REN21, 2012). (Table 22) The fact that the majority of the worlds top solar PV firms are located in China reflects a fundamental shift in the market: While Europe remains the top cons umer of solar energy and products, production has shifted, and continues to shift, to Asia. In addition to Chinese companies, Taiwanese and Indian production are also expected to become more significant (REN21, 2012). 17

PAGE 18

Together, China and Taiwan produce alm ost half of all solar PV cells worldwide (Sharma, 2011), but Europe is also a large solar exporter (Groba, 2011). After China, the worlds top PV producers are Japan, Germany, Taiwan, and the U.S. In Chinas PV market, exports are extremely important, acco unting for 95 percent of total production. PV production growth in China has been astounding: The country produced onethird of the worlds solar cells in 2008 but currently produces almost 60 percent (Choudhury, 2012). China also provides 77 percent of total global solar water heater production. Chinese production is not very concentrated, with 900 PV manufacturers. Meanwhile, new policies (e.g., feed in tariffs) are expected to spur further development of both China and Indias domestic PV cell markets (P latzer, 2012). Overall, there were about 500 firms worldwide in the PV subsector in 2009 (Kirkegaard, 2010). More recently, others have estimated the existence of more than 1,000 firms globally (Platzer, 2012). In overall solar manufacturing, six of the t op10 companies are Chinese; European and Japanese producers have been pushed out of the top (Choudhury, 2012). In the area of concentrated solar power, the top two markets are the United States and Spain (OECD/IEA, 2011). Outside the main markets in Asia, Europe, and the U.S., other emerging economies are seeing growth in their CSP industries, helped along by governments, NGOs, and multinational organizations, as well as favorable weather conditions. These upand coming CSP countries include Chile, India, Morocco, Saudi Arabia, South Africa, and the UAE ( Gonzlez 2012). The CSP subsector is more concentrated than the PV market. The CSP industry is marked by vertical integration, with companies participating in everything from R&D to production and operati on of facilities (REN21, 2012). Key companies involved in this 18

PAGE 19

sector include Abengoa (Spain), BrightSource Energy (United States), GE (United States), and AREVA (France). In the solar heating sector, five of the largest Chinese firms1 have played major roles in the market. In Europe, meanwhile, mergers and acquisitions in the face of the recent economic downturn have further consolidated the market (REN21, 2012). Cost Structure Costs play an integral role in how the international solar market operates, and overall the trend is clear: They have been declining across the board for all types of solar technologies. Nonetheless, these technologies have not yet reached cost parity with traditional energy sources; moreover, the initial capital investments required remain high ($100200 million for a 100MW plant, Susman, 2008), even while maintenance costs are low (Byrne 2010). Capital costs were even higher for solar PV in the 1970s, however, at $3035 per watt, compared with $45 per watt today (Timilsina, 2011). Despite the decrease, initial capital requirements are often cited as a barrier to the technology being used more widely (Sharma, 2011). Nonetheless, projections show that investment costs may be reduced between 30 percent and 40 percent over the next 10 y ears (OECD/IEA, Deploying Renewables, 2011). Although costs are also falling in the CSP subsector, they are not as competitive as in the PV sector (OECD/IEA, Deploying Renewables, 2011). At the consumer level, solar PV prices have dropped to $2 per wat t (compared to around $1 or less for traditional energies), representing a 5060 percent decline (UNEP, 2010). In the U.S., solar PV operation costs have fallen an average of 3.6 percent yearly in the last 10 1 Linuo New Materials, Sangle, Micoe, Himin, and the Sunrain Group. 19

PAGE 20

years (Kahn, 2009). Many PV manufacturers have been seeking to reduce costs by expanding output; it is estimated that for every doubling of output, costs can be reduced by 18 to 20 percent (Susman, 2008). Costs are estimated to fall by 10 percent per year until 2020, meaning that solars per watt cost could reach $1, or close to parity with fossil fuels. Another pricing advantage in the solar market, particularly when considering international trade in solar related products, is that in general, tariffs are very low in this sector worldwideapproximatel y 8 percent in developing nations and near zero or zero in developed nations (Algieri et al., 2011). The European Photovoltaic Industry Association predicts that the cost of solar electricity will drop by half in the next 10 to 15 years; analysts expect pr ice parity with traditional electricity in five to 10 years, meaning that solar PV will become even more competitive (QMS Partners, 2009). The current and expected price drops in solar PV are due in part to the extreme competition engendered by the supply demand imbalance in the market (to be discussed subsequently), as well as from technological improvements and economies of scale (McCrone, 2012, Solarbuzz 2010). Ease of Entry and Barriers to Entry Currently, the biggest factors affecting entry are government support and high capital costs. This is due to the fact that, while prices have been falling swiftly and steadily, the solar market is not yet developed enough to survive on its own. Indeed, generous government subsidies have recently come to an end i n Europe, and that, combined with the economic downturn, has forced companies to consolidate and made new entrants less likely. Along with the need for public support comes the issue of grid integration. Even if solar energy reaches consumer price parity w ith traditional electricity generation, and 20

PAGE 21

even if investment costs fall to a level where government subsidies are no longer necessary, integrating solar energy into existing grids or building new ones requires additional government monies and support (Johansson et al., 2012). The CSP subsector faces challenges in addition to investment costs and grid integration. Because of the large tracts of land needed for solar panels to collect energy, the NIMBY (Not in My Backyard) phenomenon can also be an issue (J ohansson et al., 2012). Availability and Pricing of Substitutes In many cases, substitutes for solar energy may also be complements. Hybrid energy systems where renewable energy is used to supplement traditional fossil fueled generation, or several different types of renewable energies are used together to produce energy depending on conditions are growing in popularity (REN21, 2012). Nonetheless, while solar energy is gaining overall market share, other renewables, especially hydro and wind power, are more popular forms of renewable energy (OECD/IEA, Deploying Renewables, 2011). Hydro power comprises almost 84 percent of the worlds renewable energy and has grown as an energy source by 50 percent since 1990 (OECD/IEA, Deploying Renewables, 2011). Wind power is the worlds secondmost used renewableenergy source, with production having increased by a massive 870 percent between 2000 and 2009 (OECD/IEA, Deploying Renewables, 2011). As with solar power, China has taken a lead in developing many of these new hydro and wind energy projects. Perhaps one of the reasons that water and wind power have become more widespread is due to their competitive pricing. Given favorable conditions where the resource is readily and steadily available and where the market i s sufficiently developed both renewable sources of energy are either pricecompetitive with, or very 21

PAGE 22

close to becoming pricecompetitive with, energy generated from traditional fossil fuels (OECD/IEA, Deploying Renewables, 2011). A recent study found that the international demand for solar PV products is both income and price elastic (Algieri et al., 2011). The authors find that income elasticity is higher than price elasticity and conclude that foreign income is therefore a major factor in increasing sol ar exports. As hybrid energy systems continue to increase in popularity, and as prices for solar modules and energy continue to fall, it seems that competition from more the traditional and well developed renewable sources of water and wind will be less of a factor. Input Market Perhaps the most important current issue in the international solar market is the gross supply demand imbalance that has plagued the PV market in recent years. In the early 2000s, prosolar policies in countries like Germany, Spain, Japan, Italy, and the United States drove up demand for solar cells, which are made from solar grade polysilicon (Hayward, 2011). The spike in demand for solar grade polysilicon quickly led to a supply shortage. Attracted to the newly high profit margins in the sector, in 2008, many producers entered the market (Kirkegaard, 2010). Due mostly to large, cheap Chinese manufacturers, production increased rapidly, and the polysilicon market was quickly oversupplied. Consequently, polysilicon prices dropped prec ipitously, 40 percent per year; this bolstered demand but also significantly reduced profit margins (Aanesen, 2012). The financial ramifications of this glut are large: One analyst estimated that as of 2011, production equipment worth about $8 billion was sitting on suppliers order books (Colville, 2011). 22

PAGE 23

The fallout of the supply demand imbalance is still being felt, and the market is changing rapidly as it adjusts. Prices remain depressed as of late 2012. While global solar demand is expected to increase in the coming years, the increases will be smaller since several of the programs encouraging solar production and consumption in Europe and the United States have expired or are set to expire. Moreover, some Chinese and Taiwanese firms are trying to reduce costs via expansion to take advantage of economies of scale, adding even more to the oversupply. Due to all of these factors, the solar PV subsector seems to be moving toward consolidation, with recent rounds of bankruptcies, mergers, and partnerships (Platzer, 2012). However, consolidation has been occurring mostly in the West, as Asian manufacturers continue to expand individually (Choudhury, 2012). Demand Europe dominates demand in the global solar market, mostly driven by generous feedin tariff policies (Byrne, 2010). 2 Countries like Germany, France, Italy, and the Czech Republic encourage solar production via such programs, and as a result, 77 percent of world demand for PV technology has come from Europe. Demand in the United States has been pos itively influenced by a number of national and state programs, though the central approach found in some European countries is lacking. The U.S. and Europe represent more than 75 percent of global demand, with Asia representing about 25 percent (Byrne, 2010). In the U.S., more than threequarters of sales are in California and New Jersey, both of which have incentives and policies designed to encourage solar energy production and consumption (Susman, 2008). 2 A feed-in tariff offers set payments to renewableenergy producers. 23

PAGE 24

In terms of total area of solar collectors install ed worldwide, China, Germany, Turkey, and India top the list, followed by the U.S., Mexico, India, Brazil, Thailand, South Korea, and Israel (Timilsina, 2011). In terms of consumption, the top markets are Germany and Italy, although Japan may overtake Ital y due to expanded incentives enacted in July 2012 (Bloomberg News, 2012). It is estimated that the most demand in OECD countries will be in rooftop solar installations, while ground installations are expected to be more prevalent in poorer nations (Aanesen 2012). Demand is growing, particularly in China and India, due to those nations rapid GDP growth. Between 2007 and 2009, 70 percent of the worlds growth in energy demand came from China and India. Worldwide, demand for solar energy has been growing at 35 to 40 percent per year, and similar rates are expected in the future (QMS Partners, 2009). Markets most responsible for this global demand growth are Germany, Spain, Japan, and the U.S. To a lesser extent, India, China, and South Korea have also been de mand drivers (Byrne 2010). Projections It is estimated that with adequate policies to encourage production and consumption, solar PV cou ld provide 45 percent of the worlds energy by 2040 (Byrne, 2010). According to one estimate, solar thermal energy, a ty pe of collector, is expected to expand 10fold by 2030; other estimates expect the same increase by 2020, with solar thermal providing 4 percent of the worlds energy by 2040 (Byrne, 2010). By 2050, 6 percent of global energy production capacity may be in CSP, increasing significantly after 2020, when analysts expect costs to fall further (Byrne 2010). Or, CSP capacity could grow by 450 percent by 2017 (OECD/IEA, 2012). CSP faces increasing competition from solar PV, as well as complications with permitting and grid connection, 24

PAGE 25

which will affect its future place in the market. The U.S., Spain, and China are expected to lead the increase in CSP production (OECD/IEA, 2012). Meanwhile, solar thermal technology capacity is expected to grow by 155 percent by 2017, led by China, Germany, the United States, Turkey, and India (OECD/IEA, 2012). Overall, the OECD expects that solar PV will be competitive with retail electricity before 2020 (OECD/OEA, 2011). The continued market imbalance, as well as fierce competition due to it, will continue to push prices down even further (EPIA, 2012). Other estimates predict that by 2040, solar energy overall will supply 11 percent of the worlds energy 6 percent PV, 4 percent solar heating and cooling, and 1 percent CSP (Byrne, 201 0). Some projections are not as optimistic, putting global solar energy production at 2.5 percent of the total (Poullikkas, 2010). These large variances can be accounted for by differences in assumptions regarding policies, market structures, and costs. Re venuewise, the solar PV industry could reach $100 billion by 2014 (Sharma, 2011). As China and other Asian countries expand production, some OECD producers may be crowded out, a phenomenon that is already occurring (Groba, 2011). Meanwhile, demand for solar PV energy is expected to grow the fastest in China, India, Southeast Asia, Latin America, the Middle East, and North Africa (EPIA, 2012). In Latin America, Brazil has recently implemented new policies designed to promote solar PV. As a result, leading industry analyst publication Solarbuzz is predicting regional growth in Latin America in solar PV of more than 350 percent in 2012 alone. By 2016, 6 percent of global solar PV demand could be from Latin America and the Caribbean (Barker, 2012). 25

PAGE 26

In the mids t of the current and expected global expansion of the solar industry, uncertainty is being caused by several issues: First, subsidy and incentive cuts in countries whose previous policies drove global solar PV demand (Germany, Italy, and the U.S.) have the potential to reduce overall demand; at this point, it is too soon to know the impact of the policy changes, but analysts note that some companies are now looking for markets that can thrive without government support (Mendolia, 2012). Second, solar is fac ing increased competition from shale gas production, particularly in the United States (Platzer, 2012). Third, the supply demand imbalance shows no signs of abating in 2012. In fact, Chinese producers are planning to expand capacity by 19 percent in 2012, after upping capacity by 57 percent by the end of 2011 (Choudhury, 2012). Fourth and finally, a trade war related to the supply demand problem is brewing between China and the United States. American solar producers have accused Chinese cell manufacturers of illegal dumping, and in May 2012, the U.S. Department of Commerce made a preliminary ruling imposing anti dumping tariffs (Agencies, 2012). These tariffs may be a boon to U.S. producers, but they could cause the overall installation costs of PV systems to rise (Platzer, 2010). In response, China is investigating the U.S. for its solar subsidies and possible dumping and could impose tariffs of its own (Agencies, 2012). In the meantime, Chinese producers are looking to transfer production to Taiwan or Sout h Korea to avoid the tariffs (Colville, 2011). In this dissertation, I am examining outgoing U.S. solar technology only. As the global market analysis shows, even though the U.S. is not one of the main manufacturers of solar technology, it is, along with t he Japan, the most prolific 26

PAGE 27

developer of solar technology in terms of patents filed, and it remains one of the major players in the global market. Based on this market analysis, and based on the fact that my dataset covers the years 1952 2011, I expect hig h quality solar technology to be more important in more developed nations since that is where demand has been most concentrated until very recently. Table 21. Solar PV o perating capacity by c ountry Ranking Country Capacity (%) 1. Germany 35.6 2. Italy 18.3 3. Japan 7.1 4. Rest of World 6.9 5. Spain 6.5 6. United States 5.7 7. China 4.4 8. France 4.1 9. Other EU 4.1 10. Belgium 2.9 11. Czech Republic 2.8 12. Australia 1.9 Adapted from REN21. 2012. Renewables 2012 Global Status Report (Page 48, Figure 12). REN21 Secretariat, Paris. Table 22 Market s hares of w orlds t op solar PV m anufacturers Ranking Company Country Market Share (%) 1. Suntech Power China 5.8 2. First Solar United States 5.7 3. Yingli Green Energy China 4.8 4. Trina Solar China 4.3 5. Canadian Solar Canada 4.0 6. Sharp Japan 2.8 7. SunPower United States 2.8 8. Hanwha SolarOne China 2.7 9. Tianwei New Energy China 2.7 10. Hareon Solar China 2.5 11. LDK Solar China 2.5 12. JA Solar China 2.4 13. Jinko Solar China 2 .3 14. Kyocera Japan 1.9 15. REC Norway 1.9 16. Other Rest of World 51 Adapted from REN21. 2012. Renewables 2012 Global Status Report (Page 48, Figure 13 ). REN21 Secretariat, Paris. 27

PAGE 28

CHAPTER 3 LITERATURE REVIEW Little has been written about patent qual ity as it specifically relates to the international transfer of environmental technology. However, the literature on patent quality, patent citations, intellectual property, and technology diffusion in general is well developed. These studies attempt to answer some key questions. First, how related are citations to quality? Second, what is the best way to measure quality using citations? The literature dealing specifically with environmental technologies and IPRs examines what effect strong patent laws may have on technology inflows. Dozens of studies have used forward patent citations to measure patent quality. However, before examining the key research in this area, it is necessary to define patent citations themselves and explore some of their characteris tics. Citations come in two types: backward and forward. Every time a patent is filed, the author and patent office officials list prior art, which includes previous patents that may be similar to or relevant to the current patent filers technology. Let t he current patent being filed = X. We can refer to all the prior art contained in X as backward patent citations of X. However, going forward, if other patents cite X as prior art, we can refer to them as forward citations of X. A 2001 NBER analysis of wor ldwide patent data (of which the dataset in this study is a subset) showed that forward patent citations occur over long periods of time. Specifically, 50 percent of patents will receive citations within 10 years of filing, 25 percent more will receive cit ations within 20 years of filing, and 5 percent more will receive citations within 50 years or more after filing. This means that if newer patents are included in the data, they most likely will not reflect the correct forwardcitation effect 2 8

PAGE 29

simply becaus e these patents are not old enough to have received all of the citations they will likely garner (Hall et al., 2001). Van Zeebroeck (2011) offers a simpler remedy to th is time issue: Count citations received by patent applications within a given period of time Now that we have examined patent citations and their characteristics, we can explore what previous research has discovered about their relationship to patent quality. Trajtenberg (1990) authored one of the first studies showing the positive correlati on between citations and quality. Since then, patent citations have been shown to be a reliable indicator of a patents quality ( Lanjouw and Schankerman, 2004); studies finding a positive link between forward patent citations and patent quality include Har hoff et al. (2003) and Marco (2007). In addition, m any researchers use the raw count of forward patent citations to measure quality. These include: Harhoff et al. (1999), Fallah et al. (2009), and Rosenkopf and Nerkar (2011). Norback et al. (2011) use the raw count, but they weight the number of patent citations received by a linear time trend following Hall et al. (2005). Acosta et al. (2009) scale citations by year and by stock of available knowledge, i.e., stock of available patents that a patent could s ite, and by sector to control for time and industry differences. Weighting citation data or using other methods to account for the age of the patents can be important. Forward citations suffer from the problem of truncation because citations can continue t o occur at any time in the future. This means that newer patents have fewer citations not necessarily because they are less useful but simply because they are younger. In addition, the frequency of both patenting and citing has increased, particularly in t he 1980s, so it is possible that more citations may be picking up this general trend rather than anything specific to the value 29

PAGE 30

of a particular patent. Another problem that may arise is that technologies from different industries (e.g., computers vs. drugs ) are patented and cited at different rates. Therefore, focusing on a specific sector can help ameliorate some of these problems. Indeed, Popp (2006) finds empirical support for the idea that allowing for different behavior across technologies is important for climate change mitigating technologies. Thus, one can conclude that examining data sector by sector, or technology by technology, is apt to yield the most accurate results. Despite the fairly rich literature on patent citations and patent quality, v ery little empirical work has been done examining these issues for green technologies. Acosta et al. (2009) provides the first and only, to my knowledge, analysis of this kind. Examining European environmental patents and using weighted citations as a qual ity measure, the authors find that patents from institutions are of a higher quality than those from individuals. They also find that green patents from the United States and Japan are cited more frequently than European patents. Finally, their analysis shows that patents that can be used in multiple sectors are more likely to be cited than patents that have very specific, limited uses. Another recent study of patent citations and environmental technology (Pillu and Kolda 2009) examines 11 energy technolog ies in France, Germany, Japan, the United Kingdom, and the United States to determine what factors induce innovation in this industry. The authors use patent citations to help construct a proxy for the available stock of knowledge that inventors can use to develop new innovations; they weight the stock of patents by their productivity, i.e., citations. The authors find that both high 30

PAGE 31

energy prices and the availability of knowledge (i.e., patent citations) encourage innovation. The literature confirms a robust positive relationship between patent citations and patent quality. This allows us to explore another key issue: Is there also a positive correlation between patent quality and patent filings? The available research shows the answer to be maybe. Little research has been done on patent quality and patent flows; however, several studies have explored the relationship between patent quality and patent valuation. One of the earlier studies in this area (Scherer 1984) showed that for U.S. firms, higher quali ty patents are worth more. Later studies have also confirmed that high quality patents are also worth more, including Hirschey and Richardson (2001, 2004) and Lanjouw and Schankerman (2004). Chen and Chang (2010) find that in the U.S. pharmaceutical indust ry, only some indicators of quality are positively associated with firm value. Lanjouw and Schankerman (1999) find that among U.S. manufacturing firms, higher quality patents are more likely to be renewed, and firms are more likely to sue when highquality patents are infringed upon. Despite the strong relationship between patent citations and patent quality, examining these two factors alone as determinants of patent flows is not enough. Researchers must also consider country and industry characteristics, which can affect technology diffusion as well. Almost all studies exploring the relationship between intellectual property and technology diffusion use a set of independent variables to control for national factors that may affect the decision to patent (B ranstetter et al., 2006, 2007; Evenson and Kanwar, 2001; Javorcik, 2004; Kanwar, 2009; and Maskus et al., 1995, 2001, 2005, 2005). To control for market size, researchers may use GDP, per 31

PAGE 32

capita GDP, or population. When dealing with innovation diffusion, i t is also essential to measure a countrys capacity to absorb new technologies; various measures of human capital are used, including years of secondary or tertiary education, or the population employed in hightech or R&D sectors. Studies also want to acc ount for a nations economic relationship to the rest of the world, so they might control for membership in a trade bloc or other trade agreements. These studies have also found that controlling for industry can yield better results. For example, researchers have looked at the different effects IPRs can have in traditional manufacturing vs. more hightech sectors such as chemicals and pharmaceuticals (Javorcik, 2004). Overall, it is important to consider a wide array of factors in addition to patent quality that may affect patent flows. Finally, it is also useful to discuss the literature specifically related to intellectual property rights protection and green technologies. Namely, do stronger IPR protections engender more environmental innovation? Barton ( 2007) makes one of the first attempts to examine the relationship between IPR protection and environmentally sound technologies (ESTs). In photovoltaic s, he concludes that patents might not present an obstacle to access for developing nations due to the level of competition induced by the high number of businesses in the industry worldwide. Dechezleprtre, Glachant, Mnire ( 2010) analyze data from 66 countries between 1990 and 2003 to determine whether higher IPR protection increases the transfer of ESTs. The authors find a statistically significant, positive relationship between the strength of a countrys IPR laws and patents filed in wind, solar, hydro, cement, building, and methane. They find no statistically significant relationship in biomass, geother mal, waste, and fuel injection. They find a statistically significant negative relationship in ocean and light. Popp et al. 32

PAGE 33

(2011) examines how technological innovations, represented as increases in a global technology stock, affect the use of renewable energy technologies in four areas: wind, solar photovoltaic, geothermal, and electricity from biomass and waste. They find a small, statistically significant positive effect of increased knowledge on renewable energy investment. When broken down by technolog y, a statistically significant positive effect is found only for the wind and biomass sectors. The literature in this area shows that while using patent citations is a tried and tested measure of patent quality, much work remains to be done at disaggregat ed levels. A review of relevant research also shows that more empirical work remains to be done in determining whether a positive relationship exists between patent quality and patent flows. Finally, although some have explored the relationship between IPR laws and the flow of green technology, it remains a relatively new area of study, and more work can be done to determine whether the solar sector behaves like other industries with respect to citations and IPR laws. In this dissertation, I hope to move th e literature forward by exploring answers to these questions. 33

PAGE 34

CHAPTER 4 PATENTS: A RELIABLE MEASURE OF INTERNATIONAL TECHNOLOGY TRANSFER? For decades, researchers have widely used patent filings to measure technology flows among countries; nonetheless, i t is worth exploring how reliable patents are as a measure of international technology transfer. A few key questions emerge: First, what constitutes technology transfer ( also refe rred to as technology diffusion)? Second, what are the most common measures o f technology transfer? And finally, which one of these measures best gauges diffusion across borders? Albors Garrigos et al. (2009, p. 156) call technology transfer an active process, during which technology traverses the borders between two entities, including nations, firms, or people. This definition reflects a process that is broad, and that can be intentional or unintentional. Steiner et al. (2009, p. 18) go further, noting that technology transfer must also include the capacity to assimilate, imple ment, and develop a technology, which ultimately leads to its consolidation in the receiving country. For the purposes of this study, I define technology tr ansfer as the process by which technology moves from one country to another in such a form that it can be assimilated or implemented in that country. The process of measuring technology transfer, however, has bedeviled researchers for decades. There are no perfect, direct gauges, but several are common: R&D expenditures, FDI flows, trade flows, licensing, and patent counts (Park 2007). I discuss each of these in turn before examining patent counts and the relative advantages they hold over other measurement methods in the case of the research contained in this dissertation 34

PAGE 35

Expenditures on research and development have often been used to measure innovation and technology transfer. The most basic issue with this measure is the fact that it is by definition an input in the technology development process, while ITT measures output (Lanjouw et al., 1998). D at a availability and accuracy can also be problematic First, data are unavailable for many firms, nations, and years, particularly in the developing world. Moreover, when they are available, they are not necessarily recorded and collected consistently over time, which further decreases their usefulness (Lanjouw et al., 1998). Finally, they are not dis aggregated, so it is not possible to analyze these data sector by sector (Dechezleprtre, 2010) as I do here. Keller (2009) argues that R&D expenditures consti tute a very noisy measure since returns to R&D can vary drastically over time and across firms, institutions, and countries. FDI and trade flows are sometimes used to measure international technology transfer. For the former, FDI, particularly in the R&D s ector, can represent the acquisition of new technology in a country. Trade flows of new goods, intermediate or final, can also indicate the adoption of new technology. The central problem with using FDI and trade flows to measure ITT is that the data avail able are highly aggregated ( Dechezleprtre, 2010). It is difficult to find extensive data that break down FDI into investment in distribution, manufacturing, and R&D. Therefore, while a country may see a spike in investment inflows, if it is due to the construction of a new textile factory that uses existing technology, that does not represent technology transfer. For example, Sawhney and Kahn (2011) find that U.S. FDI in outflows in the wind and solar sectors to both developed and developing countries result in increased exports of solar and wind technologies to the U.S. from those nations. The data were gathered using the North 35

PAGE 36

American Industry Classification System (NAICS), which allows FDI flows to be disaggregated by sector. However, sector disaggregat ion alone cannot pinpoint the type and amount of FDI dedicated to R&D; therefore, using FDI flows to measure innovation in the solar and other green sectors remains problematic. Moreover, this study seems to indicate a feedback effect between FDI and trade in green sectors, indicating that neither may be appropriate to use to isolate the effect of innovation in environmental technologies Second, FDI and trade flows are indirect measures ( Dechezleprtre, 2010). Even if the incoming investment is directly r elated to R&D, or imports are hightechnology inputs or products, it is difficult to measure to what extent these transfers spill over into the larger economy. Will research done within an MNC subsidiary spread to the rest of the country as a whole? Does i mporting a new high tech green product result in new technology being made available more widely? It is neither certain nor easily quantifiable. Indeed, Popp (2011, 2012) notes that the effectiveness of both trade flows and FDI as conduits of technology tr ansfer depends largely on a nations capacity to absorb technology. While this is true of all modes of technology transfer, it is equally applicable to the green energy sector. Licensing is probably the best way to measure technology transfer. When a firm pays for a license for a technology, this indicates that the technology is actually being used (transferred), while also attaching an exact monetary value to that technology (Nelson, 2009). Researchers have used royalty and licensing fees to examine whether strengthening IPR laws increases cross country licensing (see Maskus and Yang, 2005). However, as with other measures of ITT, the problem here is one of data 36

PAGE 37

availability. The only licensing data available are aggregated, so researchers have no way of kn owing whether the fees are being paid to use new technology or not. Licensing data are truly useful only if they are accessible at the industry or firm level, and these are not available on a wide scale. Gathering industry or firm level licensing data req uires conducting surveys, which can be costly and difficult, and may yield low responses. As an example, Steiner et al. (2009) conducted a licensing survey of 500 organizations involved in cleanenergy technologies and had a response rate of only 30 percent. In addition to the low response rates, survey data also require researchers to identify and correct for any possible selection issues. So far, we have seen that for the purposes of examining the solar energy sector, the data available for R&D expenditur es, FDI flows, and trade flows are inadequate to describe and quantify flows of international technology transfer accurately. Moreover, these measures of ITT have not been shown to be positively correlated with the quality of innovations, a key part of my research here. While licensing may be the preferred measure, such data are also unavailable in a form that is useful to researchers hoping to track and analyze ITT in green energy sectors What is left, then, are patent counts as a way to gauge levels of t echnology transfer. We can first examine the advantages that patents have over the measures discussed previously; then, we can discuss problems with using patent counts and ways that those issues can be mitigated. Patents have been used to quantify innovat ion for more than 50 years.1 For the purposes of this study, t hese data have several advantages when compared with the 1 Schmookler and Brownlee (1962) were one of the first to use patent counts to quantify innovation. They used patents to create an index of inventive activi ty employed to measure capital -goods patents and valueadded in selected industries. They noted, even at that early date, the problems with aggregated data, as well as the fact that patent characteristics can differ across industries In one of the first studies of its kind, Comanor and Scherer later 37

PAGE 38

other technology and innovation measures discussed previously. First, patent data are extensive and easily available (Griliches, 1990). D ata can be found for almost every country, sometimes going back to the 19th century. Patent applications contain a plethora of useful information not found elsewhere regarding the nationality of the inventor, where the invention occurred, where the patent was filed, the type of technology represented in the patent, and information on patent and publication citations (Lanjouw, 1998). This means that, unlike licensing or FDI data, patent counts can be disaggregated not only by industry but by sectors within i ndustries, allowing for extremely specific and accurate analyses. This is extremely important in the greenenergy industry, which has many sectors and subsectors; for example, the solar sector alone contains more than 90 separate International Patent Clas sification (IPC) codes, according to the World Intellectual Property Organization. No other type of data covering the solar sector offers the same breadth and depth as patent data. Second, since the patent application process is expensive and complex, the very act of filing a patent indicates that a technology has value and usefulness (Dechezleprtre, 2010). Indeed, during the last 200 years, very few major inventions have not been patented (Oltra, et al., 2008). In addition, empirical evidence supports th e claim that filing patents in other countries signals a willingness to deploy that technology in the recipient nation, and that worldwide, firms read and use patent applications to improve their own technologies (Maskus, 2004, p. 23). Hence, we know t hat patent data provide information that is both wide and deep, while also revealing the value of the technology contained within patents. Finally, because patents make found a positive correlation between patent applications and the introduction of new products (Comanor and Scherer, 1969). 38

PAGE 39

technology public and anyone can copy technology embodied in a patent once it expires, patents were in effect designed as agents of technology transfer; as a result, they are an ideal technology transfer measure. Nonetheless, problems do exist with using raw patent counts to measure technology transfer. Most obviously, not all technology is patented, nor is it even patentable (Griliches, 1990). Since patenting requires inventors to make public their technology, they may prefer secrecy to patenting. Other technology, such as know how and learning by doing, is tacit and therefore unable to be patented. While patent data may undercount or miss some technologies, on the whole, researchers agree that most economically valuable patents are filed ( 2010). Moreover, empirical evidence shows a positive correlation between tacit knowledge and the knowledge contained in patents (Dechezleprtre, 2010). Of course, the act of filing a patent does not necessarily mean that technology transfer has occurred. In fact, firms may file a patent for completely different reasons. While imitation prevention is the foremost reason for filing a patent, companies may also seek patent protection to block a rival from developing a similar or related invention, or to use as leverage in negotiations or lawsuits (Cohen et al., 2000). Still, others note that because the application process is costly and cumbersome, inventors are unlikely to patent unless they believe they will produce and/or use the technology where the patent is filed (Dechezleprtre, 2010). Although other motivations for filing patents exist, because patenting requires making public the technology, inventors must assume that filing a patent will be more economically advantageous than not doing so. 39

PAGE 40

Patent value can also present a problem in patent data. It is widely known that the vast majority of patent s are of very low value, and that a small proportion of patents account for most of the total value of patents (Dechezleprtre, 2010; Keller, 2009; Oltra et al., 2008). This is also true in the solar industry, where on average, only about 25 percent of pat ented solar technologies worldwide are exported (Dechezleprtre, 2011), indicating that the majority have a lower value.2 Therefore, using raw patent counts to quantify technology transfer is highly inadvisable because there is a high probability that the patents being counted have low value, and thus account for very little, if any, actual technology transfer. Lanjouw et al. (1998) propose weighting patent counts by data on patent renewals and the number of countries where a patent is filed to measure the patents value more accurately. Doing the latter is extremely common. Patents filed in multiple countries can be assumed to be of even greater value, and indeed, this coincides with evidence showing that exported technologies are of the highest value of al l technologies (Lanjouw, 1998). Weighting patent counts by citation data can also correct for this problem (Dechezleprtre, 2010 and Keller, 2009) and is the method I use here. Phalin (2012) confirmed a robust, positive relationship between patent citations and patent quality in the solar sector. Therefore, by using patent citations as a quality measure in my regression on patent counts, I can be relatively confident that I am measuring patents of higher value. In other words, because the quality measure al lows me to capture patents whose technology is more likely to be employed, I am more closely capturing a measure of technology transfer. 2 This is compared with about a 30 percent export rate in the wind sector. 40

PAGE 41

The last problem with patent data is that the propensity to patent varies widely across industries (Dechezleprtre, 201 0). Patents are most likely to be file d in the pharmaceutical, chemical, and car industries (Oltra et al., 2008). Hence, if we examine aggregated patent data and see an increase in patenting over time, this may indicate more innovation, or it could indicat e a higher propensity to patent. The simplest way to correct this issue is to use patent data disaggregated by sector or industry (Basberg, 1987), which is one reason I restrict my analysis to the solar energy sector of the cleanenergy technology industry Of course, this correction could present a disadvantage because sector specific results may not be generalizable. However, what may be lost in generalizability may be gained in accuracy, so it seems to be a tradeoff worth making. Overall, patent data hav e proved to be a widely used and dependable source of data and information for researchers examining international technology transfer in the cleanenergy industry. This is because 1) they allow for specific, disaggregated analysis, unlike FDI and trade fl ows; 2) they are easily accessible, unlike licensing data; and 3) they are more reliably and consistently collected and maintained than data on R&D expenditures. After reviewing the most common measures of technology diffusion R&D expenditures, FDI flows, trade flows, licensing, and patent counts and their possible use for examining the solar sector, a general conclusion can be reached: Yes, problems exist with using patent data to gauge technology transfer, but they are the best available measure in this c ase compared with all the others, especially once certain issues are corrected. To rephrase the well known Churchill quote: It has been said that patent counts are the worst way to measure technology transfer except for all the others that 41

PAGE 42

have been tried. Basberg observed in 1987 that, We have a choice of using patent data cautiously and learning what we can from them, or not using them and learning nothing about what they can teach us. This advice still holds true today. 42

PAGE 43

CHAPTER 5 THEORETICAL AND ECONO METRIC FRAMEWORK This work is based on the model of Gallini et al. (2001), who base their work on Eaton and Kortum ( 1994, 1995), Kortum and Lerner (1997), and Rafiquzzaman and Whewell (1998). Gallini et al. analyze aggregate data of patents filed in Canada from Germany, the United Kingdom, and the United States. Eaton and Kortum (1994) model the creation of new inventions and their international diffusion. In their model, the value of a patent depends on its quality, q, a random variable drawn from a cumulative distribution. They derive the following threshold condition: is the value of filing a patent with quality q from country i in country n; is the value of not filing a patent with quality q from country i in country n. The patent will be filed as long as Three country characteristics directly affect this threshold: the lag time it takes for the technology to be adopted in country n, the strength of patent protection laws in country n, and the cost of patenting in country n. These can be proxied empirically by a measure of human capital, an index of patent rights, and filing fees or the need for translation, respectively. In a later version of this paper (Eaton and Kortum, 1995), the authors expand the model to include the determinants of technology diffusion, i.e., the probability that an invention from country i will be adopted in country n. We let diffusion from country i to country n depend on: (1) whether n and i are the same country or not, (2) the distanc e between n and i, (3) the level of human capital in n (the adopting country), and (4) the level of country ns imports from I relative to ns GNP. 43

PAGE 44

Gallini et al. (2001) follow Eaton and Kortum to derive a model measuring the propensity to patent. Their s pecification is as follows: Where is the number of patents filed in destination country j by the source country i; is the innovation effect, or the total number of patentable inventions (which is unobservable); i s the probability that an invention from country i will be high quality enough for the patent filing to be profitable in country j; is the strength of patent protection in country j; is a set of indicators controlling for the ec onomic environment in j (i.e., GDP, human capital); is a set of indicators describing the relationship between i and j (i.e., distance, trade flows); and is the cost of filing a patent in country j. Taking logs, t heir econometri c specification is as follows: As above, Pijt is the numbers of patents from the source country, i, filed in the destination country, j (Canada), in year t. nit is the amount spent on R&D in i in year t. sjt is the strength of patent protection in j as m easured by the Ginarte and Park Index (Ginarte and Park, 1997) xjt is a set of variables measur ing human capital, GDP, and an index measuring the effectiveness of js antitrust laws. zijt describes the relationship between i and j, including: distance, di stance squared, and log(js imports from i / real GDP). cjt control s for the cost of patenting in j, including fees and a dummy variable indicating whether translation is required. Finally the authors include time and country fixed effects. 44

PAGE 45

I add to the G allini et al. model in several ways. First, in this model as in those of Eaton and Kortum and Kortum and Lerner quality is randomly drawn from a distribution. I add a quality variable on the right hand side: a weighted measure of the total number of patent citations from i in year t. Second, I perform a disaggregated analysis, breaking down the patent data and examining only solar technology. This is important because not all results will be the same across industries and technologies. My specification is a s follows: In my analysis, the source country, i is the United States. I use 16 OECD nations as destination countries: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, Netherlands, Norway, Spain, Sweden, Switzerland, and the United Kingdom. In addition, I add a variable controlling for the existence of prorenewable energy policies in country j 45

PAGE 46

CHAPTER 6 VARIABLES AND DATA The dataset includes the United States as the source country of solar patents and 84 nations as destination countries where the patents may be filed. The dependent variable, Pijt, is the number of solar patents filed in country j from country i (the United States) in y ear t. Data for this variable were downloaded from Espacenet, the patent database of the European Patent Office. This database contains information on patents, including filings and citations, from more than 100 countries worldwide. To gather the relevant information required for this study, I first assembled all the International P atent Classification (IPC) codes that relate to solar technologies from the World Intellectual Property Organizations (WIPO) IPC Green Inventory list, which was created to allow researchers to identify environmentally sound technologies more easily. Us ing these IPC codes, a script was written in the computer language C++ that, when executed, downloaded automatically from Espacenet data for each solar related patent between the years 1952 to October 2011. ( Appendix A contains a more detailed description of the computer program and process.) The pieces of data for each patent include: the patent application number, the patent application country, the patent application date, other countries where the patents were filed, and forward patent citations. Using this data, I create the dependent variable, Pijt, which is the total number of solar patents from country i ( the U S ) filed in country j in year t. Thirty six observations from Canada were dropped from the sample because in the EPO database they had origin/destination years listed as 00000000. Note that due to the models specification, this is an aggregate measure of the number of solar patents, rather than an examination of individual patents. 46

PAGE 47

My variable of interest, qual is a proxy for the aggregate quality of the patents being filed in country j from the United States in year t. Qual is derived from the total number of citations received by all patents from the U.S. filed in country j in year t. However, I cannot use the raw total number of citations received by these patents. This is because the total number of citations reflects both the number of patents filed in a country as well as the quality of these patents; i.e., the more patents filed, the more citations there will be regardless of quality. Therefore, the total number of citations is a proxy for both the number of patents and the average patent quality, and, as such, would be subject to upward bias in these regressions. To deal with this bias, I divide the number of citations of the patents filed from the U.S. in country j in year t by the number of patents filed from the U.S. in country j in year t. This yields a ratio of citations to patents that functions as a proxy for the aggregate quality of patents filed in a particular country in a particular year, which serves the purpos e of this analysis much better. Therefore, qual can be thought of as a measure of citations per patent. I also need to be concerned with the fact that it takes patents 20 years to receive 75 percent of the citations they will ever likely receive. Therefore, I run two different sets of analyses: one on the full dataset, which includes solar patents from the years 1952 2011, and another on a trimmed version of the dataset, which includes solar patents from only 1991 and e arlier. In the trimmed dataset, we know that these patents will have received the majority of the citations that they are likely to receive. As theory predicts, I expect the coefficient on qual to be positive and statistically significant. Another variabl e of interest is ipr This is the Ginarte and Park index, which measures how strongly patent rights will be protected in a given country (Ginarte and 47

PAGE 48

Park, 1997) .1 Using data from 110 countries between 1960 and 1990, Ginarte and Park created an index that has since become the benchmark measure most economists use for a countrys level of patent protection. The G&P Index covers five aspects of a countrys patent law: 1) extent of [laws] coverage, 2) membership in international patent agreements, 3) provi sions for loss protection, 4) enforcement mechanisms, and 5) duration of protection (Ginarte and Park, 1997). Scores for each are given between 0 and 1, and a weighted average yields a total score between 0 and 5 for every five years from 1960 to 2005. Th e expected results for this variable are ambiguous. Most studies examining high income nations and aggregate patent data find a positive, statistically significant relationship between IPR strength and patents filed. However, when broken down by national i ncome and/or industry, these results do not always hold. I include gdp as an independent variable to control for market size. This variable, taken from the Penn World Tables, is total PPP converted GDP in millions of 2011 dollars (Heston et al., 2011). I hypothesize that larger markets will be more likely to draw patents because there is a higher chance of profitability in more developed economies. Thus, I expect the coefficient on gdp to be positive. A measure of human capital, humk is also included as an independent variable. This variable measures the destination countrys ability to absorb new technologies and innovations. A larger stock of human capital will signal that a country is better equipped to deal with new technologies. I expect a higher lev el of patenting in countries with more human capital, and thus a positive coefficient on the humk variable. The data for the variable humk are average years of tertiary education every five years beginning in 1 I thank Dr. Walter Park, who generously provided me the latest edition of the index, which includes rankings up to 2005. 48

PAGE 49

1950 and ending in 2010. These data are found i n the Barro and Lee dataset on worldwide educational attainment (Barro and Lee, 2010). Variables that describe the relationship between the United States and country j are also important when controlling for exogenous factors. I include dist which is sim ply a measure of direct line distance between Washington, D.C., and country js capital, per Gallini et al. These distances can be found in Fitzpatrick and Modlin (1986). Because countries that are nearer to one another tend to have higher trade and closer economic relationships, I expect the coefficient on this variable to be negative; i.e., the farther the distance, the fewer patents filed. Another variable describing the relationship between the U.S. and country j is bilateral trade flows, imps Countries with higher trade flows exchange more products and technology, so I expect the coefficient on this variable to be positive. The data for imps comes from the Feenstra and Lipsey NBER United Nations Trade Data 1962 2000 dataset. I also need to control for the cost of filing a patent in country j. This is a difficult variable to proxy because so little data exist. I could easily find information on current patent costs in each of the countries in the dataset and assume that costs are constant over time, as did Gallini et al. However, since the 1980s, patent costs have risen in Japan and the United States, while they have fluctuated at the European Patent Office (de Rassenfosse and van Pottelsberghe, 2012). Data for other nat ions are not readily available. Th e first variable I use to control for cost is cost This variable is the cost of filing a patent in constant US 2000 dollars in Japan, the United States, and the European 49

PAGE 50

Patent Office between 1980 and 2007.2 It is used only in the analysis of OECD nations Because I do not have data on each of the individual European countries in my dataset, I use the EPO numbers as a proxy to measure filing costs in these nations. This is not ideal, but excluding a cost measure would be worse for the analysis than having no measure, even a blunt one, at all. Another disadvantage of this variable is that I must exclude Canada and Australia when I use it since I have no comparable data for these countries. I expect the coefficient on cost to be negative; as the cost of filing a patent rises, fewer will be filed. Note that the cost variable includes filing and other fees required by offices, but not translation fees. Another option is to exploit the fact that the cost of translation fees for patents can range in the thousands of dollars and therefore represent a significant portion of the overall cost of filing patents abroad (European Commission, 2010). There is n o way to obtain specific information on translation fees since they are generally done by private companies; however, I can create a dummy variable, lang equal to 1 if the course country (U.S.) and destination country share an official language (i.e., if translation is not required). Information for this variable was found in the CIA World Factbook. This is a blunt measure, but it does have the advantage of bringing more national specificity to the analysis. Although not an ideal gauge of cost, excluding a cost measure would be worse for the analysis than having no measure at all I expect the coefficient on the variable lang to be positive; a variable equal to 1 indicates that translation is not needed, which means overall patent filing costs will be much lower for those nations. 2 I thank Dr. Gaetan de Rassenfosse and Dr. Bruno van Pottelsberghe de la Potterie for generously sharing with me their data on patent fees. 50

PAGE 51

So far, I have included independent variables that control for patent quality, the econom ic environment in the destination country, the relationship between the source and destination countries, and the cost of filing patents in the destination country. However, since this analysis concerns solar technology, it is also important to consider wh ether any policies in the destination country regarding renewable energy may also encourage solar technology inflows. A wide range of policies can be used to encourage alternativeenergy R&D and production, including feedin tariffs, subsidies, and tax inc entives. Rather than creating a separate variable for each of these policies, I have created a dummy variable, renew equal to 1 if prorenewableenergy policies existed in destination country j in year t 1. I lag this variable because these policies often do not begin having effects immediately. I constructed this variable using the International Energy Agencys World Energy Outlook Policy Database. I expect the coefficient on this variable to be positive; prorenewableenergy policies are likely to encour age inflows of solar technology and attract such innovation to the destination countries. For the complete dataset of all 84 countries, I also add a meteorological indicator, sun which captures average hours of sunshine per year. This data, from the World Meteorological Organization, was accessed via the United Nations Data Explorer. I expect a positive sign on sun since nations with more sunlight on average can be expected to produce more solar technology. However, there may be some ambiguity in this vari able, particularly if production is being off shored to a country with less sunlight. In addition, so many other factors determine production and use of solar technology, such as infrastructure and general economic performance, that meteorological data may capture only a small sliver of the solar technology decision process. Second, extensive, 51

PAGE 52

worldwide solar/cloudcover data are not as widely available as some other data points ; therefore, including the meteorological data causes more than 100 observations to be dropped, reducing the accuracy of the results. In other words, there is a tradeoff between adding a plausible (but not necessarily central) factor in the solar technology decision and accuracy of the results as a whole. Finally, t here is also possible selection bias; i.e., countries that dont have adequate sun data may lack data because of poor infrastructure or poor reporting standards which could bias the results. 52

PAGE 53

CHAPTER 7 RESULTS: OECD COUNTRIES Almost 80 percent of greenenergy patents are filed by six nations Japan, the United States, Germany, France, the United Kingdom, and South Korea (Latif, Maskus, et al., 2011). Thus, it can be seen that technology in this sector is fairly concentrated at a national level. Figure 71 shows the top 20 destination countries for U.S. solar patents between 1952 and 2011. Japan, China, Canada, Australia, and Germany comprise the top five. In this section, I focus on a select group of OECD countries per Gallini et al. This has the advantage of controlling som ewhat for cross country heterogeneity since these nations are similar in market size and economic background. The dataset includes the United States as the source country of solar patents and 16 OCED nations as destination countries where the patents may b e filed: Australia, Austria, Belgium, Canada, Denmark, Finland, France, Germany, Italy, Japan, the Netherlands, Norway, Spain, Sweden, Switzerland, and the United Kingdom. I first examine my dataset trimmed to include only the years 1991 and earlier. This is to account for the fact that most patents receive 75 percent of all citations they will ever receive within 20 years of being filed. Thus, limiting the years in the analysis helps control for the problem that newer patents have fewer citations not neces sarily because of lower quality, but because of their age. Looking at the summary statistics in Table 7 1, we see that in an average year, the U.S. will file about 30 solar patents in country j. However, almost one third of all patents filed in the U.S. wi ll not be filed elsewhere. About 40 percent of patents will be filed in two to 20 countries. The variable quality is a raw count of the citations received by the patents in this dataset. Note that it 53

PAGE 54

differs from qual which is weighted by the number of patents filed. On average, all the patents filed from the U.S. in j in year t will receive a total of about 282 citations. Almost 36 percent of the pijt pairs receive no citations. The measure of IPR strength, ipr has a mean of 2.98, above average in the Gi narte and Park index; in this case, an average rating would be 2.5 since the index is from 0 to 5. Next, I need to determine which econometric model works best for the data at hand. Gallini et al. use a log linear specification to measure the propensity to patent. However, the dependent variable is a count, and a significant portion of them are zeros. It can be argued that OLS specifications are better matched to continuous, as opposed to discrete, data. Moreover, when data on the dependent variable cont ain a large portion of zeros, as do the data here, it may be better to use a model that takes this into account. The first model to consider using count data is the Poisson model. However, Poisson requires E(y|x) = Var(y|x), i.e., that the mean equals the variance. This is unlikely in the current case; thus, a negative binomial specification, which allows and corrects for differences in the variance, should be better suited to this analysis. Indeed, 2 value in results from Tables 7 3 and 74 indicates that the data are not Poisson, that they are overdispersed, and that a negative binomial specification is appropriate. I now analyze the results of the negative binomial regressions. Looking first at Table 7 3, Column I, the coefficient on qual the weighted measure of quality, which is the variable of interest, is positive and statistically significant. Recall from Section 6 that qual is a ratio of citations to patents that serves as a proxy for the aggregate quality of patents filed in a particular country in a particular year This indicates that the difference 54

PAGE 55

in logs of expected counts of pijt would increase by about .05 units for a one unit change in the quality ratio, while holding other variables in the model constant. Alternately stated, a one unit change in the weighted quality ratio would cause pijt to increase by about 5%. Using the summary statistics found in Table 7 1, we can c alculate the effect at the mean: A oneunit change in qual will lead to 1.5 more patents being filed in a given year. We can also calculate the effect within one standard deviation, which would be almost 9 patents. ( I calculate this effect by multiplying t he coefficient on qual by the standard deviation of qual and multiplying this number by the mean of pijt.) The coefficient on the variable that measures patent law strength, ipr is negative and statistically significant at the 1 percent level. Results show that the difference in logs of expected counts of pijt would decrease by about .68 units for a oneunit increase in the IPR index. In other words, a oneunit change in the IPR index would cause pijt to decrease by about 68%. The effect at the mean trans lates to a fall of almost 24 patents. We can also calculate the effect within one standard deviation, which would be about 12 patents This is a large and unexpected effect. While others have also found a negative result ( Gallini et al.) some have found a positive and statistical ly significan t relationship between IPR strength and patents filed in the solar industry ( Dechezleprtre et al.). The results here could be explained by other factors. For instance, i t could be that in the solar industry in developed nations, strengthened IPR laws act as a deterrent to competition by ensuring market share for established firms which discourages patent flows. Although this hypothesis requires further testing some analysis has already been done concerning the differ ent effects that strengthened IPR laws could have on an importing country Using aggregate data, Maskus and Penubarti (1995) found that 55

PAGE 56

stronger IPR laws may reduce imitation and encourage firms to increase exports to the country, thus causing a market ex pansion effect; otherwise, such laws could reduce imitation and encourage firms to raise unit price, thus having a market power effect. The authors found that the market expansion effect outweighed the market power effect when examining data on OECD exports to the developing world, but it may be that the effects differ when broken down by industry. In addition, it could be that in highly developed economies, harmonization of laws and enforcement renders the need to file patents less pressing. Human capit al has an unexpected negative, statistically significant, and large effect. Specifically, a oneunit change in the human capital measure would cause pijt to decrease by about 214%. We can also calculate the effect within one standard deviation, which would be about 13 patents This may be due to the fact that there is little variation in the number of years of schooling in this set of OECD countries; the summary statistics show that 68% of citizens in the countries included here will have between .11 and .7 1 years of schooling beyond high school, which is a negligible difference in terms of the real world effect of accumulation of human capital. As expected, distance has a negative and statistically significant effect on the number of patents filed, though t he magnitude is small: A onemile increase in distance results in a .0017 percent fall in the number of solar patents filed. However, note that the standard deviation of distance is more than 3300, so the effect within one standard deviation of the mean wi ll be almost 17 patents. gdp and imps both have statistically significant and positive effects on the number of solar patents filed. The effect of gdp within one standard deviation of the mean is almost 22 patents imps is only marginally 56

PAGE 57

statistically sig nificant; its effect within one standard deviation of the mean is almost a 17patent increase.1 The dummy variable indicating the existence of prorenewableenergy policies in the destination country, renew is positive as expected but is not statistically significant. It may be that my current proxy is not adequately picking up the effect that I want to measure. Alternately, there may be a feedback effect whereby higher innovation in renewable energies causes such policies to be created, not the other way around. The coefficient on cost has a statistically significant, negative effect on the number of patents filed. Results indicate that a oneunit change in cost would cause pijt to decrease by about .02%. This translates into a near 12patent decrease with in one standard deviation of the mean. Column II of Table 7 3 examines the negative binomial specification including the lang dummy variable. The coefficient on qual is comparable to the previous regressions both in terms of statistical significance and si ze. Moreover, the results for the other variables are similar as well. The coefficient on lang though positive as expected, is only marginally statistically significant. Finally, Column III of Table 7 3 examines the negative binomial specification includi ng the lang dummy variable as the only cost measure. In this case, all variables except dist and renew are statistically significant, though humk is in the wrong direction. As predicted, lang has a positive and statistically significant effect. The results in Table 7 3, Column III, indicate that the difference in logs of expected counts of pijt would increase by a factor of about 1.1 if the dummy equals 1. 1 Because I was concerned about collinearity between distance and imports, I also ran regressions with each variables separately; results did not c hange much. 57

PAGE 58

Full Dataset For comparison, we can also consider the negative binomial results of the full dataset; w e see that they are in fact similar to those for the trimmed analysis. Looking first at Table 7 4, Column I, the coefficient on qual is positive, statistically significant, and similar to the trimmed results at about .059. Though the coefficient on ipr rem ains negative in this specification, it is no longer statistically significant. Human capital has an unexpected negative effect, while distance is also positive, but statistically insignificant. Renew is also statistically insignificant, but the coefficien t on cost does have a statistically significant, if small, negative effect on the number of patents filed. Column II of Table 7 4 examines the negative binomial specification including the lang dummy variable. The coefficient on qual is comparable to the p revious regressions both in terms of statistical significance and size. Moreover, the results for the other variables are similar as well. The coefficient on lang though positive as expected, is not statistically significant. Finally, Column III of Table 7 4 examines the negative binomial specification including the lang dummy variable as the only cost measure. In this case, all variables are statistically significant, though humk and dist are in the wrong direction. Robustness of Results As a check on the above results, I also run log linear specifications on both the trimmed and full datasets. The variable of interest, qual is positive and statistically significant across all specifications. The main difference is that the magnitudes are much larger in t he linear specifications. For example, Table 7 6 shows log OLS results from the trimmed dataset. The coefficient on qual indicates that a 1 percent increase in the measure of quality causes a .38 percent increase in the number of patents filed. In other words, a 100 percent increase in the quality ratio leads to 38 percent increase in 58

PAGE 59

the number of patents filed. Looking again at the summary statistics, we see that the mean of qual is about 5.12, while the standard deviation is about 5.85. Thus, we see that a 100 percent increase is likely, and that this is therefore an economically significant coefficient in this specification as well. We also see a larger magnitude effect for gdp in comparison with the negative binomial specification. Meanwhile, the effect of ipr which is negative and statistically significant, is smaller. The results suggest that a 1 percent increase in the IPR index causes a 2.47 percent fall in the number of patents filed. Looking at the mean and standard deviation of IPR, we nonetheles s see that this is an economically significant result in this specification as well. Not as many of the control variables are statistically significant in linear specifications. Moreover, the signs of the coefficients on humk cost renew and gdp change d epending on which variables are used to measure cost. The complete results for these specifications are reported in Tables 7 5 7 7. However, because of the large number of zeros in the dependent variable, the results of the negative binomial regressions ar e likely to be a more accurate characterization of the relationship between quality, IPR strength, and the propensity to patent. 59

PAGE 60

Figure 71 Top 20 destination countries for U.S. solar patents, 1952 2011 60

PAGE 61

Table 7 1 OECD s umma ry statistics, t rimmed Variable Obs. Mean Std. Dev. Min. Max. pijt 680 29.64706 77.30466 0 666 quality 680 281.6294 817.1486 0 8100 qual 680 5.12054 5.853727 0 50 ipr 544 2.984651 0.603163 2.008333 4.675 gdp 662 317168.6 688220.7 4513.926 5946800 hum k 680 0.27806 0.211903 0.0331 1.303 dist 680 6609.176 3311.276 0 15943 imps 480 5913972 1.11E+07 57469 7.61E+07 cost 180 4166.978 1824.601 246 6824 lang 680 0.235294 0.424495 0 1 renew 374 0.032086 0.176463 0 1 Table 7 2 OECD s ummary statistics, f ull Variable Obs. Mean Std. Dev. Min. Max. pijt 1020 54.27353 190.8957 0 2483 quality 1020 276.4696 780.3543 0 8100 qual 1020 4.390836 5.380992 0 50 ipr 884 3.527008 0.859096 2.008333 4.875 gdp 968 691027.5 1587278 4513.926 1.44E+07 humk 1020 0.40784 0.303943 0.0331 1.5598 dist 1020 6609.176 3310.464 0 15943 imps 624 9593222 1.90E+07 57469 1.56E+08 cost 420 4823.283 1858.418 246 8025 lang 1020 0.235294 0.424391 0 1 renew 714 0.212885 0.409634 0 1 61

PAGE 62

Table 7 3 OECD n egative b inomial r esults, t rim med I II III Variable cost cost/lang lang constant 6.14048* 5.802057* 5.358989* [0.949836] [0.9553224] [0.4785652] qual 0.0497978* 0.0503595* 0.0501218* [0.0140127] [0.0138216] [0.0121826] ipr 0.6815216* 0.7907816* 0.9709525* [0.2140033] [0.2185659] [0.1375253] humk 2.135974* 1.848331* 2.193245* [0.6405463] [0.6530437] [0.4859487] gdp 0.00000105** 0.00000132* 0.00000132* [0.000000436] [0.000000454] [0.000000205] dist 0.0001739** 0.0000984 0. 0000183 [0.00008] [0.000088] [0.0000248] imps 0.0000000509*** 0.0000000265 0.0000000211* [0.000000027] [0.0000000294] [0.00000000791] renew 0.396122 0.4264614 0.0844464 [0.3125041] [0.3081512] [0.3072135] cost 0.0002106* 0.0001846* [0.0000532] [0.0000543] lang 0.541377*** 1.096371* [0.2865833] [0.1764086] N 168 168 352

PAGE 63

Table 7 4 OECD n egative b inomial r esults, f ull I II III Variable cost cost/lang lang constant 2.243044* 1.973077** 3.795624* [0.7921048] [0.8095166] [0.3751833] qual 0.0 586222* 0.057468* 0.0563394* [0.0169586] [0.0168175] [0.0129992] ipr 0.1221819 0.1870611 0.6144548* [0.1815693] [0.185547] [0.1059338] humk 2.845379* 2.566862* 2.387045* [0.5360646] [0.5682592] [0.4133247] gdp 0.00000 138* 0.00000154* 0.0000009217* [0.000000301] [0.000000321] [0.000000118] dist 0.0001154*** 0.0001621** 0.0001015* [0.0000698] [0.0000774] [0.0000196] imps 3.84E 09 0.0000000152 0.0000000209* [0.0000000158] [0.0000000177] [0.00000 000408] renew 0.2745408 0.3174615*** 0.505246* [0.182232] [0.1841768] [0.1742732] cost 0.0000846** 0.0000685 [0.0000435] [0.0000449] lang 0.3927209 1.422557* [0.2817994] [0.1721787] N 294 294 496

PAGE 64

Table 7 5 OECD b aseline OLS r esults Trimmed Full Variable constant 7.898576 30.01547** [13.64942] [12.74285] qual 0.2543524** 0.2261105** [0.1175648] [0.1172358] ipr 10.09929* 13.65478* [2.771582] [2.769846] humk 9.99337 25.54223* [9.079358] [8.862276] gdp 0.0000177 ** 0.0000184* [0.00000854] [0.00000556] dist 0.0075039* 0.0109371* [0.0015391] [0.00135] imps 0.000000771 0.000000659** [0.000000519] [0.000000266] renew 4.370742 1.913827 [2.970277] [2.767707] cost 0.0022476* 0.00 0647 [0.0006367] [0.0006155] N 168 294 R 2 0.8359 0.8493 Standard errors in brackets *= statistically significant at 1% level; **= statistically significant at 5% level; ***= statistically significant at the 10% level 64

PAGE 65

Table 7 6 OECD l og l inear r esults, t rimmed I II III Variable cost cost/lang lang constant 3.558483 1.304101 7.115385* [7.446854] [7.953427] [1.673547] qual .3789241* .3745841* .218637** [.1294908] [.127848] [.0887525] ipr 2.473575* 2.830319* 3.481379* [.5872381] [.6467856] [.347222] humk .167515 .1698834 .3246087** [.1955187] [.19943] [.1331725] gdp .5412406* .5169336* .0348758 [.1951259] [.196241] [.1551628] dist .4538205 .1143857 .161428 [.7010568] [.7 832732] [.117386] imps .1748 .1709159 .7921719* [.1732275] [.1744482] [.1406844] renew .3796127 .4143905 .0609414 [.4239422] [.4257975] [.3480032] cost .5219366* .4449594 ** [.1692918] [.1748984] lang .43152 99*** .1475781 [.2599005] [.1340851] N 148 148 327 R 2 0.6270 0.6328 0.5964 Standard errors in brackets *= statistically significant at 1% level; **= statistically significant at 5% level; ***= statistically significant at the 10% level 65

PAGE 66

Table 77 OECD l og l inear r esults, f ull I II III Variable cost cost/lang lang constant 14.34548** 17.85084* 8.294074* [5.999005] [6.27317] [1.636412] qual 0.3104596** 0.3017453** 0.1858412** [0.1269877] [0.1265791] [0.0855511] ipr 2.52938* 2.755392* 3.412448* [0.6213028] [0.6486045] [0.3122358] humk 0.2706155 0.3078322*** 0.2052258 [0.1779701] [0.1810047] [0.1246531] gdp 0.460278** 0.4978671** 0.0870124 [0.1966467] [0.2011078] [0.1573013] dist 1.431803** 1.865468* 0.2953518** [0.5694547] [0.6272655] [0.1190165] imps 0.2006617 0.1365887 0.8442886* [0.1797459] [0.1892359] [0.1446801] renew 0.3259215 0.3750733 0.332537 [0.2463205] [0.2482332] [0.2309321] cost 0.2461272*** 0.1855691 [0.1429778] [0.144494] lang 0.3977977*** 0.1751574 [0.2143608] [0.1470642] N 243 243 440 R 2 0.5262 0.5306 0.5291 Standard errors in brackets *= statistically significant at 1% level; **= statistically significant at 5% level; ***= statistically significant at the 10% level 66

PAGE 67

CHAPTER 8 RESULTS: ALL COUNTRIES I first examine the dataset trimmed to include only the years 1991 and earlier. This is to account for the fact that most patents receive 75 percen t of all citations they will ever receive within 20 years of being filed. Thus, limiting the years in the analysis helps control for the problem that new patents have fewer citations not necessarily because of lower quality, but because of their age. Trimmed Dataset: High Income Group In the previous section, I analyzed a subset of 16 OECD nations. Here, I expand the dataset to include 84 high, upper middle and lower middle income nations. I begin first by examining the results for all highincome natio ns in my dataset, which I define as a per capita GDP of at least $20,000. I chose $20,000 because, with the exception of Mexico and Turkey, most OECD nations have per capita incomes no lower than around $20,000. Therefore, this level of GDP is less likely to result in the heterogeneity that would occur if I defined highincome as the World Bank does, at about $12,500 per capita GDP. In my datasets highincome group, Luxembourg has the maximum per capita GDP, at $80,119; Poland has the lowest, at $20,334 (I MF, 2012). The average per capita GDP of the highincome group is $37,367. The summary statistics and results can be seen in Tables 81 and 8 5 Column I. Looking at Table 81 we see that in an average year, the U.S. will file about 15 solar patents in cou ntry j. However, about 55 percent of all patents filed in the U.S. will not be filed elsewhere. About 31 percent of patents will be filed in two to 20 countries. The variable quality is a raw count of the citations received by the patents in this dataset. Note that it differs from qual which is weighted by the number of patents filed. On 67

PAGE 68

average, all the patents filed from the U.S. in country j in year t will receive a total of about 139 citations. Almost 60 percent of the pijt pairs receive no citations. The measure of IPR strength, ipr has a mean of 2.61, above average in the Ginarte and Park index. I now analyze the results of the negative binomial regressions, run in the same way as the previous section, which are found in Table 8 5 Column I. As with my earlier results, the sign on qual is positive while the sign on ipr is negative. The coefficient on qual is statistically significant at the 10% level. These results indicate that the difference in logs of expected counts of pijt would increase by approximately .018 units for a oneunit change in the aggregate quality ratio, while holding other variables in the model constant. In other words, a oneunit change in the aggregate quality ratio would cause pijt to increase by about 1.8%. Using the summary st atistics in Table 8 1 we can calculate the effect at the mean: A oneunit change in qual will lead to about 0.27 more patents being filed in a given year. We can also calculate the effect within one standard deviation, which would be 2.07 patents. This is a smaller effect than that found in the subset of OECD nations, where the quality measure caused pijt to increase by about 5%. The coefficient on ipr while negative as before, is not statistically significant. We can also compare the other independent v ariables in Table 8 5 Column I with the OECD trimmed results, seen in Table 7 3 Column III. Humk the measure of human capital, is negative as before but not statistically significant. gdp is positive and statistically significant, as before. d ist is negat ive but statistically insignificant, whereas it was positive and statistically significant previously. i mps renew and lang have the same signs as before. 68

PAGE 69

In addition to adding more countries to my dataset, I also add a new variable, sun a meteorological measure of average hours of sunlight per day per country. A scatterplot of sun against the dependent variable, pijt, reveals a nonlinear relationship across both income groups. As a result, it is not surprising that when I add the sun variable to the regr ession by itself, it does not perform well and causes the other variables to perform worse also. (Appendix B contains these results.) Because several important factors affect the propensity to patent solar technology in a country, of which available sunlig ht is only one, I interact the sun variable with ipr humk and gdp The reasoning here is that if a country has abundant sunlight but little legal structure, infrastructure, or income, the solar energy available wont matter much. These results can be seen in Table 8 6 Column I. The most striking difference here is that the aggregate measure of quality, qual while positive as before, is no longer statistically significant. However, two of the three interaction terms are positive and statistically signific ant. These results may indicate that when it comes to solar technology, sun availability in concert with higher levels of human capital and stronger IPR laws is a more important factor than quality alone. As with the results in the previous section, the co efficient on ipr continues to be negative and statistically significant. As expected, the need for translation, represented by the variable lang, reduces the number of patents filed by a factor of about 0.7. At the mean, this translates to approximately 10 fewer patents per year. Surprisingly, the variable on renew the dummy indicating whether a country has prorenewableenergy policies in place, is negative and marginally statistically significant. It may be that my current proxy is not adequately picking up the effect that I want to measure. Alternately, there may be 69

PAGE 70

a feedback effect whereby higher innovation in renewable energies causes such policies to be created, not the other way around. Trimmed Dataset: Lower Income Group We can now examine the tri mmed results for lower income nations, those with a per capita GDP below $20,000. Hungary has the highest per capita GDP in this group, at $19,591; Zimbabwe has the lowest, at $487. The average per capita GDP for this group is about $9,346, which the World Bank defines as upper middleincome. The summary statistics for these data can be found in Table 8 2 The characteristics of the highincome vs. lower income group exhibit striking differences. Looking at Table 8 2 we see that in an average year, the U.S will file about 1.17 solar patents in country j (compared with 15 patents for the highincome group). However, almost 88 percent of all patents filed in the U.S. will not be filed elsewhere (compared with about 55 percent in the highincome group). Just over 7 percent will be filed in two to 20 countries (compared with about 31 percent in the highincome group). On average, all the patents filed from the U.S. in country j in year t will receive a total of about 11 citations (compared with 139 citations in the high income group). Almost 90 percent of the pijt pairs receive no citations (compared with 60 percent in the highincome group). The measure of IPR strength, ipr has a mean of 1.43, well below the Ginarte and Park index average of 2.5, and also well below the highincomegroup average of 2.61. Looking at the regression results themselves, found in Table 8 7 Column I, we see that the main variable of interest, qual is positive and statistically significant at the 10% level, indicating that the difference in logs of expected counts of pijt would increase by about .014 units for a oneunit change in the aggregate quality ratio, while holding 70

PAGE 71

other variables in the model constant. In other words, a oneunit change in the aggregate quality ratio would cause pijt to increase by about 1.4%. Using t he summary statistics in Table 8 2 we can calculate the effect at the mean: A oneunit change in qual will lead to about 0.016 more patents being filed in a given year. We can also calculate the effect within one standard deviation, which would be about 0.17 patents. While these results are marginally statistically significant, in terms of economic significance, the positive effect of quality on the propensity to file solar technology patents in lower income count ries is minimal. The other variable of interest, ipr is positive but not statistically significant. These results may reflect issues of data availability. Of the 48 lower income nations in my dataset, 12 do not have IPR data available1; 10 others, mostly former Soviet or Soviet allied countries, do not have data available until 1995 and are thus not included in the trimmed results.2 The other independent variables, also seen in Table 8 7 Column I, perform similarly to those in the highincome regression. H umk is negative and statistically significant, while it was also negative but statistically insignificant for the high income group. Gdp and dist are positive and negative, respectively, and both are statistically significant, as before. The sign on imps changes, indicating that higher trade between countries reduces the propensity to patent by an extremely small amount, an unexpected result. The coefficient on renew is negative as it was in the highincome group, but it is not statistically significant. A gain as expected, the need for translation, represented by the variable lang reduces the number of patents filed by a factor of 1 Armenia, Croatia, Cuba, Estonia, Georgia, Kazakhstan, Kyrgyzstan, Macedonia, Moldova, Slovenia, Tajikistan, Montenegro. 2 Results available from 1985 for China; results available from 1995 for Bulgaria, Czech Republic, Hungary, Lithuania, Pol and, Romania, Russia, Slovakia, and Ukraine. 71

PAGE 72

about 1.64, larger than the factor of .48 found for the highincome group. At the mean, this translates to almost 2 fewer paten ts per year. Now we can examine the results with the sun variable interacted with gdp humk and ipr found in Table 8 8 Column I. Qual is again positive and statistically significant, but as before, the economic significance is scant. In this specificatio n, the coefficient on ipr remains negative but becomes statistically significant. However, sun_ipr the interaction term, is positive and statistically significant, indicating that a combination of sun and more robust IPR laws may lead to more solar patent s being filed. However, because the coefficient on the interaction term is so small, this positive effect, while statistically significant, does not seem to have any tangible economic significance. We see the same type of results for humk which is negativ e and statistically significant alone but has a positive, statistically significant, but minimal effect when interacted with sun Interestingly, sun by itself is negative and statistically significant. It could be that other, more important, factors influence the solar technology decision and therefore outweigh sun availability alone. The other independent variables perform similarly to those in Table 8 7 Column I, the regression without the sun interaction terms. Full Dataset: High Income Group Examining t he dataset trimmed to years 1991 and earlier is important to control for the problem that newer patents have fewer citations not necessarily because of low quality but because of young age. However, restricting the dataset also limits data availability, pa rticularly for one of my variables of interest, ipr Therefore, it is worth running the same regressions as above on the full dataset to see if we can discern significant differences. 72

PAGE 73

The summary statistics can be seen in Table 8 3 We see that in average year, the U.S. will file about 28 solar patents in country j (compared with 15 for the trimmed dataset). However, almost 49 percent of all patents filed in the U.S. will not be filed elsewhere (compared with about 55 percent for the trimmed dataset). 28 percent of patents will be filed in two to 20 countries (compared with about 31 percent in the trimmed dataset). On average, all the patents filed from the U.S. in country j in year t will receive a total of about 139 citations (the same as in the trimmed dataset). About 55 percent of the pijt pairs receive no citations (compared with almost 60 percent in the trimmed dataset). The measure of IPR strength, ipr has a mean of 3.18, well above the trimmed average of 2.61. I now analyze the results of the negati ve binomial regressions, found in Table 8 5 Column II. As with my earlier results, the sign on qual is positive while the sign on ipr is negative. The coefficient on qual is statistically significant at the 5% level, indicating that the difference in logs of expected counts of pijt would increase by about .023 units for a oneunit change in the aggregate quality ratio, while holding other variables in the model constant. In other words, a oneunit change in the aggregate quality ratio would cause pijt to in crease by about 2.3%. Using t he summary statistics in Table 8 3 we can calculate the effect at the mean: A oneunit change in qual will lead to about 0.65 more patents being filed in a given year. We can also calculate the effect within one standard deviation, which would be about 4.5 patents. This coefficient is again much smaller than both the full and trimmed results from the subset of OECD countries, where the effect for both was around 5%. The coefficient on ipr while negative as before, is again not statistically significant. 73

PAGE 74

We can also compare the other independent variables in Table 8 5 Column II with the previous OECD trimmed results, seen in Table 7 4, Column III. Humk the measure of human capital, is negative and statistically significant. gdp is positive and statistically significant, as before. d ist is positive and statistically significant. I mps is positive and statistically significant across all tables; renew is positive and statistically significant in Table 7 4 Column III, but negative and statist ically insignificant in Table 8 5 Column II. Finally, lang exhibits the same sign and significance across all results. Now we can examine the full highincome results with the sun interaction terms, found in Table 8 6 Column II. The most striking difference here is that ipr while negative as before, becomes statistically significant, as it was in Table 7 4, Column III. Meanwhile, the interaction of sun_ipr is positive and statistically significant. Gdp is positive and statistically significant, w hile sun_gdp is negative and marginally statistically significant. Humk is negative and statistically significant, while sun_humk is positive and statistically significant. Overall, these results seem to indicate that sun availability in concert with other factors, such as legal systems, infrastructure, and education, may encourage solar technology transfer rather than any of these factors alone. As expected, the need for translation, represented by the variable lang reduces the number of patents filed by a factor of about 0.5. The variable on renew while negative, is no longer statistically significant. Full Dataset: Lower Income Group We can now examine the full results for lower income nations. The summary statistics for the se data can be found in Tabl e 8 4 In an average year, the U.S. will file about 2.7 patents in country j (compared with 1.17 solar patents in the trimmed data, Table 3). However, similar to the trimmed lower income group, almost 84 percent of all 74

PAGE 75

patents filed in the U.S. will not be filed elsewhere. 9.3 percent will be filed in two to 20 countries (compared with just over 7 percent in the trimmed lower income group). On average, all the patents filed from the U.S. in country j in year t will receive a total of about 11 citations (the same for the trimmed lower income group). Almost 87 percent of the pijt pairs receive no citations (compared with 90 percent in the trimmed lower income group). The measure of IPR strength, ipr has a mean of 2.05, below the G&P average of 2.5 but above t he trimmed lower income average of 1.43. Examining the regression results, found in Table 8 7 Column II, we see that the signs on the main variables of interest, qual and ipr remain positive and negative, respectively, as before, but that neither is stat istically significant. The lack of statistical significance may reflect issues of data availability. Even though the full dataset includes observations from China and former Soviet nations, the 12 missing countries not included could reduce the accuracy of the results.3 The other independent variables, also seen in Table 8 7 Column II, perform similarly to those in the trimmed lower income regression. The only difference is that imps is not statistically significant. Now we can examine the results with the sun variable interacted with gdp humk and ipr found in Table 88 Column II. In this specification, q ual remains positive and statistically insignificant, but ipr keeps its negative sign while gaining statistical significance. As with the trimmed lower i ncome group, the sun_ipr interaction is positive and statistically significant. However, this is the only interaction term that is statistically significant in these results. Interestingly, sun by itself is negative and statistically 3 Armenia, Croatia, Cuba, Estonia, Georgia, Kazakhstan, Kyrgyzstan, Macedonia, Moldov a, Slovenia, Tajikistan, Montenegro. 75

PAGE 76

significant. Compared with other specifications, the independent variables in the Table 8 8 Column II results do not yield many statistically significant results. Overall, the results for higher income nations perform similarly to the results found in the subset of 16 OECD coun tries. The results here differ, however, from the only other empirical analysis that I am aware of examining the effect of stronger IPR laws on the patenting of environmental technologies. Dechezleprtre et al. (2011) conduct sector specific regressions for several environmental industries in 96 countries between 1995 and 2007. According to their analysis, stronger IPR laws have a positive effect that is statistically significant at the 1% level on solar patenting abroad. Several factors may account for the differences between my results and those of Dechezleprtre et al. First, the time frame and national composition of our data differ. I conduct analyses on both highand middleincome nations from 1952 1991 and 1952 2011 separately, while they analyze all countries together over a shorter time period. In addition, my data analyze only solar technology outgoing from the United States. Second, their analysis does not include the measure of patent quality that mine does. Indeed, when I run the analysis on my data without using the quality measure, the ipr variable becomes positive and marginally statistically significant for the highincome countries and negative for the lower income countries (but only statistically significant using the full data set). Third, Dechezleprtre et al. use a patent breadth measure of their own construction, which I do not include. Fourth, they use the Park & Lippoldt IPR index rather than the G&P index. Finally, they did not use interaction terms with the meteorological indicators With these differences taken together, it is not surprising that my results differ. 76

PAGE 77

Table 8 1 Trimmed, h igh i ncome n ations summary statistics Variable Obs. Mean Std. Dev. Min. Max. pijt 1440 14.57917 55.03398 0 666 qua lity 1440 139.1736 578.1591 0 8100 qual 641 8.230194 7.804564 0 68.5 ipr 928 2.61091 .7387082 0 4.675 humk 1400 .2344377 .1925699 .0281 1.303 gdp 1164 194521.8 538851.2 157.7524 5946800 dist 1400 7843.714 3445.941 0 15943 imps 890 3762704 8631188 2880 76100000 renew 974 .026694 .1612705 0 1 lang 1440 .2222222 .4158841 0 1 sun 1240 1957.592 542.626 1157.1 3353.55 Table 8 2 Trimmed, l ower i ncome n ations summary statistics Variable Obs. Mean Std. Dev. Min. Max. pijt 2120 1.174057 6.427171 0 107 quality 2120 10.98538 65.2599 0 1165 qual 255 9.22992 10.22286 0 83 ipr 967 1.432921 .7254176 0 4.341667 humk 1840 .1299278 .1215751 .0016 .904 gdp 1285 74662.89 176090.9 82.74619 1706318 dist 1960 8362.551 3541.677 1823 16360 imps 1099 990912.5 2442438 1 25300000 renew 1062 .0028249 .0530993 0 1 lang 2120 .1509434 .3580782 0 1 sun 1480 2284.929 442.3077 1317.562 3 468.708 77

PAGE 78

Table 8 3 Full, h igh i ncome n ations s ummary statistics Variable Obs. Mean Std. Dev. Min. Max. pijt 2160 28.12454 134.7981 0 2483 quality 2160 138.5278 552.8843 0 8100 qual 1103 6.317243 6.956223 0 68.5 ipr 1544 3.182611 .9967091 0 4.875 humk 2100 .3630812 .2930161 .0281 1.5598 gdp 1794 415509.3 1208561 157.7524 14400000 dist 2100 7843.714 3445.53 0 15943 imps 1183 6 231174 14700000 2880 156000000 renew 1614 .1765799 .381431 0 1 lang 2160 .2222222 .415836 0 1 sun 1860 1957.592 542.5531 1157.1 3353.55 Table 8 4 Full, l ower i ncome n ations summary statistics Variable Obs. Mean Std. Dev. Min. Max. pijt 3180 2.705346 23.98405 0 726 quality 3180 10.85975 59.85566 0 1165 qual 512 6.283733 8.560475 0 83 ipr 1653 2.055712 1.114057 0 4.541667 humk 2760 .2021941 .2030974 .0016 1.587 gdp 2146 194938.3 632307.7 82.74619 11300000 dist 2940 8362.551 3541.375 1823 16360 imps 1518 1953428 7348557 1 140 000000 renew 1602 .0892634 .2852126 0 1 lang 3180 .1509434 .35805 0 1 sun 2220 2284.929 442.2579 1317.562 3468.708 78

PAGE 79

Table 8 5 High i ncome n ations Variable I Trimmed II Full constant 2.830954* 2.520893* [.4477508] [.3077731] qual 0.0181586*** .0229523** [.0105798] [.01017] ipr 0.1546193 .1134582 [.136787] [.0897001] humk 0.5335707 .8676658* [.4366192] [.3011774] gdp 0.00000144* 0.000000941* [0.00000026] [0. 00000013] dist 0.00000747 .0000319** [.0000229 ] [.000016] imps 0.0000000209** 0.0000000154* [0.00000000839] [0.00000000411] renew .5076545** .035951 [.2506399] [.160591] lang 0.4824089* .7153815* [.1501773] [.1239607] N 457 644

PAGE 80

Table 8 6 High i ncome n ations w/ sun i nteraction t erms Variable I T rimmed II Full constant 9.385993* 8.117742* [1.455864] [1.075063] qual 0.0130424 .0208639** [.0103745] [.0097136] ipr 1.656895* 1.696221* [.5644635] [.4494788] humk 7.140312* 4.290176* [1.896204] [1.458858] gdp 0.00000103 0.00000327* [0.00000131] [0.000000915] dist .0000356 .0001097* [.0000354] [.0000263] imps 0.000000012 0.0000000344* [0.0000000265] [0.0000000124] renew .3990825*** .1183232 [.2351678] [.1496096] lang .69931* .5220402* [.1803713] [.1463313] sun .0034245* .0030976* [.0007041] [.0005033] sun_gdp 0.000000000434 0.000000000814*** [0.000000000681] [0.000000000487] sun_humk .0028144* .0015041** [.0008204] [.000608 4] sun_ipr .0008029* .0008275* [.0002868] [.0002213] N 396 557

PAGE 81

Table 8 7 Lower i ncome n ations Variable I Trimmed II Full constant 3.958188* 3.264327* [.3140219] [.2173395] qual .0138862*** .0073805 [.0082447] [.0076096] ipr .1000573 .0965665 [.1079751] [.0795142] humk 2.424992* 1.681088* [.5809949] [.3399262] gdp 0.00000119* 0.000000968* [0.000000224] [0.0000000892] dist .0002782* .0001806* [.0000254] [.0000169] imps 0.000000072* 0.00000000107 [0.0000000188] [0.0000000 0316] renew .2100385 .0133917 [.7352274] [.2606343] lang 1.642622* 1.426293* [.1863629] [.1370179] N 180 274 81

PAGE 82

Table 8 8 Lower i ncome n ations w/ sun i nteraction t erms Variable I Trimmed II Full constant 9.117117* 7.547661* [1.204876] [.7837144] qual .018779** .011745 [.0087269] [.0078011] ipr 1.23887** 1.500062* [.6244091] [.3439177] humk 13.22754** 1.012846 [5.407465] [3.490371] gdp 0.00000318 0.000000393 [0.00000237] [0.00000105] dist .0002424* .0002014* [.0000346] [.0000242] imps 0.0000000503** 0.00000000248 [0.0000000214] [0.00000000343] renew .0395182 .2416033 [.6674709] [.2485642] lang 1.875209* 1.741561* [.191813] [.1486871] sun .002239* .0018089* [.00048] [.0003371] sun_gdp 0.000000000942 0.000000000581 [0.00000000101] [0.000000000446] sun_humk .0048127*** .0000832 [.0028049] [.0017197] sun_ipr .0005203** .0005896* [.0002251] [.0001401] N 155 233 2 387.04 536.43 82

PAGE 83

CHAPTER 9 CONCLUSION In this dissertation, I examine the relationship between patent quality and the international transfer of solar technology. I also explore the relationship between IPR laws and ITT. The analysis includes a subset o f OECD members, as well as high and lower income nations. By examining the countries in these income groupings, I can determine if patent flows behave differently for these nations. To understand global technology flows in the international solar market, understanding the structure of the international solar market is helpful. First, this analysis noted that there is a chasm between countries that supply solar technology versus nations that manufacture solar technology. Specifically, most solar patents are from Japan and the United States, while most production occurs in China and the rest of Asia. Second, there is a divide between those nations that demand solar energy versus countries that supply it. Producers include Asian countries, while most demand comes from Europe and the United States. However, this is changing, with nations like China and Brazil implementing policies designed to encourage solar energy consumption. The literature shows that using patent citations is a proven method of measuring the overall quality of patents. However, accuracy of results is enhanced when the analyses can control for industry; hence, the importance of using disaggregated data, and the reason why this dissertation examines only one subsector of the greenenergy indust ry. When examining ITT, it is important to establish whether the chosen method of measuring technology transfer is the best for the analysis. Via a literature discussing the different measures of technology transfer, I show that patent counts can be a viable 83

PAGE 84

measure of technology transfer. While they do present some problems, these can be corrected by examining disaggregated patent flows and using a patent quality measure. Moreover, other methods of measuring technology transfer, including R&D expenditures and FDI flows, have proven to be even more problematic than patent counts. Therefore, even though the measure used here is not perfect, it is one of the better methods available to track levels of technology diffusion. The results of this analysis show that, on the whole, patent quality is a factor in the international transfer of solar technology; the variable of interest qual was positive and statistically significant in 19 out of 22 regressions. Although results of another study have shown a positive relationship between IPR laws and patent flows in the solar sector, my results show that when an aggregate quality measure is included along with IPR, IPR strength no longer has a positive effect; the variable of interest ipr was negative in 21 of 22 regress ions, and statistically significant in 16 of those. This may be due to the fact that globally, solar technology, even the newer PV variety, is fully developed and easily available, so IPR rights do not play a large role in this technology (Kirkegaard, 2010). However, this may change on the frontier of solar R&D, which today comprises solar nanotechnology ( Kirkegaard, 2010). Results were also fairly consistent between higher income and lower income nations. In addition, I added a meteorological variable to t he larger dataset to determine whether nations with more hours of sunlight on average see more incoming transfers of solar technology. My results show that on its own, sunlight is not a statistically significant indicator of solar patent filings. However, when interacted with other variables, such as the IPR measure, GDP, and human capital, sun has a positive, 84

PAGE 85

statistically significant relationship with incoming solar technology transfers. This may indicate that solar resources alone are not a deciding fact or in producing solar technology; other factors, such as income, infrastructure, and legal systems, may need to be developed first to attract solar technology. Overall, the results here confirm the importance of disaggregating data when examining internati onal technology transfer. This analysis shows that when it comes to IPR protection, solar technology is not the same as other technologies and sectors. Moreover, I have been able to show for the first time that quality is a factor the international diffusi on of solar technology. 85

PAGE 86

APPENDIX A COMPUTER PROGRAM USED TO GATHER DATA Below is a stepby step list of the process used to gather data on the patents and citations used in this study. 1. The program read the list of IPC codes 2. For given year and month rang e the program a. Accesse d web pages using the following templates http://worldwide.espacenet.com/searchResults?page=0&IC=[CODE]&DB= EPODOC&PD=[YEAR][MONTH]&locale=en_EP&ST=advanced&compact =false http://worldwide.espacenet.com/searchResults?page=0&IC=H01L31/00& DB=EPODOC&PD=197001&locale=en_EP&ST=advanced&compact=fals e b. Downloaded the HTML code and stor ed it on the local hard disk c. If the file did not contain all the results, the search was narrowed and broken down into days. In that case, the program downloaded H TML using the following template http://worldwide.espacenet.com/searchResults?page=0&IC=[CODE]&DB= EPODOC&PD=[YEAR][MONTH][DAY]&locale=en_EP&ST=advanced&co mpact=false d. If in one day t o o many patents were filed for the results to be returned on a single web page, corresponding files were downloaded manually 3. For each search result HTML file, the following actions were performed a. Ignored file s with more than 15 search results, as these files were broken down into smaller files as described above in 2 b. Follow ed each link in the search results and download its HTML content c. Analyze d each downloaded file to search for a more button, which linked to additional information. If the phrase wa s present, follow ed it and download complete data d. Followed the link View list of citing documents and downloaded its HTML content ( T his option may no longer be available on the website) 4. For each downloaded citations file the program then a. Check ed if the file had a reference to a next button (that is it checked if the citations w ere listed on more than one page) b. If needed, download ed the file with the next portion of citations c. Red id part s a. and b. until all citations were downloaded 5. Then the program create d a CSV data file patents.csv and 86

PAGE 87

a. Loaded each stored patent file and read and extract ed the data from the file b. Append ed the data into corresponding columns of the CSV file 6. Then the program create d a CSV data file citations.csv and a. Loaded each stored citation file and read and extract ed the data from the file b. Append ed the dat a into corresponding columns of the CSV file 7. At each step, the program check ed for the integrity of the downloaded data by ensuring that the entire file was downloaded, that it really contain ed data rather than an error message etc. After this procedure, the files were r eady to be imported into dataanalyzing software (MS Access, Stata ) The specific code used in this process is available upon request, as are the downloaded files themselves. 87

PAGE 88

APPENDIX B RESULTS WITH SUN VARIABLE ONLY Table B 1 High i ncome n ations w/ sun o nly Variable I Trimmed II Full constant 2.427767* 1.946772* [.5364347] [.3902793] qual .015032 .0202603** [.010917] [.0101822] ipr .091001 .0767611 [.1419479] [.093954] humk .7740828 .8 743347** [.499311] [.3435773] gdp 0.00000203* 0.00000184* [0.000000456] [0.000000247] dist .0000478 .0001184* [.0000338] [.0000257] imps 0.0000000305 0.0000000411* [0.0000000277] [0.0000000131] renew .4924884** .0987498 [.2443889] [.1583828] lang .691433* .6826475* [.1747335] [.1447772] sun .0000139 .0000676 [.0001197] [.0001034] N 396 557

PAGE 89

Table B 2. Low er i ncome n ations w /s un o nly Variable I Trimmed II Full constant 6.604848* 4.13132* [.7864616] [.3201093] qual .0165569*** .0077132 [.0089892] [.0083272] ipr .048417 .2546875* [.1156631] [.0937474] humk 6.505824* 1.248246* [1.783716] [.4369271] gdp 0.000000775* 0.000000838* [0.000000226] [0.000000085] dist .0002605* .0001376* [.0000344] [.0000219] imps 0.0000000379*** 0.00000000377 [0.0000000196] [0.00000000342] renew .1463312 .2903299 [.7009319] [.2808021] lang 1.6 75434* 1.503355* [.1848643] [.1601738] sun .0008577* .0004201* [.0001907] [.0001108] N 155 233 89

PAGE 90

REFERENCES Aanesen, Krister, Stefan Heck, and Dickon Pinner Solar Power: Darkest before dawn, McKinsey on Sustainability & Resource Productivity ( May 2012) Acosta, Daniel Daniel Coronado, and Ana Fernndez, Exploring the quality of environmental technology in Europe: evidence from patent citations Scientometrics 80, n o. 1 (2009) 133 154. Agencies, Global Solar Market Will Suffer from Tit for Tat Trade War Clean Biz Asia ( July 22, 2012) Albors Garrigos, Jos Jos Luis He rvas Oliver and Antonio Hidalgo Analysing high technology adoption and impact within public s upported high tech programs: An empirical case, Journal of High Technology Management Research, 20 no. 2 ( 2009), 153 168. Algieri, Bernardina; Antonio Aquino, and Marianna Succurro, Going Green: Trade Specialization Dynamics in the Solar Photovoltaic Sector Energy Policy 39 (2011) 7275 7283. APCO Worldwide, Market Analys is Report: Chinas Renewable Energy Industry APCO Worldwide Report November 2010. Barker, Michael, Policy changes in Brazil poised to drive PV growth in South America, Solarbuzz ( March 21, 2012) Barker, Michael, US Solar Market Entry Solarbuzz ( October 19, 2011) Barro, Robert and Jong Wha Lee A New Data Set o f Educational Attainment in the World, 19502010 NBER Working Paper No. 15902, April 2010. Barton, John, Intellectual Property and Access to Clean Energy Technologies in Developing Countries: An Analysis of Solar Photovoltaic, Biofuel and Wind Technologies ICTSD Issue Paper No. 2, 2007. Basberg, Bjorn L., Patents and the measurement of technological change: A survey of the litera ture, Research Policy 16 (1987) 131 141. Bloomberg News, Japan Poised to Become SecondBiggest Market for Solar Power The New York Times ( June 18, 2012) Branstetter, Lee, Raymond Fisman, and C. Fritz Foley, Do Stronger Intellectual Property Rights Increase International Technol ogy Transfer? Empirical Evidence from U.S. Firm Level Panel Data, The Quarterly Journal of Economics 121, no. 1, ( 2006) 321 349 90

PAGE 91

Branstetter, Lee, Raymond Fisman, C. Fritz Foley, and Kamal Saggi, Intel lectual Property Rights, Imitation, and Foreign Direct Investment: Theory and Evidence, NBER Working Paper No. 13033, April 2007. Byrne, John, Lado Kurdgelashvili, Manu V. Mathai, Ashok Kumar, Jung Min Yu, Xilin Zhang, Jun Tian, Wilson, Rickerson, and G ovinda R. Timilsina, World So lar Energy Review: Technology, Markets, and P olicies Center for Energy and Environmental Policy World Bank Report May 2010. Chen, YuShan and KeChiun Chang The Relat ionship Between a Firms Patent Quality and Its Market Value The Case of the U.S. Pharmaceutical I ndustry Technological Forecasting and Social Change, 77, no. 1, ( 2010) 20 33. Choudhury, Nilima, China Dominates Top Ten Global Solar Manufacturers PVTec h ( March 27, 2012) CIA World Factbook ( https://www.cia.gov/library/publications/the world factbook/ ) Cohen, Wesley M. Richard R. Nelson, and John P. Walsh Protecting Their Intellectual Assets: Appropriability Conditions and Why US Manufacturing Firms Patent (or Not), NBER Working P aper No. 7552, February 2000. Colville, Finlay, NPD Solarbuzz: Top10 PV cell producers in 2011, PVT e ch (January 23, 2012) --------, Finla y, PV equipment backlog becomes an US$8 billi on elephant in the room PVTech ( November 15, 2011) Comanor, W., and F.M. Scherer, F.M ., Patent statistics as measures of t echnical change, Journal of Political Economy 77, no. 3 ( 1969) 392 398. Dechezleprtre, Antoine, Matthieu Glachant, Ivan Ha Nick Johnstone, and Yann Mnire, Invention and Transfer of Climate Change Mitigation Technologies: A Global Analysis Review of Environmental Economics and Policy 5, no. 1, ( 2 011) 109 130 Dechezleprtre, Antoine, Matthieu Glachant, Ivan Nick Johnstone, and Yann Mnire, Invention and Transfer of Climate Change Mitigation Technologies on a Global Scale: A Study Drawing on Patent Data, CERNA Working Paper Series, Working Paper No. 201001, January 2010. Dechezleprtre, Antoine, Matthieu Glachant, and Yann Mnire, What Drives the International Transfer of Climate Change Mitigation Technologi es? Empirical Evidence from P atent Data CERNA, Centre dEconomie Industrielle Working Paper No. 201003, 2011. 91

PAGE 92

De Rassenfosse, Gaetan, and Bruno van Pottelsberghe de la Potterie, On the Price Elasticity of Demand for Patents Oxford Bulletin of Economics and Statistics 74, no. 1 (2012) 58 77. Eaton, Jonathan, and Samuel Kortum, International Patenting and Technology Diffusion, NBER Workin g Paper No. 4931, November 1994. -------, Trade in Ideas: Patenting and Product ivity in the OECD, NBER Working Paper No. 5049, March 1995. European Commission, Proposal for a Counci l Regulation on the Translation Ar rangements for the European Union Patent European Commission Regulation, 2010. European Patent Office Espacenet Patent Database ( http://worldwide.espacenet.com/advancedSearch?locale=en_EP ). European Photovoltaic Industry Association, Global M arket Outlook for Photovoltaics Until 2016, EPIA Report, May 2012. Evenson, Robert and Sunil Kanwar, Does Intell ectual Property Protection Spur Technological Change? Yale University Ec onomic Growth Center Discussion Paper No. 831, June 2001. Fallah, M. Hosein, Elliot Fishman, and Richard R. Reilly Fo rward Patent Citations as Predictive Measures for Diffusion of Emerging Technologies, PICMET Proceedings, August 2 6, 2009. Feenstra, Robert C. and Robert E. Lipsey, NBER United Nations Trade Data, 1962 2000. Fitzpatrick, Gary L., and Marilyn J. Modlin, Direct Line Distances: International Edition ( New York, NY: Scarecrow Press, 1986). Gallini, Nancy Jonathan Putnam, and Andrew Tepperman, Intellectual Property Rights and the Propensi ty to Patent, in Intellectual Pr operty Rights and Innovation in the KnowledgeBased Economy J. Putnam, ed. (Ottawa Ont.: Industry Canada, 2006). Ginarte, Juan and Park, Walter, Determinants of Patent Rights: A Cross National Study Research Policy 26 (1997) 283 301. Gonzlez Bea, New Editio n of the Concentrated Sol ar Thermal Power Markets Report Released Solar Thermal Magazine ( August 8, 2012) 92

PAGE 93

Griliches, Zvi, Patent Statistics As Economic Ind icators: A Survey Part I, NBER Working Paper No. 3301, March 1990. Groba, Felix, Environment al Regulation, Solar Energy Technology Components and International TradeAn Empirical Analysis o f Structure and Drivers, World Renewable Energy Congress Proceedings, 2011. Hall, Bronwyn H. Adam B. Jaffe, and Manuel Trajtenberg, The NBER Patent Citatio ns Data File: Lessons, Insights, and Methodolog ical Tools, NBER Working Paper No. 8498, October 2001. -----, Market Value and Patent Citations, RAND Journal of Economics 36 no. 1 ( 2005) 16 38. Harhoff, Dietmar, Francis Narin, F.M. Scherer, and Katr in Vopel Citation Frequency and the Value of Patented Inventions, The Review of Economics and Statistics 81, n o. 3 (1999), 511 515. -----, Citations, Family Size, Oppos ition, and the Value of Patent Rights Research Policy 32 (2003) 1343 1363. Harhoff, Dietmar, Measuring and estimating patent value, WIPO OECD Workshop on Statistics in the Patent Field, 2003. Climate Policy and Technologi cal In novation and Transfer: An Overview of Trends and Recent Empirical Results, OECD Envi ronment Working Papers, No. 30, 2010. Hayward, Jenny and Paul Graham, Developments in technology cost drivers dynamics of technological change and market f orces, Commo nwealth Scientific and Industrial Research Organisation Paper March 2011. Heston, Alan Robert Summers, and Bettina Aten, Penn World Table Version 7.0, Center for International Comparisons of Produ ction, Income and Prices at the University of Pennsylvania, May 2011. Hirschey, Mark and Vernon Richardson, Valuati on Effects of Patent Quality: A Compariso n for Japanese and U.S. Firms Pacific Basin Finance Journal 9, n o. 1 ( 2001) 65 82. -----, Are Scientific indicators of Patent Quality Useful to Investors?, Journal of Empirical Finance, 11, no. 1 ( 2004) 91 107. Holm, Dieter, Renewable Energy Future for the Developing World: White Paper International Solar Ene r gy Society White Paper 2005. 93

PAGE 94

Hopwood, David, Which countries are winning the race to invest in renewables? Renewable Energy Focus Blog ( April 27, 2011) International Energy Agency RD&D Statistics ( http://data.iea.org/IEASTORE/DEFAULT.ASP ). International Monetary Fund, World Economic Outlook Database ( http://www.imf.org/external/pubs/ft/weo/2012/01/weodat a/index.aspx ). Jaiswal, Anjali, Ambitious Solar Program in India Driving Prices to Impressive Lows NRDC Press Release, April 25, 2012. Javorcik, Beata Smarzynska, The Composition o f Foreign Direct Investment and Protection of Intellectual Property Rights: Evidence from Transition Economies European Economic Review 48 (2004) 39 62. Johansson, Thomas B. Ana nd Patwardhan, Nebojsa Nakicenovic, and Luis Gomez Echeverri, eds., Global Energy Assessment (Cambridge, U.K.: Cambridge University Press 2012 ). Kahn, Shayle, Emerging Trends in the U.S. Solar Mark et, Enterprise Florida and GTM Research Paper November 2009. Kanwar, Sunil, Intellectual Property Protection and T echnology Transfer: The Case of Overseas R&D Delhi School of Economics C entre for Development Economic s Working Paper No. 166, April 2009. Keller, Wolfgang, International Trade, Foreign Direct Investment, and Technology Spillovers, NBER Working Paper No. 15442, October 2009. Kirkegaard, Jacob Funk Thilo Hanemann, Lutz Weischer, and Matt Miller, Towar d a Sunny Future? Global Integration in the Solar PV Industry, World Resources Institute and Peterson Institute for International Economics Working Paper No. 106, May 2010. Kortum, Samuel and Josh Lerner, Stronger Protecti on or Technological Revolution: What is Behind the Recent Surge in Patenting? NBER Working Paper No. 6204, September 1997. Kumar, Nagesh, Intellectual Property Protection, Mar ket Orientation and Location of Overseas R&D Activities by Multinational Enterprises World Development 24, n o. 4 ( 1996) 673 688 Lall, Sanjaya, Indicators of the Relative Importance of IPRs in Developing Countries, ICTSD UNCTAD Issu e Paper No. 3, 2003. 94

PAGE 95

Lanjouw, J .O., and M. Schankerman, The Quality of I deas: Measuring Innovation with Multiple Indicators NBER Working Paper No. 7345, September 1999. -----, Patent Quality and Research Productivity: Measuring Innovation with Multiple Indicators Economic Journal 114 no. 495 (2004) 441 465. Latif, Ahmed Abdel, Keith Maskus, Ruth Okediji, Jerome Reichman, and Pedro Roffe Overcoming the Impasse on Intellectual Pro perty and Climate Change at the UNFCCC: A Way Forward International C entre for Trade and Sustainable Development, Pol icy Brief No. 11, November 2011. Lian, Ray, ChinaThe Largest Global PV Market in 2012? Solarbuzz ( April 3, 2012) Marco, Alan C., The Dynamics of Patent Citations Economics Letters 94, no. 2 ( 2007) 290 296. Maskus, Keith, Encouraging International Technology Transfer ICTSD UNCTAD Issue Paper No. 7, May 2004. Maskus, K., and L. Yang, Intellectual Property Right s and Licensing: An Econometric Investigation, in Intellectual Property and D evelopment: Lessons from Recent Economic Research, C. Fink, and K. Maskus, e ds. ( Washington DC: The International Bank for Reconstruction and Development/The World Bank 2005). Maskus, Keith, The Role of Intellectual Property Right s in Encouraging Foreign Direct Investment and Technology Transfer in Intelle ctual Property and D evelopment: Lessons from Recent Economic Research, C. Fink, and K. Maskus, e ds. ( Washington DC: The Internatio nal Bank for Reconstruction and Development/The World Bank 2005 ). Maskus, Keith and Mohan Penubarti, How TradeRe lated Are Intellectual Property Rights? Journal of International Economics 39, (1995) 227 248 McCrone, Angus, Solar Power Project M&A Hits New Record in 2011, Solar Thermal Magazine (July 19, 2012). Mendolia, Michael, The Sol ar Market: An uncertain year ahead, IBT Partners Blog ( February 19, 2012) Movellan, Junko, US Solar Photovoltaic Market More than Doubles in 2011, Ex ceeding 2 GW Solarbuzz ( 2012) Mufson, Steven, Chinas growing share of solar market comes at a price, The Washington Post ( December 16, 2011) 95

PAGE 96

Nelson, Andrew J., Measuring Knowledge Spillovers: What patents, licenses, and publications reveal about innovation diffusion, Research Policy 38 (2009) 994 1005. Norback, Pehr Johan, Lars Persson, and Rog er Svensson, Creative Destruction and Productive Preemption, Center for Economic Policy Research Di scussion Paper Series No. 8281, March 2011. OECD/IEA Medium Term Renewable Energy Market Report, International Energy Agency Report, 2012. OECD/IEA, Deploying Renewables 2011: Best and Future Policy Practice, International Energy Agency Report 2011. OECD/IEA, World Energy Outlook (Paris: International Energy Agency 2011). OECD/IEA Renewable Energy Technologi es: Solar E nergy Perspectives, International Energy Agency Report, 2011. Oltra, Vanessa R en Kemp, and Frans de Vries, Patents As a Measure for Eco Innovation, MEI Project Report June 3, 2008. Oxford Analytica Solar energy to boom amid global market tension ( July 27, 2012) Park, Walter G., Int ellectual Property Rights and International Innovation, in Intellectual Property, Growth and Tr ade (Frontiers of Economics and Globalization, Volume 2) Keith E. Maskus, ed. (Bingley, UK: Emerald Grou p Publishing Limited, 2007) Phalin, Amanda J., Paten t Quality and the Internat ional Transfer of Environmental Technology: An Investigation of the Solar Technology Sector University of Florida Department of Economics 2nd Year Paper 2012. Pillu, Hugo and Gilles Kolda, Induced Innovation and International Technological Opportunity in the Field of Energy: Evidenc e from World Patent Citations, Equipe de Recherche en Analyse des Systmes de Mondialisation Economique Paper, June 2009. Platzer, Michaela D., U.S. Solar Photovoltaic Manufac turing: Industry T rends, Global Competition, Federal Support, Congressional Research Service Report June 13, 2012. Popp, David, The Role of Technological Change in Green Growth, NBER Working Paper No. 18506, November 2012. 96

PAGE 97

-----, International Technology Transfer, Climate Change, and the Clean Developme nt Mechanism Review of Environmental Economics and Policy 5, no. 1 ( 2011) 131 152 Popp, David, and Neelaksi Medhi Technology and the Diffusion of Renewable Energy Energy Economics 33, (2011) 648 662 Popp, D., They Don't Invent Them Like They Used To: An Exam ination of Energy Patent Citatio ns Over Time Economics of Innovation and New Technology 15 no. 8 (2006), 753 776 Poullikkas, Andreas, Technology and market future prospects of photovoltaic systems International Journal of Energy and Environment 1, no. 4 ( 2010) 617 634. Rafiquzzaman, Mohammed, and Lori Whewell, Recent Jumps in Patenting Activities: Comparative Innovative Performance of Major Industrial Countries, Patterns, and Explanations Industry Canada Research Publications Program Working Paper No. 27, December 1998. REN21, Renewables 2012 Global Status Report (Paris: REN21 Secret ariat 2012) Rosenkopf, Lori and Atul Nerkar, Beyond Local Search: Boundary Spanning, Exploration, and Impact in the Optical Disk Industry, Strategic Management Journal 22, (2001), 287 306. QMS Partners, Solar Energy Market, Thin Film Technology, 2009: A Global Market Research Report, 2009. Sawhney, Aparna and Matthew E. Kahn, Understanding crossnational trends in high tech renewable power equipment exports to the United States, NBER Working Paper No. 17217 July 2011. Sawin, Janet L., and Eri c Martinot, Renewabl es 2010: Global Status Report, Renewable Energy Policy Network for the 21st Century Report September 2010. Scherer, Frederic, Using Linked Patent and R& D Data to Measure Interi ndustry Technology Flows in R&D, Patents, and Productivity Zvi Griliches, ed. (Chicago: Univer sity of Chicago Press, 1984) Schmookler, Jacob, and Oswald Brownlee, Dete rminants of Inventive Activity, American Economic Review 52, no. 2 (1962), 165 1 76. Sharma, Atul, A comprehensive study of solar power in India and World, Rene wable and Sustainable Energy Reviews 15 (2011) 1767 1776. Solarbuzz Solar Energy Market Growth ( 2010) 97

PAGE 98

Solar Energy Industries Association, Solar Industry Data: U.S. Market Installs 506 MW in Q1 2012 ( 2012) Steiner, Achim, Benot Battistelli, and Ricardo M elndez Ortiz, Patents and Clean Energy: Bridging the Gap Between Evidence and Policy Final Report, UNEP, European Parliament, and ICTSD Report 2009. Susman, Gerald I., Evolution of the Solar Energy Industry: Strategic Groups and Industry Structure, PICMET Proceedings, 2008. Timilsina, Govinda R., Lado Kurdgelashvili, and Patrick A. Narbel, A Review of Solar Energy: Markets, Economics, and Policies, The World Bank Development Research Group Environment and Energ y Team, Policy Research Working Paper No. 5845, October 2011. Trajtenberg, Manuel, A Penny for Your Quotes: Pa tent Citations and the Value of Innovations, The RAND Journal of Economics 21, no. 1 ( 1990 ) 172 187 United Nations Data Explorer ( http://data.un.org/Data.aspx?d=CLINO&f=ElementCode%3a15). UN Environment Programme, Global Trends in Green Energy 2009, UN Environment Programme Repor t, July 2010. U.S. Department of Energy, 2010 Solar Technologies Market Report, November 2011. Van Zeebroeck, Nicolas, The puzzle of patent value indicators, Economics of Innovation and New Technology 20, no 1 ( 2011) 33 62. Venezia, John, and Jef f Logan Weighing U.S. Energy Options: The WRI Bubble Chart WRI Policy Note ( July 2007) Williams, Leslie, Patenting solar energy innovations PV Magazine January 2009. World Intellectual Property Organization, IPC Green Inventory ( http://www.wipo.int/classifications/ipc/en/est/ ). W orld I ntellectual P roperty O rganization, Patent Based Technology Analysis Report Alternative Energy, WIPO Report, 2009. World Resources Institute, Statement: A Climate Deal Comes Together in Durban, WRI Press Release, December 11, 2011. 98

PAGE 99

BIOGRAPHICAL SKETCH Amanda J. Phal in earned a B.A. in French and international s tudies from Vassar College and an M.A. in international economic a ffairs from George Washington University. She worked as a journalist covering international news for 10 years be fo re pursuing further study in economics at the University of Florida. She earned an M.A. in e conomics from UF and will complete her Ph.D. in May 2013. 99