Differences in diffusion of Renewable Energy Technologies in
OECD and BRIICS countries: The moderating role of
environmental policy stringency
Master Thesis
MSc International Business & Management
Anna Visser a.visser.32@student.rug.nl Student number: S3199819
University of Groningen, the Netherlands Faculty of Economics and Business
ii "Vision without action is just a dream, action without vision just passes the time, and vision
with action can change the world."
iii ABSTRACT
This study gives more insight in the potential drivers for promotion of Renewable Energy Technology (RET) diffusion for both emerging and developed countries. Despite the big potential of RET, the diffusion of it remains low in both sets of countries. The innovation system view will be used to see the differences in RET diffusion between these sets of countries. In addition, it will show the importance of environmental policy stringency on the innovation system of a country. This research contributes to both the environmental and energy economics literature and environmental policy literature. By using panel data from 2003 till 2012 of 27 OECD countries and 6 BRIICS countries, it is empirically confirmed that emerging countries are not lagging behind developed countries in their RET diffusion, but different innovation systems occur. Specifically, developed countries have a greater enabling environment to radically change the innovation system. Emerging countries lack this enabling environment which is not favourable for the promotion of renewables. The innovation systems result in different technology trajectories and sustainable growth paths. In addition, results show that more stringent environmental policies will increase the promotion of RET diffusion. In order to become the pioneer in RET and account for the big increase in carbon emissions, emerging countries need to restructure their current energy system.
iv TABLE OF CONTENTS
LIST OF FIGURES AND TABLES ... v
LIST OF ABBREVIATIONS ... vi
1. INTRODUCTION ... 1
2. LITERATURE REVIEW ... 4
2.1 Renewable Energy Technologies (RET) ... 4
2.2 Mechanisms to avoid climate change ... 4
2.3 Perspectives on slow RET diffusion ... 6
2.4 Determinants RET diffusion ... 7
2.4.1 Socioeconomic environment ... 8
2.4.2 Technological environment ... 9
2.4.3 Physical infrastructure ... 10
2.5 Moderation effect: Stringency of environmental policies ... 10
3. METHODOLOGY ... 13
3.1 Sample ... 13
3.2 Data collection ... 13
3.2.1 Dependent Variable: RET diffusion ... 13
3.2.2 Explanatory variables ... 14
3.2.3 Moderating effect: environmental policy stringency ... 15
3.2.4 Control variables ... 16
3.3 Method and empirical model ... 17
4. RESULTS ... 20
4.1 Pre-analysis and descriptive statistics ... 20
4.2 Results PCSE model ... 22
4.2.1 Results determinants RET diffusion... 25
4.2.2 Results moderating effect stringency of environmental policies ... 26
4.3 Robustness checks: OLS and Fixed effect model ... 27
5. DISCUSSION ... 29
5.1 Discussion ... 29
5.2 Limitations and future research recommendations ... 33
6. CONCLUSION ... 35
REFERENCES ... 36
APPENDICES ... 42
Appendix A: Normality Test RET Diffusion ... 42
Appendix B: Hausman Test ... 43
Appendix C: Details specification Tests ... 44
Appendix D: Multicollinearity checks ... 45
Appendix E: Correlation matrix including all variables... 46
Appendix F: Descriptive statistics RET diffusion ... 47
Appendix G: OLS results different models ... 49
v LIST OF FIGURES AND TABLES
Figure 1: Conceptual model of hypotheses……… 12
Formula 1: Econometric model……….. 19
Table 1: Descriptive statistics all variables ……… 20
Table 2: Descriptive statistics EPS indicator ………. 21
Table 3: PCSE results developed countries ……… 23
Table 4: PCSE results emerging countries ………. 24
Table 5: Summary of estimated effects different techniques ………. 28
Appendices Figure A: Distribution of Renewable Energy Technology Diffusion ……… 42
Table A: Descriptive statistics of dependent variable RET Diffusion ………... 42
Table B: Hausman Test ………. 43
Table C: Specification tests……… 44
Table D: Multicollinearity Statistics: ……… 45
Table E: Correlation matrix including all variables ……….. 46
Figure F1: Average RET Diffusion developed and emerging countries ……… 47
Figure F2: RET diffusion emerging countries ……… 47
Figure F3: RET diffusion developed countries c_id <20 ………... 48
Figure F4: RET diffusion developed countries c_id>19 ……… 48
Table G1: OLS results developed countries ………... 49
Table G2: OLS results emerging countries ……… 50
Table H1: Fixed effect model results developed countries ……… 51
vi LIST OF ABBREVIATIONS
Concepts
CO2 Carbon Dioxide Emissions EPS Environmental Policy Stringency FDI Foreign Direct Investment GDP Gross Domestic Product
PACE Pollution Abatement and Control Expenditure R&D Research & Development
RE Renewable Energy
RET Renewable Energy Technology WEF World Economic Forum
Statistical concepts
CTV Critical Tolerance Value
FE Fixed Effect
OLS Ordinary Least Squares
PCSE Panel Corrected Standard Errors VIF Variance Inflator Factors
Organizations and others
BRIICS Brazil, Russian Federation, Indonesia, India, China and South Africa EIA Energy Information Administration
EU European Union
IEA International Energy Agency
1 1. INTRODUCTION
In the last twenty years the global energy sector is subject to a number of changes. Due to a rapid global social and economic growth, the demand for energy is increasing. This is especially the case in emerging countries who will account for 90% of the energy demand growth in 2035 (OECD, 2011). In order to meet this energy demand most countries depend on conventional fossil fuels, such as oil, natural gas, and coal, which results in a significant increase in carbon emissions. Eventually, the supply of these fossil fuels will be depleted, which will endanger energy security. In order to reduce global carbon emissions and dependence on fossil fuels, energy systems within countries need to change. One way to change is investing in energy efficiency technology, in which the energy intensity will be reduced but the carbon infrastructure remains unchanged (Pfeiffer & Mulder, 2013). Another way to change is to transition towards a low-carbon economy by increasing the capacity of renewable energy technology (RET), which is called RET diffusion. The RET diffusion process will diminish the environmental burden of economic activities (Jaforullah & King, 2015) and create socioeconomic benefits (e.g. creation of jobs) for countries (Omer, 2009).
Although different countries have implemented mechanisms to increase the adoption of RET, the global RET diffusion is too slow and still in its early stages. Countries around the globe face difficulties in the diffusion of RET because of commercialization barriers (e.g. underdeveloped infrastructure, and poor economies of scale) and market barriers (e.g. institutional barriers, high upfront cost, absence of level of playing field, and split incentives) (Jaffe, Newel, & Stavins, 2005). The slow RET diffusion can be explained by two different perspectives. Firstly, the neo-classical economic paradigm sees the market failures, that result from the above mentioned market barriers, as the main barrier for RET diffusion (Arrow, 1962). This view shows that market failures result in too little investment in both innovation and diffusion of RET. The neoclassical view states that in order to encourage RET diffusion, public intervention in energy efficiency technologies should focus on balancing the RET prices (e.g. R&D subsidies). This will lead to close to the market innovations instead of radical innovations in the existing energy systems (Johnstone et al., 2010).
2 technology. The RET diffusion has been slow because the current innovation systems are not in favour of the radical change needed. In order to develop and diffuse a new technology, a new technological system needs to be build up which is significantly different from the existing technological system. The environment of the innovation system can be divided into socioeconomic, technological, infrastructural, and institutional factors, which influence the current innovation system (Jacobsson & Johnson, 2000). The innovation system view shows that public intervention is crucial to encourage certain technology trajectories, to create an enabling environment for RET and to show commitment for more sustainability which goes far beyond only market failures.
In addition, scholars have shown the importance of environmental policy stringency. In order to speed up the diffusion of RET, lots of countries (both emerging and developed) signed the Kyoto protocol, where binding agreements on environmental strategies have been signed (UNFCC, 2010). The policies with the aim to decrease the carbon emissions within countries became more stringent, which shows the sustainability commitment of a country and influences different parts of the innovation system (Johnstone, Hascic & Popp, 2010). Therefore, more stringent environmental policies may encourage RET diffusion and less stringent environmental policies may discourage RET diffusion.
Thus, the innovation system view states there are many other aspects in an environment than only the market aspects which are determining the diffusion of RET. It shows that the configurations in the environment and the build-up of new innovation systems within countries need to change to have more RET diffusion. To show how much an innovation system needs to change, this study will show how the existing innovation systems look like and how it currently influences RET diffusion. This study tries to find the differences in innovation systems between emerging and developed countries. Therefore, this study answers the following question:
Is the diffusion of RET similar in OECD and BRIICS countries and how important is the institutional commitment?
locked-3 into fossil fuel energy technologies which is not visible in emerging countries. Furthermore, this study argues that more stringent environmental policies, which are visible in developed countries, gives greater pressures in the innovation systems towards RET diffusion.
I test my theoretical model on a sample of 6 emerging countries (BRIICS), and 27 developed countries (OECD). A longitudinal research design is carried out to find the influence of socioeconomic, technological, physical, and institutional environment on RET diffusion. A panel data regression is conducted from 2003 to 2012. This time frame has been chosen because of availability of data and because of a worldwide significant development in RET diffusion. Since autocorrelation, heteroscedasticity, and contemporaneous correlation is detected in the data, the Panel Corrected Standard Error (PCSE) model is used.
This study contributes to the existing literature in several ways. Firstly, environmental and energy economic literature primarily focused on energy efficiency technologies instead of RET diffusion. This study focuses on RET diffusion since an increase in RET capacity is needed to radically change existing energy systems. Secondly, there is a general lack of research in environmental economics literature on RET diffusion, especially on emerging countries. This study gives more insight in the potential of emerging countries by showing the existing innovation system and the dynamics of transformation processes. This insight is needed because expectations show that emerging countries will become a pioneer in RET. By showing the existing innovation systems and ways to change it into an enabling environment for RET diffusion, this study contributes to the macroeconomic literature. Thirdly, this study empirically shows from the innovation system perspective the different influences of socioeconomic, technological, and infrastructural factors on RET diffusion for both sets of countries. Lastly, the critical role of institutional commitment on the innovation systems within countries will be shown. It will shed light on the importance of stringent environmental policies for RET diffusion. The newly developed EPS index of the OECD is a contribution to the existing literature. These in-depth studies are almost not existing in the current literature.
4 2. LITERATURE REVIEW
2.1 Renewable Energy Technologies (RET)
Since the globalization, the energy demand has been increasing rapidly. Currently, the primary energy sources of most countries to meet the energy demand are conventional fossil fuels like coal, oil, and natural gas. Recent concerns with the current energy regime led to a wide interest in renewable sources. Reasons for these concerns are the insecurity of the energy supply, potential depletion of conventional fossil fuels, increasing CO2 emissions, and environmental burdens because of conventional fossil fuels (IEA, 2006). These concerns have led to a global action towards different ways of producing, delivering, and consuming energy, namely from renewable sources. Diversification of RET such as the adoption of biomass, tides, geothermal, biofuels, and solar and wind energy occurred (OECD, 2010).
According to the IEA (2006), the RET can be distinguished into three generations: (1) first-generation technologies which are matured, examples are hydropower, biomass, and geothermal; (2) second-generation technologies which are facing an increase in development, examples are solar, wind, and new forms of biomass; and (3) third-generation technologies which are in the early development stages like modern solar, modern geothermal, and ocean. The use of more renewable sources can create substantial environmental and socio-economic benefits (Zhao, Tang & Wang, 2013). It can reduce the greenhouse gas emissions, use more local resources for energy supply, create more jobs, increase access to energy, reduce the dependence on fossil fuels, and improve the energy security (Rifkin, 2011; Omer, 2009).
In contrast with the big potential of RET, literature shows that it is still accounting for a small fraction of the world’s primary energy supply (Masini & Menichetti, 2013). Numbers show that in 2010, the total net electricity power generated from renewable sources was less than 21% of the global total net electricity generation in the world (Zhao et al., 2013). The biggest barrier towards an increasing number of generation from RE sources is the high uncertainty and the high fixed costs associated with RE projects in comparison to conventional energy resources (Menanteau, 2000). Therefore, lots of governmental support has been introduced to decrease the uncertainty of RE projects and create an enabling environment for RET.
2.2 Mechanisms to avoid climate change: energy efficiency vs. RET diffusion
5 efficiency technology or to reduce the carbon intensity of energy usage (Holdren, 2006). Firstly, countries can invest more in energy efficiency technology, with the potential of fast adoption because of cost savings for relevant actors. Rose and Joskow (1990) found that adoption of more environmental friendly technology by U.S. electric utilities was positively correlated with the fuel prices. These kinds of energy efficiency innovations are more gradual and environmental regulations are most of the times not needed because of the cost saving benefits. The downside of more energy efficiency technology is that it is not reducing the carbon infrastructure of the energy systems in countries since efficiency technologies still depend on fossil fuel energy sources. Therefore, energy efficiency technologies will not radically change the innovation system within countries.
6 different RET varies over time and locations since existing systems are changing. The above discussion shows that different technological trajectories are expected to be followed.
2.3 Perspectives on slow RET diffusion
Although different countries implement different mechanisms to increase the adoption of RET, the global RET diffusion is too slow to deal with the current climate change. Figures show that the diffusion process of RET is a very slow and gradual process (OECD, 2010). Although the EU increased their share from 5.7% in 1994 to 8.2% in 2007, this is still too slow in comparison with the ambitions of the policy makers (OECD, 2010). In general, the share of renewables is higher in emerging countries in comparison to EU countries, but both diffusion rates still remain low (OECD, 2010).
Literature shows two different perspectives on the slow diffusion, namely the neo-classical view and the innovation system view. The neo-neo-classical view highlights the market failures as the main barrier to an energy transition (Arrow, 1962). These market failures are categorized by several market barriers (e.g. high upfront costs, absence of level of playing field, high uncertainty, and split incentives within the different stakeholders) (Jaffe, Newel, & Stavins, 2005). Jaffe et al. (2005) show that the ‘invisible hand’ as coined by Adam Smith allows too much pollution externalities and too little investment in both innovation and diffusion of RET. Reason for this is the economic incentive of firms in which they try to reduce their internal costs, in which the external costs of pollution are not included. The external costs of pollution are not for the firms but for the society as a whole since current regulations do not restrict firms to take the external costs into account. Therefore, the market fails to put a price on the external environmental pollution (Jaffe et al., 2005). In order to create an enabling environment, institutional efforts like implementation of innovation and environmental policies, need to be taken. From the neoclassical point of view, governments need to get the prices right and implement public policies to balance the low investment in R&D by firms. The limitation is that it does not give any direction where subsidies and public policies must be focused on (Smith, 2000). Intervention in balancing the prices of new technologies will lead to more close to the market innovations (Johnstone et al., 2010). These innovations will not radically change current energy systems since the technological systems remain unchanged when public intervention is only focused on price changes of a certain technology instead of system changes.
7 view is focusing on the innovation system of a country and argues that the direction, speed, and success of innovations are influenced by the environment in which innovations are developed and deployed (Freeman, 1995; Lundvall, 1992). The environment of the new technology is categorized by innovation systems which are the configurations of actors, rules, and physical infrastructures and their relationships (Negro et al., 2012). The goal of innovation systems is to develop and diffuse the new technology and technological knowledge (Jacobsson & Johnson, 2000). In other words, the structure around a new technology is called the innovation system. In order to develop and diffuse a new technology, which is a radical change in the case of RET, a new technological system that significantly differs from the existing technological system, needs to be build up. The new structures and the processes within these structures (‘system functions’) determine if the new technology will be a success or a failure. Since there is strong interrelatedness between the existing energy system and the economic system and because of the inertia of technological systems, the RET diffusion is slow (Hughes, 1983). These ‘system failures’ hamper the development of new technological innovation systems which are needed for more RET diffusion (Jacobsson & Johnson, 2000). More specifically, system failures can be categorized into socioeconomic/market, technological/capability, physical infrastructure, and institutional factors. Building a new technological innovation system requires more radical innovations which results in greater development times and greater chance of failure (Van de Ven et al., 1999). The difference with the neoclassical view is that the innovation system view states many other aspects in an environment than only the market aspects which are determining the slow diffusion of RET. It can be seen that a choice for a certain technology is not only led by the choice of an individual firm, but resides from an ‘innovation system’ which both encourage and diminish the choice of a certain technology (Jacobsson & Johnson, 2000). Using innovation system approaches in the policy intervention has a greater potential of identifying where the public support should go to and which of the environmental factors show a weak performance in the new technological system (Alkemade et al., 2011). The view sheds light on the dynamics of innovation system transformations.
2.4 Determinants RET diffusion
8 diffusion. In line with the innovation system view, the environment can be divided into socioeconomic, technological, infrastructural, and institutional factors (Jacobsson & Johnson, 2000). This environment will be used to see differences in existing innovation systems.
2.4.1 Socioeconomic environment
The socioeconomic environment reflects the overall business environment and development including the social and economic conditions of a country. Most of the developed countries literature suggest that environmental concerns like CO2 emissions are incentives for RET diffusion (Sadorsky, 2009; Van Ruijven & van Vuuen, 2009). CO2 emissions reflect the environmental damage caused by economic development. The diffusion literature further expects that a greater level of income leads to more diffusion of RET because it is easier to account for the high regulatory costs and there are more resources available to invest in RET (Marques, Fuinhas & Pires Manso, 2010). Thus, wealthier countries can afford to be more concerned with the environment and use more renewables. Furthermore, the energy needs of a country show how much demand there is for energy. When the needs are rising there is more pressure on the supply side, both fossil as non-fossil fuels. Therefore, either a positive or negative effect on RET diffusion is expected. Since the rapid industrialization, emerging countries face a significant increase in GDP, energy needs, and CO2 emissions which results in more traditional technology trajectories and makes it difficult to diffuse RET (O’Conner, 1996). Particularly, the socioeconomic development increases the share of fossil fuel energy sources to meet the demand, which is reflected in the inverted Environmental Kuznets curve and the need logic (e.g. O’Connor, 1996). An increased share of fossil fuel sources is the result of the economic development incentives in the existing energy system of emerging countries in the short run (Romano et al., 2016). In the long run this relationship can be different, because economic development will lead to more resource availability to develop new RET and there will be more normative pressures from other countries to pollute less. Furthermore, emerging countries are faced with lower opportunity costs when the RET comes from developed countries which will lead to higher adoption of that technology.
The above discussion shows that developed countries have more means to speed up the RET diffusion. A higher level of socioeconomic development will increase the environmental concerns and may lead to more adoption of RET. Therefore, I hypothesize:
9
2.4.2 Technological environment
The technological environment shows which energy technologies play a role in the business environment and the ability of a country to work with these technologies, which is reflected in the competitiveness in RET. Through both internationalization and development in human capital, the technological environment enables more RET diffusion. Firstly, internationalization, especially FDI inflow, is an important driver for RET adoption. In particular, FDI inflow allows countries to use foreign knowledge to advance the RET development and to follow technological trajectories found in foreign countries. This technical learning results in more mimic activities in RET development, technological capabilities, and environmental incentives (Nelson, Peterhansl, & Sampat, 2004). It is expected that more internationalization activities will result in more RET diffusion.
Secondly, the decision of a country to adopt a technology depends on the capability and capacity of the people in a country to work with the particular technology. Benhabib and Spiegel (2005) show that the human capital also facilitates the adoption of RET and the associated productivity growth. Human capital is defined as “the skills, knowledge, competences and other attributes embodied in individuals that are part of economic activity” (OECD, 2001: 18). It is expected that more human capital endowments will result in more RET diffusion. Developed countries are seen as the ‘leading markets’ in RET because of their amount of FDI inflow and human capability endowments (Popp, Hascic, & Medhi, 2011). In contrast, there are some mixed thoughts on emerging countries. On the one hand, emerging countries have lower technological capabilities which are expected to negatively influence the RET diffusion since most of the R&D efforts are focused on advances in energy-intensive technologies (Bodas Freitas et al., 2012). On the other hand, emerging countries have a latecomer advantage in which they tend to improve the RET of developed countries (Mathews, 2007). The latecomer advantage results from lower opportunity costs and technological learning through internationalization activities (Lanjouw & Mody, 1996). Emerging countries are not yet locked into a conventional energy system, like developed countries (Watson & Sauter, 2011).
The above discussion shows that developed countries have more capabilities and enabling environments to speed up the RET diffusion. A higher level of technological development will improve the enabling environment for RET diffusion. Therefore, I hypothesize:
10
2.4.3 Physical infrastructure
The physical infrastructure of the existing energy regime shows how reliant a country is on a particular energy source. Although RET has a huge potential, most countries still rely heavily on conventional energy technologies. Huang et al. (2007) show that the strong relationship between conventional energy technologies and the economy is a result of lobbying activities of important stakeholders. The results of these lobbying activities is path dependence in which more incumbent technologies are widely accepted and user externalities are important (Arthur, 1994a). This ‘lock-in effect’ occurs because the current energy system has a stabilising influence which prevents RET diffusion and radical change to occur (Klitkou et al., 2015). Factors that contribute to these locked-in trajectories are sunk costs, learning effects, transaction costs involved in the current technology, formal regulations, and network externalities (Arthur, 1994b; Sandén and Azar, 2005; Geels et al., 2004). The lock-in effect will not encourage RET diffusion (Foxon, 2002). The major difference between the emerging and developed countries is the reliance of RET (Bodas Freitas et al., 2012). Since the 1990s, emerging countries relied less on RET as a result of a decreased use of traditional RET like firewood and animal waste, whereas developed countries relied more on RET as the result of more investment in new and sustainable RET (Bodas Freitas et al., 2012). Although the RET reliance increases, the returns from the existing conventional energy regime may not be encouraging to change towards more RET for both sets of countries (Kuchler, 2010). Especially in developed countries the proportion of energy generation from conventional resources will lead to less RET diffusion.
Thus, it is expected that countries that rely more heavily on conventional energy technologies are less likely to have more environmental incentives and participate less in RET diffusion. Therefore, I hypothesize:
Hypothesis 3: A higher reliance on fossil fuel technologies is negatively related to RET diffusion.
2.5 Moderation effect: Stringency of environmental policies
11 policies. The most common environmental policies that are introduced are both price-based mechanisms (market based) and command- and control mechanisms (non-market based), to encourage RET diffusion (Menanteau, Finon, & Lamy, 2003). In particular, countries implemented mandatory production quotas, tariff systems, and tradable certificates (Johnstone et al., 2010). The implemented policies consequently affect different factors of the environment like resource allocation, capital investment, labour intensity, innovation incentives, and production processes (Albrizio, Kozluk, & Zipperer, 2016). Johnstone et al. (2010) show that most market-based mechanisms which are less stringent environmental policies, tempt to encourage innovation in technologies that are closest to the market, which is not encouraging a radical change in the innovation system. The stringency of environmental policies determines the change in costs for energy production, and eventually can lead to more innovation at the firm-level to decrease these costs (Johnstone et al., 2012).
Since each country faces a different innovation system, the stringency of the environmental policies has different effects. Some economic theories show that the more stringent environmental regulations, the more companies face costs to abide to these regulations, which will result in more relocation of their economic activities to places with less environmental regulations (Albrizio et al., 2016). In contrast, Porter and van der Linde (1995) show that the more stringent environmental regulations are, the more productive and innovative a country is. Particularly, stricter environmental regulation will give pressure to individual firms to create and adopt RET which will positively influence the economy and the global competitiveness. Most of the literature shows that emerging countries have less stringent environmental policies than developed countries. Emerging countries are primarily encouraging installed capacity of existing technologies, which are first generation RET or conventional technologies, instead of new technologies, which are second and third generation RET, seen in developed countries (Bodas Freitas et al., 2012). There are several reasons why emerging countries encourage installed capacity of existing technologies. Firstly, the lower economic development is not favouring RET diffusion, since there are high upfront and transaction costs (Kedia, 2016). Secondly, emerging countries rely more on existing technology trajectories to meet the rapidly rising demand (Bodas Freitas et al., 2012). Lastly, there is a lack of sufficient technical, geographical, and commercial information from institutions which is not encouraging RET diffusion (Beck & Marinot, 2004).
12 innovation system, this proxy will be used as a moderating effect. In particular, the proxy will be used to see if the relationship between the existing innovation system and RET diffusion is positively moderated by more stringent implemented environmental policies. It is expected that the reliance on fossil fuel technologies will diminish and that innovation systems will more rapidly change into new technological systems which favour RET when there are more stringent environmental policies. Therefore, I hypothesize:
Hypothesis 4: More stringent environmental policies positively moderates the relationship between the socioeconomic environment and RET diffusion.
Hypothesis 5: More stringent environmental policies positively moderates the relationship between the technological environment and RET diffusion.
Hypothesis 6: More stringent environmental policies positively moderates the relationship between the reliance on fossil fuel technologies and RET diffusion.
Figure 1 shows the conceptual model of the above literature review, corresponding hypotheses, and the control variables of this study.
FIGURE 1
13 3. METHODOLOGY
This research aims to expand the existing knowledge on the global RET diffusion patterns by providing empirical evidence. Since the institutional environment is critical in the RET diffusion, this study will empirically show what the influence is of more environmental stringent policies on determinants of RET diffusion by using a new stringency index of the OECD. Moreover, this study will give more insight in the potential differences between the environments of emerging countries and developed countries and their fight against the global climate change.
3.1 Sample
In order to find the differences in innovation systems, this study will focus on two sets of countries, namely emerging countries and developed countries. The emerging countries included in this study are the 6 BRIICS countries: Brazil, Russia, India, Indonesia, China, and South Africa. The developed countries included in this study are 27 OECD countries: Australia, Austria, Belgium, Canada, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Japan, South Korea, Netherlands, Norway, Poland, Portugal, Slovak Republic, Slovenia, Spain, Sweden, Switzerland, Turkey, United Kingdom, and United States. The 8 OECD countries not included in this study are Chile, Estonia, Iceland, Israel, Latvia, Luxembourg, Mexico, and New-Zealand. This is due to missing data on the environmental stringency index. The timeframe from 2003 to 2012 has been chosen because of data availability and the significant development in RET diffusion throughout these years (IEA, 2006). Furthermore, many environmental policies were introduced after the signing of the Kyoto protocol in 1997 (Popp et al., 2011). Since the effect on the environmental policies is not directly visible, 2003 has been used as the starting point to see if the binding agreement indeed led to more environmental concerns in the innovation system.
3.2 Data collection
3.2.1 Dependent Variable: RET diffusion
14 study will measure the diffusion of RET as the contribution of renewable energy generation, as a percentage of the total energy generated. There are two important reasons why the percentage has been chosen over the absolute number of deployment. Firstly, because of the environmental incentive in using RET, renewables need to increase over the use of fossil fuel technologies which can best be shown by a percentage. Secondly, environmental policies often reflect a certain target for the share of RET. The measure will show information of the energy generation for the renewable sources biomass, biofuels, geothermal, solar, tides, and wind. Hydropower generation has been excluded from the measure because the generation processes of these, often large plants, are viewed as unsustainable and non-renewable in the literature (Brunnschweiler, 2010). The dataset is collected from the Renewables Information Database of the International Energy Agency (IEA).
3.2.2 Explanatory variables
To measure the effects of factors of the environment on the RET diffusion, the innovation system view will be used. According to this view the environment that is influencing the RET diffusion can be distinguished into the socioeconomic, technological, physical infrastructure, and institutional factors (Jacobsson & Johnson, 2000). The data will be collected from the World Development Indicators (WDI) database of the World Bank. The socioeconomic environment will be measured through several variables. Firstly, the income level of a country will be measured by the GDP per capita in current US dollars. Secondly, the CO2 emissions will be measured as CO2 emissions in metric tons per capita. Thirdly, the
energy needs will be measured through the energy consumption in kilogram of oil equivalent
per capita. The per capita proxies are used to see the standard of living effect and to find the incentives of the innovation systems for different countries (Carley, 2009; Dong, 2012). The technological environment will be distinguished in the internationalization of a country and the capability of a country to work with the RET (human capital). The internationalization of a country will be measured by the Foreign Direct Investment (FDI) net inflow as a percentage of the GDP to see how much knowledge spill-overs occur (Del Rio Gonzalez, 2009). The human capital will be measured by tertiary school enrolment as a percentage of gross enrolment. Tertiary has been chosen, because these technical advanced workers are needed to show the capability of adoption and learning in a country (Guerrieri, Luciani, & Meliciani, 2010). The physical infrastructure of a country reflects the reliance on the conventional
energy sources and will be measured as the electricity production from oil, gas, and coal
15
3.2.3 Moderating effect: environmental policy stringency
This study will add the institutional commitment as moderating effect. Theoretical work has shown that environmental policies can encourage incentives for technological innovation (Downing & White, 1986; Milliman & Prince, 1989; Jaffe et al., 2005), but literature on the effect of stringency of environmental policies remains limited. The limited research is due to the difficulty to measure the stringency of environmental policy since many regulations take the form of production constraints instead of explicit price changes (Johnstone et al., 2012). So far, research has used poor cross-country proxies for environmental policy stringency (Botta & Kozluk, 2014; Brunel & Levinson, 2016).
At the macroeconomic level, the Pollution Abatement and Control Expenditure (PACE) has been used a lot as a proxy of stringency of environmental regulation (Lanjouw & Mody, 1996). Brunel and Levinson (2016) show there are several methodological problems with the use of the PACE proxy which are related to the unobserved heterogeneity and endogeneity. Particularly, they show some identification problems and difficulties for respondents to allocate the costs of environmental assets. The most important issue with the PACE index is the comparability problems across countries and over time (Kozluk & Zipperer, 2014).
Another used proxy of stringency of environmental regulation are the composite indices which summarize the multidimensional regulations in one proxy to try to solve the multidimensionality issues of environmental regulations (Kalamova & Johnstone, 2011). The most used composite index proxy is the World Economic Forum (WEF) index, which reflects the perceptions on the stringency of environmental regulation (Johnstone et al., 2010). The limitation of the WEF index is that it does not give data on the real stringency of policies but is constraint to the perceptions of people on the stringency.
16 dimensions of the environmental policies are reduced into one proxy which is comparable across countries (Albrizio et al., 2016). To support the confidence of the practical implementation of the EPS index, Botta and Kozluk (2014) show highly significant correlations with other proxies of stringency. Therefore, the moderating variable stringency of environmental policies will be measured through the Environmental Policy Stringency (EPS) index of the OECD. This measure takes the value between 0 and 6, where higher numbers are associated with more stringent environmental policies. The new EPS index will be used to avoid the methodological issues of the PACE and WEF index (Brunel & Levinson, 2016; Johnstone et al., 2010; Kozluk & Zipperer, 2014).
3.2.4 Control variables
Aside from the main effects which are included in this study, other variables exist which are known to influence the RET diffusion. This study will control for these variables which are not the main objective of the study. Firstly, the energy imports will be measured by the net imports as a percentage of the total energy consumption from the WDI database. This measure will show the dependency on foreign countries to meet their energy demand. Research shows that the higher the dependency of a country on foreign countries to meet their energy demand, the more domestic RET development (Dong, 2012), hence, the more RET diffusion. Secondly, the financial development of a country will be measured by the domestic credit to the private sector as a proportion of GDP from the WDI database, which controls for the impact of financial sector development on the RET share (Zhao et al., 2013). Since RETs have relatively high costs, any access to financial resources can create more opportunities for RET development (Brunnschweiler, 2010), hence, the more RET diffusion. Thirdly, the
demographic structure of a country will be measured as the share of female of the population
from the WDI database. Literature shows that women have a higher environmental incentive than man because they feel more compassion for the world (Torgler & Garcia-Valinas, 2007), which increases the RET diffusion in a country. Lastly, this study will control for the
ratification of the Kyoto protocol. The ratification dates are taken from the website of the
17 the country ratified the Kyoto protocol before 2003, and the value of “0” when the protocol was ratified after 2003, to show a countries’ commitment. Since this study focuses on the time span from 2003 till 2012, it will show what happened with the RET diffusion after the Kyoto protocol was ratified.
3.3 Method and empirical model
Researching the RET diffusion and the factors influencing it requires a longitudinal research design (Johnstone et al., 2010; Popp et al., 2011). A panel data regression allows greater variability, lower co-linearity between the variables, more efficiency in the estimates of data, and control for possible endogeneity. Prior research primarily focused their panel data regressions on the EU level (Marques & Fuinhas, 2012). The aim of this study is to show the differences between two country groups, namely emerging countries and developed countries, and show the potential importance of the stringency of environmental policies on the environment. The sample will be split into two subsamples to find the differences, namely; developed and emerging countries. Developed countries represents most of the OECD countries and emerging countries are the BRIICS countries.
The sample is characterized as time-series and cross-sectional. This type of data does not take into account the different characteristics of individual units (countries). To see the relationship between the explanatory variables and the dependent variable, a regression analysis is appropriate when the data is continuous. The dependent variable RET diffusion is in percentages and will be treated as continuous data, like most research suggests. The following steps will be taken (Marques & Fuinhas, 2012): (i) observe the nature and quality of the data, (ii) test for the presence of heteroscedasticity, contemporaneous correlation, and panel autocorrelation, (iii) if the errors are not independent and identical distributed, the econometric analysis with PCSE estimators needs to be conducted.
Normally, the standard Ordinary Least Squares (OLS) regression is used to test the relationship between independent and dependent variables. Several tests have been conducted to see if the dependent variable, RET diffusion, is normally distributed, which can be found in Appendix A. The histogram and the descriptive statistics show that RET diffusion is not normally distributed.
18 contemporaneous correlation, and panel auto-correlation (Marques & Fuinhas, 2012). Using panel data which can perform upon complex error compositions, addressment of heteroscedasticity, contemporaneous correlation, and panel auto-correlation is required. Contemporaneous correlation and panel auto-correlation is expected because lots of countries in the sample are exposed to the same policy guidance on RET. Furthermore, there is the risk of unobserved heterogeneity, which refers to the unobserved country characteristics, which are likely to be correlated with the variables included in the model. Several tests have been conducted to see if heteroscedasticity, contemporaneous correlation, and panel auto-correlation is an issue and is provided in Appendix C. The Modified Wald test, Wooldridge test, Pesaran test, and Frees’ test show there is enough evidence for these concerns. The tests show that common used panel estimators, like random or fixed effects, are not appropriate models in the context of both correlation, heteroscedasticity, and contemporaneous correlation (Marques & Fuinhas, 2012; Polzin et al., 2015).
19 The econometric model used in this study for both sets of countries can be seen in Formula 1.
FORMULA 1
Econometric model
RESHit = β1GDPit + β2CO2it + β4FDIit + β5TECit + β6RCTit + β7MOD1 + β8MOD2 +
β9MOD3 + β10MOD4 + β11MOD5 + Xitδ + εit
20 4. RESULTS
4.1 Pre-analysis and descriptive statistics
Before conducting empirical techniques to test the hypotheses, multicollinearity was checked based on the Variance Inflator Factors (VIF), Critical Tolerance Values (CTV) and the Pearson’s correlations. The VIF and CTV calculations with further details can be found in Appendix D. Appendix E shows the correlation matrix. Multicollinearity was detected
between CO2 emissions and Energy Consumption. Since CO2 emissions has a higher correlation with RET diffusion, Energy Consumption has been excluded from the model.
The descriptive statistics of all variables included are presented in Table 1.
TABLE 1
Descriptive Statistics all variables
21 differ per country. The technological capabilities statistics are as expected since some countries face lower tertiary education rate than others. The reliance on conventional energy sources show a big difference between the countries, with a minimum score of 0.39% and a maximum score of 98.48%. The EPS variable shows that there is no country in the sample who scores the maximum score of 6 on stringency of environmental policies since the maximum score is 4.133. This score indicates that in order to rely more heavily on RET, policy makers still have room to stringent their environmental policies. The descriptive statistics show that the mean of the EPS variable is 2.167987 and the median is 2.295833. Since the EPS variable is normally distributed the mean will be seen as the midpoint. The data show that over time, the environmental policy stringency in all OECD and BRIICS countries has been increasing. The data also show that environmental policies remain more stringent in developed countries than in emerging countries, which is indicated in Table 2.
TABLE 2
Descriptive statistics EPS indicator
Note: Emerging countries are notified by EC. Developed countries are notified by DC.
Within the developed countries the Nordic countries have the highest stringency of environmental policies and the Southern countries have the lowest stringency of environmental policies. Most of the other countries score around the mean score. Furthermore, the energy imports variable shows also a big difference between countries in their reliance on foreign countries for energy supply. The financial development numbers are in line with the expectations of the above discussion. There is not a big difference in the percentage of female population in the sample. Lastly, the Kyoto protocol represents a dummy variable which shows that more countries ratified the Kyoto protocol after 2003.
22 behind in the RET diffusion compared to most of the developed countries, since the average RET diffusion rate is the same as the average of most developed countries.
4.2 Results PCSE model
The findings of the PCSE regression model for developed countries are presented in Table 3. Table 4 shows the findings of the PCSE regression model for emerging countries. Since the dependent variable, RET diffusion, is a percentage variable, the β in the equation stands for the change in percentage point more reliance on renewable energy sources.
The first model in both findings (model 1 and 9) is the baseline model with solely the control variables included. The findings show that all the control variables have a significant positive influence on RET diffusion for developed countries. It shows that an increase of 1 percentage point in energy imports, financial development and female population, increases the RET diffusion percentage, holding other variables constant. If countries ratified the Kyoto protocol after 2003 there is an increase of 5.6% in RET diffusion. In comparison, the emerging countries show that only two of the control variables (financial development and female population) have a significant negative effect on RET diffusion. It shows that an increase of 1 percentage point in financial development and female population, decreases the RET diffusion percentage. The second model (model 2 and 10) tests the explanatory variables. The third model (model 3 and 11) includes the moderating effect of EPS on GDP. The fourth model (model 4 and 12) includes the moderating effect of EPS on CO2. The fifth model (model 5 and 13) includes the moderating effect of EPS on FDI. The sixth model (model 6 and 14) includes the moderating effect of EPS on technological capability. The seventh model (model 7 and 15) includes the moderating effect of EPS on reliance on conventional technology. The eighth model (model 8 and 16) constitutes the full model.
23 TABLE 3
PCSE results developed countries
24 TABLE 4
PCSE results emerging countries
25
4.2.1 Results determinants RET diffusion
The socioeconomic environment is measured by the variables GDP and CO2. For both sets of countries it is shown that GDP has a small positive influence on the RET diffusion, with a significance level of 0.1%. It is a small influence since GDP is measured as GDP per capita in current US$, which indicates that an increase of 1$ in the GDP per capita, increases the RET diffusion with 0.00008% for developed countries and 0.0003% for emerging countries. Since this effect is in US$ per capita it can be a poor explanation for the RET diffusion in the sets of countries which will be discussed in the discussion section. The results are not in line with the expectation that a greater level of income of developed countries will lead to more RET diffusion, since the effect is bigger for emerging countries which was expected to have a negative relationship. Reason could be that the higher standards of living of developed countries lead to more energy demand which will not encourage the use of renewables. The CO2 emissions show that for both emerging and developed countries an increase of CO2 emissions per capita will lead to a decrease in RET diffusion of 0.825% (p<0.01) and 0.632% (p<0.001), respectively. This is not in line with the expectation that developed countries, who are wealthier, have more environmental concerns in mind which leads to more use of renewables. Therefore, hypothesis 1 is partially supported. For both sets of countries an increase in the level of income is positively related to RET diffusion and an increase in CO2 emissions is negatively related to RET diffusion. Most surprising is that higher environmental concerns in developed countries, reflected in CO2 emissions, are not encouraging RET diffusion.
26 The reliance on fossil fuel technologies shows some surprising results. Due to lock-in effects and lobbying activities it was expected to see a negative relationship between reliance on fossil fuel technologies and RET diffusion for developed countries. However, the results show a positive relationship of 0.077% (p<0.001), which means that more reliance on conventional technologies leads to more RET diffusion which is reflecting the environmental concerns. The results of the emerging countries did not show a significant result. Hypothesis 3 is rejected for developed countries.
4.2.2 Results moderating effect stringency of environmental policies
The results of the separate models who run the interaction effects show no significant results for emerging countries and only two significant results for developed countries. Therefore, the full model will be used to show the results for the moderating effect of EPS. The overall models shows an adjusted R2 of 0.91 for the emerging countries and 0.51 for developed countries.
Firstly, the results of the moderating role of environmental policy stringency on the relationship between the socioeconomic environment and RET diffusion support hypothesis 4. Looking at interaction effect of the EPS on the relationship between GDP and RET diffusion, results show a significant negative moderating effect of 0.002 (p<0.05) for emerging countries. The negative effect is caused by the lower stringency in environmental policies of emerging countries which weakens the positive relationship between GDP and RET diffusion. Developed countries show a significant positive moderating effect of 0.0001 (p<0.001), which is in line with the expectation that more stringent environmental policies, the more the level of income will lead to RET diffusion. The interaction effect of the EPS on the relationship between CO2 and RET diffusion shows a significant positive moderating effect of 1.358 (p<0.01) on the negative relationship for emerging countries. The negative relationship will strengthen because emerging countries have lower stringency of environmental policies. Developed countries results show that the relationship between CO2 emissions is significant negatively moderated by the EPS with 0.911 (p<0.001). The negative relationship between CO2 and RET diffusion weakens when the environmental policies get more stringent.
27 interaction effect of the EPS on the relationship between technological capability and RET diffusion did not show significant effects for emerging countries. Developed countries results show that the positive relationship between technological capability is significant (p<0.001) positively moderated by the EPS with 0.190.
Lastly, the results of the moderating role of environmental policy stringency on the relationship between the reliance on fossil fuel technologies and RET diffusion support hypothesis 6. The interaction effect did not show significant effects for emerging countries. For developed countries the positive relationship between the reliance and the RET diffusion is positively moderated by the EPS with 0.080 (p<0.01).
4.3 Robustness checks: OLS regression and Fixed effect model
The OLS regression and FE model have been conducted to check the robustness of the results. To simplify the results, Table 5 shows the coefficient signs with the significances to see what kind of relationship the variables of interest have on RET diffusion. Only the models that are used in the interpretation of the results from the PCSE model are presented. The full analyses of the OLS regression and FE model can be found in Appendices G and H, respectively. The results of the OLS and FE models are consistent with the results of the PCSE model.
28 TABLE 5
Summary of the estimated effects different techniques
29 5. DISCUSSION
5.1 Discussion
The results show that the emerging countries are not lagging behind in RET diffusion compared to developed countries. In particular, emerging countries who encourage RET diffusion rely on average the same on renewables as developed countries. The results do show the differences between the sets of countries in their innovation system and the influence on RET diffusion. Across emerging countries, a great number of heterogeneity occurs in the intensity of the use of renewables. Furthermore, several factors of the environment regarding emerging countries are significantly influencing RET diffusion. GDP per capita has a positive influence on the RET diffusion and both CO2 emissions and FDI have a negative influence on RET diffusion. For developed countries all variables except for FDI found a significant relationship with RET diffusion, whereas GDP per capita, technological capability, and reliance on conventional technologies has a positive influence and CO2 has a negative influence on RET diffusion.
30 2009). The results suggest that pressures for environmental quality and global climate changes were not leading to more RET diffusion.
Focusing on the technological environment, a negative effect of internationalization was found for emerging countries. The result is not in line with the expectation that emerging countries will use the RET development of developed countries as a ‘blueprint’ for their development (Lanjouw & Mody, 1996). The results show that the level of internationalization does not result in environmental friendly attitudes in which they use knowledge on RET development of foreign countries. In particular, the results suggest that environmental concerns are not the priority of global firms. No significant effects were found for developed countries, which was not expected since these countries have the enabling environment to use foreign knowledge to advance their RET development. The national capabilities of emerging countries have no significant results, which was unexpected because most of the emerging countries are focused on development in energy-intensive technologies which results in a negative relationship with renewables (Bodas Freitas et al., 2012). The national capabilities of developed countries show a positive effect, which shows evidence that R&D efforts were put into RET diffusion. The developed countries results suggest that search capacity and skills are increasing the RET diffusion, which is in line with the enabling environment expectation (Benhabib & Spiegel, 2005).
Focusing on the physical infrastructure, there are no significant results for emerging countries, which was not expected since the emerging countries are not yet locked into conventional technologies and relied more on RET in the past. Developed countries show no significant results for the expected locked-in and lobbying effect (Huang et al., 2007). The results show more support for the environmental incentive instead of the lobbying effects, which is surprising because it was expected that industry stakeholders would have more influence on the use of conventional technologies (Romano et al., 2017).
Thus, varying degrees of environmental commitment are visible. Emerging countries’ innovation systems, who’s economic growth is energy intensive, will show greater emphasis towards conventional energy technologies and less to RET. This is because these countries do not have the means and enabling environment to pursue more innovative technological trajectories. Developed countries do have the means and the enabling environment but do not show enough environmental incentives needed for innovative technological trajectories due to lock-in effects.
31 environment for RET diffusion. The results show that the level of stringency in environmental policies have different effects on the environment. Developed countries, who have a higher score on the EPS index, indeed do encourage RET diffusion. This encouragement is visible in the strengthening effect in the relationship between GDP, technological capability, and reliance on conventional technologies with RET diffusion. The unexpected negative relationship between CO2 and RET diffusion is also weakened by more stringent environmental policies. The results for developed countries are in line with the expectation of Porter and van der Linde (1995) that more stringent environmental policies in a country will lead to more productivity and innovation of renewables. It shows that more stringent environmental policies affect the business performance indirectly by encouraging firms to adopt an environmental strategy, which will change the innovation system.
On the other hand, this study shows that emerging countries, who have a lower score on the EPS index, indeed not encourage RET diffusion. In particular, the discouragement is visible in the weakening effect of the lower EPS score on the positive relationship between GDP and RET diffusion and the strengthening effect of the lower EPS score on the negative relationship between CO2 emissions and RET diffusion. Thus, the current level of EPS does not increase RET diffusion but instead encourages the installed capacity of already existing technologies. It shows that emerging countries currently do not have the enabling environment to promote RET diffusion. The result is in line with previous literature which showed that institutions in emerging countries generally lack in providing sufficient technical, geographical and commercial information, and strategies which are needed for more RET diffusion (Beck & Marinot, 2004). Institutions in emerging countries are not changing the stringency of their environmental policies since the current low level of stringency attracts economic activities and is strongly linked to the socioeconomic development (Cai et al., 2016). In particular, the current environment of emerging countries attracts economic activities from developed countries with higher environmental stringency because emerging countries have less environmental rules and regulations, which gives lower costs for individual firms. Furthermore, the current environment shows that socioeconomic development is closely linked to the reliance on conventional technologies. The consequence is that traditional technology trajectories are still being followed, which makes it difficult to introduce sustainable growth paths. The traditional trajectories result in a further increase in inequalities between emerging and developed countries (Kuchler, 2010).
32 quality and productivity, policy makers of emerging countries need to encourage RET diffusion by more stringent environmental policies. The strengthening of the environmental policies can be done in several ways. Firstly, since GDP per capita has been rising rapidly, more attention needs to be given to the quality of life which involves the environmental protection of the society. Policy makers in emerging countries should acknowledge the growth in income and demonstrate a stronger commitment for environmental concerns because their CO2 emissions are rising rapidly. Policy makers should acknowledge the economic incentive in their economic activities and try to link it with environmental concerns in order to change the innovation system. Since more developed countries require more environmental standards, emerging countries need to adopt accordingly if they want to play a key role in the global business environment of renewable energy. Instead of attracting less sustainable economic activities of developed countries, emerging countries should adopt the environmental standards of developed countries. By introducing more stringent environmental policies the business environment will change which will increase the environmental growth paths in which RET will mature and become competitive.
33 energy, use of local resources, and reduced dependency on conventional technologies can be reached (Omer, 2009).
5.2 Limitations and future research recommendations
There are a couple of limitations related to the design of this study and the nature of the data used. Future research should incorporate these limitations. Firstly, the GDP per capita variable has been used to measure the influence of the economic development on RET diffusion. Some unexpected results have been found for emerging countries. Reason for this can be the use of the per capita variable, which can be a poor explanation for RET diffusion. The limitation of this proxy is that it does not look at the absolute economic size of the country but at the standard of living of its population. According to Huang et al. (2007) looking at the absolute economic size of countries indicates a different relationship since it shows the wealth of the overall country, which is recommended for further research.
Secondly, although the EPS indicator improves the multi-dimensionality problems of other proxies it still has some limitations in assessing and comparing stringency across countries due to high level of country specific elements. The indicator overlooks areas like water, natural resources, or waste and it is not taking into account the soft policy instruments (e.g. tax incentives). Future research could include other proxies of environmental policy stringency for reasons of robustness.
Thirdly, the sample size of emerging countries is relatively small which can be a potential explanation of some insignificant results. Future research should use a larger sample size for emerging countries to check the results.
Fourthly, although the results show that more stringent environmental policies will increase RET diffusion, the economic performance of firms have not been analysed. Future research could show if the increase of RET diffusion also involves an increase in economic performance. It is also interesting to know what the actual investment of firms is in RET diffusion to see if the above conclusion that stringent environmental policies leads to more environmental concerns are indeed true and if any differences among firms occur. So future research should focus more on firm-level outcomes than on aggregated country-level results.
34 the influence of environmental policy stringency on industrial competitiveness across countries. This could lead to more specific recommendations for the environmental policies that should be implemented by policy makers.
Sixthly, this study uses the contribution of all renewables (excluding hydropower) in the total generation of energy. Since there is no distinction between different technologies, only general recommendations are given for the overall RET diffusion. Future research could make this distinction to see the different effects of relevant environmental policies on the diffusion of a particular technology. The effects will differ from the general recommendations in this study since the costs of different RET varies between locations because of resource availability. Subsequently, it is interesting to show which RET contributes to more productivity in different sets of countries.
