Has the developing world forgotten about renewables in its pursuit of economic growth?

(1)

Has the developing world forgotten about renewables in its

pursuit of economic growth?

University of Groningen

Faculty of Economics and Business

Master’s Thesis International Economics and Business

Author: M.H. (Marnix) Meijer

Student ID: S3258726

E-mail address: m.h.meijer.3@student.rug.nl

Supervisor: Dr. C.J. Jepma

(2)

2

Abstract

Whereas renewable electricity (RE) has been increasing in relative importance in developed countries, this does not hold true for most developing countries. Considering that

developing countries have not committed themselves to binding targets to reduce CO2

emissions within the Kyoto protocol, the question arises whether the developing world lacks stimuli to increase the relative share of RE while pursuing economic growth. As such, this paper aims to shed more light on the impact of economic growth on the relative share of RE; a research subject left largely untouched. Using a panel analysis that includes data of 70 countries over the period 1990-2014, I find clear empirical evidence that an increase in economic growth negatively impacts the relative share of RE in general. Subsequently, by slicing up the dataset, the results suggest that this negative impact lasts longer in developing countries. Lastly, by separating periods in which countries experienced strong economic growth from the rest, I find evidence that economically strong growing countries form an exception to the finding of this negative impact.

(3)

3

Tables and graphs

Graph 1 – The share of renewable electricity of the total electricity generation in percentages (1990-2015) ... 6 Table 1 - Summary Statistics ... 20 Table 2 – The impact of economic growth on the relative share of renewable electricity: All countries ... 25 Table 3 – The impact of economic growth on the relative share of renewable electricity: Developing countries ... 27 Table 4 – Regression results: Interaction effects regarding the level of income hypothesis ... 28 Table 5 – Regression results: Interaction effects regarding the strong economic growth

hypothesis ... 30

List of abbreviations

CO2: Carbon dioxide

GHC: Greenhouse gasses

IEA: International Energy Agency IMF: International Monetary Fund RE: Renewable Electricity

(5)

5

I. Introduction

A steep increase in the global usage of electricity and the subsequent rise in CO2

emissions have brought us on the verge of a serious global warming problem. In their latest publication report, the Intergovernmental Panel on Climate Change (IPCC) once again argued that they expect dramatic consequences if the rise in temperature cannot be kept below 1,5 degrees (IPCC 2018). The consensus that climate change is a problem that needs to be tackled on a global level is growing. As a result, the majority of the countries have pledged to address

their environmental problems.1

The United Nations Framework Convention on Climate Change (UNFCCC), adopted on 9 May 1992 (UNFCCC 1992) and signed by 165 different countries, is the first

framework that recognized that reducing CO2 emissions and thus the prevention of

undesirable effects of climate change is an international responsibility. This framework is extended by the Kyoto protocol in 1997, which recognizes that capacities to combat climate change differ on a country level (UNFCCC 1997). This implies that most of the responsibility

for reducing CO2 emissions is put on developed countries as they are held responsible for the

majority of the current levels of greenhouse gases (GHG) in the atmosphere. As a result,

almost only developed countries have committed themselves to binding targets,2 while almost

all developing countries have non-binding targets.

Since the start of the UNFCCC, generating electricity from renewable sources has been viewed by many countries and international organizations as a crucial element of

mitigating CO2 emissions (Carley 2009; Marques and Fuinhas 2012). Almost all developed

countries, as classified by the International Monetary Fund (IMF 2016), have started to

address their electricity related CO2 emissions by increasing the deployment of RE in absolute

numbers. However, increasing RE in absolute terms does not necessarily correspond to a

reduction in CO2 emissions because it may be that electricity generated from conventional

sources increases alongside. Therefore, renewables need to displace fossil fuels. In other words: the relative share of RE is what matters instead of its absolute amount. This is a distinction that is remarkably often overlooked in literature. Whereas the relative shares of RE have been increasing in most developed countries in recent years (European Commission

1_{The best-known examples are probably the Paris agreement of 2015, signed by 195 countries to keep the rise of}

global temperature on top of pre-industrial levels below two degrees Celsius by all means (UNFCCC 2015) and the subsequent agreement designed at most recent conference (COP24) on how to implement the former agreement (UNFCCC 2018).

2_{Australia, the European Union (and its 28 members states) Belarus, Iceland, Kazakhstan, Liechtenstein,}

(6)

6

2017), this is not true for most developing countries. This raises the question whether developing countries lack stimuli to increase the relative share of RE as they are not

committed to binding targets to reduce CO2 emissions. Or as the title reads: Has the

developing world forgotten about renewables in its pursuit for economic growth?

It is argued by various studies (e.g. Arrow et al. 1995; Ekins 1997; Hung and Shaw 2002; Stern 2004) that economic policies are not substitutes for environmental policies.

Considering that almost none of the developing countries are subjected to binding CO2

reduction targets, the question arises whether developing countries which experienced economic growth may have done so at the expense of their relative shares of RE. In other words: whether economic growth and subsequent increases in electricity demand are mostly

met by generating electricity from conventional sources (hence increasing CO2 emissions).

Graph 1 shows examples of large developing countries that experienced periods of economic

growth recently, whose shares of RE have stayed constant or even declined.3

Graph 1 – The share of renewable electricity of the total electricity generation in percentages (1990-2015)

Source: Data from World Data Bank Database

In general, the impact of factors that influence RE may change when RE is considered from a relative instead of an absolute perspective. This is because the impact on electricity generated from conventional sources also matters when RE is considered from a relative perspective. Whereas it is generally accepted that economic growth, measured by the annual GDP per capita growth rate indicator of the World Data Bank in this paper, stimulates RE in absolute amounts (Sadorsky 2009; Chang et al. 2009) there is little consensus on its impact on

3_{China made significant progression recently, yet it already had experienced strong economic growth for over}

twenty years. 0 5 10 15 20 25 30 35 40 45

(7)

7

the relative share of RE of a country’s total electricity production.

Against this background, this paper aims to shed new light on the discussion of RE deployment on a country level and subsequently the lagging results in the developing world. Therefore, I formulated the following research questions: What impact does economic growth have on the relative share of RE? Does the level of income of a country (and thus also its

potential commitment to binding CO2 targets) affect the specificities of this impact? And does

the strength of economic growth affect the specificities of this impact?

Using a panel analysis that includes data of 70 countries over the period 1990-2014 (extracted from the World Data Bank and the IEA), in combination with a variety of estimation methods including interaction terms, a number of findings emerge. First, I find clear empirical evidence that an increase in economic growth negatively impacts the relative share of RE in general. Second, after making a distinction between developed- and developing countries using the identification strategy of the IMF, the results suggest that this negative impact lasts longer in developing countries compared to developed countries. Lastly, by separating periods in which countries experienced strong economic growth from the rest, I find evidence that economically strong growing countries form an exception to the finding the negative impact of economic growth.

This paper is structured as follows: Section 2 provides an insight on the reviewed literature relevant for this paper and presents the hypotheses to be tested. Section 3 describes the process of data collection, while section 4 explains the methodology used to test the hypotheses. Section 5 discusses the empirical results of the analysis. And finally, section 6 concludes the paper.

II. Literature review and Hypothesis tests

First, I will examine the relevant literature regarding the potential impact of economic growth on RE. I have divided this review into three sub-sections: 1) the positive view; 2) the negative view and; 3) the neutral view. Afterwards, I will take a closer look at the following two potential dynamics that may affect the specifics of this impact: 1) the level of income of a country and; 2) the average strength of economic growth.

2.1 Literature review

(8)

8

these fields of study will be relevant also for the topic of this paper, these will be reviewed as well. Achieving a positive switch in a country’s energy mix implies that a country substitutes away from environmentally-harmful- towards less environmentally-harmful energy inputs (Stern 2004). Strictly speaking, a switch from coal towards natural gas would already fit the description. Nevertheless, in accordance with the majority of the reviewed literature, this switch is defined as a switch from generating electricity from conventional sources towards renewable sources in this paper. In history, several switches in energy mixes (also known as energy transitions) have occurred. In their paper dedicated to energy transitions in history, Solomon and Krishna (2011) argue that the first major energy transition was that from wood to coal. The second major energy transition emerged at the beginning of the 20th century as petroleum and natural gas grew in importance. Therefore, the transition from fossil fuels to renewable energy is referred to as the ‘third energy transition’ (Solomon and Krishna 2011). However, the latter differentiates itself from the former ones because it is not a technology driven transition (e.g. coal being a more efficient source for energy than wood), but a morally driven transition to counteract climate change. Because of this, several challenges hinder this transition. Whereas the literature is rather unanimous regarding most of these challenges (e.g. lack of climate change awareness and market barriers), opinions on whether economic growth hinders or stimulates the current energy transition remain ambiguous.

2.2 The negative view

(9)

9

of environmentally friendly practices by the actors, while economic growth brings immediate advantages (Padilla and Serrano 2006).

The second main argument why economic growth may negatively impact the share of RE relates to relative high production costs of the latter (Cerović et al. 2017). Most sources for RE, such as the wind and the sun, are free and maintenance of its facilities is generally relatively cheap. However, the high upfront expenses of building and installing RE power plants are the costs creating financial barriers (Reddy and Painuly 2004; Union of Concerned Scientists 2017). According to Kariuki (2018), this explains to a large extent why the average costs per kilowatt-hour of RE is still relatively high in many countries. Additionally, locations suitable for generating RE are generally spread across a large area compared to those for fossil fuel power plants (Kariuki 2018). As a result, the needed power lines and costs concerning siting (e.g. permits, negotiating the locations) likely increase costs or even turn RE projects unprofitable. Simultaneously, countries that experience economic growth need to react quickly to prevent blackouts as a consequence of rising electricity demand (De Janosi and Grayson 1992; Bowden and Payne 2009; Cheng and Zhang 2009). Therefore, these countries are inclined to use low cost conventional technologies that can be deployed relatively fast. As a consequence, financial resources that could be invested in RE are shifted away towards economic growth enhancing purposes. India is an often-cited example because it has repeatedly insisted that developed countries should bear the majority of the burden for tackling climate change. In 2014, India´s environment minister, Prakash Javadekar, told The New York Times that alleviating poverty was the country´s first priority, which would involve the opening of new coal-powered production plants if needed (Davenport 2014). Because coal is a relatively cheap fossil fuel and India’s coal-fired power industry proves to be ‘too big to fail’ this is exactly what happened since then according to The Economist (2018). As a result, India kept hold of its strong economic growth rate at the expense of the share RE in its electricity mix.

(10)

10

facilities for transmission and the distribution networks as well as the equipment needed for RE power companies. This lack of equipment forms an infrastructural challenge for the deployment of RE and forces countries to import it (Luthra et al. 2015). Because imported equipment is relatively expensive and time-consuming, most countries prefer to invest in existing conventional techniques to meet rising electricity demand as a consequence of economic growth (Kariuki 2018). Third, low reliability, because of poor quality of servicing and maintenance of (imported) equipment, diminishes customer confidence in RE technologies. As a result, an equipment failure is likely to happen because of a lack of spare parts and adequate repairing skills hence hindering the adoption of RE technologies (Richards et al. 2012). Last, economic growth takes place mostly in and around cities, which implies that the majority of the increase in electricity demand originates in urban areas (Dhakal 2009). Inability to adequately connect large expansions of RE into the electricity grid shies investors away as it diminishes their confidence (Eleftheriadis and Anagnostopoulou 2015). Addittionally, significant amounts of electricity are lost during the transmission from the RE production point to final consumption (i.e. urban areas) as suitable locations for its generation are generally spread across a large area hence increasing its travel distance. This may also

hamper the willingness to invest in its technologies (Atwa et al. 2010).

The last main argument suggests that insecurity about the stability of economic growth may increase uncertainty about the future (Painuly 2001). As a consequence, investments with a low payback period may be preferred to avoid too much risk exposure. Because RE technologies bear high production costs and payback insecurities relative to fossil fuels in many countries, investors may shy away in an unstable macro-economic environment when time is limited. Based on these arguments, there is sufficient reason to hypothesize that economic growth may have a negative impact on the relative share of RE. Therefore, the first hypothesis is as follows:

Hypothesis 1: Economic growth decreases renewable electricity as a share of the total electricity production.

2.3 The positive view

(11)

11

Verbeke and Clercq (2006) tested the argument regarding the willingness of the society to address pollution in their paper and confirmed that economic growth increases the demand for environmental quality. One of the channels through which this ‘income effect’ flows is that attention shifts away from material well-being towards renewable technologies. Because economic growth increases incomes on average, people are willing to pay more for electricity

on the condition that it comes from renewable sources.4 Marques et al. (2010) found evidence

in favor of this argument and argued that economic growth increases the availability of financial resources to be invested in RE. They looked at the differences between European countries and found a significant positive effect of an increase in incomes and being able to

bear the costs of deploying RE.

Second, RE is still in an unfavorable position in most countries as its technologies are relatively new and need to compete with well-established conventional technologies (Menanteau et al. 2003). This implies that RE technologies need improvements to effectively compete with these conventional technologies because they do not function at an optimal level yet. This optimal level can be gradually reached by the process of learning by using or learning by doing (Dosi 1988). Thus, investments in technology improvement and incentive policies are needed to make RE technologies competitive. Economic growth may make the necessary funding available hence is hypothesized to have a positive impact on the relative share of RE (Menanteau et al. 2003; Foxon 2010). In his paper on the economics of climate change, Stern (2008) supports this argument as he calculated that the annual costs of 1-2% GDP for investing in cleaner technologies is needed to achieve global environment goals.

Third, a proper functioning credit market may facilitate the deployment of RE. This is because an improper functioning credit market distorts the access to credit (Painuly 2001). As a consequence, the following barriers can hinder the deployment of RE: high transaction costs, inability to take on long term loans and high financing costs (Hussain 2013). Economic growth is argued to spur credit market development (Khan et al. 2001; Vazakidis and Adamopoulos 2009; Mishra et al. 2009). This implies that a better-developed credit market, by means of economic growth, will facilitate the deployment of RE.

Last, improvements in the enforcement of regulations and stringent environmental standards, as another consequence of the demand for environmental quality, can eliminate several barriers to RE deployment. Examples of these barriers are: a lack of mechanisms/institutions to provide information, lack of legal/regulatory framework, red tape

4_{This argument is an extension of Maslow’s hierarchy of needs theory as economic growth enables people to}

(12)

12

and corruption (Painuly 2001). This argument is similarly related to economic growth as the first argument, as people demand more attention to be paid to noneconomic aspects of their well-being as they become richer (Grossman and Krueger 1995).

In their research to the electricity mix of China, Lin et al. (2016) concluded that the relative share of RE in the electricity mix is positively impacted by economic growth. Their reasoning follows the above mentioned arguments as they argue that economic growth creates the needed human and financial capital to invest more in the adaption of RE compared to conventional methods. Simultaneously, people are more willing to adopt RE because of their increased income and standard of living. Similar results are found by Nakicenovic and Swart (2000), and by Chang et al. (2009) in their analyses on economic growth, energy prices and renewable energy, using a panel dataset across all OECD member countries.

The above discussion implies that there are also well-founded arguments why economic growth, in contrast to the hypothesis formed by the negative view, increases the relative share of RE. This results in the second hypothesis:

Hypothesis 2: Economic growth increases renewable electricity as a share of the total electricity production.

2.4 The neutral view

In addition to the positive and negative view, there are also researchers who opt for a neutral view. In their research regarding the determinants of RE growth, Aguirre and Ibikunle (2014) found no evidence that economic growth is a main driver for its deployment. On the one side, they argue that countries with significant economic growth can permit to be more

concerned with CO2 emissions related issues and the environment. On the other side, the

opening of new frontiers for accessing fossil fuel deposits because of the development of new

technologies may cancel out the increase in concerns regarding the CO2 emissions as

mentioned above. Chang et al. (2015) also conclude that economic growth has no impact on the deployment of RE when looking at the evidence from a heterogeneous panel of G7 countries. Payne (2009) and Sari et al. (2008) focused specifically on the United States and did not find a relationship between economic growth and RE nor non-renewable electricity. In general, because there are sound arguments in favor of both a positive and a negative impact of economic growth on the relative share of RE, it seems only self-evident that these factors may cancel out against each other. Therefore, the third hypothesis is as follows:

Hypothesis 3: Economic growth does not impact renewable electricity as a share

(13)

13

2.5. Level of income hypothesis

A strand of literature hypothesizes that the level of income of a country forms an important dynamic that affects the impact of economic growth on RE. In general, much of the positivism as discussed in the positive stems from the concept of the environmental Kuznets curve (EKC). This curve implies that there is an inverted U-shape relation between economic growth and environmental degradation (Barbier et al. 1996). As a result, the environment is especially harmed by economic growth in an early stage as the intensified agricultural sector and industrialization result in an increase in the electricity demand. The rates of resource depletion start to exceed the rates of resource regeneration and the levels of waste creation

and CO2 emissions increase. However, eventually a peak in environmental degradation will

be reached, and any further economic growth will lower the environmental impact of economic activity. This is because the impact of positive factors (e.g. de-industrialization and

diminishing marginal utility of income5) increasingly outweighs the impact of negative factors

(e.g. increased demand for electricity and congestion). This implies that the relationship of the society with the environment improves when a certain threshold of the level of income is surpassed (Barbier et al. 1996). This concept gained much attention after being discussed in the World Bank’s World Development Report (1992) and since then has gained support from numeral studies (Bandyopadhyay and Shafik 1992; Beckerman 1992; Panayotou 1993; Lomborg 2001; Dasgupta et al. 2002; Gallagher 2003).

Additionally, it is argued that developing countries may also lack incentives to reduce

CO2 emissions as they are not committed to binding targets. In other words: the non-binding

targets of the Kyoto protocol may be insufficient to persuade developing countries to meet increases in electricity demand caused by economic growth with RE instead of fossil fuels. This argument is supported by Aguirre and Ibikunle (2014), as they found no evidence that voluntary environmental-friendly policies from governments stimulate RE deployment.

However, there are also arguments that suggest that developed countries are more negatively impacted by economic growth compared to developing countries. Biomass forms the main source of RE in most of the developing countries (IEA 2006). In addition, the enforcement of regulations is generally weaker in developing countries. As a result, these countries can simply extract additional electricity from biomass to meet its demand. In contrast, most developed countries are unable to do so. Considering that the RE technologies

5_{As incomes increases, the benefits of similar additional gains (i.e. €1.000) diminish. This is because the benefit}

of the first €1.000 is very high compared to an increase in income from, for example, €50.000 to €51.000. As a result, gaining an additional benefit loses its importance when you are surrounded with environmental

(14)

14

commonly used in developed countries (e.g. sun and wind) are often driven by subsidies, are sensitive to policy and consist of long construction times and have relative high production costs, developed countries may be more inclined to generate electricity from conventional sources during periods of economic growth.

All in all, to examine whether the income level of a country significantly affects the specificities of the impact of economic growth on the relative share of RE, I will also test the following hypothesis:

Hypothesis 4: The level of income of a country affects the impact of economic growth on renewable electricity as a share of the total electricity production. 2.6 Strength of economic growth hypothesis

The hypothesis that the strength of economic growth significantly affects the specificities of its impact on RE is the last dynamic that will be examined in this paper. As mentioned in the introduction, especially in periods that countries experience strong economic growth, they seem to struggle with deploying RE. The argument that the countries that

experience economic growth need to react quickly to meet the subsequent steep increases in

electricity demand, lies at the heart of this dynamic. Generally, it is both time-consuming and capital intensive to meet an increase in electricity demand with RE. Also, a general lack of RE technologies may incline a country to prefer conventional energy sources when an increase in electricity demand emphasizes the need of a quick respond (Mohammed et al. 2013). As a result, it is hypothesized that countries are able to meet this increased demand up to a certain height with RE, but that they switch back to fossil fuels when the economic growth and the subsequent increase in electricity demand surpasses this threshold:

Hypothesis 5: The strength of economic growth affects its impact on renewable electricity as a share of the total electricity production.

III. Data

In this section, I will describe the data I used to proxy the variables that are included in the analysis. Additionally, I will discuss the other factors that are expected to affect the relative share of RE.

3.1 Dependent variable

Considering that a reduction in electricity produced from conventional sources is

(15)

15

per country as the dependent variable. Therefore, this paper differs from the majority of the reviewed literature as it examines RE in relative- instead of absolute terms. To represent this variable, I use the RE output (as percentage of total electricity output) indicator of the World Data Bank, which has been annually collected since 1990 and is denoted as Share_RE in the model. Within this indicator, Wind, Solar PV, Hydro and Biofuels are considered as renewable electricity.

3.2 Independent variables 3.2.1. Economic growth

As I am interested in the impact of economic growth on the relative share of RE, I use the GDP per capita growth rate indicator of the World Data Bank to proxy economic growth. This annually collected data is based on constant 2010 U.S. dollars and is denoted as EG in the model. Additionally, I use a total of nine lagged variables as a robustness check. They correspond to the economic growth rates of prior years’ up to ten years back per country. These variables are denoted as EG (t-1) and so forth, with the number in the brackets representing what year in the past the lagged variable represents. The reason I include lagged variables is twofold: First, as discussed in the negative view, RE generally has high production costs (Cerović et al. 2017). Therefore, its deployment needs: a stable macro-economic environment as instability may increase risk and uncertainty (Painuly 2001), time to train personnel to maintain and operate renewable energy structures and time to develop facilities for transmission and distribution networks as well as equipment for power companies (Richards et al. 2012). This implies that the impact of economic growth on the relative share of RE might not occur within one year. It could also imply on average, a stable economic growth of several years is needed to have a significant impact on the relative share of RE. Second, it also eliminates the possibility for reverse causality. More specifically, the eventual results might indicate that economic growth impacts the relative share of RE, while the opposite could also be true as argued by some literature. Lagged variables control for this potential issue as it is unlikely that past values of economic growth are impacted by current values of the relative share of RE. To be able to test whether the impact of economic growth is affected by the level of income of a country (hypothesis 4), I will sort countries according

to whether a country is identified as a developed or a developing by the IMF.6 A similar

strategy will be applied to test the strong economic growth hypothesis, which will be explained in detail in the methodology section.

(16)

16

3.2.2. Control variables

Whereas economic growth may be a factor that explains the relative share of RE in a country, the latter is also a function of many other factors. To control for these factors and thus to aim for a minimized omitted variable bias, a set of control variables will be included in the model. Trade openness is often mentioned in the reviewed literature as a variable that may exert a positive impact on the relative share of RE (Lin et al. 2016). The rationale behind this claim is that trade openness, and subsequently the additional movement of products and services between countries, requires an effective use of electricity (Omri and Nguyen 2014). Also, trade openness stimulates economic activity, which is hypothesized to put more emphasis on the enforcement of regulations and stringent environmental standards (Grossman and Krueger 1995; Painuly 2001). Additionally, trade openness may also stimulate the deployment of RE by means of technology transfer (Omri and Nguyen 2014) since RE technologies are generally relatively new and need to compete with well-established conventional electricity technologies (Menanteau et al. 2003). To measure trade openness, I use the annual proportion of imports of goods and services as percentage of GDP per country indicator of the World Data Bank. Despite the fact that illegal transactions and goods carried by travelers across borders are not included and thus potentially distort trade statistics, this indicator is widely used in the reviewed literature to proxy for openness to foreign trade. Within the model it is denoted as Trade_openness.

A second channel through which RE technologies can be transferred and thus its

deployment can be stimulated, especially from developed- to developing countries, is foreign direct investment (FDI). Peterson (2008) concludes that FDI is by far the largest fund when it comes to greenhouse gas mitigation in developing countries by means of technology transfer. Similar results are found by Omri and Nguyen (2014) on a global scale. Contrastingly, Lin et al. (2016) found results that imply that the large influx of FDI into China had a significantly negative impact on the relative share of RE. Despite these ambiguous results, FDI may have a significant effect on the relative share of RE and is therefore included in the model. FDI (similarly denoted in the model) is measured by using the FDI annual net inflows per country in current U.S. dollars indicator of the World Data Bank and it is defined as the sum of equity capital, reinvestment of earnings, and other capital in the reporting economy. Because countries in the dataset are heterogeneous in several dimensions, I use the log-scale of FDI to examine the effect of relative instead of absolute changes.

The switch towards RE is largely driven by the concern of the greenhouse effect

(17)

17

production (United States Environmental Protection Agency 2017). On the one hand, rising

CO2 emissions might strengthen the call for environmental protection (Omri and Nguyen

2014) and thus for a switch from fossil fuels towards RE. On the other hand, the demand for a

switch towards RE may weaken if CO2 emissions in fact decrease. I follow the reviewed

literature in general and the approach of Marques et al. (2010) specifically by including per

capita CO2 emissions in the model. The variable is denoted as CO2 and controls for the

impact of CO2 emissions on the relative share of RE. Marques et al. (2010) consider this to be

a good proxy for environmental concerns as rising per capita CO2 emissions imply that more

electricity is produced from conventional sources. However, I have extracted my data from the World Data Bank instead of Eurostat because I have chosen not to limit my paper to Europe. For reasons similar to FDI, the log-scale of CO2 is used.

(18)

18

In addition to these economic factors, literature regarding renewable energy (e.g. Menz and Vachon 2006; Marques et al. 2010; Aguirre and Ibikunle 2014) emphasizes that differences in population density may explain a significant share of the heterogeneity in RE deployment between countries. A well-known example is the so-called ‘NIMBY’ (not in my back yard) phenomenon. This implies that many residents strongly oppose RE facilities despite being beneficial for the majority of the population (Van der Horst 2007). It may come as no surprise that this effect is present relatively more in countries with a higher population density (e.g. the Netherlands). I use the population density indicator of the World Data Bank as proxy for this. They calculate this variable by dividing the land area in square meters by a country’s population in a specific year and it is denoted as Pdensity within the model of this paper. Also here the log-scale is used.

Geographical conditions may matter for a country’s potential for RE. Literature regarding renewable energy (e.g. Menz and Vachon 2006; Marques et al. 2010; Aguirre and Ibikunle 2014) emphasize that differences in geographical conditions (e.g. unequal amount of sun hours) may explain a significant share of the heterogeneity in RE deployment between countries. However, as it is generally accepted that these factors hardly change over time hence are time invariant. Because I use a fixed effects model (discussed in methodology), any time invariant country specific differences are captured by the constant variable. This eliminates the need to control for time invariant factors hence geographical factors are omitted.

3.2.3. Policy control variables

(19)

19

variable is denoted as Feed_in and represents whether a country has adopted a feed-in tariff. This price-based incentive consists of a guaranteed tariff that is paid to its producers, regardless of the electricity system’s demand situation (Winkler et al. 2016). The second dummy variable includes policies that correspond to both renewable portfolio standards (RPS) and quota systems. These policies closely resemble one another as they are both capital based mechanism. In both schemes, electricity suppliers are obliged to produce a certain percentage of their electricity from renewable sources. By generating RE, companies earn certificates that they can sell to electricity suppliers along with the electricity. The percentage that needs to be generated from renewable resources is generally increased on an annual basis. The idea is that the market will find a way itself to comply with the amount of RE that is mandated (Winkler et al. 2016). Within the model, these policies correspond to the variable “RPS & quota”. The third dummy corresponds to policies regarding grants and subsidies. Public institutions can use these price based mechanism to create financial incentives to stimulate an investment on a project base; often subjected to tight targets and standards that have to be met (Kitzing et al. 2012). Within the model, these policies correspond to the variable “Grants & subsidies”. The data for these variables is collected from both the Renewables Global Status Rapport (REN21 2016) and the IEA Policies and Measures database.

All in all, the total sample includes data collected from 70 countries, of which 22 are classified as developed countries and the rest are classified as developing countries by the IMF (Appendix Table A). I chose data from 1990 to 2014 as measuring period for two distinct reasons: First, because data regarding the relative share of RE has only been collected since 1990. Second, 2014 is chosen as the last year in the measuring period because many countries lacked data on 2015 for several indicators, which would imply that the dataset would shrink significantly in size. Therefore, to simultaneously have data for a large set of countries and a

strongly balanced dataset7 to guarantee reliable results, I selected the period 1990-2014. Table

1 provides a summary of the variables and Appendix Table C provides a brief description of the variables and their sources.

7_{Nevertheless, some observations dropped out of the final results. This is because the value of an indicator has}

(20)

20

Table 1 - Summary Statistics

Variables Observations Mean Std. Dev. Minimum Maximum

Share_RE 1750 38.46497 30.82164 0 100

EG 1750 1.996645 3.679374 -19.05683 30.35658 Trade_openness 1750 34.65976 16.47769 4.631322 100.5974 FDI 1750 1.32e+10 4.21e+10 -2.51e+10 7.34e+11 CO2 1750 4.44743 4.340092 .0172763 20.17875 Price_non_RE 1750 47.5808 34.43367 12.76 111.63 Electricity_con 1750 3795.704 4760.373 34.95031 25590.69 Pdensity 1750 128.215 165.7343 2.221353 1224.593 Feed_in 1750 .2805714 .4494068 0 1 RPS & quota 1750 .0805714 .2722536 0 1

Grants & subsidies 1750 .2502857 .4333014 0 1

EG (t-1) 1750 1.985687 3.736357 30.35658 -19.05683 EG (t-2) 1750 1.995603 3.85312 30.35658 -19.05683 EG (t-3) 1750 1.991916 3.921164 30.35658 -19.39828 EG (t-4) 1750 1.946589 3.936344 30.35658 -19.39828 EG (t-5) 1750 1.871205 3.957401 30.35658 -19.39828 EG (t-6) 1750 1.979801 3.9266 30.35658 -19.39828 EG (t-7) 1750 1.895338 4.012191 30.35658 -19.39828 EG (t-8) 1750 1.727608 4.109718 30.35658 -19.39828 EG (t-9) 1750 1.617323 4.154697 30.35658 -19.39828 Source: see Appendix Table D

IV. Methodology

In this section I will introduce the model and its different versions that I use to examine the impact of economic growth on the relative share of RE. Subsequently, I will introduce two dummy variables that correspond to: (1) the income level of a country (i.e. developed or developing) and; (2) the strength of a country’s economic growth. By applying these dummy variables, it is possible to examine whether the specificities of the impact of economic growth depend on these factors.

4.1 Empirical model

(21)

21

sample data of this paper contains 70 countries (i=1… i=70) with the annual time series spanning from 1990 up to 2014 (t=1… t=25). This approach has several advantages: First, it recognizes that the sample data includes different countries and years. A time-series or cross-sectional approach could be applied as well but the benefits of using a panel data set would be reduced as it does not recognize cross-sectional units and time differences. This is because panel data allows to control for country-specific differences that are not captured by the model because they are either difficult to measure (e.g. culture or business practices) or do change over time but not across cross-sectional units (e.g. international agreements). Thereby, it corrects for a potential omitted variable bias as it accounts for individual heterogeneity and contains less collinearity between the individual variables. Second, the number of observations increases significantly as data is not only collected per cross-sectional unit, but also for several years. This implies that it contains a greater variability of data and more degrees of freedom hence improving the accuracy of the model parameters and the overall robustness of the results. Last, this approach enables me to slice up the database, which is needed to test hypothesis 4 and 5.

The individual heterogeneity can be either fixed or random, depending on what is expected. The difference is whether it is assumed that country specific effects are correlated with the other control variables in the model (fixed effects) or not (random effects). All reviewed literature regarding renewable energy used the fixed effects approach as it is expected that country-specific differences (e.g. geographical factors and qualitative differences in policies) correlate with the various independent variables. To examine which of the two approaches fits the model best, I performed the Hausman test (Appendix Table E). This test favors the fixed effects approach, as its p-value falls below 0.05. Moreover, it is

expected that country-specific differences (random error 𝜀𝑖t in the model) are correlated with

(22)

22

Table G), which can result in biased estimators, the standard errors will also be clustered. As for multicollinearity, I have constructed a correlation matrix for all variables (Appendix Table

H). A correlation suggests a bivariate linear relationship between two explanatory variables,

potentially biasing the estimators (Hill et al. 2012). The results reveal that the variables lnCO2 and lnElectricity_price do correlate to some extent. To test whether this causes a multicollinearity problem, I have applied the Variance Inflation Factor (Appendix Table I). Because individual values are below 10 and the mean value is below 5, the estimators are not significantly inflated and thus I can safely use the model. Altogether, the estimate equation looks as follows:

Share_RE𝑖𝑡 = 𝛼𝑖 + 𝛽1EG𝑖𝑡 + ΣβkXk +𝛽2Trade_openness𝑖𝑡 + 𝛽3FDI i𝑡 + 𝛽4CO2𝑖𝑡 +

𝛽5Price_non_RE𝑖𝑡 + 𝛽6Electricity_con𝑖𝑡 + 𝛽7Pdensity𝑖𝑡 + δFeed_in𝑖 + δRPS & quota 𝑖 +

δGrants & subsidies𝑖 + 𝜀it

Where Share_RE𝑖𝑡 the relative share of RE of country i in year t is, EG𝑖𝑡 is the

economic growth of country i in year t and Xk is a vector of lagged variables of EG𝑖𝑡. The rest

of the model corresponds to the control variables and the error term. The first results regarding the full regression will provide an answer to hypothesis 1, 2 and 3:

1. Economic growth decreases renewable electricity as a share of the total electricity production.

2. Economic growth increases renewable electricity as a share of the total electricity production.

3. Economic growth does not impact renewable electricity as a share of the total electricity.

Subsequently, to test the level of income hypothesis, the countries will be grouped

(23)

23

as developed. Also here, volatility (in the revenues from exporting oil) is the main argument for adopting this criterion. The third and last criterion used is the degree to which a country is integrated into the global financial system (IMF 2018). Thus, because strict criteria (economic or otherwise) are avoided, I expect that the applied method to categorize countries fits the aim of this hypothesis well as it organizes its data meaningfully.

By applying this dummy variable, it is possible to integrate interaction effects between this level of income dummy variable and the economic growth variables. These will capture the interaction effect between the status of an economy and economic growth on the relative share of RE. If significant, it would confirm the hypothesis that the level of income significantly affects the specificities of the impact of economic growth. Additionally, I will run the initial model twice: once including only the developed countries and once including only the developing countries. Altogether, it aims to answer the following hypothesis:

4. The level of income of a country affects the impact of economic growth on renewable electricity as a share of the total electricity production.

With the full sample data now being split between developed- and developing countries, I will go one step further by extending the results of both groups of countries while being separated from one another. I will do this by making a distinction between countries that experienced a strong economic growth and the rest. For developed countries, I have chosen to use two percent as a hard cutoff as economic growth rates above this threshold are generally considered to be exceptionally strong. For developing countries, this cutoff is set on three percent as economic growth rates above two percent are more common and thus less

exceptionally in developing countries.The rationale behind this extension, as discussed in the

literature view, is that the impact of economic growth may be affected by its strength. This led to the fifth hypothesis:

5. The strength of economic growth affects its impact on renewable electricity as a share of the total electricity production.

(24)

24

experienced a relative stable and strong economic growth have a value of 1 (see Appendix B and C for a complete overview). The logic behind this step is that a lot can happen in 25 years, hence grouping countries in five-year periods improves the robustness. Subsequently, I interact this level of income dummy with the economic growth. Thereby, I am able to capture any potential interaction effect.

V. Results

In this section, I will start with presenting the results of the model that includes all countries and with analyzing how the results are affected by adding lagged variables of economic growth. After this, the economic growth variable (and its lagged versions) will be interacted with the level of income dummy to examine whether this leads to divergent results. At last, this exercise will be repeated, only then using the strength of economic growth dummy and developed- and developing countries being separated from one another.

5.1 Empirical results: All countries

Table 2 presents the results from the estimated regression.8 The present economic growth rate

variable (EG (t)) is insignificant in all ten variants of the model. This implies that economic growth of a specific year does not have a significant impact on the relative share of RE in the same year. However, this finding is unsurprising as it is expected that the impact of economic growth on the relative share of RE will not occur within one year. On the contrary, it is expected that economic growth will have a significant impact in a delayed manner and this expectation is supported by the results of models 3 up to 10. When including the economic

growth rate variables of one year earlier,9 two years earlier, three years earlier, four years

earlier and five years earlier, these variables turn out significant. This suggests that economic growth rates from one year up to five years in the past have a significant impact on the relative share of RE on a country level. In general, when economic growth increases by one unit (thus a 1% higher economic growth rate) the relative share of RE decreases by its corresponding coefficient (see Table 2 for exact values). Because all the significant lagged economic growth variables have negative values, they are in favour of hypothesis 1, which claimed that economic growth decreases RE as a share of the total electricity production. Therefore, I conclude that the results are in favor of hypothesis 1 hence both hypothesis 2 (positive impact) and 3 (no significant impact) are rejected.

8_{Model 1 contains no lagged variables of economic growth. The subsequent models add the lagged variables one}

by one up to a total of nine.

(25)

(26)

26

5.2 Level of income hypothesis

To test the level of income hypothesis and thus whether especially developing countries struggle to deploy RE in periods of economic growth, I ran the same model as in section 5.1 but now including only the developing countries. The results can be found in Table 3. These results confirm that an increase in economic growth has a significant negative impact on the relative share of RE (a delayed effect of four up to six years) in developing countries. This suggests that developing countries lack incentives to increase the relative share of RE when economic growth increases. In other words: it seems that the non-binding targets of the Kyoto protocol are insufficient to persuade developing countries to meet rising demand of electricity, as a consequence of an increase in economic growth, with RE instead of electricity generated from conventional sources.

To examine whether this finding is different when comparing with developed countries, I have interacted the economic growth variables with a dummy variable in the model that includes all countries. This dummy variable takes the value of 1 for a developed

country and a value of 0 for a developing country. For ease of interpretation, I have located

the outcomes of these interaction terms next to the original economic growth rate variables.

The results are shown in Table 4. Looking at the interaction terms (column 22), all are insignificant. This suggests there is no significant difference between developed- and developing countries. Although this does not have to be all decisive for interpreting interaction effects because it may still be valuable to analyze the directions, I cannot make any statement on the basis of full statistical certainty and therefore I refrain to do so.

As a robustness check, I also ran the initial model including only the developed countries (Appendix Table J). Here, only the variable representing the economic growth rate of four years earlier (EG (t-4)) remains significant. This implies that the relative share of RE in developed countries is also negatively impacted by an increase in economic growth, but that this impact is less persistent as it fades away after one year.

(27)

(28)

28

Table 4 – Regression results: Interaction effects regarding the level of income hypothesis

Variable (21) Slope-indicators (22)

Economic growth + control variables

EG (t) -0.0259635 (0.081411) EG (t)#level of income -0.1332241 (0.1438118) EG (t-1) -0.0455018 (0.0604044) EG (t-1)# level of income -0.1585667 (0.1101255) EG (t-2) -0.0446503 (0.0533785) EG (t-2)# level of income -0.1425827 (0.1129078) EG (t-3) -0.1269462** (0.0524745) EG (t-3)#level of income -0.1301211 (0.0973459) EG (t-4) -0.0956024* (0.0547256) EG (t-4)# level of income -0.1358742 (0.1180652) EG (t-5) -0.1754114*** (0.0562215) EG (t-5)#level of income -0.088969 (0.1398157) EG (t-6) -0.0149479 (0.0601501) EG (t-6)# level of income -0.0470918 (0.1068456) EG (t-7) 0.0278293 (0.0648278) EG (t-7)# level of income -0.0849949 (0.128926) EG (t-8) 0.0050469 (0.0584364) EG (t-8)# level of income (0.0904861) -0.0785884 EG (t-9) -0.0329868 (0.0674465) EG (t-9)#level of income (0.1191029) -0.1909358 Trade_openness 0.0320746 (0.09593) lnFDI -0.2709673 (0.3031197) lnCO2 -18.92949*** (5.313024) Price_non_RE 0.0426465* (0.0223407) lnElectricity_con 7.096222** (3.647013) lnPdensity -21.30632*** (10.01248) Policy control variables

Feed_in -2.624307 (1.858415) RPS & quota 2.998953 (2.032911) Grants & subsidies 0.8682921 (1.421456) Constant 62.13523 (51.80367) Observations 1167 R-squared 0.4079 Number of countries 70

(29)

29

5.3 Strong economic growth hypothesis

As explained, the results so far suggest that the negative impact of an increase in economic growth lasts longer in developing countries. To test whether the initial strength of economic growth in a country affects the specifics of this negative impact, I apply the same method as in section 5.2. However, this time I use the strength of economic growth dummy instead. Recapping the content of this dummy: for both the developed- and developing countries the observation period of 25 years is split in five separate time series (e.g. 1990-1994, 1995-1999, 2000-2004, 2005-2009 and 2010-2014). A period of strong economic growth corresponds to an average of two percent or higher for developed countries. This threshold lies at three percent for developing countries. Averages are calculated by taking the average of economic growth rates of the relevant five-year period plus the economic growth rates of the five years preceding this period. I do this because the earlier results found a lagged impact of economic growth on the relative share of RE. Developed- and developing countries are regressed separately, which implies that the results will shed new light on whether there are significant differences between countries from the same level of income group. For ease of interpretation, I have again located the outcomes of the interaction effects next to the original economic growth variables. The results can be found in Table 5.

To start with the developing countries; three interaction terms are significantly positive. This means that an increase in economic growth has a less negative (or even positive) impact on the relative shares of RE in developing countries that experience strong economic growth on average compared to developing countries that do not. The results of the robustness check (Appendix Table K, model 39 and 40) support this finding. This is because the economic growth variable is significantly positive for the developing countries that experience economic growth. In other words: an increase in economic growth positively impacts their relative share of RE. On contrast, this coefficient of the economic growth variable is significantly negative for the rest of the developing countries. Similar results hold for developed countries in model in Table 4 and these are supported by the robustness check (Appendix Table K, model 37 and 38).

(30)

30

Table 5 – Regression results: Interaction effects regarding the strong economic growth hypothesis Variable (23) Advanced economies Slope-indicators (24) Variable (25) Emerging economies Slope-indicators (26)

Economic growth + control variables EG (t) 0.0340457

(0.1220728) EG (t) # Strong EG growth (0.1324256) 0.0126712 EG (t) (0.0667912) -0.0041134 EG (t) # Strong EG growth (0.1294769) 0.0633 EG (t-1) 0.0155563 (0.152709) EG (t-1) # Strong EG growth -0.0767225 (0.2051157) EG (t-1) -0.0824079 (0.0715758) EG (t-1) # Strong EG growth 0.233048** (0.1038172) EG (t-2) 0.0410913

(0.0900301) EG (t-2) # Strong EG growth (0.1546447) 0.1370641 EG (t-2) -0.0971146* (0.067258) EG (t-2) # Strong EG growth 0.2317683** (0.0962708) EG (t-3) -0.0312644 (0.0826827) EG (t-3) # Strong EG growth 0.0294151 (0.1441633) EG (t-3) -0.0923863* (0.0694266) EG (t-3) # Strong EG growth 0.2066824* (0.116584) EG (t-4) -0.1825236** (0.0899877) EG (t-4) # Strong EG growth -0.0615158 (0.1863638) EG (t-4) -0.1058754** (0.0657344) EG (t-4) # Strong EG growth 0.1100107 (0.1209466) EG (t-5) -0.1944877** (0.1008893) EG (t-5) # Strong EG growth 0.1111526 (0.1368337) EG (t-5) -0.0966814 (0.0771247) EG (t-5) # Strong EG growth 0.0576207 (0.1225415) EG (t-6) -0.1652633* (0.0972488) EG (t-6) # Strong EG growth 0.2731984** (0.1260016) EG (t-6) 0.0884473 (0.0937428) EG (t-6) # Strong EG growth -0.0661111 (0.1208587) EG (t-7) -0.2292633**

(0.092945) EG (t-7) # Strong EG growth 0.3366091** (0.128057) EG (t-7) (0.0900803) 0.0547779 EG (t-7) # Strong EG growth (0.1199172) -0.0457094 EG (t-8) -0.1085915 (0.0909969) EG (t-8) # Strong EG growth 0.1630042 (0.1278577) EG (t-8) 0.0815826 (0.0793115) EG (t-8) # Strong EG growth -0.0537122 (0.1410267) EG (t-9) -0.0163 (0.1205593) EG (t-9) # Strong EG growth 0.2407784 (0.1578473) EG (t-9) -0.0027212 (0.0692145) EG (t-9) # Strong EG growth -0.0429502 (0.0910656) Trade_openn ess 0.00487 (0.0766934) Trade_openn ess 0.0061378 (0.0498039) lnFDI 0.1966684 (0.1930155) lnFDI -0.596446*** (0.2252849) lnCO2 -35.56363*** (6.125709) lnCO2 -17.00991*** (3.9637) Price_non_ RE -0.0211961 (0.0131183) Price_non_ RE 0.00273517 (0.0185952) lnElectricity_

price (6.320084) 11.44533* lnElectricity_price (2.670293) 2.161512 lnPdensity 2.945846

(9.655763)

lnPdensity -8.353736 (12.0711) Policy control variables

(31)

31

VI. Conclusion and final remarks

The title and research question of this paper reads: Has the developing world forgotten about renewables in its pursuit for economic growth? Already at an early stage it became clear that it is the relative share of RE what matters instead of its absolute amount to properly answer this question. This is because an increase in the latter does not necessarily correspond

to a reduction in CO2 emissions, which is needed to counteract climate change. Thereby, this

paper differentiates itself from the reviewed literature by focusing on the relative share of RE and because it uses a more extensive panel data set (70 countries from all continents with data from 1990 up to 2014) than any of the reviewed studies. After presenting the estimated results of the model and examining both the level of income and strength of economic growth dynamics in the previous section, I am able to answer the five constructed hypotheses and subsequently the research question.

First, there is no clear consensus in the reviewed literature about the impact of economic growth on RE in general and almost none of the studies focuses specifically on its relative share. This raises the question whether economic growth and the subsequent increase in electricity demand are met relatively more by generating electricity from fossil fuels (hence increasing CO2 emissions) or with RE. Using the full dataset, I find clear empirical evidence that an increase in economic growth negatively impacts the relative share of RE. This suggests that the arguments described in section 2.2 (e.g. economic policies tend to ignore environmental consequences, relative high production costs, technical barriers and instability of economic growth) are empirically supported by the results. Therefore, I reject hypotheses 2 and 3 as the results are in accordance with hypothesis 1, which implies that an increase in economic growth counteracts the deployment of RE relative to non-renewable electricity.

Second, to gain more insight into this finding, I have sliced up the dataset by classifying a country either as developed or as developing. The rationale behind this step is the fact that the relative shares of RE have been increasing in most developed countries during the sample period (European Commission 2017). In other words: I hypothesized that especially developing countries struggle to deploy RE in periods of economic growth. The findings of this paper support the arguments for this hypothesis because they suggest that the negative impact on the share of RE caused by an increase in economic growth lasts longer in developing countries compared to developed countries. This finding is in line with the EKC concept (Grossman and Krueger 1991) and the argument that developing countries lack incentives, probably partly explained by not being committed to binding targets, to reduce

(32)

32

developing countries will not slow anytime soon. Change is needed so that the deployment of RE, relative to electricity from conventional sources, is not hampered by economic growth in these countries.

Based on these findings, I recommend the following policy measures. First,

committing developing countries to binding targets to reduce CO2 emissionsis a necessity as

(33)

33

However, the results of section 5.3 (strong economic growth hypothesis) suggest that countries that experience strong economic growth on average (either developed or developing) are an exception to this rule. For these countries, an increase in economic growth, on top of its already strong economic growth, has a less negative or even positive impact on the relative share of RE. There might be several explanations for this finding. First of all, the arguments explained in the positive view (section 2.3) may counteract the negative impact of economic growth (as found by the previous results) in strong growing economies. For example: the impressive economic growth of China during the last decades has led to a rapid increase of smog problems in several urban areas. As a result, the willingness to address pollution and thus the willingness to counteract the rapidly increasing smog problem by means of switching to RE may have risen abnormally fast (Lin et al. 2016). Another possible explanation is that a stronger economic development may accelerate the structural change towards the less polluting tertiary sector as it is argued that an increase in income stimulates the demand for services and lowers the demand for agricultural and industrial products (Sheram and Soubbotina 2000). In general, it intuitively makes sense that when the economic growth reaches certain strength, countries can permit paying relatively more attention to environmental concerns. For the results of this paper, this implies that policies and future research should be aimed at countries that do experience (increases in) economic growth but

have not yet reached strong economic growth rates in general.

Additionally, several limitations should be noted. First, from a CO2 reduction point of

view, the relative share of RE is an imperfect proxy hence a relative increase of RE may occur simultaneously with an absolute increase in non-renewable electricity. As a result, an increase

of the relative share of RE does not necessarily correspond to a reduction in CO2 emissions.

Second, the R-squared does not reach above 0.5 in the majority of the models, which means that a variety of factors explaining the relative share of RE are missing. This may bias the results and thus harm the robustness of the findings of this paper. Last, using dummy variables can be subjective because a hard cut between what represents a value of 1 (or 0) must be made. For example, economic growth in developing countries was considered strong if it was above three percent. Obviously, this threshold could also be set higher or lower, which may alter the results. Therefore, it can be beneficial for future research to examine these dynamics using variant variables instead.

(34)

34

the developing world has forgotten about renewables in its pursuit of economic growth. Therefore, I expect that further research into the factors that hinder the deployment of RE relative to non-renewable electricity in developing countries would be a fruitful continuation of this paper. Additionally, models that better measure the differences between binding and

non-binding targets regarding CO2 reduction could also be valuable as they may strengthen

the argument that also developing countries should commit themselves to binding targets. Last, the RE output (as percentage of total electricity output) indicator of the World Data Bank I used within this paper considers biomass as a source for RE. However, opinions on whether biomass is a clean alternative for fossil fuels are ambiguous because it is argued that biomass is uneconomic and harmful for the environment as it is a pollution-intensive source of energy (Bradley 1998). Excluding biomass as a source for RE may be more consistent

with the ultimate goal of reducing CO2 emissions.

Climate change is real and combating global CO2 emissions is arguably the biggest

(35)

References

Aguirre, M., Ibikunle, G. (2014). Determinants of renewable energy growth: A global sample analysis. Energy Policy, 69, pp.374-384.

Arrow, K,. Bolin, B., Costanza, R., Dasgupta, P., Folke, C., Holling, C., Pimentel, D. (1995). Economic growth, carrying capacity, and the environment. Ecological

Economics, 15(2), pp.91-95.

Atwa, Y., El-Saadany, E., Salama, M., Seethapathy R. (2010). Optimal renewable resources mix for distribution system energy loss minimization. IEEE Transactions on Power Systems, 25(1), pp.360-370.

Bandyopadhyay, S., Shafik, N. (1992). Economic growth and environmental quality: time-series and cross-country evidence. World Bank Working Paper WPS 904. Washington DC:World Bank Publications.

Barbier, E., Common, M., Stern, D. (1996). Economic growth and environmental

degradation: the environmental Kuznets curve and sustainable development. World development, 24(7), pp.1151-1160.

Beckerman, W. (1992). Economic growth and the environment: Whose growth? Whose environment? World development, 20(4), pp.481-496.

Bowden, N., Payne, J. (2009). The causal relationship between US energy consumption and real output: a disaggregated analysis. Journal of Policy Modeling, 31(2), pp.180-188. Bradley, R. (1998). Renewable Energy: Not Cheap, Not Green? Strategic Planning for Energy

and the Environment, 17:3, pp.15-21

Carley, S. (2009). State renewable energy electricity policies: an empirical evaluation of effectiveness. Energy Policy, 37(8), pp.3071-3081.

Cerović, L., Maradin, D., Mjeda, T. (2017). Economic effects of renewable energy technologies. Naše gospodarstvo/Our economy, 63(2), pp.49-59.

(36)

Chang, T., Gupta, R., Inglesi-Lotz, R., Simo-Kengne, B., Smithers, D., Trembling, A. (2015). Renewable energy and growth: Evidence from heterogeneous panel of G7 countries using Granger causality. Renewable and Sustainable Energy Reviews, 52,

pp.1405-1412.

Cheng, XM., Zhang, XP. (2009). Energy consumption, carbon emissions, and economic growth in China. Ecological Economics, 68(10), pp.2706-2712.

Dasgupta, S., Laplante, B., Wang, H., Wheeler, D. (2002). Confronting the environmental Kuznets curve. Journal of Economic Perspectives, 16(1), pp.147-168.

Davenport, C. (2014). Emissions From India Will Increase, Official Says. The New York Times. [online] Available at: https://www.nytimes.com/2014/09/25/world/asia/ 25climate.html [accessed 2018 Oct 19].

De Janosi, P., Grayson, L. (1992). Patterns of energy consumption and economic growth and structure. Journal of Development Studies, 8(2), pp.241-249.

Dhakal, S. (2009). Urban energy use and carbon emissions from cities in China and policy implications. Energy Policy, 37(11), pp.4208-4219.

Dosi, G. (1988). The nature of the innovative process. Technical Change and Economic Theory. London: Pinter Publishers.

Ekins, P. (1997). The Kuznets curve for the environment and economic growth: examining the evidence. Environment and Planning A: Economy and Space, 29(5), pp.805-830. Eleftheriadis, I., Anagnostopoulou, E. (2015). Identifying barriers in the diffusion of

renewable energy sources. Energy Policy, 80, pp.153-164.

European Commission. (2017). Fossil CO2 & GHG emissions of all world countries. [online]

Available at: http://edgar.jrc.ec.europa.eu/overview.php?v=CO2andGHG1970-2016&sort=des8 [updated 2017 Oct 30; accessed 2018 Oct 15]

Foxon, T. (2010). Stimulating investment in energy materials and technologies to combat climate change: an overview of learning curve analysis and niche market support. Philosophical Transactions of the Royal Society of London A: Mathematical, Physical and Engineering Sciences, 368(1923), pp.3469-3483.

Gallagher, K. (2003). Development of cleaner vehicle technology? Foreign direct

Has the developing world forgotten about renewables in its pursuit of economic growth?