Table of contents
List of Figures
List of Tables
List of Abbreviations
2 Literature Review and Deduction of Hypotheses
2.1 Literature Review
2.1.1 Overview of the literature reviewed
2.1.2 Conclusion on the literature reviewed
2.2.1 Hypothesis 1
2.2.2 Hypothesis 2
2.2.3 Hypothesis 3
2.2.4 Hypothesis 4
2.2.5 Hypothesis 5
2.2.6 Hypothesis 6
2.2.7 Hypothesis 7
2.2.8 Hypothesis 8
2.3 The sample of countries – a brief overview of their particular energy markets
2.4 Interrelations and integration obstacles of the four energy markets
3 Empirical Analysis – Interconnection between Renewable Energies and Economic Growth
3.1 Depiction of the sample and methods of data collection
3.2 Implementation of the sample
3.2.1 Selection of variables and parameters
3.2.2 Applied analysis methods and their boundary conditions
4.1 Hypothesis 1
4.2 Hypothesis 2
4.3 Hypothesis 3
4.4 Hypothesis 4
4.5 Hypothesis 5
4.6 Hypothesis 6
4.7 Hypothesis 7
4.8 Hypothesis 8
5.1 Discussion of the results
5.2 Theoretical and practical implications of the results
6 Conclusion and Outlook
List of Figures
Fig. 1: Electricity production from hydroelectric sources per country and year
Fig. 2: Electricity production from renewable sources per country and year
Fig. 3: Clusters of hydropower reservoirs in Brazil
Fig. 4: Net Energy Imports in % of total energy use per country and year
Fig. 5: Key years in electricity sector reforms and global environment initiatives
Fig. 6: Data plot on structural break ARG – representative for Hypotheses 1-4
Fig. 7: Data plot on structural break BRA – representative for Hypotheses 1-4
Fig. 8: Data plot on structural break COL – representative for Hypotheses 1-4
Fig. 9: Data plot on structural break ARG – Hypothesis 5
Fig. 10: Data plot on structural break BRA – Hypothesis 5
Fig. 11: Data plot on structural break CHI – Hypothesis 5
Fig. 12: Data plot on structural break COL – Hypothesis 5
Fig. 13: Data plot on structural break BRA – Hypothesis 6
Fig. 14: Data plot on structural break CHI – Hypothesis 6
Fig. 15: Data plot on structural break COL – Hypothesis 6
Fig. 16: Data plot on structural break ARG – Hypothesis 7
Fig. 17: Data plot on structural break CHI – Hypothesis 7
Fig. 18: Data plot on structural break COL – Hypothesis 7
Fig. 19: Seasonal component of hydroelectricity generation upon electricity prices - Brazil
Fig. 20: Monthly evolution of hydroelectricity generation in selected regions and domestic electricity prices between 2003 and 2011 - Brazil
Fig. 21: Flow rate mean annual cycle for the Paraná River at Itaipú and the Tocantins River at Tucuruí - Brazil
Fig. 22: Global cost estimates from renewable power generation, 2005-
List of Tables
Table 1: Outcome of Hypothesis 1: coefficients, statistical significance, and Δ R²
Table 2: Outcome of Hypothesis 2: coefficients, statistical significance, and Δ R²
Table 3: Outcome of Hypothesis 3: coefficients, statistical significance, and Δ R²
Table 4: Outcome of Hypothesis 4 (including hydroelectricity): coefficients, statistical significance, and Δ R²
Table 5: Outcome of Hypothesis 4 (excluding hydroelectricity): coefficients, statistical significance, and Δ R²
Table 6: Outcome of Hypothesis 5: coefficients, statistical significance, and Δ R²
Table 7: Outcome of Hypothesis 6: coefficients, statistical significance, and Δ R²
Table 8: Outcome of Hypothesis 7: coefficients, statistical significance, and Δ R²
Table 9: Outcome of Hypothesis 8: coefficients and statistical significance
Table S1: Correlation matrix of Hypothesis 1
Table S2: VIF values of Hypothesis 1
Table S3: Durbin-Watson statistics of Hypothesis 1
Table S4: Outcome of Hypothesis 1 after elimination of multicollinearity and autocorrelation: coefficients and statistical significance
Table S5: F-Test on regression model significance of Hypothesis 1
Table S6: Goldfeld-Quandt-Test on heteroscedasticity of Hypothesis 1
Table S7: Chow-Test on structural break of Hypothesis 1
Table S8: Correlation matrix of Hypothesis 2
Table S9: VIF values of Hypothesis 2
Table S10: Durbin-Watson statistics of Hypothesis 2
Table S11: Outcome of Hypothesis 2 after elimination of multicollinearity and autocorrelation: coefficients and statistical significance
Table S12: F-Test on regression model significance of Hypothesis 2
Table S13: Goldfeld-Quandt-Test on heteroscedasticity of Hypothesis 2
Table S14: Chow-Test on structural break of Hypothesis 2
Table S15: Correlation matrix of Hypothesis 3
Table S16: VIF values of Hypothesis 3
Table S17: Durbin-Watson statistics of Hypothesis 3
Table S18: Outcome of Hypothesis 3 after elimination of multicollinearity and autocorrelation: coefficients and statistical significance
Table S19: F-Test on regression model significance of Hypothesis 3
Table S20: Goldfeld-Quandt-Test on heteroscedasticity of Hypothesis 3
Table S21: Chow-Test on structural break of Hypothesis 3
Table S22: Correlation matrix of Hypothesis 4 (including hydroelectricity)
Table S23: VIF values of Hypothesis 4 (including hydroelectricity)
Table S24: Durbin-Watson statistics of Hypothesis 4 (including hydroelectricity)
Table S25: Outcome of Hypothesis 4 (including hydroelectricity) after elimination of multicollinearity and autocorrelation: coefficients and statistical significance
Table S26: F-Test on regression model significance of Hypothesis 4 (including hydroelectricity)
Table S27: Goldfeld-Quandt-Test on heteroscedasticity of Hypothesis 4 (including hydroelectricity)
Table S28: Chow-Test on structural break of Hypothesis 4 (including hydroelectricity)
Table S29: Correlation matrix of Hypothesis 4 (excluding hydroelectricity)
Table S30: VIF values of Hypothesis 4 (excluding hydroelectricity)
Table S31: Durbin-Watson statistics of Hypothesis 4 (excluding hydroelectricity)
Table S32: Outcome of Hypothesis 4 (excluding hydroelectricity) after elimination of multicollinearity and autocorrelation: coefficients and statistical significance
Table S33: F-Test on regression model significance of Hypothesis 4 (excluding hydroelectricity)
Table S34: Goldfeld-Quandt-Test on heteroscedasticity of Hypothesis 4 (excluding hydroelectricity)
Table S35: Chow-Test on structural break of Hypothesis 4 (excluding hydroelectricity)
Table S36: Correlation matrix of Hypothesis 5
Table S37: VIF values of Hypothesis 5
Table S38: Durbin-Watson statistics of Hypothesis 5
Table S39: Outcome of Hypothesis 5: coefficients and statistical significance
Table S40: F-Test on regression model significance of Hypothesis 5
Table S41: Goldfeld-Quandt-Test on heteroscedasticity of Hypothesis 5
Table S42: Chow-Test on structural break of Hypothesis 5
Table S43: Correlation matrix of Hypothesis 6
Table S44: VIF values of Hypothesis 6
Table S45: Durbin-Watson statistics of Hypothesis 6
Table S46: Outcome of Hypothesis 6 after elimination of multicollinearity and autocorrelation: coefficients and statistical significance
Table S47: F-Test on regression model significance of Hypothesis 6
Table S48: Goldfeld-Quandt-Test on heteroscedasticity of Hypothesis 6
Table S49: Chow-Test on structural break of Hypothesis 6
Table S50: Correlation matrix of Hypothesis 7
Table S51: VIF values of Hypothesis 7
Table S52: Durbin-Watson statistics of Hypothesis 7
Table S53: Outcome of Hypothesis 7 after elimination of multicollinearity and autocorrelation: coefficients and statistical significance
Table S54: F-Test on regression model significance of Hypothesis 7
Table S55: Goldfeld-Quandt-Test on heteroscedasticity of Hypothesis 7
Table S56: Chow-Test on structural break of Hypothesis 7.
List of Abbreviations
illustration not visible in this excerpt
Climate change and greenhouse gas (GHG) emissions, as well as the United Nations (UN) convention UNFCCC of Rio de Janeiro (1992) and the Kyoto Protocol of 1997, are synonyms that reflect global community’s arising awareness of anthropogenic global warming (Brechin, 2003). This is mainly caused by too much GHG emissions, in particular CO2. Politicians attempt to set binding benchmarks for CO2 reduction in all countries, and to align them with domestic economic and political interests. On this awareness, authors like Salazar (2008), Chang et. al. (2009), Chang and Soruco Carballo (2011), Chien and Hu (2007), Chien and Hu (2008) Gan et. al. (2007), Luecke (2011), Ruiz-Mendoza and Sheinbaum-Pardo (2010) and Sheinbaum et. al. (2011) base their analyses of changes in the global energy market while considering renewable energy (RE) sources and implications for energy policies and regulations. The aforementioned climatic changes, and the attempt to mitigate their economic and environmental impacts, are owed to the recognition that humans are to a large extent responsible for these developments. People have a demand for energy and electricity that can be gained from either non-renewable or renewable natural resources. Non-renewable resources are those whose global occurrence is limited, their depletion depending on overall demand accordingly. These resources accounted for 90% of global energy consumption in 2008, thus largely dominating RE whose share is expected to slowly increase from 10% in 2008 up to 14% in 2035 (EIA, 2011a). Between 1990 and 2008, world energy consumption increased by 43%, and it is supposed to increase by further 53% by 2035 (Ibid.). Acaravci and Ozturk (2010), Ang (2007), Chang and Soruco Caraballo (2011) and Kuntsi-Reunanen (2007) identified increasing energy consumption (EC) from non-renewable resources as crucial for CO2 emissions which are produced when fossil fuels such as coal, natural gas and oil are burned. Their findings and EIA expectations are backed by UN world population projections assuming that the world population will increase by 47% between the years 2000 and 2050, and then reaching 8.9 billion, which implies the same number of potential energy consumers (UN, 2004). Such steady growth makes clear that energy is not a simple commodity which per se would be available in unlimited amounts. Thus, the origin of primary energy, and electric energy derived from it, gain in importance. Luecke (2011) even asserts “a dramatic evolution” of the global energy market and a shift to an environmentally friendly and sustainability-oriented energy market. Since Luecke (2011) speaks of an “energy market”, it is obvious that energy is a good whose production and consumption are directly interrelated with follow-up economic activities. Hence, Apergis and Payne (2010a) rightly point out that in spite of all discussions regarding mitigation of GHG emissions and the search for appropriate alternatives, the compatibility of corresponding policies and regulations with economic growth (EG) always plays an important role. This is also seen as an imperative for the South American and the Latin American and Caribbean regions in general, as demonstrated by Sadorsky (2012) and Sharma (2010). The former adverts that South America records fast and strong growth in terms of EC, economic output, and trade activities in the last two decades. According to him, potential causalities between these factors have not been examined yet. Sharma (2010), who analyzes the energy-economic growth nexus for six world regions, figures out that the Latin American and Caribbean regions have the highest “…average growth rate of fossil fuel energy consumption” in percentage of total EC among these six regions (Sharma, 2010, p. 3568). It is predicted that by the year 2100 climate change and resulting environmental problems will have cost implications equal to 137% of the GDP of 2011 if no measures are taken to mitigate effects of climate change (Chang and Soruco Carballo, 2011).
These evidences strongly suggest a need to deal with the interrelation of EC and EG in the South American region and with alternatives to fossil energy which are compatible with EG and climate protection. Therefore, the following research question can be formulated: Can renewable energies act as an acceleration factor upon the development of emerging South American economies? In accordance with the topic of my thesis, the research question refers to the theory that RE may contribute to EG. This notion derives from corresponding approaches that were chosen, for instance, by Apergis and Payne (2011a), Chien and Hu (2008), or Sadorsky (2009a). In the following, different approaches which addressed this research question in a similar way in previous studies will be discussed. Hypotheses will then be formulated which cover relevant aspects of the aforesaid theory enabling to answer at the end some of the research questions posed here.
This thesis is supposed to contribute to economic research in the energy sector through narrowing the focus on the RE markets of a few economically and politically comparable South American countries. Identification of critical success factors, essential for the evolution of RE and the relation of the latter with EG, will complement my research study. Finally, I attempt to put forward certain recommendations on energy policies and investments which may have practical implications for decision-makers in the selected South American countries.
The introductory Chapter 1 will be followed by a literature review which is based on manuals released by Anglo-Saxon universities (Edith Cowan University, 2008; Taylor, 2012; University of North Carolina (UNC), 2012). To structure this review, I merge thematic and methodological approaches which I consider to be more appropriate than a mere chronological, methodological, or thematic review according to UNC (2012). From findings in this review, I will derive eight hypotheses which address the research question and the underlying theory. To put the hypotheses in a more global context, I present an analysis of the energy markets of the four South American countries and their interrelations to complement Chapter 2. Chapter 3 is dedicated to data allocation and to implementation of the sample. This includes description of the quantitative methodology applied upon the empirical secondary data, and definition of the given assumptions and conditions which are required to validate the results. The econometric examination of the hypotheses in Chapter 4 is followed by the discussion and interpretation of the results in Chapter 5, focusing on the relevance of the renewable energy proxies. Concluding remarks and an outlook will be given in the Chapter 6.
2 Literature Review and Deduction of Hypotheses
2.1 Literature Review
2.1.1 Overview of the literature reviewed
In the papers reviewed here, all authors agree on the view that energy source alternatives must exhibit positive GHG emission balances, i.e. emissions which are significantly lower than those of fossil fuels. With respect to concrete alternatives, two viewpoints have been identified: on the one hand, to increase the use of existing nuclear technology for energy generation, and on the other hand, to enhance the share of RE in the total energy supply. To avoid blurred usage of the term renewable energy, Meisen and Krumpel (2009) point out that RE sources can be defined in a narrower and in a broader sense. The latter is adopted by the EIA, IEA, and the World Bank, and, apart from hydro, solar, tidal and wind energy, also includes combustible renewables and waste which embrace “solid biomass, liquid biomass, biogas, industrial waste, and municipal waste”. Given the fact that relevant data for empirical analysis will mainly stem from these institutions, the broader definition will be adopted in this thesis. A sharp distinction must also be made between energy consumption and electricity consumption (ELC), two variables which are highly relevant for the subject of this thesis. While EC encompasses all sources energy is gained from, ELC considers only electric energy, which can originate from various sources such as hydroelectric power stations, solar panels, wind power plants, but excluding fossil fuels such as oil or gas. The latter are usually not used for electricity generation, but for gasoline production or heating.
World’s high dependence on oil, which became particularly evident in the two oil price shocks in the 1970s, and in its function as a driving force of overall energy costs, is the starting point of Lee’s and Chiu’s (2011) analysis of the interdependence between “nuclear energy consumption, oil prices and economic growth”. They aim to demonstrate on a sample of industrialized countries that nuclear energy can become a long-term substitute for oil to better meet general economic needs and guarantee stable energy prices and constant energy supply. They admit that this approach has not received much attention in the past, but had been taken up mainly by Apergis and Payne (2010b), Schurr (1983), Wolde-Rufael (2010), Wolde-Rufael and Menyah (2010), Yoo and Jung (2005) and Yoo and Ku (2009). Some other related publications are in press. From their empirical analysis, they cannot conclude that nuclear energy can substitute oil, or would at least be complementary in all regarded countries in a similar way. Thus, it could not generally be proved that nuclear energy might be a long-term alternative for mitigating GHG emissions and compensating increasing oil prices.
Furthermore, the nuclear incident in Fukushima/Japan of 11th of March 2011 showed that nuclear technology has considerable drawbacks when it gets out of technical control. Moreover, final and safe places for long-term storage of uranium waste have not been identified yet. Notwithstanding, a broader review of nuclear energy-related literature would be necessary in order to reach a final conclusion on Lee and Chiu’s approach. But their results, in conjunction with the nuclear incident in Japan, give good reasons to shift the focus from the first alternative, nuclear energy, to the second alternative, RE. Critical observers might argue that dooming nuclear energy and glorifying bioenergy, which is entirely based on renewable resources, is undue and too simple. Potential drawbacks should not be neglected. The construction of hydropower plants draws on huge amounts of land mass, having strong impact on the natural course of rivers, and often disturbing the biodiversity in a certain region. Biofuels are usually made of crops that for their growth claim a lot of land, which actually would be used for food production if subsidies did not make biofuel production more attractive to farmers, a situation that raises many ethical concerns (Ruiz-Mendoza and Sheinbaum-Pardo, 2010; Zuluaga and Dyner, 2007). In spite of the relevance of a critical discussion on secondary effects associated with the extension of RE, an analysis of those issues would be beyond the scope of my thesis. That is why I adopt the approach of ECLAC and GTZ making a clear distinction between renewability which “…is defined as an attribute of the energy source…”, and sustainability which “…is defined as an attribute of the way the source is used” (ECLAC and GTZ, 2004, p. 57). Henceforth, secondary effects and potential drawbacks as described before are considered as issues of sustainability, but the scope of the thesis will be renewability of energy sources. Chien and Hu (2007), Reijnders (2006) and Zuluaga and Dyner (2007) are aware of the aforementioned critical points, and admit that RE technologies would at present not be competitive without public subsidies. But, as Chang and Soruco Carballo (2011) commented, global warming caused by GHG emissions is a matter that will affect economic activities worldwide. This requires considering alternatives which may not promise countable monetary benefits in the short term, but which could prevent losses. The last aspect is a notion taken up by Awerbuch and Sauter (2006) which will be analyzed in more detail when methodologies are discussed.
The geographical area of research interest underlying my thesis work is South America. In order to classify the South American energy market, its strengths and weaknesses, risks and opportunities within a global context, I will analyze which other regions were of interest to present studies, and how authors did define such regions. I will then pass on to the methodological part and analyze how related previous studies approached the examination of their subjects. Before discussing empirical findings of previous studies, dealing with the interrelation of RE with economic activities, at first stage, the focus will be on energies in general, not differentiating between renewable and non-renewable ones. The reason for this is that potential impact of RE on economic activities is hard to conceive if energy in general were irrelevant to them. Hence, in the first step, I will examine whether, and to which extent, relations between energy and economic activities exist before narrowing down the scope to RE in the second step.
The definition of a region depends first of all on the character of the analysis that is used to specify whether it is a quantitative or a qualitative study. In the past, five different types of approaches were applied to define a geographical area of interest in quantitative studies: general panel data, panel data covering a certain political region, panel data covering a certain geographical area, focus on a selected number of countries, or an in depth-analysis of one single country.
The general panel data approach was chosen by Chien und Hu (2008), Narayan and Popp (2012) and Sharma (2010). Chien and Hu (2008) set up their quantitative models on the basis of 116 countries, while they did not consider any country specifics, but covered a large sample size which is potentially good for statistical validity. On the basis of 93 countries, Narayan and Popp (2012) carried out their empirical survey, and they also showed results for the different continents and for some selected countries. Sharma (2010) divides the total size of samples into regions and continents. She argues that panel studies are underrepresented in the literature giving her special motivation to broaden the scope of the analysis.
Empirical studies on RE in terms of economic activities in a predefined political region were amongst others carried out by Apergis and Payne (2010c), Chang et. al. (2009), Chien and Hu (2007), and Yoo and Lee (2010). Their panel of countries embraces the 26 OECD member states. The two latter authors further included in their publications 19 non-OECD member states to figure out any significant differences between both sample groups based on their level of development. OECD member states are mainly developed countries, whereas the second panel group is composed of developing and emerging countries. In the OECD-related studies, statistical validity is gained through a high number of sample countries whereas the period analyzed plays a minor role. In contrast to those studies, Apergis and Payne (2009) and Apergis and Payne (2010d) need to take into account chronology since they examined economic-energetic interdependences of the Commonwealth of Independent States (CIS). Thus, they can only consider data as of 1991, therefore facing a left-side time constraint. The approaches of Apergis and Payne (2011a) and of Sadorsky (2009a), which are composed of a panel of data of 16 or 18 emerging countries from all over the world, fit into the same category as the OECD-related studies since the term emerging country is used in a politico-economic context. While Pao and Tsai (2010) focused on the largest and most powerful emerging countries, the so-called BRIC states, Narayan and Smyth (2008) analyzed an energy-economic growth nexus for the G7 states. Yoo (2006) does the same for the ASEAN region. Such country groupings often serve to underscore the economic and political influence and relevance of those countries. Furthermore, they describe the shifting of power balance between developed and non-developed countries, and between the occidental and non-occidental world.
Panel data covering a certain geographical region were collected by Apergis and Payne (2010a), Chang and Soruco Carballo (2011), Wolde-Rufael (2006), and Zilio and Recalde (2011). The latter, as well as Chang and Soruco Carballo (2011), examined a panel of 20 and 21 countries from Latin America and the Caribbean, taking into account a period of more than 30 years. Hence, the sample comprises more than 600 observations (number of countries multiplied by the time period) giving a reliable data base. Wolde-Rufael (2006) carried out a study with a similar sample size of 17 African countries. Apergis and Payne (2010a) focused on Eurasia whereby they contribute to close a geographical research gap since that region is usually not of primary interest for researchers studying the energy market. At this point, it becomes clear that grouping countries according to geographical and/or political criteria is a reasonable method to analyze the energy market from different perspectives on a common statistical basis.
A kind of intermediate approach has been applied by Apergis and Payne (2011b) and Narayan and Smyth (2009). For their analysis, they selected a small number of countries, six or seven countries, respectively, which are located in Central America or in the Middle East. A single-digit number of countries were also selected by Ruiz-Mendoza and Sheinbaum-Pardo (2010), Sadorsky (2012), Sheinbaum et. al. (2011), and Yoo and Kwak (2010), but compared to Apergis and Payne (2011b) and Narayan and Smith (2009) they selected countries in a certain region based on economic, legal, or political commonalities. Ruiz-Mendoza and Sheinbaum-Pardo (2010) analyzed the impact of electricity sector reforms on CO2 emissions and RE in Argentina (ARG), Brazil (BRA), Colombia (COL), and Mexico. Since these four countries realized similar liberalization reforms of their energy markets at the beginning of the 1990s, a standardized starting point for analysis is given. This allows for better comparison of the developments in the aftermath of the reforms. Further, the fact that these four countries represent a major part of the population and of the GDP of the Latin American and Caribbean regions was an important criterion for limiting the scope of analysis. Sheinbaum et. al. (2011) also took into consideration socio- and macro-economic indicators for the selection of sample countries, among others the GDP per capita, HDI and the population’s annual growth rate. More implicitly, such indicators were taken into account by Yoo and Kwak (2010) since they focused on all major South American countries where they identified a well-developed electricity infrastructure as critical for further economic development. An in-depth quantitative or qualitative analysis of one country´s energy sector was centered on either the entire energy market or a specific energy source. The latter was taken up by Kissel and Krauter (2006), Martins and Pereira (2011), and Watts and Jara (2011) who analyzed wind and solar power in BRA in terms of RE policies, and carried out a “Statistical analysis of wind power in Chile”. In comparison to that, Recalde (2011), and Peláez-Samaniego et. al. (2007) examined the energy markets of ARG and Ecuador, without having on scope one specific energy source. These in-depth approaches facilitate the detailed analysis of a country`s energy market and help to clarify if a certain electricity-generating technology is appropriate and promising for a certain country, e.g. electricity generation through wind turbines in Chile (CHI). A potential drawback of qualitative analysis of single countries is lower statistical validity since the numbers of observations are smaller, and/or sector-specific data are not available. For this reason, Recalde (2011) and Peláez-Samaniego et. al. (2007) approach their research questions by descriptive rather than inductive statistical methods.
The general problem of data availability was observed in many papers. Since data are often not freely available, or do not perfectly meet individual needs, proxies are used as resort. This route was chosen by Chien and Hu (2008) for example. To setup a multivariate model based on the GDP expenditure approach they searched for data on the gross domestic investments in their sample countries in 2003. Since such data were not available, they used CF in % of GDP as proxy retrieved from the World Bank.
Some factors affecting the energy market are not reasonably quantifiable for purposes of inductive statistics. This is mainly the case when the impact of political decisions, laws, or subsidies on the energy market is examined. Then, qualitative studies are carried out which are more meaningful. Nonetheless, descriptive statistics are used to underscore or visualize qualitative statements. Such an approach was followed by Gan et. al. (2007), and by Kissel et. al. (2009). They analyzed how policy instruments and related laws, such as tax incentives, GHG emission certificates, or feed-in tariffs, contribute to the establishment of RE on energy markets in Europe, in the US, and in South America. Since the success of such policy instruments and laws is linked to benchmarks to be achieved, the aforementioned authoring groups do the following: they take into account the current situation and the projected scenario, thereby analyzing under which conditions the latter is realistic, and then discuss the principal advantages and disadvantages of a certain instrument. Further, they argue positive or negative discriminating impacts of such instruments upon certain energy sources. They use charts and tables to underline and visualize their data. A top-down approach is followed by Guzowski and Recalde (2010) in this context. Before they discuss which RE technologies are potentially promising, given the existing resources and geographic conditions, they analyze the current electricity markets and consumer habitudes. So, they examine if there is potential demand for RE that would justify investments into certain technologies providing supply RE. A broader perspective is presented by Gischler and Jansen (2011), and Luecke (2011). The former discuss how decentralized electricity generation (DG) based on RE could be used in the emerging Latin American and Caribbean markets, explaining which conditions in terms of legal framework and public promotion measures must be fulfilled to make DG successful. Further, they draft how DG could change the energy market in a positive way especially in developing countries. Caspary (2009), Dutra and Szklo (2008), Geller et. al. (2004), Luecke (2011), and Ruiz et. al. (2007) set up current-potential comparisons for existing RE, reflecting about their cost-effectiveness, and they ask which promotion measures (e.g. subsidies or public-private partnerships) must be taken by the private and public sector to establish RE on the energy market, in order to make these energies competitive and to stimulate the economy through job creations.
Energy as a generic term with regard to economic activities, in particular EG, was subject of numerous scientific studies in the past. This is underscored by Ozturk (2010) who dedicated a study to this field of research carrying out a survey that compiles the main relevant literature. Ozturk depicts in tabular form which causal relationships were studied, which methodology was applied, and what was the given period analyzed in the papers published. He enriches his survey with a summary of the main results indicating which ones are conflicting, and which ones turn out to be common sense. It can be stated that the causal relationship between GDP and EC, respectively ELC, was of particular interest to the authors who analyzed periods of sixteen years (Ramcharran, H., 1990; Yu, E.S.H. and Jin, J.C., 1992) and fifty-seven years (Payne, J.E., 2009). In the following, the causal relationships that are underlying the aforementioned quantitative studies will be presented, and the methodology applied will be discussed. The main purpose of this section, at a first stage, is to identify relationships that are relevant for approaching the research question of the thesis. At a second stage, I will analyze which of these relationships are adoptable and/or to which extent they need to be customized. In the last step, the applicability of present methodologies to my thesis work will be the main subject.
According to Ozturk (2010), empirical studies were carried out on two main causalities: The causal relationship between EC and GDP on the one hand, and ELC and GDP on the other hand. Most studies assume implicitly that GDP is usually measured as real GDP per capita and considered as a proxy for EG, whereas EC or ELC is indicated in different units, but Yoo (2006) explicitly point out to this fact. Ozturk (2010) adds that while using GDP it should be kept in mind that this parameter is not always measured correctly, mainly in developing countries, giving a possible explanation for unreliable results. Most models reviewed consider further variables such as CF, exports, household income, labor force (LF), or they differentiate between RE and non-RE, or electricity. Those quantitative studies encompass different aggregation levels of countries (single, multiple, or large panels) as described before. The two main causal relationships can be classified into four groups: no causality between both variables, two uni-directional causalities where EC depends upon GDP and vice versa, and bi-directional causality where both variables condition one another (Apergis and Payne, 2011a, Apergis and Payne, 2011b; Ozturk, 2010; Chang and Soruco Carballo, 2011; Squalli, 2007; Yoo and Kwak, 2010). These authors are representative for all the others who implicitly assumed such causalities, since they formulated them explicitly and educed from them four hypotheses that have general implications for energy policy makers: the neutrality hypothesis, the conservation hypothesis, the growth hypothesis, and the feedback hypothesis. The neutrality hypothesis refers to the observation that there is no causal relationship between EC or ELC and EG. Hence, no matter whether energy policy takes conservative or expansive measures upon EC or ELC, there is no impact on GDP. The conservation hypothesis and the growth hypothesis are defined for unidirectional causalities. The conservation hypothesis applies if changes in GDP bear on EC or ELC, which implies that policies promoting EC or ELC have no impact upon GDP. The growth hypothesis in this context applies if EC or ELC significantly contributes to changes in GDP, in particular to GDP growth being in line with capital and labor. Consequently, stable and sufficient energy supply at affordable prices is a growth factor (Bassi et. al., 2009; Ghosal, 2000). The bi-directional causality refers to the feedback hypothesis implying that EC or ELC and GDP affect each other at the same time.
The causality between EC and GDP has been examined only by Narayan and Popp (2012), Ozturk et. al. (2010), Payne (2009), Yoo (2006), Yoo and Kwak (2010), and Zilio and Recalde (2011). Narayan and Popp (2012) analyzed this causality for 93 countries splitting them into seven panels according to regions. Their intention was to figure out if for these panels a long-run impact of EC on GDP does exist. For six out of 17 Latin American and Caribbean countries analyzed, they detected a negative Granger Causality between GDP and EC meaning that decreasing GDP will cause an increase in EC, and vice versa. For most of the 93 countries they do not examine such a relation. The so-called Granger Causality which they applied in their empirical work is an econometric concept that is used by the big majority of all studies examining the EC and ELC – GDP causality or related nexuses (Tables in Ozturk, 2010). The basic idea of the Granger Causality test in this context is to analyze whether EC or ELC has a significant impact on GDP, or vice versa, contributing to the prediction of EG and future EC and ELC. Since this test is usually applied upon time series, the Granger Causality is especially used for long-run examinations, assuming that the past predicts the future. Further, Granger assumes a stationary process. This means that the time series is supposed to be time-invariant having the same expected value and the same variance at each time point. The covariance of two random variables must be time-invariant and meet the criteria for a stationary process. To prove if a time series is stationary, unit root tests are used. If the existence of a unit root is identified, a time series is non-stationary. Alternatively, in several papers, the Augmented Dickey-Fuller Test (ADF) is applied. It does not directly prove if a time series is stationary, but it examines the existence of a unit root. At the end, both approaches give an answer to the question of whether a time series is stationary or not. Time series often experience trends or seasonal influences which could be misinterpreted as so-called spurious regressions. Spurious regression occurs if there is a long-term equilibrium among at least two non-stationary variables, which means that they are cointegrated. To detect such spurious regressions, the test upon cointegration is applied. When the result is positive, such trend or seasonal influences are first eliminated before regression models are applied to the raw data. Therefore, so-called error correction models are used in order to minimize the loss of statistical information. They enable to differentiate between long- and short-term Granger Causality, and hence mitigate the problem of trend data and seasonally afflicted data (About.com, 2012; Drobetz, 2003; Schulze, 2004; Von Auer, 2007). For complexity reasons, the ADF-test, error correction models, Granger Causality, and unit root tests will not be applied in this thesis.
Ozturk et. al. (2010) chose a sample of 51 countries that they split into three income groups (low, middle, and upper). In contrast to Narayan and Popp (2012), they could not identify a significant relation between EC and GDP for any of the three income groups. Payne (2009) made a similar observation for the US because his results support Granger’s neutrality hypothesis. Based on the Energy Environmental Kuznets Curve (EKC) hypothesis, which assumes that EG and environmental deterioration are interrelated in the long-run, Zilio and Recalde (2011) carried out a survey of 21 Latin American and Caribbean countries, obtaining results that do not support the hypothesis of a long-term relationship between both variables either. Apart from the time periods analyzed, data used as proxies for the two variables differ in terms of units. While Narayan and Popp (2012), and Zilio and Recalde (2011) count GDP per capita in constant year 1990 USD, Ozturk et. al. (2010), and Payne (2009) use constant year 2000 USD. The two latter measure total primary energy supply, as a proxy for EC, in Kt of oil equivalent per capita while the two others use Kg of oil equivalent per capita. These differences need to be taken into account when results are interpreted, or studies are compared. As such analyses do not focus necessarily only on two variables, inclusion of further variables as proxies for EC and ELC and/or EG can make models more meaningful. Apergis and Payne (2010d), Chang and Soruco Carballo (2011), and Sadorsky (2009b) consider CO2 emissions in addition to EC and GDP, since these emissions are strongly linked to EC (Kuntsi-Reunanen, 2007). All three studies evidence that these three variables are interrelated. In the case of the CIS, Apergis and Payne (2010d) even identify a long-run equilibrium relationship whereupon Russia’s exclusion from, or inclusion in, this sample is crucial since the outcome changes. While EC and EG were complementary, the inclusion of Russia in the panel turns these two variables into an opposite relation. The authors advert to the fact that countries such as Russia in the CIS region are regionally dominant in terms of economic size and EC, and might distort the result so that it is reasonable to carry out two examinations as done in this case. This evidence is potentially relevant for studies including BRA since the latter has a dominant role in South America (Chapter 2.3).
Chang and Soruco Carballo (2011) remark that only in four out of twenty Latin American and Caribbean countries examined no causality between CO2 emissions and GDP can be observed. Hence, the converse argument is that CO2 reduction or energy conservation policies will affect GDP growth. This notion is supported by Pao and Tsai (2010) who examined these relations in the case of the BRIC states. They found causality between GDP as dependent variable, and CO2 emissions and EC both in the short-run and in the long-run. But they add that energy policies such as the aforementioned ones do not necessarily bear negatively on GDP as long as they are paralleled by investments in energy supply capacities and by an increase of energy efficiency. All of them count CO2 emissions in Kt. Apergis and Payne (2009a), Apergis and Payne (2009b), Apergis and Payne (2011b), Chien and Hu (2007), Narayan and Smyth (2008), and Sadorsky (2012) extended the scope to macroeconomic variables such as CF, LF, and trade balance (TR) determining factors. They all apply methodologies such as cointegration test, error correction model, and Granger Causality. Apergis and Payne (2009b), as well as Apergis and Payne (2011b), focused on the Central American region analyzing the relationship between EC, or renewable EC, and three macroeconomic factors, namely GDP, LF, and CF. By definition, renewable EC is equal to renewable electricity net consumption measured in millions of kilowatt hours (MWh). In both cases, the authors identify a long-term equilibrium among the four variables, and Granger causality in the long- and in the short-run. While they examine unidirectional causality between EC and GDP supporting the growth hypothesis, they identify a bidirectional relationship for renewable EC. They explain the latter observation through economic potentials provided by RE. They claim that public subsidies such as direct financial drawings, that promote investments into RE technologies, or tax benefits, that incentivize the energy sector to modernize, may contribute to economic stimulation. Furthermore, they assert that the high price volatility of fossil resources can be balanced through RE whose price level is more predictable, therefore reducing cost burdens for the economy. When examining the same variables for the CIS countries, Apergis and Payne (2009a) find bidirectional causality in the long-run, whereas in the short-run, the growth hypothesis is supported by the EC-GDP nexus. The feedback hypothesis states that energy policies reducing EC in the short-run cannot have a sustainable impact on EG since the economy adapts to changing conditions.
An approach more concentrating on how, rather than if, renewable energy impacts on GDP was chosen by Chien and Hu (2007). Since they predetermine a unidirectional causality running from the three independent variables of EC, capital stock, and LF to GDP, Granger Causalities, are not examined. They apply the concept of technical efficiency to 45 OECD and non-OECD economies. Their basic idea is that an economy is the more technically efficient the less energy input is needed to produce the same output of goods. Hence, through energy efficiency, the economy’s productivity can be augmented. Countries are either net energy importers or exporters. EI are negatively recorded in the TR. Thus, higher technical efficiency leading to lower EI improves the TR. Chien and Hu (2007) conclude from their survey that RE can improve notably the energy balance of non-OECD countries as technical efficiency is lower there. This EC-TR nexus is also taken up by Sadorsky (2012) who focused on seven South American countries using the same methodological techniques as Apergis and Payne (2009a, 2009b, and 2011b). He argues that much literature focuses on the EC-GDP (Ozturk, 2010) and the export-GDP nexus (a.o. Ahmad and Harnhirun, 1995; Kónya, 2006), but less attention is paid to the EC-export nexus. His approach comprises the integration of both nexuses by analyzing the EC-export nexus. He ascertains long-run bidirectional causality between both variables, leading him to the conclusion that energy policies reducing EC induce a decrease of trade volume, but will ultimately affect GDP, too. That’s why he draws a direct linkage between unlimited export of energy and wealth in South America. At this point, it can be stated that increasing importance is attached to energy as a tradable good.
An important methodological aspect of structural break is brought up by Narayan and Smyth (2008). This statistical concept basically means that regression parameters are inconsistent over time and change abruptly, which is particularly relevant since time series data are influenced by events happening in certain years (Von Auer, 2007). Through their analysis of the relation between EC, CF, and GDP, Nayaran and Smyth (2008) detect structural breaks, for instance, in several Western countries in the course of the oil price shocks in the 1970ies, and the stock market crash in 1987. Hence, the testing on structural breaks is thought to be appropriate also for this thesis as the data used will cover a longer period of time.
Another major nexus examined is the relationship between ELC and EG (Ozturk, 2010). This approach was taken up, amongst others, by Ciarreta and Zarraga (2010), Narayan and Smyth (2009), Ramcharran (1990), Squalli (2007), Wolde-Rufael (2006), Yoo (2006), Yoo and Kwak (2010), and Yoo and Lee (2010). With the exception of Ramcharran (1990) and Yoo and Lee (2010), all authors examined data upon Granger Causality and cointegration. While the former carries out a regression analysis for Jamaica applying a few common statistical tests such as the test upon auto-correlation and the F-test, the latter use so-called EKCs to approach this nexus. Hence, the methodologies used are in general the same for the EC- and the ELC-EG nexus. ELC is measured in KWh per capita. For both the EC-EG nexus and the ELC-EG nexus, it was realized that causalities valid for a certain region or a political union of countries are very difficult to find. Rather, countries running similar energy policies exhibit the same causalities. Only five out of eleven OPEC states show a unidirectional causality running from ELC to GDP. Squalli (2007) suggests that by implication of non-uniform energy and economic policies across countries, causalities are diverging. These observations are also made by Wolde-Rufael (2006), Yoo (2006), and Yoo and Kwak (2010) who obtain mixed results by examining both unidirectional and bidirectional causalities for African, ASEAN and South American countries. In Malaysia and Singapur, ELC contributes to GDP growth, whereas in Indonesia and Thailand it does not. The explanation of Yoo (2006) is that in the two latter countries electricity is used to satisfy basic human consumption needs, rather than for economic activities which could stimulate EG. Hence, the country-dependent composition of electricity consumers is also relevant in this context. This notion is supported by Wolde-Rufael (2006) who ascertains Granger Causality for 12 out of 17 African countries of which only three show unidirectional causality running from ELC to GDP. He remarks that the share of ELC in total EC amounts to even less than four percent in the examined countries. Whether this evidence also applies to EC, might be an interesting question. For the seven South American countries examined, Yoo and Kwak (2010) conclude that in all sample countries, except for Venezuela, an increase of ELC leads to EG. For Venezuela, this causality runs bidirectional, while for Peru no Granger Causality is detected. The authors advert to loans granted by the IADB to ARG, BRA, and COL aiming to promote electricity generation projects. National energy policies in the aforementioned countries enabled this arrangement with the IADB. Furthermore, the share of industry, in particular heavy industry, at GDP is considered as a relevant factor since such industrial economic activities consume much electricity. Hence, the authors note the relevance of investments into the infrastructure to ensure sufficient electricity supply.
Advanced models were put forward by Ciarreta and Zarraga (2010) and Narayan and Smyth (2009) who include a third variable enhancing the scope of analysis, whereas Sari and Soytas (2007) and Yoo and Lee (2010) replace GDP by income per capita in constant 2000 USD as a proxy for EG. The latter split their sample of countries into groups of OECD and non-OECD members, as well as in developed and developing countries, whereby OECD member states are mostly developed countries. Their main findings are that ELC per capita in OECD countries starts decreasing when a certain income per capita is reached, thus proving the existence of EKCs in those countries, whereas in developing countries, increasing ELC induces growth of income per capita. This evidence coincides with the findings from the EC-EG nexuses demonstrating that developing and emerging economies benefit in particular from EC, given that energy policies at least do not hinder its increase and that some industry does exist. Based on their results, Sari and Soytas (2007) argue that in some developing countries EC might be a factor that drives national income growth even more than do CF and/or LF. Ciarreta and Zarraga (2010) consider energy prices as a third variable for their analysis of twelve European countries in terms of the ELC-EG nexus. They argue that the weak relation between energy prices and ELC, as well as the negative unidirectional causality running from ELC to GDP, are implications from policies destined to creating a common and integrated market in Europe. Regional specifics also influence the results of Narayan and Smyth (2009) who additionally consider energy exports for their panel of countries from the Middle East. They detect bidirectional relationships between the three variables, whereby the strongest causality runs from GDP to ELC, but not vice versa. Thus, they conclude that the countries examined are predominantly important oil exporters, but not electricity exporters that would care about energy conservation since energy is cheap there. Narayan and Smyth (2009) urge those countries to a more sustainable way of thinking recommending them to invest more in their electricity infrastructure and to implement energy conservation policies to enhance technical efficiency. Therefore, EG and sustainability may be achieved in a combined manner.
Numerous studies deal with the nuclear EC-EG nexus using the same methodologies as applied to the EC/ELC-EG nexuses, mainly Granger Causalities and cointegration tests (a.o. Apergis and Payne, 2010b, Lee and Chiu (2011), Menyah and Wolde-Rufael (2010), Wolde-Rufael (2010), Wolde-Rufael and Menyah (2010), Yoo and Jung (2005), and Yoo and Ku (2009). They all come to the conclusion that there are causalities between nuclear EC and EG, either unidirectional or bidirectional, either positive or negative. Notwithstanding, this nexus will be disregarded given the overall purpose of this thesis which is to examine the potential impact of RE on EG.
Following up the aforementioned approaches of Apergis and Payne (2011b), and Chien and Hu (2007), their findings will be deepened by other studies in a first step. In a second step, alternative approaches will be taken into account. Apergis and Payne (2010a, 2010c, and 2011a) analyzed the causality between renewable EC and EG in the Eurasian region, in selected OECD countries, and for a couple of emerging economies. As in previous studies, GDP, CF, and LF are used as proxies for EG. The methodologies applied are the same as for the EC-EG nexus. All of them detect, at least in the long term, bidirectional causality running from REC to EG, and vice versa, thus supporting the feedback hypothesis. A conclusion drawn from countries and regions is that promoting RE through public subsidies and RE policies is a beneficial and sustainable investment leading to more energy autonomy of the countries. But those policies should be tailored to the available RE resources in a certain country. In other words, each country needs to examine for itself which RE technology is most promising, and allocate financial resources and enact specific policies accordingly. This notion is also argued by Chang et. al. (2009) who pursue a different approach to examine the REC-EG nexus. They analyzed the impact of energy price changes upon RE development while further differentiating in lower and higher EG. The Consumer Price Index (CPI) serves as proxy for changing energy prices. These authors’ main finding is that countries experiencing higher EG respond to increasing energy prices by the consumption of a higher amount of RE. By contrast, countries having lower EG do not show a strong attitude towards consuming more RE when energy prices are rising. The authors conclude that high EG countries are in an economic environment providing them with financial capacities to mitigate impacts of increasing energy prices upon EG through an extended use of RE sources. Those sources are only available after previous investments. That is why Chang et. al. (2009) recommend making corresponding investment decisions only after careful consideration, as delineated before by Apergis and Payne (2010a, 2010c, and 2011a). These findings support the notion that investments in RE as part of energy policies can be considered as a strategic step to induce EG, insofar country specifics are taken into account to avoid negative impacts upon the economy.
Following the approach of Sari and Soytas (2007) and Yoo and Lee (2010), Sadorsky (2009a) analyzed the relation between renewable EC, instead of general EC, and income for a sample of 18 emerging countries. Further, he extended the model to electricity prices examining how renewable EC reacts to changes in electricity prices, using a subsample of ten countries where sufficient data are available. Starting point for his empirical survey is the assumption that “…emerging economies are the ones that are going to experience the greatest increase in energy demand and carbon dioxide emissions” (Sadorsky, 2009a, p. 4021). His main findings are that income ultimately exerts influence upon renewable EC, and that renewable EC responds to changes of the electricity prices much more strongly than does the total electricity market. Moreover, he identifies significantly high long-term income elasticities leading him to the conclusion that renewable EC rises proportionally stronger to small increases of income. Sadorsky (2009a) deduces from these findings that the extension of renewable energy supply (RES) has to be proportionally faster than income growth in order to meet the demand for electricity and to avoid deceleration of EG.
Chien and Hu (2008) apply a very macroeconomic approach that deviates less from the previous studies in terms of parameters selected than it does in terms of methodology. Their purpose was to examine the impact of RE upon EG expressed by GDP for a sample of 116 countries. RE in this case means renewable EC measured in million Kt of oil equivalent. Using the so-called expenditure approach, they identify household consumption, government consumption, CF, and TR as relevant determinants for GDP in the basic model. According to them, RE influences that model in two ways: first, the extension of RE coming along with the emergence of corresponding industries requires investments which are reflected by rising CF. Second, they assume that locally produced RE makes countries more independent from EI leading to an import substitution effect which bears upon the TR. If CF and net exports increase, GDP will grow as well. Their empirical survey suggests that CF is significantly positively influenced by RE, whereas significant influence upon TR cannot be confirmed. They point out that RE influences GDP only indirectly through CF. From their results they conclude that CF-related policies are more effective than TR-related policies to increase GDP through RE.
A more integrated approach following Chien and Hu (2008) is applied by Sharma (2010) who analyzed the influence of different energy variables upon EG for a sample of 66 countries. According to her, the classical growth model using GDP as proxy for EG consists of four economic variables: CF, inflation, labor, and trade. She strengthens the validity of this model by using six different variables as proxies for energy: (1) EC measured in Kg of oil equivalent per capita, and (2) in Kt of oil equivalent; (3) electric power consumption in KWh; (4) electricity production in Kt of oil equivalent; (5) energy production in Kt of oil equivalent and (6) fossil fuel EC in % of total EC. She evaluates seven different models for each of the four defined world regions (a.o. Latin America and the Caribbean) applying a dynamic panel data methodology. The first model considers the classical growth model only. Models two to seven enhance the first model by one of the six proxies for energy. Therefore, Sharma (2010) aims to find out which energy variable exerts statistically significant influence upon EG in which region. For Latin America and the Caribbean, she identifies as critical factors only CF, inflation, and EC. All the other proxies for energy bear positively on EG, yet not significantly. She explains these results through the uneven distribution of energy resources in the countries of this region. Nonetheless, this region possesses ample energy resources in general. For example, some countries are very strong in fossil energy production and consumption, such as Ecuador and Venezuela in terms of oil, while others have core competences in the hydropower sector, such as BRA and Paraguay. While the two latter consume and produce large amounts of electricity, Ecuador and Venezuela have a proportionally higher share of fossil energy. For this reason, several proxies for energy are not significant for a panel including all Latin American and the Caribbean countries. She deduces from these results that countries in this region are well advised to enact electricity conservation policies since these will not have negative impacts upon EG while general energy conservation policies would have. For the total sample, she concludes that the six energy proxies bear positively on EG although they are not statistically significant.
In comparison to all other studies performed before, Awerbuch and Sauter (2006) do not attempt to examine the type of impact that renewable EC, or related variables, have upon EG, but rather they aim to make a qualitative statement as to which extent RE can contribute to EG. They ask how RE can help to avoid GDP losses with regard to the widely accepted nexus that an increasing oil price negatively impacts on GDP. The notion is that the oil price for fossil energy sources, such as coal and gas, is highly volatile causing a systematic risk and leading to uncertainty. Therefore, costs for the economy rise when firms face constraints in their long-term planning. Further, they state that an increasing oil price leads to decreasing output and salaries due to rising costs for input factors. Therefore, inflation might be induced with concomitant increase of interest rates. Hence, RE is supposed to diversify the energy demand of an economy and to mitigate impacts of a volatile oil price. The authors argue that investments in RE in terms of public subsidies and private investments will be outweighed by avoiding GDP losses, since for the private sector lower costs result from the uncertainty that is characterized by higher tax payments. For large net oil exporters such as Venezuela, this approach may not apply. To calculate the Avoided GDP Loss, the authors multiply the percent oil price change by the oil-GDP elasticity by GDP. Awerbuch and Sauter (2006) conclude that so-called avoided GDP losses induced through enhanced usage of RE will have an overall positive effect on the economy’s performance. Notwithstanding, they refrain from drawing from these results any conclusions on the making of energy policies, since EG and resulting welfare gains cannot be clearly attributed to the oil-GDP effect.
Numerous qualitative and quantitative studies deal with RE or energy in general and potential causalities associated with EG in different regions of the world. Factors and parameters are analyzed which are hard to quantify, such as energy policies and the legal and political environment in general. But, predominantly, quantitative methods are applied that give statistically proven results upon the existence and the type of causality. Hence, a quantitative analysis based on secondary data, which are obtained from institutions such as ECLAC, EIA, IEA, ONS, and World Bank, will form the core of my thesis work. For quantitative analyses, I will mostly use inductive, rather than descriptive, statistics. Therefore, approaches concentrating on the energy sector of one South American country, or analyzing one specific technology, as applied by Kissel and Krauter (2006), Peláez-Samaniego et. al. (2007), Recalde (2011) and Watts and Jara (2011), are excluded since they use descriptive statistics only. Universal evidences are derived from panels composed of a two-digit number of Latin American countries, as used by Chang and Soruco Carballo (2011), and by Zilio and Recalde (2011). This is problematic since country specifics, such as heterogeneous resource endowment, or energy policies, are not taken into account wherefore results can be distorted. Sharma (2010), Squalli (2007), Wolde-Rufael (2006), Yoo (2006), and Yoo and Kwak (2010) demonstrated, inter alia, that these problematic issues also exist for panels of neighboring countries, or for countries in a certain region. Thus, this thesis will focus on a single-digit number of countries, similar to the above-mentioned studies of Pao and Tsai (2010), Ruiz-Mendoza and Sheinbaum-Pardo (2010), Sadorsky (2012), Sheinbaum et. al. (2011), and Yoo and Kwak (2010).
Considering economic, legal, and political commonalities, as well as the corresponding RE potential, four countries are selected for in-depth analysis within the scope of this thesis: Argentina, Brazil, Chile, and Colombia. These four countries were pioneers in the Latin American and Caribbean region when they launched reforms of their electricity sector at the beginning of the 1990s, whereas this happened in CHI only ten years before (Guzowski and Recalde, 2010; Ruiz-Mendoza and Sheinbaum-Pardo, 2010). Moreover, these four countries share similarities in their cultural, linguistic, and economic development. They share commonalities in the relevance of hydropower as source for renewable electricity. BRA and COL already have deployed this technology to a certain extent, whereas ARG and CHI have significant potential. However, they applied diverging policy models to promote renewable EC (Arango et. al., 2006). It will be of interest to analyze along which roads they reached the same goal, namely an increase of sustainability and energetic self-sufficiency at affordable prices, after an experience of about 20 years in the transformation process towards more sustainable energy and ELC through extension of RE usage. From an ex-ante point of view, the available raw data for this period should produce a sufficient number of observables allowing to get meaningful results.
Zilio and Recalde (2011) point out that ARG, BRA, CHI, and COL are the most relevant countries in their sample of 21 Latin American and Caribbean states that have institutionalized the attempt to increase their energy demand from renewable sources. Currently, they gain RE primarily from biomass, biofuel, and hydropower. For BRA, solar energy is also relevant (Luecke, 2011). Other South American states, such as Paraguay and Uruguay, are relatively small countries, and in case of Paraguay, we could assume ex-ante some statistical bias since 100% of internal electricity demand is met by hydropower. A similar view applies to electricity exports which also stem from this energy source only (ECLAC and GTZ, 2004). In this context, Meisen and Krumpel (2009) remark that many Latin American countries heavily depend on hydropower for electricity generation. Hence, as in the case of BRA, hydropower may have a dominant position among RE sources, and therefore could potentially distort results. An examination that either includes or excludes hydropower might be reasonable when the impact of RE upon EG is analyzed. Economic as well as legal and political commonalities will be discussed at a later stage. Yoo and Kwak (2010) are closest to the intended scope of my work since they have selected only seven countries for analyzing the ELC-EG nexus by making use of different statistical tests for each country. Nevertheless, by the 5th of June 2012, no study was found that would focus on these four countries exclusively, with the aim of analyzing a potential impact of RE upon EG in emerging South American economies. This is the research gap that should be closed by this thesis.
Reflecting the different approaches which were regarded before, the expenditure approach used by Chien and Hu (2008) seems to be appropriate as starting point for my analysis, since these authors use a fundamental and established macroeconomic concept for the calculation of GDP as proxy for EG, and they reinforce this model by the inclusion of renewable EC. To put their model on a more meaningful basis, at least two different proxies for RE should be included according to the notion of Sharma (2010). However, this depends on data availability. LF, income, or inflation as used in other studies are conceivable parameters which also need to be examined. This has been endorsed by studies performed on Germany (Easterlin, 1968; Hillebrand, 2006; Lehr, 2008). The clear hypotheses formulated by Chien and Hu (2007) and by Sharma (2010) are good examples of how to structure the empirical survey. Annual data will be mainly gathered from the World Development Indicators (World Bank, 2011). With regard to methodologies, the Ordinary Least Squares (OLS) method will underlie the estimations of regression parameters. Several basic tests used to prove statistical significance and validity will follow up, based on pertinent literature (Albers and Skiera, 1998; Schild, 2012; Schira, 2006; Von Auer, 2007). Apergis and Payne (2010d) advert to potential distortion of results which can arise if data from one country dominate the sample. This could be an issue in the case of BRA, given its economic size, if it were part of a panel data regression. Further concepts that should be considered for this thesis are introduced in order to test upon structural breaks (Narayan and Smyth, 2008) and seasonalities (Coelho et. al., 2006; Lima and Lall, 2010a; Lima and Lall, 2010b; Ribeiro and da Silva, 2010). The overarching idea of this thesis is to integrate several concepts into the empirical analysis to approach the research question in an optimal way. Similar to Ruiz-Mendoza and Sheinbaum-Pardo (2010), the empirical part will be preceded by country-specific analyses which are supposed to unravel how the energy sectors are composed, what are the dominant energy sources, at which stage a particular country is in the RE evolution process, and what are possible obstacles to further extension of RE technologies. In this regard, descriptive statistics is used as a tool for visualization. Though I am aware of the need to narrow down the scope of my thesis work, I believe that one sub-chapter should be dedicated to some political aspects following Salazar (2008) and Linkohr (2006), as in this way energy policies in the four analyzed countries become more comprehensible. Hence, a broad quantitative survey, combined with some qualitative analysis where appropriate, is thought to provide the required outcome of the research topic.
2.1.2 Conclusion on the literature reviewed
Some previous studies focused on nuclear energy, some on RE, and some others on energy in general. Some of them examine electricity in particular, while many others take into account all types of energy. Whatever approach is followed, the uniting feature is the awareness of the finiteness of fossil energy sources, and the need to put energy on the scope of economic considerations so that it does not become a potential obstacle to economic development and growth. Such reflections have to be at the core of the debate on increasing energy demand, which is expected to continue rising along with world population growth, and on the need to reduce climate-harmful GHG emissions. In accordance with the literature reviewed by myself, and to a broader extent by Ozturk (2010), it becomes clear that the question of whether and how RE and ELC are interrelated with EG nowadays attracts attention gradually. Discussion on this subject has been revived during the last ten years, shown as a reaction to surveys that were raised in almost all regions of the world. This brings me to the conclusion that this topic is of utmost relevance to all countries. However, the degree of relevance depends individually on the resource endowment and on the level of development.
South American countries were part of large panel studies, were on the scope of political or geographical regional panels, or were analyzed individually. Previous studies either had a strong quantitative scope that embedded results and recommendations on energy policies and were insufficient in the political and legal context, or they were centered on qualitative analyses that yielded insufficient information on causalities. At this point, my thesis work is supposed to hook into the subject and to procure new evidences for South America by following a more integrated approach than in previous studies. I believe that a strong quantitative analysis, complemented by ample information on the underlying political and legal background, will approach the research question in an optimal way. This notion is underlined by the findings of Aravena (2008), Arnson and Varat (2008), Landerretche (2008), Mares (2008), O’Keefe (2008) and Vasquez (2010) who analyzed conflicts and potential for cooperation in that region. No other study before focused on ARG, BRA, CHI, and COL only. All four countries follow similar energy policies that they had launched in the 1980s and 1990s. Furthermore, they share a similar ideology in economic policy in comparison with some leftist regimes in that region such as Bolivia, Ecuador, and Venezuela (Linkohr, 2006; Vasquez, 2010). Due to a number of commonalities, these four countries have been selected for my analysis.
The validity of quantitative results heavily depends on the availability and quality of data, as well as on the appropriateness of the applied methodology. As was done in most previous studies, I will use freely available data which have been gathered from the World Bank’s database and from other international institutions. Data from the EIA about world net geothermal, solar, wind and wood and waste electricity installed capacity/electric power consumption/electric power production, ECLAC’s renewable share at total energy supply or World Bank’s combustible renewable and waste in % of total energy can serve as proxies for RE. Hence, there is no lack of data available for my work. Using in particular Granger Causalities, cointegration tests, unit root tests, and error correction models, all previous studies found some causality between EC or ELC and EG. Studies concentrating on renewable energy or electricity reached similar results, and from these results, energy policy recommendations were formulated. While the causalities can be the same in different countries, energy policies must be tailored individually in order to not constrain EG. It is obvious that energy- or electricity-exporting countries implement less restrictive energy policies than do countries which depend on corresponding imports.
I further extract from the literature that RE is seen as an attractive opportunity to meet rising energy demands especially in developing and emerging countries, since it enables them, particularly in rural areas, to supply themselves with energy or electricity through wind and solar power, for instance. Technical installations for such alternative energy-producing systems can be realized in a decentralized manner as they do not require high investments in pipelines to be connected with large power plants. To cope with the complexity of the subject and the scope of my work, multivariate statistics built on OLS regressions and supplemented by some statistical tests to examine the hypotheses will be used to meet the methodological claim of this thesis. However, to facilitate statistical evaluations, I will refrain from applying panel data regression.
Through my thesis work, I aim to gain a clear view of each country, in order to answer the question of whether RE can act as an acceleration factor for the countries’ economic development and growth. At the end, I expect to be able to give recommendations on the conditions and appropriateness of energy policies.
2.2.1 Hypothesis 1
Hypothesis deduction: Macroeconomic technical efficiency describes the economic efficiency of a country to produce an economic output depending on the relations between input factors used (Chien and Hu, 2007). In this context, the defined input factors are CF, EC, and LF determining the output measured as GDP. Chien and Hu argue that evolving RE entail the formation of a new industry which in turn creates new employment opportunities. RE poses a potential investment opportunity that attracts capital. LF was found to complement CF (Apergis and Payne, 2010a; Ibid.). In the past, there was a positive, statistically significant impact of renewable EC upon EG as observed both in the short and in the long term run (Apergis and Payne, 2009a; Apergis and Payne, 2010c; Apergis and Payne, 2011a; Apergis and Payne, 2011b). To address the aspect of technical efficiency, Chien and Hu (2007) take several proxies for renewable EC into account. Therefore, they cover the three categories of RE as defined by the IEA. Non-RE is also considered (Chien and Hu, 2007). Due to limited data availability, I will not consider the share of geothermal, solar, tide and wind fuel in RE. Instead of the share of hydro fuel as a fraction of RE, I will analyze the share of renewable fuel in total fuel consumption. From this deduction derives the following hypothesis:
Hypothesis: Renewable energy has a positive relation with macroeconomic technical efficiency.
Variables used: GDP in current USD, LF in absolute numbers, CF in current USD, traditional EC (total primary EC minus RE consumption) in thousand tons of oil equivalent, share of renewable fuel consumption (RFC).
2.2.2 Hypothesis 2
Hypothesis deduction: The expenditure approach encompasses GDP as dependent variable to be estimated, and household consumption, government consumption, CF and TR as independent variables, where household and government consumption are summed up to final consumption expenditure (FC) (Chien and Hu, 2008). An increase of domestically produced RE in total energy supply is deemed to lower EI and therefore to bear upon the TR. The latter influences economy’s FC since EI have an impact on the level of domestic energy and electricity prices, and, implicitly, on domestic EC as well (Ibid.). Influence of RE upon CF is assumed (Chien and Hu, 2007; Ibid.). Since an indirect influence of RE upon all components of the expenditure approach may play a role, the model has been expanded by an additional variable. From this deduction derives the following hypothesis:
Hypothesis: Renewable energy has a positive relation with economic growth subject to the expenditure approach.
Variables used: GDP, FC, CF, external balance on goods and services all given in current USD, renewable electricity generation (REG) in billion KWh.
2.2.3 Hypothesis 3
Hypothesis deduction: In theory, EC and the TR are interrelated in different manners. If exports increase, the demand for production factors such as capital, LF, and energy increases as well to meet the demand from abroad. On the other hand, energy is an important production factor particularly in industrial manufacturing which may impact on output if not supplied in sufficient amounts and/or at affordable prices (Chien and Hu, 2007; Narayan and Smyth, 2009; Sadorsky, 2012). This nexus is strengthened by Chien and Hu (2007) who further differentiate energy consumption/demand since they incorporate RE in their analysis. They argue that RE is predominantly domestically generated, and therefore the dependence from EI from abroad diminishes along with an improving TR. Hence, if EI have significant negative impact on the TR, RE could improve it insofar it is a significant factor in the enhanced expenditure approach. The original model of Chien and Hu (2007) has been adapted by expressing EI in % of total energy use (EC), and replacing renewable EC by EC in kilotons (Kt) of oil equivalent. Therefore, the approaches of Chien and Hu (2007) and of Sadorsky (2012) have been combined. From this deduction derives the following hypothesis:
Hypothesis: Energy imports have a negative relation with economic growth subject to the expenditure approach.
Variables used: GDP, FC, CF, external balance on goods and services all given in current USD, EC in Kt of oil equivalent, net EI in % of total EC.
2.2.4 Hypothesis 4
Hypothesis deduction: In all four countries, hydroelectricity production plays an important role. The share of electricity production from renewable sources including hydroelectricity ranged from 27.8% (ARG) to 83.8% (BRA) in 2009 (Fig. 1). Excluding hydroelectricity, the share declines to a range going from 1.16% (COL) to 7.17% (CHI) in the same year (Fig. 2). Hence, the idea underlying this hypothesis is to figure out if electricity production from renewable sources has significant impact on EG making a distinction of case: in the first case, hydroelectricity is included in the analysis, whereas in the second case it is excluded. The need for distinction is underlined by the awareness of Meisen and Krumpel (2009). They state that the strong dependence on hydro energy of some countries such as BRA makes them vulnerable in terms of energy security due to the inherently volatile energy supply from renewable sources. Hence, first the null hypothesis, saying that electricity production from renewable sources including hydroelectricity has no significant relation with EG, has to be examined. If this cannot be rejected further analysis is not meaningful, given the low share of non-hydro sources at REG. From this deduction derives the following hypothesis:
Hypothesis: Electricity production from renewable sources has a significant relation with economic growth if hydroelectricity is excluded.
Variables used: GDP, CF, external balance on goods and services all given in current USD, electricity production from renewable sources in KWh (including/excluding hydroelectricity), net EI in % of total EC.
2.2.5 Hypothesis 5
Hypothesis deduction: While twenty years ago a neutral relation was found between EC and income, recent studies have shown contradictory results in terms of renewable EC (Sadorsky, 2009a; Sadorsky, 2009b; Sari and Soytas, 2007; Yoo and Lee, 2010; Yu and Jin, 1992). Sadorsky (2009a, b) identified income and GHG emissions as significant determinants of renewable EC. They further remark that emerging countries are those which are experiencing the highest increase in EC and GHG emissions. This raises the question of how energy needs can be satisfied. To answer this question, Sadorsky (2009b) sets renewable EC per capita as the dependent variable, and analyzes how income measured in real GDP per capita, GHG emissions, and the real oil prices bear on that. The two latter independent variables reflect the external pressure to use cleaner and more sustainable energy, or to change energy demand habits, assuming that increasing oil prices will entail increasing RE demand. Yoo and Lee (2010) gather from their analysis that in developing and emerging countries ELC and, implicitly, energy demand grow in parallel with increasing income, challenging countries to ensure corresponding energy supply and security. Sari and Soytas (2007) provided evidence that in some countries energy is a more important determinant of EG than LF. For that reason, LF, as used in the models of Sari and Soytas (2007) and Yu and Jin (1992), will not be included in this hypothesis. From this deduction derives the following hypothesis:
Hypothesis: The level of income has a positive relation with renewable energy consumption.
Variables used: Renewable EC in MWh per capita (net geothermal, solar, wind and wood and waste electric power consumption), GDP in current USD per capita, CO2 in Kt per capita, real oil prices.
2.2.6 Hypothesis 6
Hypothesis deduction: Several studies that analyzed the impact of RE on GDP consider LF as independent variable of the EG model. They place less emphasis on the question of how RE bears upon employment using the number of LF as a proxy variable (Apergis and Payne, 2009a, b; Chien and Hu, 2008; Sharma, 2010). Since for the four countries of scope such an analysis was not found in the literature this hypothesis is supposed to contribute to the closure of a research gap following analyses undertaken on Germany’s RE market (Hillebrand et. al., 2006; Lehr, 2008). The authors identified investments in renewable energy (CF) as significant factor for employment increase. Therefore CF and RE generation will be considered as independent variables. According to the Phillips Curve, there is an inverse relation between inflation rate (using the CPI) and unemployment, while GDP as proxy for EG reflects the general market situation (Blanchard, 2006). Hence, these two variables are further treated as two independent variables in this model based on different approaches with reference to LF. From this deduction derives the following hypothesis:
Hypothesis: Renewable energy has a significant relation with employment increase.
Variables used: LF in absolute numbers of people, GDP in current USD, CF in current USD, inflation rate (CPI) in %, RE generation in billion KWh.
2.2.7 Hypothesis 7
Hypothesis deduction: Global warming caused by GHG emissions has not only ecologic implications but is also expected to entail obstacles to economic activities in different manners. EC is considered as one of the key factors of those emissions (Chang and Soruco Carballo, 2011; Pao and Tsai, 2010). Since those emissions primarily result from the use of fossil energy, and given the fact that the Latin American and Caribbean regions show the highest average growth rate of fossil EC among the emerging regions in the world, energy conservation and GHG reduction policies are of utmost relevance to South America. Promoting the use of RE is part of these policies (Sadorsky, 2012; Sharma, 2010). Previous studies used in particular the Granger cointegration method to analyze the causal relation between EG and EC, proposing four hypotheses. The resulting null hypothesis of Hypothesis 7 corresponds to the neutrality hypothesis according to the Granger Causality models. If H0 is rejected, the conservation and the growth hypotheses (uni-directional causality) can be considered for the interpretation of results, though this does not mean that H1 would in turn be accepted (Chang and Soruco, Carballo, 2011; Pao and Tsai, 2010; Sadorsky, 2012; Yoo and Kwak, 2010). From this deduction derives the following hypothesis:
Hypothesis: Energy conservation and GHG reduction policies have a positive relation with economic growth.
Variables used: GDP in millions of constant year 2000 USD, EC in Kt of oil equivalent, CO2 emissions in thousand metric tons.
2.2.8 Hypothesis 8
Hypothesis deduction: In countries such as BRA where primary EC comprises 29% hydroelectricity (2010), which accounts for 83.8% of total electricity production, the supply of electricity gained from renewable sources probably influences the market price of electricity in general (EIA, 2011c; The World Bank, 2011b). The higher the domestic electricity supply, the lower are corresponding imports, and the lower is vulnerability through volatile foreign electricity prices. These factors bear in turn upon the TR. Changing electricity imports also imply changing domestic electricity prices (Chien and Hu, 2007). Several authors have dealt with the seasonality of hydropower generation in South America, mainly in BRA. Hydroenergy is generated both in the North and in the South of the country, but the hydropower potential is not evenly distributed across the country. Hydroelectricity plants are predominantly concentrated in the South and South-East (Lima and Lall, 2010a; Fig. 3). Precipitation differs considerably among the regions over the year, resulting in changing water circuits, and this bears in turn upon the electricity generation (Coelho et. al., 2006; Garreaud et. al., 2009; Ribeiro and da Silva, 2010). Cost implications and economic consequences in terms of GDP changes have been argued about for a number of years (Guido Tapia Carpio and Olimpio Pereira Jr, 2006; Lima and Lall, 2010b). From this deduction derives the following hypothesis:
Hypothesis: The price of electricity gained from renewable sources is subject to seasonal influences.
Variables used: electricity price in total BRA, hydroelectricity generation in South East/Center and West BRA in GWh.
2.3 The sample of countries – a brief overview of their particular energy markets
ARG had in 2010 a total population of about 40.4 million generating a GDP of 368.74 billion current USD (The World Bank, 2011b). Currently, the country’s energy supply is mainly composed of natural gas and oil accounting for 46% and 40%, respectively, of total energy supply which is mainly generated domestically (Sheinbaum et. al., 2011). Among the South American countries, ARG is the largest producer of natural gas (EIA, 2011b). In 2009 RE had a minor share of 9.4% at total energy supply (ECLAC, 2011). As by the end of the 1980s ARG turned into a net energy exporter, but strong EG in the aftermath of the national bankruptcy in 2001/02 were coming along with rising energy demand, the country became increasingly dependent on EI (EIA, 2011; Fig. 4). In the following years, a mismatch between energy supply and demand, volatile energy prices and insufficient investment in the electricity infrastructure repeatedly caused supply interruptions bearing negatively upon the economy (Guzowski and Recalde, 2010; Larsen et al., 2004; Recalde, 2011). In 2006, the largest energy consumer was the manufacturing industry accounting for 35% of total EC (Sheinbaum et. al., 2011). In terms of RE sources, Meisen and Krumpel (2009) certify ARG’s high potential for hydropower and geothermal energy in the Andean region as well as for wind power in the country’s south. The potential for biomass-based energy exists mainly in the North and in the Center. Despite manifold potentials, only hydropower has really evolved, and accounted for 27.8% of total electricity production in 2009 (Fig. 1). Renewable sources in general, excluding hydropower, had a share of 1.4% at total electricity production in the same year, gained mainly from biomass (Fig. 2; Ruiz-Mendoza and Sheinbaum-Pardo, 2010).
At the beginning of the 1990s, ARG carried out far-reaching liberalization reforms of the economy coming along with the total privatization of the electricity sector. In the aftermath, fossil resources rebounded while hydroelectricity heavily lost share at total electricity production (Recalde, 2011; Ruiz-Mendoza and Sheinbaum-Pardo, 2010; Fig. 1 and 5). In 1998 (Law 25.019) and 2007 (Law 26.190), two laws were enacted which particularly promote the use of RE. The first law fosters the generation of wind and solar energy through so-called feed-in tariffs. These tariffs are charged from non-renewable electricity producers rewarding the feed-in of solar and wind energy. The second law stipulates to increase the share of RE sources at the total ELC up to 8% within ten years, coming along with incentive programs (Guzowski and Recalde, 2010; Ruiz-Mendoza and Sheinbaum-Pardo, 2010). Moreover, ARG has promoted the production of soybean-based biodiesel production, being globally among the five largest producers (Sheinbaum et. al., 2011). In the course of the 2001/02 crisis, the government started intervening the market through the implementation of final tariff regulations, price caps and subsidies for certain energy suppliers. That raised uncertainty among market participants regarding the economic, legal, and political framework, and, as a consequence, lowered investments in RE technologies (Guzowski and Recalde, 2010; Haselip and Potter, 2010; Recalde, 2011). Hence, after a decade of a totally liberalized energy market, ARG’s energy policy now offers energy security at affordable prices.
BRA had in 2010 a total population of about 194.4 million generating a GDP of 2.09 trillion current USD (The World Bank, 2011b). In 2010, the country’s primary EC mainly consisted of oil and other liquids including biofuel (39%), hydroelectricity (29%), and other renewables (21%). Coal, natural gas and nuclear energy nowadays play a minor role (EIA, 2011c). Pending large-scale oil exploitation along the Brazilian coast carried out by the state-controlled oil company Petrobrás, together with the country’s efforts made in the RE sector (in particular biofuels and hydropower), endorse the government’s attempt to make the country self-sustaining in terms of energy supply (EIA, 2011c; Geller et. al., 2004; Sheinbaum et. al., 2011). The FFV technology  allows cars to be refueled with an arbitrary composition of bioethanol or gasoline and enables drivers to use the cheaper fuel giving them a kind of hedging instrument against price volatilities of one or the other fuel type on the world market (De Freitas and Kaneko, 2011). BRA has become the world’s largest bioethanol exporter, its energy import balance in % of total EC is tending to be balanced in the near future (Sheinbaum et. al., 2011; Fig. 4). In spite of duplication of its share in the last 20 years, reaching 5.22% in 2009, RE excluding hydropower could not seriously jeopardize the dominant position of the latter in terms of electricity generation. In 2009, hydropower accounted for 83.8% of total electricity production (The World Bank, 2011b; Fig. 1-2). Notwithstanding, other renewables such as biomass, solar and wind power are promising alternatives. These can help to further diversify Brazil’s energy supply, and, in particular, the electricity market, thereby increasing energy security (Geller et. al., 2004; Martins and Pereira, 2011). In terms of seasonal availability, hydropower and wind power complement one another (Kissel et. al., 2009). Biomass in terms of bioethanol and biodiesel already contributes significantly to energy supply (Martins and Pereira, 2011).
BRA has set up numerous programs to save energy in general, and to promote energy-saving technologies. The electricity sector reform of 1993 opened the market for private investors, but semi-public ownership constellations were more common, and regional needs were taken into account so that privatization underwent a more deliberate and gradual process as compared to ARG and CHI (Ruiz-Mendoza and Sheinbaum-Pardo, 2010). This reform was preceded by the National Alcohol Program (Proalcool) of 1977 and the National Electricity Conservation program (PROCEL) of 1985, marking the beginning of a more sustainable energy policy in BRA (Sheinbaum et. al., 2011). The PROEOLICA of 2001 promoted wind energy installations in rural regions (Ibid.). A major national program to promote the use of biomass, small-sized hydropower and wind power (PROINFA) was launched in 2002, aiming to increase their share at total Brazilian electricity supply up to 10% by 2022 (Dutra and Szklo, 2008). In a first phase of the PROINFA program, inflation-adjusted feed-in tariffs were granted to producers of the aforementioned technologies. In the second phase, a bidding system was implemented where construction permits for biomass, solar or wind power plants were primarily granted to the most cost-effective suppliers (Dutra and Szklo, 2008; Kissel and Krauter, 2006). PROINFA was preceded by PRODEEM (1994-2001) which fostered the same renewable technologies, including solar energy, on the regional and municipal level (Ruiz et. al., 2007). Hence, BRA pursues an integrated energy policy approach where superior national interests are merged with a liberalized energy market, and actors are driven through incentive-setting by public promotion programs.
CHI had in 2010 a total population of about 17.1 million generating a GDP of 212.74 billion current USD (The World Bank, 2011b). In 2009, the country’s primary EC was mainly composed of crude oil (42.7%), wood and others (20.5%), carbon (16%) and natural gas (12%) (MdEC, 2010). Crude oil and natural gas were imported at 98% and 90%, respectively, in 2006 (García et. al., 2011). The high dependency on foreign supply is further underlined by the fact that EI accounted for 67.7% of total EC in 2009 making the country vulnerable to price volatility of such commodities (Ibid.; The World Bank, 2011a). Roughly 90% of subsequently generated energy is consumed by the industrial sector (Gebremedhin et. al., 2009). A particularity of the Chilean power infrastructure is that it its divided in four systems which are completely separate from each other. The two major ones, SIC and SING, cover the demand of more than 90% of the total population (Ibid.). Among the renewables, only hydropower contributes significantly to total energy production, with 8.7% in 2009 (MdEC, 2010). Apart from (mini-) hydroelectric power stations, solar and wind power plants are deemed to become particularly promising to make CHI´s EC more sustainable and independent of imports from abroad. The country’s North, including the Atacama Desert, suits very well the installation of large-scale solar panels since CHI’s prime economic activity, the mining, is also located there having an abundant need for electricity. The coastline stretches across 6400 km offering considerable opportunities for cheap onshore wind power plants, since for these the lowest investment costs in USD/kWh are required among the RE sources in CHI (Gebremedhin et. al., 2009; Ortega et. al., 2010; Watts and Jara, 2011). An insufficient amount of cultivable land makes the production of corn- or soybean-based biodiesel less attractive than in ARG for instance (García et. al., 2011)
In Latin America, CHI was the cutting-edge country in terms of electricity market (deregulation) reforms at the beginning of the 1980s, coming along with privatizations. Heavily affected through the lack of hydro-based electricity supply in 1998/99, and through natural gas export interruptions from ARG in 2004, the Chilean government has pushed the extension of RE in recent years (Haselip and Potter, 2010; Murillo and Le Foulon, 2006; Ortega et. al., 2010; Recalde, 2011). By law, 5% of total electricity generation must come from non-conventional renewable energy sources  by 2010, gradually rising up to 10% by 2024 (Ortega et. al., 2010). The first goal was met in 2009 (Fig. 2). By law No. 20,257, energy suppliers are forced to feed a certain quote of renewable electricity into the network, thereby receiving further incentives to increase its share through feed-in tariffs (Guzowski and Recalde, 2010; Watts and Jara, 2011). Hence, CHI promotes the extension of RE to become more independent from EI, and to enhance energy security which proved to be the country’s neuralgic point in the past and could otherwise become a potential obstacle for EG.
COL had in 2010 a total population of about 46.3 million generating a GDP of 288.19 billion current USD (The World Bank, 2011b). In 2008, the country’s EC was composed of crude oil (42%), hydroelectric power (31%), natural gas (18%) and coal (9%) (EIA, 2011d). Other renewables accounted only for about 1.15% in the same year, while hydropower had a share of 82.3% at total electricity production, followed up by thermal electricity (Ruiz Mendoza and Sheinbaum-Pardo, 2010; The World Bank, 2011b). In comparison to the three other countries on scope, COL has been a net energy exporter since more than 30 years, except for a period of two years at the beginning of the 1980s. In 2009, COL’s energy exports were more than twice as much as its imports (Fig. 4). The main drivers of this very positive energy export balance were coal and oil which are abundantly available in COL. The country possesses South America’s largest proven coal reserves (EIA, 2011d). To gauge potentials of the different RE in COL, Caspary (2009) takes into account the geographic conditions, investment costs, and potential cost improvements and undertakes projections into the future. Large-scale hydropower plants, which already drive COL’s electricity generation, are predicted to keep their dominant position in the medium run, given the low investment and electricity production costs. Small-scale hydropower is also deemed competitive. COL’s windy North is thought to suit for large-size wind power installations that are expected to become the country’s most competitive RE sources by 2030. Biomass, geothermal, and solar-based energy are not considered as competitive (Caspary, 2009).
 This Flexible-Fuel Vehicle (FFV) engine technology was plugged in 95% of all passenger vehicles sold in BRA by July 2010 (De Freitas and Kaneko, 2011).
 Non-conventional RE sources comprise the classical RE such as biomass, hydro, solar and wind but small-scale.
- Quote paper
- Thomas S. Konrad (Author), 2012, Renewable Energies as a Growth Factor in Argentina, Brazil, Chile and Colombia, Munich, GRIN Verlag, https://www.grin.com/document/369520