2.1. Contrasting results on the impacts of support policy2.2.3. The energy dependence rate2.2. Influences of the structure and dynamics of energy markets2.3. Macroeconomic determinants2.3.1. Income per capita2.2.2. Electricity prices and the relative costs of RE and conventional energy sources2.3.2. Access to funding i2.3.3. The openness to international trade l2.4. Political and institutional determinants3.2. Support policies in Europe and Latin America3. Data and method3.1. The dependent variabletable_1value 1 for the variable Ex_pol (for a year t) this means it has applied one of the following instruments: FIT, RPS, AUC or FIS. Similar measures for RE policy have been used in previous studies, including [52], and [25].3.3. Other explanatory variables3.4. The econometric model3.4. The econometric model3.4. The econometric model4. Results and discussion4.1. General results4.1. General results4.1. General results4.1. General resultsTable 2 Summary of results - global model.Table 2 SSummary of results - global model.4.2. About RE policy effectiveness in Latin America and Europe4.2. About RE policy effectiveness in Latin America and Europe4.2. About RE policy effectiveness in Latin America and Europe4.2. About RE policy effectiveness in Latin America and EuropeTable 35. Conclusion and policy implicationsTable 4 Summary of results - Europe.Table 4Notes: Significance levels: *** (1%), ** (5%), * (10%); (S): Statistically significant, (NS): Non-significant.Table 5Table 5Table 5CREdiT author statementDeclaration of competing interestAcknowledgements & FundingAppendix A. Supplementary dataReferencesReferencesReferencesReferences9Z'Am2Ley J 0U\1OYcmwOther indirect instruments, including tradable CO2 emissions permit systems, seem to have a positive impact, but only for the most mature technologies.2.2.3. The energy dependence rate Countries dependent on external energy supply need further to develop local sources of production, including RE. Reducing energy dependence is an energy policy objective and one of the main arguments for increasing RE in both developed and developing economies [44]. As Marques and Fuinhas [23] pointed out, the expected theoretical rela- tionship between these two variables is positive: the higher the country’s dependence on external supply, the greater will be the incentive to deploy RE. The Kilinc-Ata [21] study used data from 1990 to 2008 from Europe and the US. They showed that FITs, tenders and tax incentives are effective for boosting the deployment of RE, while quotas are not. Finally, Kim and Park [33] included a single policy variable: FIT. The results of the various models they tested showed a positive and signifi- cant influence of FIT. 2.2.4. Power sector dynamics: the annual rate of consumption growth y f g The more electricity demand increases or, the higher the prospects for its growth, the more a country will need to invest in new power capacity, including RE. Besides, the need for new investments in the electricity sector is linked to the average age of existing plants. When fossil fuel plants reach the end of their useful lives, they must be replaced by equivalent new capacity (except in the case of an equal drop in de- mand). Although a dynamic power sector should positively influence the spread of RE, the empirical evidence concerning that link is not conclusive. l In summary, econometric studies tend to agree on the effectiveness of policies in general but are less conclusive on the differences between specific instruments. More empirical evidence is required, considering a more recent sample that includes developing countries.2.2. Influences of the structure and dynamics of energy markets Beyond the incentive policies, context-specific variables can have a significant effect on RE diffusion. However, the empirical evidence is inconclusive on several points. In this section, we focus on variables related to the structure, functioning, and evolution of the energy sector (and particularly the electricity subsector). 2.2.5. The level of CO2 emissions from fuel combustion Fighting climate change by reducing energy-related GHG emissions is one of the main reasons for developing RE. Hence, the need for a control variable that captures the emission levels in each country. Most studies include a variable for CO2 emissions (from the energy sector) per capita. Aguirre et al. [44] recall the well-known fact that environmental concerns should stimulate investment in RE. 2.2.1. The energy mix: competition or complementarity between technologies The existing electricity mix may condition the dissemination of RE technologies in several ways. On the one hand, countries with high proportions of hydropower and nuclear-based electricity in their energy mix may be less concerned with developing new low-carbon technolo- gies [43]. Similarly, power generation that is concentrated on a partic- ular source (e.g., nuclear) may reflect a situation of technological lock-in4 and the possible influence of lobbying to maintain their market share [31]. On the other hand, countries with high percentages of hy- droelectricity have the necessary storage capacity to help to balance the intermittency of solar and wind power. From this perspective, new RE and hydroelectric plants are complementary rather than competing technologies. Therefore, the link between these sources of electricity generation is difficult to predict.2.3. Macroeconomic determinants The better the country’s macroeconomic situation, the higher the investment level in RE, ceteris paribus. Economies have specific charac- teristics that make them more or less attractive for investment generally, and in new technologies in particular. These features include, among others, the existence of highly skilled workers, good quality infrastruc- ture, and legal and financial security [24]. Access to project financing sources at reasonable interest rates is also fundamental and, especially, concerning high initial capital-intensive investments.2.3.1. Income per capita2.2.2. Electricity prices and the relative costs of RE and conventional energy sources Several studies [23,44] included GDP per capita in their models. The underlying assumption is that high-income countries are more likely to deploy RE because they can more easily bear the costs of developing these technologies and encourage them through economic incentives. The heterogeneity in our sample of countries is essential for this vari- able, given the large gap between low-income nations, such as Bolivia or Honduras, and countries with significantly higher incomes such as Norway and Luxembourg. RE compete with other sources of electricity generation. Thus, the availability of cheap domestic fossil fuel resources for electricity pro- duction, particularly gas and coal, can affect the attractiveness of RE. The higher the price of any competing sources, ceteris paribus, the more attractive will investments in RE be [43]. Simultaneously, several au- thors [24] have discussed the importance of electricity prices: the higher the price of electricity, the more investors will be encouraged to invest in RE. Chang et al. [45] found a positive and significant relationship2.3.2. Access to funding i One of the most significant barriers to RE deployment is the high initial investment cost. The RE sector’s capital intensity is higher than for fossil fuel energy and requires a proportionately higher initial in- vestment before production can begin. Using panel data for 30 countries 3 According to Directive 2009/28/EC. 4 Learning effects and economies of scale can lead to reductions in the cost of a specific technology in some countries (e.g., nuclear power in France). Renewable and Sustainable Energy Reviews 133 (2020) 110351 G. Bersalli et al. RE capacity (for country x for year t); b) RE generation in kWh per capita [31]; c) share of RE in the electricity mix, as a percentage [44,52]; d) ratio of total RE capacity to total electricity generation [53]; e) newly installed RE capacity per inhabitant (MW/1 million inhabitants) [22, 43]. In our case, investments in new RE are measured not in monetary terms, but as physical quantities (MW). We use indicators of capacity since they are the most accurate proxy for technology deployment. We define net investment per capita in technology j, of country i, in year t as: for the period 2000–2013, Kim & Park [33] examined the relationship between the development of financial markets and RE deployment on a global scale. Their results suggest that RE technologies are diffusing rapidly in countries with well-developed (both equity and credit) financial markets.2.3.3. The openness to international trade l Pfeiffer & Mulder [31] tested the influence of foreign direct invest- ment and the level of openness to international trade. Contrary to their initial hypotheses, they found a negative relationship between both variables and the level of RE generation. From a theoretical point of view, the influence of economic openness on new energy technologies can follow contradictory rules. In principle, trade openness should foster the diffusion of new technologies through lower costs (due to increased competition), removal of barriers, international agreements, etc. [46]. However, greater openness to international competition also requires additional efforts to lower domestic product costs, which, in turn, might lead countries to develop cheaper sources of energy. Therefore, the role of international trade in RE depends on several economic and political factors specific to each country, which are difficult to represent using a single variable. CAP is the total installed capacity at the end of each year; POP is the population. The RE technologies included in the dependent variable are wind, solar, geothermal, and biomass. Large hydropower plants are excluded because hydropower is a more mature technology that is not promoted by the support policies we are evaluating. Small hydropower plants could be included because they tend to be targeted by promotion pol- icies. However, most countries’ data do not distinguish reliably between small and large hydropower plants. Thus, we exclude all of them from our analysis. The data also do not include small generation plants that are not connected to the grid since they are affected by other policies, beyond the scope of this study.2.4. Political and institutional determinants In recent years, some theoretical and empirical studies have focused on the political factors influencing the design and effectiveness of energy and environmental policy. They have highlighted two sets of political determinants: governance quality, including the institutional frame- work in which these policies are implemented, and the ideology (po- litical orientation) of the government in power. Several articles [47–51] suggested that political factors can influence both energy and environ- mental policies. Nevertheless, none of these studies refers specifically to RE diffusion. i Finally, this study includes thirty European and twenty Latin- American countries, two regions using similar policy instruments to promote RE growth. The list of countries is available in Appendix 1. Some countries of these regions were excluded because of the lack of reliable data.3.2. Support policies in Europe and Latin America Cadoret and Padovano [20] is one of the first papers investigating the influence of purely political factors on RE diffusion. The authors studied the determinants of RE’s share in final energy consumption based on panel data for 26 European countries over the period 2004–2011. Their results showed that industry lobbies have a negative influence on the deployment of RE and that left-wing governments tend to favour the deployment of RE compared to their right-wing counterparts. They also showed a positive effect of governance quality on the deployment of RE. The main explanatory variables included in our model are Ex_pol, which denotes the existence or not of support policy, and four variables for the type of instrument. We consider the main support instruments: Feed-in Tariff and Feed-in Premium (FIT), the auction system (AUC), the quota obligations with negotiable green certificates or Renewable Portfolio Standards (RPS), and the fiscal -tax- incentives (FIS). We do not include instruments implemented for short periods, such as voluntary agreements or green pricing. The selected variables reflect the intensity and diversity of the policies implemented in the two regions studied. Other political or geographical factors that can influence RE diffu- sion include population density (related to the space available to install solar and wind parks), wind strength, solar radiation, etc. These vari- ables might have a strong effect on the diffusion of a particular tech- nology but are less relevant when we study general RE policy, independently of technology. Therefore, we exclude them from our model. We exploited several data sources: IRENA/IEA (Joint Policies and Measures database), European RE-Shaping projects [54], and data from Ref. [55,56]Since we are interested in investments in new technologies for electricity generation, we focus on policies specific to the electricity sector. Our database contains 1000 observations indicating the existence or not of a policy in the country i in the year t, and the type of instru- ment. These observations were transformed into binary variables.3. Data and method Fig. 1 depicts the percentage of countries that implemented at least one support policy in the electricity sector. In 1995, 40% of European countries had a policy in place. The number of countries increased significantly after the signing of the Kyoto Protocol (December 1997) and especially after the European Commission published Directive 2001/77/EC. This increase illustrates two central features of RE support policies in the Old Continent. First, they emerged in the context of climate policies and, especially, the need to reduce greenhouse gas emissions from the energy sector. Second, European institutions have played and continue to play a central role in this area through specific regulatory frameworks and common long-term objectives. Thus, the RE development objectives of European Union member states are set at the European level, although the choice of policy instruments to achieve them are chosen nationally. Our dataset covers 50 countries (i = 1, …,i = 50) and 20 years (t = 1995, …,t = 2014) and is balanced.5 In this section, we explain our choices concerning the explained variable (3.1), the explanatory vari- ables, including a brief description of RE policy in Europe and Latin America (3.2 and 3.3), and the econometric model (3.4).3.1. The dependent variable This study seeks to explain the RE diffusion determinants and, accurately, the level of investment in RE technologies over the 20 years to 2014. The literature review showed that several different definitions of the dependent variable had been proposed: a) amount invested in new In Latin America, promotion policies have followed a different logic. These countries were much less concerned about climate policies in the 1990s and 2000s: per capita emissions were significantly lower due to 5 There are no missed observations for any country and any year. Renewable and Sustainable Energy Reviews 133 (2020) 110351 G. Bersalli et al.table_1value 1 for the variable Ex_pol (for a year t) this means it has applied one of the following instruments: FIT, RPS, AUC or FIS. Similar measures for RE policy have been used in previous studies, including [52], and [25]. lower energy consumption and the existence of high levels of hydro- electric production in most countries. Fig. 1 shows that the first policies were put in place only in the late 1990s and early 2000s. They seek to diversify electricity production by attracting private investment needed to meet growing demand. There is also no coordinated planning of the energy sector in this region. The role played by regional institutions, such as Mercosur or Comunidad Andina, is somewhat limited in all but the Central American countries where energy integration is stronger [56].3.3. Other explanatory variables According to our theoretical framework, we included several control variables in our model, accounting for the influence of the structure and dynamics of the energy system and the main macroeconomic de- terminants of RE diffusion. Table 1 presents the name, the definition, the expected influence of each variable to the dependent variable, and also the data sources. Fig. 2 depicts the diversity of policy instruments in Latin America. A specific characteristic of the region is the dominance of quantity in- struments based on auctions, used by five states in 2014: Argentina, Brazil, Uruguay, Peru, and Nicaragua. Four smaller economies - Bolivia, Ecuador, Dominican Republic, and Honduras - chose FIT. A quota sys- tem has been in place in Chile since 2008. Six countries - Colombia, Venezuela, Paraguay, Costa Rica, Jamaica, and Belize - have no eco- nomic or regulatory support policies while Mexico,6 Guatemala, El Salvador, and Panama offer only tax incentives.3.4. The econometric model To assess the effects of RE policies on RE generation, we specified our model as follows: Fig. 3 shows the evolution of support policies in Europe. The price- instruments have tended to dominate, although, in recent years, several countries have reformed their support systems and introduced Feed-in Premiums to replace Feed-in Tariffs (Denmark, Estonia, Ger- many). The quota system is in place in several countries, although Italy and the UK recently decided to withdraw it. It should be noted that auctions, which had almost disappeared, are being used again in Italy, France, the Netherlands or Russia. Also, most countries use the same instrument to promote all types of RE technologies, although with differentiated levels of support. Only a few EU member states apply instruments specific to a technology. For example, in 2014, Italy had auctions for wind and biomass and FIT for solar PV, and Denmark had auctions for offshore wind and FIP for other technologies. l where Yit is the dependent variable (annual new RE capacity) observed for country i at year t, Xit is the time-variant 1 × k regressor matrix, β is the vector of coefficients αi is the unobserved time-invariant country effect and εit is the error term. i Our analysis encompasses several steps. We first analysed the char- acteristic of our data, and we especially confirmed the absence of mul- ticollinearity through the variance inflation factor (VIF) indicator. The next steps consisted of choosing between a fixed or a random-effects model. Using a fixed-effects (FE) model, we assume that specific unob- served characteristics of individuals (in our case countries) can have an impact or a bias on the explanatory variables of our model, and we, therefore, seek to control this. The random-effects assumption is that the individual-specific effects are uncorrelated with the independent vari- ables. The opposite is valid for a FE model, which removes the effect of these time-invariant characteristics, allowing us to assess the explana- tory variables’ net effect on the dependent variable.Given the charac- teristics of the data and the literature review, our panel should correspond to a FE model. We tested this hypothesis through the Hausman test [57]. The null hypothesis implies that the preferred model is the random-effects model, and the alternative hypothesis leads to a FE model. The test rejected the null hypothesis, and in the following, we consider the FE model. In our model, the influence of support policies is estimated in two stages. First, we included the variable Ex_pol, which indicates the exis- tence or not of a support policy (regardless of the type of instrument) in each country at the end of each year. Only direct promotion policies are considered. Thus, regulatory frameworks in the electricity sector, stra- tegic plans and other similar policies that do not include direct diffusion instruments are excluded. Ex_pol is a binary variable, which takes the value 0 if none of these policies are in effect and 1 otherwise. Second, we included four variables for the type of policy tool. If a country i has the 6 Mexico implemented a RPS in 2015. G. Bersalli et al. Renewable and Sustainable Energy Reviews 133 (2020) 110351 nan Then, we conducted several tests to analyses heteroscedasticity and correlation. Firstly, we tested for cross-sectional dependence/contem- poraneous correlation, through the Pesaran cross-sectional test of in- dependence (see Ref. [58]) where the null hypothesis is that residuals across entities are not correlated. The test did not reject the null hy- pothesis, and we conclude that the data have not contemporaneous correlation. Secondly, we tested for serial correlation, applying a Breusch-Godfrey test (see Ref. [59,60]) where the null hypothesis is that there is no serial correlation. The test rejected the null hypothesis, and we conclude that the data have a possible problem of serial correlation. Thirdly, we tested for heteroscedasticity through the Breusch-Pagan test (see Ref. [61]), where the null hypothesis is that there is homoscedas- ticity. The test rejects the null hypothesis, and we conclude that there is heteroscedasticity in the data.Finally, to take into account both heter- oscedasticity and serial correlation for a fixed-effects model, we applied the Arellano’s correction [62]. In summary, after applying several tests, we have chosen a fixed- effects model with Arellano correction. Then we compare the results with those obtained using the random-effect and the PCSE models. We conducted the different steps described here first for the entire sample (50 countries) and then separately for Europe and Latin America.4. Results and discussion This section presents the results of the regressions, first on the whole sample: 1000 observations corresponding to 50 countries over 20 years; and, second, for the European and Latin American countries separately. Also, we discuss the motivation and effects of RE policy in both regions.4.1. General results Table 2 summarises the results of our primary model (fixed-effects with Arellano correction), comparing them with two alternative models (PCSE and random-effects models). The three models are significant, according to the Wald statistic. Several variables are statistically sig- nificant, given the probability attributed to them. Three of our public policy variables appear to be statistically significant: feed-in tariffs, renewable portfolio standards, and auctions. In contrast, the variable Some previous analyses [22,23] applied the Panel Corrected Stan- dards Error (PCSE) method. This technique is well suited to estimating model parameters in the presence of heteroscedasticity and correlation at panel level. However, the estimation of the variance-covariance ma- trix in PCSE depends on large T, which is not the case for our sample (T = 20). Thus, we did not privilege this method. Renewable and Sustainable Energy Reviews 133 (2020) 110351 G. Bersalli et al. Table 1 Explanatory variables. NAME/CODE DEFINITION/INDICATOR EXPECTED INFLUENCE Electricity demand growth (ele_agr) * Growth rate of electricity consumption over the last 5 years. (+) countries facing rapid growth of electricity demand need to invest in new generation capacity, including RE. Share (%) of nuclear in the electricity mix (nuk_shr)* Share of nuclear energy in the total gross electricity production. (−) a high share of nuclear, i.e., low-carbon electricity, provides less incentive to invest in RE. Share of hydropower in the electricity mix (hyd_shr)* Share of hydropower in the total gross electricity production. (−) a high share of hydro, i. e., low-carbon electricity, provides less incentive to invest in RE. Energy dependence rate (ind_shr)* Share of primary energy consumption covered by domestic primary energy production. (−) a low level of energy independence can encourage countries to further invest in RE. CO2 emissions (CO2_pcp)* CO2 emission from fuel combustion per capita (tCO2/capita). (+) high CO2 emission levels can encourage countries to invest in low- carbon energy sources. Coal production per capita (coa_pcp)* Gross annual coal production per capita (Mt/ capita). (−) higher coal production can discourage RE investment. Gas production per capita (gas_pcp)* Gross annual gas production per capita (Mm3/capita).(−) higher gas production can discourage RE investment. Income per capita (gdp_pcp)* GDP US$ at constant price and exchange rate (2005) per capita. (+) it is easier for high- income countries to invest in RE. Access to domestic credit (cdt_shr)** Share of financial resources provided to the private sector by the banking sector and other financial corporations (% of GDP) (+) well developed local financial markets can facilitate RE investment. Source of data: *ENERDATA; **World Bank WDI. countries and years, as in our study, promotion policies appear to be the first determinant for RE investment. These results are in line with those in Ref. [21,22,31], but contrasts with the findings of [43,44] which found no positive effects of RE policies. EXPECTED INFLUENCE (+) countries facing rapid growth of electricity demand need to invest in new generation capacity, including RE. Growth rate of electricity consumption over the last 5 years. Besides, the relationship between growth in electricity consumption and the dependent variable is significant but negative. It indicates that countries with higher demand growth have not invested more in RE, but have relied on other, less expensive electricity sources. For instance, countries like Bolivia and Honduras have had averaged a 6% growth in electricity demand but almost no RE investment. Likewise, countries that have invested the most in RE have simultaneously implemented ambitious energy efficiency policies, which have influenced demand. This is the case of several EU countries like Denmark and Sweden, which have had high RE investment per capita, while the electricity demand stagnated. Popp et al. [43] found similar results in a sample of 26 OECD countries. Our model’s outputs also show that a high share of nuclear power in the electricity mix has a negative relationship with investment in RE. This result can be explained by the fact that countries with high shares of nuclear power (like France) have been less concerned with the rapid development of RE because they already have a decarbonised electricity mix. The coefficient for hydroelectricity is also negative but no signifi- cant.This finding confirms the fact that the complementarity or oppo- sition between RE and hydropower is country-specific. On the one hand, a high share of hydroelectricity facilitates the integration of intermittent technologies (wind and solar) thanks to the storage capacity. On the other hand, countries with a high share of hydropower in the electricity mix (a low-carbon source) may be less concerned about the rapid deployment of other RE sources. l Concerning the influence of coal and gas production, the results are different. Contrary to our hypothesis, a high level of coal production per capita has a positive influence (statistically non-significant, however) on RE diffusion. The explanation for this finding might be that some countries with a high share of carbon in the electricity production have substantially invested in RE as a form to decarbonize the mix or replace nuclear (that is especially true for Germany). The result is negative for gas: a high gas production would be, contrary to coal, a disincentive to develop RE fast. This inverse relationship may be because many coun- tries started to replace coal with natural gas during the 1980s or 1990s to reduce pollution what created a lock-in in these technologies. How- ever, the decarbonisation of the electricity mix is reflected in some countries by the development of RE and gas simultaneously: they phase- out coal and, at the same time, use gas to preserve the grid balance, while intermittent electricity increases. i representing “fiscal incentives” is not statistically significant in any model. We also estimated a model in which the four policy-instrument variables are replaced by the single variable “Ex_pol” which indicates the existence or not of a promotion policy in year t for country x, regardless of the instrument adopted. It allows us to test our primary hypothesis about the effectiveness of RE energy policy when controlling by the macroeconomic and energy-related determinants. The results of this model show a positive and significant influence of Ex_pol on the dependent variable.Therefore, considering a sufficiently large sample of Results in Table 2 show a clear positive and significant relationship Table 2 Summary of results - global model. Variables/model Fixed-Effects PCSE Random-Effects Y = new RE capacity Coeff. SE Coeff. SE Coeff. SE Feed-in Tariffs 7.04 3.45 (+)S ** 7.41 3.15 (+)S ** 9.50 2.00 (+)S *** Portfolio Standards 16.55 4.36 (+)S *** 12.68 5.28 (+)S ** 15.20 2.91 (+)S *** Auctions 7.31 3.12 (+)S ** 5.92 3.64 (+)S * 6.44 2.83 (+)S ** Fiscal incentives 3.25 3.04 (+)NS 3.64 2.98 (+)NS 0.52 2.68 (+)NS Demand growth −0.63 0.27 (−)S ** −0.49 0.22 (−)S ** −0.80 0.20 (−)S *** Share nuclear −0.50 0.25 (−)S ** −0.32 0.11 (−)S *** −0.13 0.06 (−)S ** Share hydropower −0.12 0.10 (−)NS 0.00 0.08 (+)NS −0.08 0.05 (−)S * Energy dependence 0.42 0.23 (+)S * 0.51 0.19 (+)S *** 0.06 0.05 (+)NS CO2 per capita −4.20 1.56 (−)S *** −2.76 1.68 (−)S * −3.48 0.58 (−)S *** Coal production 4.86 3.91 (+)NS 7.12 2.57 (+)S *** 3.66 0.78 (+)S *** Gas production −0.02 0.01 (−)S * −0.02 0.01 (−)S ** −0.00 0.00 (−)NS GDP per capita 1.81 0.62 (+)S *** 1.40 0.41 (+)S *** 1.05 0.13 (+)S *** Credit access 0.04 0.08 (+)NS 0.06 0.07 (+)NS 0.07 0.03 (+)S ** Observations 50 × 20 = 1000 50 × 20 = 1000 50 × 20 = 1000 R2 0.23 0.35 0.24 Notes: Significance levels: *** (1%), ** (5%), * (10%); (S): Statistically significant, (NS): Non-significant.Table 2 Summary of results - global model.Table 2 SSummary of results - global model. * (5%), * (10%); (S): Statistically significant, (NS): Non-significant. Notes: Significance levels: *** (1%), ** (5%), * (10%); (S): Statistically significant, (NS): Non-significant. Renewable and Sustainable Energy Reviews 133 (2020) 110351 G. Bersalli et al. quantity control, and competitive pressure between producers, the financial incentives will have to be gradually allocated by competitive auctions. This evolution of incentive schemes is too recent to appear in the model. However, it is in line with the results observed in Latin America and more broadly throughout the world, with a substantial increase in the auction system as an effective RE supply scheme. This evolution of promotion policies has been accompanied by a very marked dynamic of cost reduction over the last 20 years: while the first feed-in tariffs that enabled the PV sector to take off exceeded €500/MWh [55], the average prices of the 2019 tenders for ground-based PV installations in Germany were below €50/MWh. i between income per capita (GDP per capita) and the dependent variable. That reflects that RE has deployed more rapidly in high-income coun- tries, where the economic and institutional conditions are more favourable to the early diffusion of new technologies. This result is in line with the findings of [43,53]. Finally, the availability of credit does not appear as statistically significant. Further analysis that includes additional refined variables is needed to understand the relationship between the development of equity and credit markets and RE deployment.4.2. About RE policy effectiveness in Latin America and Europe The fixed-effects model for Latin America shows a positive and sig- nificant influence of auctions and RPS, while FIT is not significant. We observe a “sub-performance” of feed-in tariffs in Latin America compared to the European experience, where it was a central instru- ment. In Latin America, only a few countries applied feed-in tariff and feed-in premium, with quite poor results. It was the case in Argentina between 1998 and 2006: public policy based on feed-in premium failed to help RE to take off, affected by an unfavourable macroeconomic and regulatory context [63]. Brazil applied FIT for a few years and then changed for an auctions system. Since 2010 auctions became the main policy instrument with, in general, positive results in countries like Brazil, Uruguay, and, more recently, Argentina. We performed specific analyses for the Europe and Latin America samples to identify possible policy effectiveness differences between both regions. i Firstly, we evaluated the significance of our main explanatory vari- ables (the RE promotion policies) through a t-test. The results in Table 3 show that the existence of a promotion policy (observations from group 1) increased the mean of the dependent variable significantly (p-value <0.05). Furthermore, the policies have a positive and significant impact in both regions but with different relative effectiveness. In Europe, for years and countries with incentive policies (observations from group 1) the additional RE capacity is ten times higher than cases without policy (group 0); the ratio is 5 in Latin America. The standard deviation in group 1 observations is also significant, suggesting differences in policy effectiveness within countries.The auction scheme has several advantages that explain its imple- mentation in Latin American countries: i) It provides a stable and well- known ex-ante revenue stream once projects have been awarded, which facilitating project funding; ii) It stimulates competition among pro- ducers, facilitates the externalization of RE costs and allows control of the total cost of the policy; iii) It is adaptable to promote technologies with different degree of techno-economic maturity; iv) Besides costs, it allows the introduction of other selection criteria (like jobs creation) among the tenders submitted according to the objectives of the public policy. Most importantly, auctions scheme fits well with the institutional design of the electricity system in most Latin-American countries [42]. Nevertheless, the downside of auctions is that it favours “big” existing actors to the detriment of new and smaller producers to enter the mar- kets. Thus, this instrument should be complemented by specific policy targeting small and decentralized RE projects. In recent years, several countries of that region have moved in that direction. Secondly, we estimated the model with a binary “existence of RE policy” variable. The impact of support policy was positive and signifi- cant for both Europe and Latin America. Finally, we introduced the four specific policy instruments in our model. As in Section 4.1, we base our analysed in the fixed-effects model with Arellano correction and compare them with the two alternative models (Tables 4 and 5). The three main policy instruments applied in Europe -FIT, RPS, and auctions-show a positive and significant influence. These results confirm what the review of energy policies for RE in Europe shows. The public policies implemented by the Member States at the European Commis- sion’s initiative in favour of the development of RE have been very effective.This is confirmed by the mid-term progress assessment report,7 which states that the EU Member States are on track to meet the renewable energy targets for 2020, i.e., 20% renewable energy in gross energy consumption.8 To this end, the European Union’s policy has been based on several successive directives setting quantitative objectives for Member States (2010, 2020, 2030) but leaving each of them free to choose the means to be used to achieve them. The very marked oppo- sition during the 2000s between price and quantity instruments has evolved towards the clear domination of the former. Since then, incen- tive mechanisms have been adapted to take into account technological progress and cost trends. Again under the impetus of the European Commission, support schemes have evolved in recent years. Indeed, to allow better integration of renewables into the electricity market, RE policies in Latin America were introduced later and started to be effective when the cost of technology was affordable enough to allow private investments while controlling the cost of the policies. The pri- mary motivation of RE policy was concerns about energy security and the diversification of the electricity mix. Most Latin American countries have an electricity system concentrated in fossil fuels (Chile, Mexico, Argentina) or hydroelectricity (Brazil, Paraguay, Uruguay). A power mix concentrated in a few sources raises risks considerably: a succession of dry years or problems with the supply of imported gas or coal can affect electricity production and its cost substantially. For that reason, coun- tries tried to diversify the electricity mix, and nuclear power and re- newables have represented the main options. However, nuclear power exists only in three countries (Argentina, Brazil, and Mexico), and is a sophisticated technology that requires enormous investment and expertise. Renewables were relatively expensive, but the substantial cost decrease and their modularity made them an increasingly attractive option during the 2000s. Even if some RE technologies are currently cost-competitive in Latin America, public policies still play an essential role in reducing the risk of such investments. Table 3 Effect of policies on RE investment - Student’s t-test by continent.Europe Latin America Group Obs. Mean SE Obs. Mean SE 0: Without policy 129 2.19 6.5 282 1.07 3.94 1: With policy 471 23.32 30.9 118 5.33 13.78 Combined 600 18.7 28.9 400 2.33 8.39Table 35. Conclusion and policy implications This paper’s primary purpose was to evaluate the effectiveness of different RE policy instruments implemented in Europe and, more recently, in Latin America, while controlling by other determinant fac- tors. We first performed a review of the literature on econometric evaluation of RE policy. The determinants explaining the different levels 7 European Commission, 2015, Report from the Commission to the European Parliament, the Council, the European economic and social Committee and the Committee of the regions, Renewable energy progress report. 8 Climate 2020 Energy Package. 8 Climate 2020 Energy Package. Renewable and Sustainable Energy Reviews 133 (2020) 110351 G. Bersalli et al.Table 4 Summary of results - Europe.Table 4 Summary of results - Europe. Variables/model Fixed-Effects PCSE Random-Effects Y = new RE capacity Coeff. SE Coeff. SE Coeff. SE Feed-in Tariffs 12.25 4.38 (+)S *** 11.67 4.42 (+)S *** 14.39 3.12 (+)S *** Portfolio Standards 23.51 5.06 (+)S *** 18.22 6.58 (+)S *** 19.68 4.20 (+)S *** Auctions 10.95 3.32 (+)S *** 8.77 5.71 (+)NS 7.42 5.09 (+)NS Fiscal incentives 8.35 7.37 (+)NS 9.66 6.10 (+)NS 7.83 5.40 (+)NS Demand growth −1.77 0.46 (−)S *** −1.50 0.50 (−)S *** −2.06 0.43 (−)S *** Share nuclear −0.66 0.26 (−)S ** −0.42 0.15 (−)S *** −0.22 0.09 (−)S ** Share hydropower −0.52 0.22 (−)S ** −0.26 0.18 (−)NS ** −0.21 0.10 (−)S ** Energy dependence 0.61 0.32 (+)S * 0.61 0.31 (+)S * 0.10 0.10 (+)NS CO2 per capita −4.02 1.91 (−)S ** −2.45 1.84 (−)NS −3.53 0.86 (−)S *** Coal production 4.53 3.82 (+)NS 6.61 2.78 (+)S ** 3.16 1.18 (+)S *** Gas production −0.02 0.01 (−)S ** −0.02 0.01 (−)S ** −0.00 0.00 (−)NS GDP per capita 1.25 0.66 (+)S * 1.01 0.43 (+)S ** 1.00 0.18 (+)S *** Credit access 0.03 0.09 (+)NS 0.06 0.07 (+)NS 0.05 0.04 (+)NS Observations 30 × 20 = 600 30 × 20 = 600 30 × 20 = 600 R2 0.26 0.34 0.25 Notes: Significance levels: *** (1%), ** (5%), * (10%); (S): Statistically significant, (NS): Non-significant.Notes: Significance levels: *** (1%), ** (5%), * (10%); (S): Statistically significant, (NS): Non-significant. Table 5 Summary of results - Latin America. Variables/model Fixed-Effects PCSE Random-Effects Y = new RE capacity Coeff. SE Coeff. SE Coeff. SE Feed-in Tariffs 0.43 1.87 (+)NS 1.15 1.24 (+)NS 1.72 1.63 (+)NS Portfolio Standards 6.37 2.28 (+)S *** 3.75 2.98 (+)NS 8.52 3.70 (+)S ** Auctions 7.05 3.02 (+)S ** 3.97 2.80 (+)NS 9.52 1.55 (+)S *** Fiscal incentives −0.08 1.27 (−)NS −1.20 1.02 (−)NS 0.21 1.47 (+)NS Demand growth −0.09 0.04 (−)S ** −0.05 0.08 (−)NS −0.11 0.10 (−)S Share nuclear 0.93 0.55 (+)S * −0.11 0.50 (−)NS −0.31 0.33 (−)S Share hydropower 0.09 0.06 (+)NS 0.18 0.07 (+)S ** 0.03 0.02 (−)S Energy dependence 0.25 0.26 (+)NS 0.31 0.17 (+)S * −0.01 0.02 (+)NS CO2 per capita 0.89 1.02 (+)NS −1.27 1.74 (−)NS −1.04 0.62 (−)S * Coal production −0.86 1.03 (+)NS −3.22 1.73 (−)S * −1.13 1.59 (+)S Gas production −0.00 0.00 (−)S 0.00 0.00 (+)NS −0.00 −0.00 (−)NS GDP per capita 2.87 1.69 (+)S 4.45 1.65 (+)S *** 1.23 0.37 (+)S *** Credits acces −0.04 0.10 (+)NS 0.01 0.09 (+)NS −0.01 0.03 (+)S Observations 20 × 20 = 400 20 × 20 = 400 20 × 20 = 400 R2 0.19 0.24 0.19 Notes: Significance levels: *** (1%), ** (5%), * (10%); (S): Statistically significant, (NS): Non-significant.Table 5 1995, 14.6% in 2014, and 19.5% in 20189. In the beginning, the incentive policies, especially subsidies like feed-in tariffs, were para- mount due to the cost gap between fossil fuel and RE projects. In Latin America, RE (excluding hydro) represented 2.5% of total electricity production in 1995, 6.42% in 2014, and 11% in 2018. The countries of this region started to introduce RE policy later than the Europeans one and, thanks to a substantial drop in the costs of RE technologies and favourable natural conditions, most RE projects in Latin America do not need direct subsidies anymore. Currently, that region benefits from one of the greenest electricity mixes in the world thanks to a strong base of hydroelectric power and an increasing share of solar and wind energy. Indeed, Latin America is the first region in the world in terms of the share of RE generation if we include hydroelectricity (the share is around 58% in Latin America, while 36% in Europe, and 25% in the world average). The public policies implemented since the 2000s have had a significant influence on the development of wind, solar, and biomass energy sources. However, the decarbonisation, first, of the electricity sector and then of the whole energy sector, has revealed various problems and requires sustained public policies over time. Several European countries have accumulated vast experience in the implementation of RE policy, including potential adverse effects. or RE diffusion across countries can be classified into four categories: support policy, factors related to the structure and dynamics of energy markets, macroeconomic factors, and political and institutional de- terminants. We identified several conflicting points and lacunas in the econometric literature. We then developed a model which is the first to integrate a significant sample of Latin-American countries. Our results converge for the influence of promotion policies in general: public policies had a positive and statistically significant effect on RE invest- ment. However, the effectiveness of these policies seems stronger in Europe than in Latin America, partially explained by the different temporality in policy implementation.Also, some differences appeared concerning the type of instrument: auctions have consolidated as the main instrument in Latin America, where the institutional conditions felicitate their implementation. We conclude on the effectiveness of the main policy instruments to promote RE, in both Europe and Latin America. Instruments like feed-in tariff or the auction scheme are essential to reduce the risk associated with RE investment, even when some technologies like wind and solar PV are already cost-competitive in several markets. On the opposite, tax incentives, on their own, are not sufficient to incentivize RE deployment, and countries should not base their technology policy only on such instruments. RE generation is increasing in both regions. In Europe, the share of renewables in electricity production (excluding hydropower) was 1% in 9 ENERDATA, 2019. Renewable and Sustainable Energy Reviews 133 (2020) 110351 G. Bersalli et al. Stronger cooperation among the policymakers in both regions would help to boost energy transition. One specific area of plausible collabo- ration is the decentralized RE solutions (like home-based solar and wind technologies) that could enhance energy access in many rural areas of Latin America. [12] del Río P, Mir-Artigues P. Support for solar PV deployment in Spain: some policy lessons. Renew Sustain Energy Rev 2012;16(8):5557–66. ii [13] Bergek A, Jacobsson S. Are tradable green certificates a cost-efficient policy driving technical change or a rent-generating machine? Lessons from Sweden 2003–2008. Energy Pol 2010;38(3):1255–71. [14] Buckman G, Diesendorf M. Design limitations in Australian renewable electricity policies. Energy Pol 2010;38(7):3365–76. Our econometric analysis has some limitations and several possible extensions. First, the available data did not allow us to estimate the influence of some determinants identified in the literature, such as the cost of other sources of power generation, the real potential for different RE sources, and political factors. Similarly, more detailed consideration of the variables for access to financing (e.g., changes in the cost of available financing) would improve our model’s explanatory power. Finally, effectiveness is not the only criterion relevant to an evaluation of energy policies.Other factors, such as static and dynamic efficiency [42,64] and the equity of such policies, are also relevant and should be considered in future research. [15] Verbruggen A. Performance evaluation of renewable energy support policies, applied on Flanders’ tradable certificates system. Energy Pol 2009;37(4):1385–94. i [16] Lipp J. Lessons for effective renewable electricity policy from Denmark, Germany and the United Kingdom. Energy Pol 2007;35(11):5481–95. [17] Liu W, Zhang X, Feng S. Does renewable energy policy work? Evidence from a panel data analysis. Renew Energy 2019;135:635–42. [18] da Silva PP, Cerqueira PA, Ogbe W. Determinants of renewable energy growth in Sub-Saharan Africa: evidence from panel ARDL. Energy 2018;156:45–54. [19] Romano AA, Scandurra G, Carfora A, Fodor M. Renewable investments: the impact of green policies in developing and developed countries. Renew Sustain Energy Rev 2017;68:738–47. [20] Cadoret I, Padovano F. The political drivers of renewable energies policies. Energy Econ 2016;56:261–9. [21] Kilinc-Ata N. The evaluation of renewable energy policies across EU countries and US states: an econometric approach. Energy Sustain Dev 2016;31:83–90. l [22] Polzin F, Migendt M, T¨aube FA, von Flotow P. Public policy influence on renewable energy investments—a panel data study across OECD countries. Energy Pol 2015; 80:98–111.CREdiT author statement Bersalli: Conceptualization, Background, Statistical analysis, Results and Discussion. El-Methni: Statistical analysis, Results and Discussion. Menanteau: Conceptualization, Writing- Reviewing and Editing. [23] Marques AC, Fuinhas JA. Are public policies towards renewables successful? Evidence from European countries. Renew Energy 2012;44:109–18. ´ [24] Del Río P, Taranc´on M-´A. Analysing the determinants of on-shore wind capacity additions in the EU: an econometric study. Appl Energy 2012;95:12–21. [25] Johnstone N, Haˇsˇciˇc I, Popp D. Renewable energy policies and technological innovation: evidence based on patent counts. Environ Resour Econ 2010;45(1): 133–55.Declaration of competing interest [26] Aquila G, de Oliveira Pamplona E, de Queiroz AR, Junior PR, Fonseca MN. An overview of incentive policies for the expansion of renewable energy generation in electricity power systems and the Brazilian experience. Renew Sustain Energy Rev 2017;70:1090–8. The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. [27] Pueyo A. Enabling frameworks for low-carbon technology transfer to small emerging economies: analysis of ten case studies in Chile. Energy Pol 2013;53: 370–80.Acknowledgements & Funding [28] Lampreia J, De Araujo MSM, De Campos CP, Freitas MAV, Rosa LP, Solari R, et al. Analyses and perspectives for Brazilian low carbon technological development in the energy sector. Renew Sustain Energy Rev 2011;15(7):3432–44. The authors acknowledge the support of the GAEL laboratory members for reading the article. The correspondent author gratefully acknowledges his former thesis supervisor Dr. Patrick Criqui and his colleagues at the IASS Potsdam institute. This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. [29] Guzowski C, Recalde M. Latin American electricity markets and renewable energy sources: the Argentinean and Chilean cases. Int J Hydrogen Energy 2010;35(11): 5813–7. [30] Moreno R, Barroso LA, Rudnick H, Mocarquer S, Bezerra B. Auction approaches of long-term contracts to ensure generation investment in electricity markets: lessons from the Brazilian and Chilean experiences. Energy Pol 2010;38(10):5758–69. [31] Pfeiffer B, Mulder P. Explaining the diffusion of renewable energy technology in developing countries. Energy Econ 2013;40:285–96. i [32] Surana K, Anadon LD. Public policy and financial resource mobilization for wind energy in developing countries: a comparison of approaches and outcomes in China and India. Global Environ Change 2015;35:340–59.Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.rser.2020.110351. [33] Kim J, Park K. Financial development and deployment of renewable energy technologies. Energy Econ 2016;59:238–50. [34] Finon D, Menanteau P, Lamy M-L. Prices versus quantities: environmental policies for promoting the development of renewable energy. Energy Pol 2003;31(8): 799–812.References [35] Butler L, Neuhoff K. Comparison of feed-in tariff, quota and auction mechanisms to support wind power development. Renew Energy 2008;33(8):1854–67. i [ ] , p , q support wind power development. Renew Energy 2008;33(8):1854–67. [36] Del Rio P. The dynamic efficiency of feed-in tariffs: the impact of different design elements. Energy Pol 2012;41:139–51. [1] Plessmann G, Blechinger P. How to meet EU GHG emission reduction targets? A model based decarbonization pathway for Europe’s electricity supply system until 2050. Energy Strat Rev 2017;15:19–32. [36] Del Rio P. The dynamic efficiency of feed-in tariffs: the impact of different design elements. Energy Pol 2012;41:139–51. [37] Dinica V. Support systems for the diffusion of renewable energy technologies—an investor perspective. Energy Pol 2006;34(4):461–80. [2] REN21 R. Global status report, REN21 secretariat. Paris: France. Tech. Rep; 2018. [3] IRENA I. Renewable power generation costs in 2017. Report. Abu Dhabi: International Renewable Energy Agency; 2018. [ ] p , p; [3] IRENA I. Renewable power generation costs in 2017. Report. Abu Dhabi: International Renewable Energy Agency; 2018. [38] Fouquet D, Johansson TB. European renewable energy policy at crossroads—focus on electricity support mechanisms. Energy Pol 2008;36(11):4079–92. [4] Percebois J. Aides publiques aux ´energies ´eolienne et photovoltaïque. Rev Fr Econ 2015;30(4):141–86. [39] Weitzman ML. Prices vs. quantities. Rev Econ Stud 1974;41(4):477 91. [40] Tanaka M, Chen Y. Market power in renewable portfolio standards. Energy Econ 2013;39:187–96. [5] Quirion P. Quels soutiens aux ´energies renouvelables ´electriques? Rev Fr Econ 2015;30(4):105–40. [41] Klessmann C, Held A, Rathmann M, Ragwitz M. Status and perspectives of renewable energy policy and deployment in the European Union—what is needed to reach the 2020 targets? Energy Pol 2011;39(12):7637–57. ´ [6] Sun P, Nie P-y. A comparative study of feed-in tariff and renewable portfolio standard policy in renewable energy industry. Renew Energy 2015;74:255–62. [7] Held A, Ragwitz M, Merkel E, Rathmann M, Klessmann C. Indicators assessing the performance of renewable energy support policies in 27 Member States. available at: www.reshaping-res-policy.eu; 2010. [42] Bersalli G. ´Evaluation et ´evolution des politiques de promotion des ´energies renouvelables: la transition des secteurs ´electriques en Am´erique Latine. Grenoble Alpes; 2017.p g p y [8] Verbruggen A, Lauber V. Assessing the performance of renewable electricity support instruments. Energy Pol 2012;45:635–44. i [43] Popp D, Hascic I, Medhi N. Technology and the diffusion of renewable energy. Energy Econ 2011;33(4):648–62. [9] Haas R, Resch G, Panzer C, Busch S, Ragwitz M, Held A. Efficiency and effectiveness of promotion systems for electricity generation from renewable energy sources–Lessons from EU countries. Energy 2011;36(4):2186–93. [44] Aguirre M, Ibikunle G. Determinants of renewable energy growth: a global sample analysis. Energy Pol 2014;69:374–84. [45] Chang T-H, Huang C-M, Lee M-C. Threshold effect of the economic growth rate on the renewable energy development from a change in energy price: evidence from OECD countries. Energy Pol 2009;37(12):5796–802. [10] Ragwitz M, Held A, Resch G, Faber T, Haas R, Huber C, et al. Assessment and optimization of renewable energy support schemes in the European electricity market: final Report. Germany: Optimization of Renewable Energy Support (OPTRES) Karlsruhe; 2007. gy [46] del Río Gonz´alez P. The empirical analysis of the determinants for environmental h l l h h d l ( ) [46] del Río Gonz´alez P. The empirical analysis of the determinants for environmental technological change: a research agenda. Ecol Econ 2009;68(3):861–78. [46] del Río Gonzalez P. The empirical analysis of the determinants for environmental technological change: a research agenda. Ecol Econ 2009;68(3):861–78. [47] Biresselioglu ME, Karaibrahimoglu YZ. The government orientation and use of renewable energy: case of Europe. Renew Energy 2012;47:29–37. [11] Mitchell C, Bauknecht D, Connor PM. Effectiveness through risk reduction: a comparison of the renewable obligation in England and Wales and the feed-in system in Germany. Energy Pol 2006;34(3):297–305. [47] Biresselioglu ME, Karaibrahimoglu YZ. The government orientation and use of renewable energy: case of Europe. Renew Energy 2012;47:29–37. Renewable and Sustainable Energy Reviews 133 (2020) 110351 G. Bersalli et al. [48] Potrafke N. Does government ideology influence deregulation of product markets? Empirical evidence from OECD countries. Publ Choice 2010;143(1–2):135–55. i Energía Renovable en Centroam´erica y Panam´a. Banco Centroamericano de Integraci´on Econ´omica; 2018. i [49] Fredriksson PG, Vollebergh HR, Dijkgraaf E.Corruption and energy efficiency in OECD countries: theory and evidence. J Environ Econ Manag 2004;47(2):207–31. [57] Hausman JA. Specification tests in econometrics. Econometrica J. Econom Soc 1978:1251–71. y g [50] Fredriksson PG, Svensson J. Political instability, corruption and policy formation: y g [50] Fredriksson PG, Svensson J. Political instability, corruption and policy formation: the case of environmental policy. J Publ Econ 2003;87(7–8):1383–405. [58] Pesaran MH. General diagnostic tests for cross section dependence in panels. 2004. [58] Pesaran MH. General diagnostic tests for cross section dependence in panels. 2004. [59] Breusch TS. Testing for autocorrelation in dynamic linear models. Aust Econ Pap 1978;17(31):334–55. the case of environmental policy. J Publ Econ 2003;87(7–8):1383–405. [51] Neumayer E Are left wing party strength and corporatism good for the [59] Breusch TS. Testing for autocorrelation in dynamic linear models. Aust Econ Pap 1978;17(31):334–55. [51] Neumayer E. Are left-wing party strength and corporatism good for the environment? Evidence from panel analysis of air pollution in OECD countries. Ecol Econ 2003;45(2):203–20. [60] Godfrey LG. Testing against general autoregressive and moving average error models when the regressors include lagged dependent variables. Econometrica J. Econom Soc 1978:1293–301. i [52] Zhao Y, Tang KK, Wang L-l. Do renewable electricity policies promote renewable electricity generation? Evidence from panel data. Energy Pol 2013;62:887–97. [52] Zhao Y, Tang KK, Wang L-l. Do renewable electricity policies promote renewable electricity generation? Evidence from panel data. Energy Pol 2013;62:887–97. [53] Shrimali G, Kniefel J. Are government policies effective in promoting deployment of renewable electricity resources? Energy Pol 2011;39(9):4726–41. [61] Breusch TS, Pagan AR. A simple test for heteroscedasticity and random coefficient variation. Econometrica J. Econom Soc 1979:1287–94. [62] Arellano M. PRACTITIONERS’CORNER: computing robust standard errors for [61] Breusch TS, Pagan AR. A simple test for heteroscedasticity and random coefficient variation. Econometrica J. Econom Soc 1979:1287–94. [53] Shrimali G, Kniefel J. Are government policies effective in promoting deployment of renewable electricity resources? Energy Pol 2011;39(9):4726–41. [62] Arellano M. PRACTITIONERS’CORNER: computing robust standard errors for within-groups estimators. Oxf Bull Econ Stat 1987;49(4):431–4. [54] Winkel T, Rathmann M, Ragwitz M, Steinhilber S, Winkler J, Resch G, et al.Renewable energy policy country profiles. Report prepared within the Intelligent Energy Europe project RE-Shaping. www.reshaping-respolicy.eu; 2011. [63] Bersalli G, Simon J-C. Vers une transition ´energ´etique des pays ´emergents: quelles politiques d’incitation aux ´energies renouvelables dans le secteur ´electrique en Argentine et au Br´esil? D´eveloppement durable et territoires ´Economie, g´eographie, politique, droit, sociologie 2017;8(2). [55] Fulton M, Mellquist N. The German feed-in tariff for PV: managing volume success with price response. DB Climate Change Advisors. Deutsche Bank Group; 2011. g g p p q g [64] Nicholls J, Mawhood R, Gross R, Castillo Castillo R. Evaluating renewable energy policy: a review of criteria and indicators for assessment. 2014. [56] BCIE. An´alisis Comparativo del Marco Regulatorio. Incentivos y Sistema Tarifario de Precios Existentes, para la compra/generaci´on de Electricidad de plantas de2#f*.6;B>I>jA]EMVW9YbCksyzzbl˘Ҙ֛Ƨ}fhIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIIII$O KhNQJ1XSI$\#kUPX]7^V]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!1AQaq!1AQaqiV <&<Ϻ==Ef]a=T҉5ԓ::6 軸<̼5OhQ=RI<'»Ko<ɼ;$Г6=nF=/&NF婼+H<ӻ!;o`<(Y<-=L;W<=^*x%*<==i<sL'T I@=nk=Z9؅ $E:#？9 =7 <Ɖ,=^[Eң<ͺِ;=;o`,<"=rb;x:e}C'¿B[{o=Q<$3M\׀7:;<''!F=뼂3RZ4R|XW)<c:&$ռA⌽$=<*+cڼ60w;cpb<%<|")睼{{VA楿>v:<;vG<a z<1:*M<5=Ypϼ'(2`PآJ:eW; υA,!={-P;ϋ<\`-=;G<Ш =b>Q6=+PM <"Ӑ:i8O;6Nsm;=E0<8[;(<ӼR =<ح=<><쟼$)<1<Լ׬>s_Ap Pͼ-: 'L<:Ҽ|<,v=%%<zk;\N4r<L< =d2ZWc|==(~dD=b1\"=B<&%=¯X=B==<8V<-<`5,=f"<@o0/x<ta=2[ν<6 3~NτQ<6 <=ɻ\w<.̻Fּ0=8<'F'L.;C=?)ѼY=μ'/<@nIM;VM=;y=\ #h;Ի1炽HZ=yT(=mǼ>jmd3Qt*;~3ϻ)<'zN=?yp<=O=v /9<´5*=_FO;u'=- F<z2<'_ S;n;-=lm=;ȫмA;`;ykW=hX<Ӱ5gDռEݼסk<469=͆2pR=Nh=H=Y/ϛŻ֥A<{o;*upr7<޼ӭ<0S`<<;P<#+q7<716<IxJ:*; iv=CFnQ<+(R;?l;$< NF L;q]Ӽ^z;. Ot<< g5<'<&4aWO|=2ټ==ɕdӼK*t_: <ؙQ;@<X=nWP=.=t;5 9J/#<ծ<α|X=w T=D;<9ʇY7=8 :b9c";P8=Z=<,@='I\^=U9><:h<; hh9q}4:Xq=:<A=BfsE<^!0]R<_ *<dV9΢~#x}ә<'%z83=\ =Q =]j "ѼT&!g<==/I" Ȣ<oE<\;S>!=3w&Ӷ*=[ᑼ;/ $]x .&Bg6޻,ۻLJbՎ=kC@=|SM<(:r[8;lAn$y,\=7' p^<*<>%uj<,<Ѱ=kO砄;XKV=\0`<w==$UJ)l?{<0{y(o*U|<엺)̻l< L: =x䴺y <2; "#"G:=Oaμ;c<弘 =0sv <*༈h<.g<=doV$v;U={!Wż'3G@;=8^OF.(|̗&T=hP&]^<ןe\<f;y=LDa<)<Z<(6bS6!<{< û<`=|`z_<5QڼA.ϼW<2$=Tu =mmŠ'c_:>O;'MO<.cP=F< % )y/<L:=$ZŃz;;(qmd=xr<jRZwC*=@ƦLͼ*c;d%(=c2=ay=@E^)$]&*<~<ݳ޻T ={:=-=m =y8 o<Ƽ(q #n9<'w;=gG>>,=/;;y;=z;Nj뒃& =Sq[<1=ߪ=c{ =A;rmM<Մ;:px[Hɠ9KE=> *kUR<{»xBLC- /;{=b+=t<~˼{i漚$#Bۼ3M6}<FۼWv)#y<J=[j<‰0Cv`R.5 Yl=u;v=`.<'\<爽1"]sMW}3*= ?;Fy hz<$p};)X=@ϼMZ-=I)5;%:g=~&B;pKF<^?\=THc=<[n8=<{Q:xA<eO팻#!x< ;Ij;;=僙s:_6kP-p;"}<7=<8VFU<$Ox޼MмA=Ko;N*<𓀼㷼;! =C-ļ⼠ Nz=7q<92=߭<{.|7<[%e_=1;d=%B<=Zy<2Ń=XV=W=B> ļt]ȼ +F߮< ;_S<0;=E9=ށ<ܼdf2B;]-=ϼ>ż<$37D@:C=Fl(=?ɼ{))@lՌ;R<:6f#: ,\<<R=?,:<,@t=4 f/ <,JF=X@;.KNq=%q=εĚZ MStQ؏:|YH<˒=6<;ؕ< N=or<5cn[Z=)= 9ɽ3

{<*A .;;*%1ž< <*`Y< $;k]ۻ9"hX=>p;hV<C=-F5=@6_d<m=:uleh<)='Mf'g,;-]9c&PfPv+=ʜA?2˻ι=|X< m<۴<~=c̹$m7!>*=(5λ:A+aLM|=T4;bi:ܦY)눻m<~>=h:C=m m^yq;C;e<:[<)<= \AR¼lE=*^:-b<;/bn![<~.;<ѺւQϪ=&=dμM<7oJǼvE=<ȀWռS =V7=LP!:Z;|J%=ZVdaIS==d=}IР`<a<A;6 <<ˢ;@/<<=?89Tx:, 9 =="=(= =<,W=6L}6=c޻dK|cDn<|Ꮌlh;+׼s>fDL<;lc8Wyq=_μxFYKH(=ab9 xT;e=|<S=Y<|<ƭe4=! 1=lrxļ#uu1;7&< =<`Z='<_]G=6{oC=QR83){ͼ<' <} 8^ $<KI靼" ;;;Dټ9=T<KW;#ջqڼbf#1<孼~l.J=ͩ<Ɍ;)aP>ׇXBz<<{HS LZ = C`Ofw=A}%<; <|<`43< %9/!ɼ9t=xO<=PWsn{Ἶӻ 0{`|vܒ<C<:b<sE<<^7hq:# cL;K=>騧 N; μ~۞TS=[ٵS=o<7ȹ= .=g=\%h ={,=k<: NƁ'߼EƼ, =W<< 2EߦjQ%n ˡ<%0<*#2k<1dP;-<~ {pл=à٩ =O-==(5NN51ú;ʕr=ܚ=@H MR=00μ$33UsU;|=,X<؈;h~b;M;N޼m#;'ʼ#9ߞ`={<4ɼ<a[ <8< Mڼ\`<̼+n<><#\s;=_/<;@=輼Y<;=;ylQ n<n#C7 5<\L==<3;=L5=gI<ռ&[=P9U=37x=?<߼<\;,V = ;4 R=H(Ǯ;]a<4=)<ԛxʻO <܏=X=>}S<~><3XZ.Rڞ=c^Jb@E79$d<{{R=; k=8=KD<(<ĕd[<r@lQ <(<]?W,฾vPX=Ƽ~)N<(<ϫ|P+(<=ǜ%m<,<Ս<-nz%[<=Kt;xRƴ:M=, <;Ʊ JqI3Z;ꖼ<0@0r;=y=͡;̄+μ;=bb;!h=-s<'P<L < ǻhx𳼢Pټ\<2.=z=Jy<7F=˗;2ڗ<͑J#\;=;rX =御;g[?"3=J(ؼj;;^cZ<,Ǜ=u`Fiq4x; &-ov]O;FD< v:ATU=6<};ǎ2(5;)/dռn5=syS,;X}ޢLh5`Kռ{B=C9;M9.X; 䢼P<~?˼zaYu<=,f8\= =(;ּk<#+lN!R<U1OԱI!|<)󂽖#=Y;<;;ҩ =N<)< /;`漨;uۼ,<_:$dLS È<@ b:<ўl#=V<($;uQ8\ߊ;6K"& &C=QGÏ֚}93`C#=SJ==;a= <1>6@;Vva=-n;ZZ<=-;G#=;P1޼s<';$޻<@;M9#? W;2s<;= >:w< A|<{<݈;c==<"=i<T) =y<>ea69*};=gN&W:5N;7<;jc<8/\=jR^C;4axg=mM=vs<=n(*C:W<=ԇ<k=I;ռΨ<Ӽ;0R%=Ĭ< T<;q-<1<^b?4< H 3:;y:<0SҼ AҼ?㻆J};g\;e+\μͷZl޼?0=IgH4=(S|=伿ϲ;q n)b=f;'TPn?<ׁ=)8R,]%<< q$7< +=~E =^J;<'=~-м;~POFi 8q]=v7;bH<4xdwn=p<:X =t<ռ[(wo;ZY!zЫ==cs=FM-]=ǹp;金38X#Qy.B=s/=`=IE;U;;lW<;=ld.JE<1=|Fܼ?g<͹a:)^ۑ:ԝaUҺ\G+)=>=n<_$_}=CL:=JW<u==[9hXC=``>;4c^sz_|5*l"w<=,[`sA%;F;CwO񼶺 M<%¼v(=== 7={<)g6'<4VE=TTsW%"d(IFO-=S7;>=$ I:!~;!);Bv=\;={3QG<7= ݓl .):](Ѻ?NUC<:u<";Z!Ezr;E`y*U;<Ċu٪ P<pf 偼1 p;)G;:6 2纼Ո18=igzExmI) =~Ż½%WoK=< &{I<ݝ=A)=ɔ=:ME=po<+TdKl;ys=r׼Y< EFː/bkТ1=p^<'9<"<@;U!H1xh=8 =̜=4;&b)=vO=ڼ{:{;2U_<@!/S;`=;O m2<^<8p =Z\<<릇<(ne>9x̼<륜1=(y<'*<*;\F0I==,"CEQQM4;sպ&=|&mfڈdYٲ<^&=OO)=0<k<c:!AC<פ2m>z <6<ߣ<(v=򅼼<.6;$n=~ٓ;Ǽhoo@\]I<%[)<=49=+;=^)^|<'8< 㻵ռb|=2Z<!M=|5,N^Ҧ~{<.=s=t =D; $;̺׿;wf<.b1Sq?Gļdz5=FrTa4zsٺk(='!K=(Vz<={`<"%=W_< ;*έc<)Ļu~D=)p=Q%=q?Q;!&H=%;.Kq;E= r=5<..=>=j)܆ >H#4 #>gՉH<4=[J;cμ'N!;Ux@1k]<ICk=8#K;#<"< vؼ>ǟ_;n/1a=WY =4 TѼl]F= l=;_ =ѺCc<1=o;;p=v P= -K< .=J<~{ǽ<K<9f_;UIx弝L95=Z<S=٭!;,I=<86<&.<5;(,̾w3 =J]=:#T9mN=a-;oU=0I/P] 5|<@ !fІ<tΑ2=h= <<4M=:n^< =<=;h"<<ZFѼ;Լw<'<-;;CW˼p-eל' A=,⼇m\=঻N:<,ջB@%<ڄ;Nca=-7W;mM!1gK<2b2&; '=ͽٌ=x\"=8{|=⅁;ɐhzHW &=;^==@=r;<>Ƌ=l![=T+Fɼ!Y".=;+qW=/ 3g<,5OlxN+<ƷR=P<2:?u=|<׻V3p)V.A< fRiNjs=:,}ǃ<8 hQ<-<_JZ<8N+=_i9;,캄P. <^;=+g>G'<8V.=s0<} ?yD=L< 5:yBɧ?1=Iĥ<:>Y*<`H=:B=wK=7V:c=|4=DJ8:-ML=PH䡅<};}"b$S`v N=ںz)=ED?Bq;q녬(ewK!=:6$J >Q=һ{l^;c|<" ȼJhd=4;ŕy9>< 2<ƨ=MRހ+k1|"ڼ{PEGک=6*.<j<ᴅ(h@<ȫ<<,<in<4<걊 -u;l;g O;l/~cyh;M<"=8w<; *W7< ̻wcHi<:1<:)騱44鼡=Q< Ѽo`<8<攪<,X;=(_< =,dU? `<1k<;=DM<;x뫼⽻<̃IF=r믝t;L<Ϻ<1+\D;==d);;{:0LU>CyIDR=H:42=dc=2ؼ<<.弭[<t RC=\<I_<D׼}o |^ ]= u(< 7C=-=Cݼy(+w=MrѻҶQ7=@<c),ǁR#.C

:ܻ:b1=?A<߼Ae j'঻#3$)ļ6e|L=k:G=Am"o=⌎ tѼc;?{=_;9C;,ٰ<=a =Z<}<|c=^:T'fջ@yIr=6va~"8-zVv&=輋Y4^1<+;*[s<: E<漤R?($SdǼuF8;z<"=&<=<޼'Ff ;+.=տq:W5TOʼNE<!AfQwL;s8<5,F{V<<_xc.So<3w=mm#=oRܻt&KȀ B=T]=r=vJdK(~=:۽?;;r<><<> =e ;!=<Đ<73:z<)>;,={ኼr;FǼə<`;/;"9,!=F^}#g }j1=M&klz<ü=A7 <]<&~< O|:u$ '<9! YLGҼ;;< f{'=T};O@姼TU?2;;<1zB=frR<<¾~`9&;<0;(<1b-2;m?i<2%<$/ =(& ],;Ŏq|== M< >ѱ<Լ?+_=5E;=I==DMG=*?<(<+]0|=.vP켾7gRt]2<9Iq=+h<帝S =?fָn#(<ͻr<&=* l@7q<_P<cGdҏ:<6:d9pJK; ]<^+65zs!6ٺ=THv;۵=K=>;=mJIU=dP=q<%S^=>^kO={o<A<*U7; 缇a="Ѕ:⼵N| V; <;,>g<WuၻF4<飻N#q~P,`9F?<?< E=43=g(Hnһu<ׇ$<Љ9q(VOC$< u򊻑3P I,=ʻ!\`[A=$fN۔=SiC =[W<,=`WU<=<<+aʹ91=q󂽡<;uK=4wGK<"-=?KMӈ<(! P  `<v;I;p<맼h;$a-=13z; _/=%)e9,H=*ϫjw{Х_; i<8ʼP8h鋽\_;Ł_mN{T/`= [5e<<)=-a=igV<;:LN|=M68#==n)Lr㵽^*</;ּT =9V<ڂ<En.= ļ~]<A;>=oQ;bѼ 9=n6f@r==@=Бd<|<<9WZe ;ڵ)Rd=:C=&e=5X<~ir<ؚļVe]<Լm J]<>Nɼ\뼝Ȝc4;*I4==;%<y;k-QxA9=~ =1==Ww5=||t;ĽWS;m;<q<ŧ Ƃ<̜;Iq=vh*%|FպFf}54<; ˿<5ϒ<‰ü ֻD*0_\H<~d<;|' 7 廧͊;;c<]$ =:풼q/Max=g1s<{X7xY(p岟'By*b:~$=7Bļ=jkJS<;=J<}l8_:H063==B =c=D OJA})<s"K=BӼdӼybR #< <> fi,<"}eڼ3(ټ$g<=1=<6_kn$)=E<*"@E!<;eT I<'+;/<;a<|z{>Xa^< ;P_ UT S=(;=<9#<քVm<="<<<лQA'<^E&+=1-C;0p4ٻF;<1n; <=]?;z;}VP>=.;={ѻצ;$1%;}tzvc<2Rh/>;P-=W/=%&P=RȊ\1;^La<{E(>м\22,+;Ģē;xy/g|3<d45=p('(<'JԼ<==/<=P8jN<~g=<];2ݼc5aNǫ7%:ϟ)ȻbHTH<* @w/ }F<<@2g Tt%=0+:;v;j;I=;ɈY=R=o07M{g=YLe=mj<2E;UXZ;Kz}<%h<􃗻QZ=Fvf5=n=;F&=~h"Z;4tcRavkeP35(<~9i<|hB<3͐%ҼwtxY=z׻Tൻ_()<%*= %R<FW<٤8==$Fj㿼м|X=Tͼ 6<:tf<;ּ=zf޼#|M\ҽ.ܼZMɣf;Xp<y;#]t*Ҽ,$ܼeH;:II;,=沼bd2<Bɼ;;ah<ɛ=$Ax<;l<Ҳ{}29>;?[=/,;ά% I=A=W==Ct<&<@\;B0:K;`p-=ҙ=v#=ndQm's3=MH=4;P<~'=5:S/;d<.];D;Q <<q=L*< y< K =vk:(Z`Ƽ졸<{U1018~0G<eC< #;Hf;w<` V;.r<@=gN=sc=2bDR#j ;5D#w=HMሎټC G:!#B~T޸Z<8t= <;e:z6=AL< < <ҁ >=UWh7=w>;~32I;]<؆G.r<TX?2<;;c<#̼?$:9==O Q;*ς;GjټnD<'ށ;8'= 1M% ,<::e:'Z;_̻QzL<+żEv= =ަ9=6x;D;wVF=M=7ra=j5D<^(0B=\a߻8o=;= T;-ܱ<B<<=TD+p;=#<;Ǽ a<0 <,d<%6N"G:0l<<&m"䃹K]Y|/=<)q'=cu<ݾ?(g=蔼[@:9{;)Xڪ=ӳ<<: =\;𓼼CR~<(mC= <#uN v5 '=ۼ絼[;U-Dĵ5=[Z=k_FX.;G40I;͜)P=a5=r=M=z=qt; <o5<';ᔻ)!:>εk<@=<4<%9z<0B;Bьa;zA=UF\=s6 _DAC</==EC8l;ۭo<;:(=;K&pt2ڼ<U Q!= <笼r:K: =3`ڼ= <<'=<4^<0<]<.xb<{0=vC`=ء=;<4鶼zû9G=X<*FX FjV^=&!0~y=D= w=X<=r>C]=ά5=7B m<w⻼㜼 Eb=ӿ󤼥A<< MSga=穼<`5r=z6==8<5<Ī ;,<]3*=ɖ~<6ֻwG; =g;Z )j@u; .(;dL< ;ۭ;Mf3;~p<<"ċj<>d\<'6Y3QԼT<<<;<~t7'<&3=q<<+W&A;+J9=Kb:dZnVbX;N =.){"XP"8'y<+s=M;N;<;ֺ%=:N<=S3=By:==>=zx<ބPg=17[F<:<5Kf<{<^,=AjCd2:/S=*%P *=m|;N.;ꫪ=u<4֔;%z0/=$h;@חOR!&d.;;kB #<ݤk<<ʢKq o߼09Ϟ<eJm=0eK{:<;;I}!+<ݰ9<;Ἁ$ <˚;Q<;Qj:;iJa1U?=^z"M=Π0V%<5}X&լ 5hhp-=P- WTy,` <e仙x=غ ȼMb.=ok1n=m=-@W:Zi=pּB<Ϯ<wҼeѪ>;<#;==kH ,hS y+7<[<;ma߿=Cr*j~<׾p=搼-R> =2=U8=qsk==, = Iv<LT5<8ɻ3;LӪ<ϴ<5< RS>p0<+ =IWͼ.=%<p=E=oW?=/{<3;ys; L@HS;v &e,<*<ƀ<U<2[%!< &=pH < hn0;v`;un<ZL?<a=%y;zY;ś^=6P=6*<Ј;ݎμ)x<;;}$;I_z9<^Vʼz<Γ2 " Ĺ=cʻ ?#^xtYX74V ^B0{<꯼%X<Ϭ.m<.(󟼃: ?cY&=]|=q3%դ<< <ЇLI<"*: g&=7 =T~ͺ뱪Lh< I(;V;E&^<gHw<[8O1 < .;Q|=፼W('W< =A@b=,jEnE=Ѯ'Wv;ނIZ ѻԼk@6/%S^<*8=>x;5Yr3<3:;< E<-W9&Ӻ; ;;汥= <)<#;=+^_:5<*<4`)Tu2r໎F=u1<d<w-9_伊0Me6~K;ω=j0=+X=S1V:;EU㹎P<6Prt<ʲ<(-缴¼)@[<B^ -?*@g.4<$$OƼ&=;<3=({]>=V:c0z%Q<&Z<<E=z;].=|üL"ü(i<=y"3ݻM5w=;= c=輀<B";?]]=ׁ;+l=,=ԕw=f?px};ܻi )6/i)oh9RV; ֻä;&: ?3)FȼrOwE&fMi=[[ ռ=׼b=>RK<; ;\;.: ;F;m<>#❮<ώ>偽oVP= <‡<4; =0B<@@=Fa<ػּْR< :xF 0ܠ&'=a$t<<ط¼P_<*#=<eج>s*C`<GgavKWu=IU܆<};e=Ɨ%;`˱;<ő19SR <:R=R9= -]<𮥻az;0ݛ|gc`ż1==R<껅mh=~(:<+Ȼ'tvk=7^j<.ͻuN=><ф9=;q<"<*ּ<,ؼcHӝGo<⹿U: ˼1o4:7($ٞ<[=MA<>ûÁ=W_=;4h=BtN =%ZۻԠ9S=Nݲp<V<<˞dn2ؓ =$k;)<ϰ[ ۻ;2<f;%<<|=+%=?ºiN= "~ϱ2u<6H=$X{FLhKATU< < <èA(;w<;F!=D$2=;v^99C{<=fܰj<56=FO=<0L';<| Ӈw==;>:<ӭoKFh;n;\` dZ;; oip< 4<זɼu;6=< !%P=+/d&|=`EqV <<; 54Ӑv;Ξ=`Ix;=N<<)N9C֩K<잎<9)=kD=v;/<+l=`0Ls';˗ƟĻbtN'9W[}U=#;@;!`=nmOUFÀJԻ8=޹F V2c<8-eqt=<\=P6a;m=}Y =dVQEtZ=\Xk=Ej:UZN(L= ^^=׵<1ڼKP2=4J=P5v>0o<(;==y= ̼JtӞb:=;K>c<ǧ<2 B̻aº^=\b;ă !=Gꣻ<ё)t`ռ= 7߼ֻ\ <ټ2e.=n%=51!HѼVo<ꬼ@=&.<t^=2<,'_gq:<r?Z<#t<=6< )< X]7=2H=049Z_C<<>c< YvFb=o$ 8ҧ,=#;b<*r:=[=@<)O)=d<SW@#żzûg7=[>=r<[H"A-8<(<<Ӛ#< 9<<\=/v<_鞼z]=:AJ0S;N޼*':>Cc89w=[Ҽ߼7¦;; <;20<'<9m=o[ƻu0 <* =S=qO-] ; ư;zG<< <ȍ< 8=U;ȍ7C)=$)<>/̹:ӷ.=;82>==<7WT=DU8=-=丼m=<0=!m<'P[;E6#<{~<+A=L »μ0<=9==vUϼӬWqp<{["ruC<Շmm)WVWs<@E뺼<򸡼}bĹ9/;\<|=Pc=Lk)<