1. Introduction1. Introduction1. Introduction2. Literature review3. Data and methodology3.1. Data3.1. Data3.1. Data3.1. Datatable_1 Gross Domestic Product (GDP) (cyclical component): The gross domestic product in constant local currency units were retrieved from the World Bank Open Data [27]. Then we applied the Hamilton [28] filter, with four lags and the differentiation parameter equal to 2, following the author's recommendation, and computed the cyclical component of GDP. An above-trend growth may generate a positive investor sentiment that raises the investment in renewable sources. On the other hand, it could increase the expectation of a reversal in the growth trend that causes renewable energy investors to postpone their projects. Thus, the expected effect of this variable on renewable energy installed capacity is ambiguous. Moreover, in this empirical investigation, we renamed the variable as (GDP_Cyc).table_3table_2(grant, loan, insurance, credit line, private development finance, equity investment, concessional loan, guarantee, and other official flows). Indeed, we define, as our explanatory variable, in the main regression whose purpose is to assess the evolution of all the non-hydroelectrical installed capacity, the total amount of finance flows, in constant 2010 United States Dollar (USD), directed at the promotion of renewable energy, except hydro- electrical, regardless of the instrument type. In this study, we named this variable as (REFF). We also build three other vari- ables that include the finance flows directed, specifically, at the promotion of wind energy (WFF), solar energy (SFF), and bio- energy (BFF). International finance flows render renewable in- vestment more affordable. They may foster their development, particularly in LAC countries, most of which do not have a suf- ficiently robust economy to support these investments on their own.(grant, loan, insurance, credit line, private development finance, equity investment, concessional loan, guarantee, and other official flows). Indeed, we define, as our explanatory variable, in the main regression whose purpose is to assess the evolution of all the non-hydroelectrical installed capacity, the total amount of finance flows, in constant 2010 United States Dollar (USD), directed at the promotion of renewable energy, except hydro- electrical, regardless of the instrument type. In this study, we named this variable as (REFF). We also build three other vari- ables that include the finance flows directed, specifically, at the promotion of wind energy (WFF), solar energy (SFF), and bio- energy (BFF). International finance flows render renewable in- vestment more affordable. They may foster their development, particularly in LAC countries, most of which do not have a suf- ficiently robust economy to support these investments on their own.3.2. MethodologyTable 1Table 1 Descriptive statistics of variables.Table 2Table 2Table 2Table 2Notes: This table was built using the Stata function multipurt. The null hypothesis for CIPS stipulates that the series have a unit root. ***, **, and * denote a statistically significant rejection of the null hypothesis at the 1 %, 5 %, and 10 % levels, respectively.we check the robustness of our results to the presence of outliers using dummy variables.4. Empirical results4. Empirical resultsTable 4Notes: The estimations were performed using the R code QRPanel.R; ***, **, and * denote a statistically significant rejection of the null hypothesis at the 1 %, 5 %, and 10 % levels, respectively.N. Silva, J.A. Fuinhas and M. KoengkanTable 5Table 6Table 7Table 8 E i iTable 8 Estimation results (with dummies) - DRen.N. Silva, J.A. Fuinhas and M. KoengkanTable 9 Estimation results (with dummies) - DWind.Table 9Table 10Table 11Table 11 Estimation results (with dummies) - DBio.Estimation results (with dummies) - DBio.5. Discussion5. Discussion6. Conclusion6. Conclusion6. ConclusionAuthors contributionsDeclaration of competing interestAcknowledgementsReferences-AXajs|wB R Y   / 6 !(/6Ct>KXerWhat is the novelty of our research? Did it fill a gap in the literature? Our approach was developed to handle the initial stages of non-hydroelectrical renewable installed capacity take-off in LAC countries. To measure new renewable energy sources' growth was used the yearly change of Gigawatts (GW) of non-hydroelectrical installed renewable capacity, and this measure was used as our explained variable. Indeed, the coherence of evolution among the new renewables is necessary to achieve an in-depth vision of the drivers of the evolution of the yearly change. The option to measure the evolution of installed capacity through the yearly change in GW, instead of the most used in literature, the growth rate, is condi- tional in it may provide more accurate insights about the causal impact of factors that promote renewable energy sources. The use of growth rates has unwanted side effects resulting from high growth rates in the first years of the sample because the initial installed capacity was relatively low in most LAC countries. We also fill the literature gap about non-hydroelectrical renewable installed capacity drivers in its first stages in LAC countries. The econometric analysis of the initial phase of the development of new renewables is essentially an absence in the literature that needs to be fulfilled. The panorama of promoting RE in Latin American (LA) countries, particularly solar panels, and the establishment of mini-grids, re- quires policymakers to pay attention to areas that do not achieve economies of scale that turn them profitable [3]. Banal-Esta~nol et al. [3] pinpoint that LA countries also have adopted universal service policies to allow low-income households to connect to grids. They also stress the value of establishing strong links between stake- holders to care for the launch of new technologies. The promotion of renewables in LA countries also has was based on measures that include tax incentives. These tax incentives workout mostly through exemptions in income tax and through sales tax/value-added taxes or via tariffs [4].Washburn and Pablo- Romero [4] pinpoint that most LA countries are using active policies to promote RE and that the intensity of promoting RE is related to their RE adoption performance. They also find that they switch from a system based on feed-in tariffs to use the auction system. y What is under study? In short, the research focuses on the evolution of non-hydroelectrical renewable installed capacity from 2001 to 2017 in 19 countries from the LAC region. The deployment of new renewables has several characteristics that point to nonlinear relationships with its drivers. Indeed, by one hand, it is expected that any process of take-off will behave in a nonlinear way with its determinants. On the other hand, new renewables were under fast technological evolution and achieving maturity or opening a new phase, for example, as energy sources like photo- voltaic become market competitive with fossil sources. Techno- logical innovations have been a significant source of change in the process toward achieving energy efficiency. Technological innova- tion has also led to economic structural changes, mainly in the production's process that influences the renewable energy market competitiveness. As the political option was more often than not to proceed with renewable energy deployment in a market-oriented environment rather than through a public incumbent, renewable energy sources' competitiveness plays a significant role. This role evolves with the deployment of new renewables themselves in a possible nonlinear way. In accordance, the study makes use of the quantile regressions econometric technique. This technique is prepared to handle the heterogeneity that is present at different quantiles of the distribution. One of the advantages of quantile estimation is that it does not require that the variables follow a normal distribution. Kruckenberg [5] pinpoint that international programmes directed to facilitate the adoption of RE technologies among less developed countries have achieved a mixed level of success. The author identifies partnerships as the dominant RE technologies programme. These programmes involve local, national, and inter- national organisations.Nevertheless, Kruckenberg [5] stress that the attention should be put on key actors' relationships more than on barriers and drivers to develop RE. Kruckenberg [6] stress that North-South partnerships for sus- tainable energy face numerous knowledge challenges in imple- menting off-grid RE technologies. The author concludes that the exchange of knowledgeepower in relationships involving sustain- able energy is vital in establishing off-grid RE technologies. Kim and Park [7] analysed the Clean Development Mechanism, created by the Kyoto Protocol to promote low-carbon development, in a panel of 64 host countries from 2001 to 2014. They found that the adoption of RE is conditional on the level of financial devel- opment of countries. More precisely, they found that the Clean Development Mechanism's effect in implementing RE is much more forceful in countries with undeveloped financial markets. The authors also conclude that promoting that kind of RE projects is beneficial to countries with underdeveloped financial markets. Renewable energy sources in developing countries, with unso- phisticated financial markets, face, as a rule, scarcity of financing through debt and equity [7]. Moreover, Zeng et al. [8] complement that in developing countries such as Brazil, Russia, India, China, and South Africa, the industry funds play an important role in renew- able energy projects, where in the early stage of renewable energy development, their financing models still need to mature. In developing countries, most renewable energy projects raise funds The article is organised as follows. Section 2 analyses the liter- ature. Section 3 reveal the data and method used in the research. Section 4 show empirical results. Section 5 discusses the empirical results. Section 6 concludes the research and present policy recommendations.2. Literature review Since the note On the Relationship Between Energy and GNP [1], the search for the link between energy consumption and economic growth has been an active topic in the field of energy economics. This research has evolved first mostly under the designation of N. Silva, J.A. Fuinhas and M. Koengkan Energy 237 (2021) 121611 from debt; these funds are provided by banks and related in- stitutions, which offer developers very low-interest loans. The use of RE technologies is severely constrained in countries with limited access to external financing [9]. of wind and solar sources of energy. They conclude for higher variability for wind than solar power generation. In this context, Brazil is well-positioned to play a significant role in renewable energy integration in LA countries. CO2 emissions growth has been connected consistently with evolution through the time of several variables. Among these de- terminants of CO2 emissions, the literature identifies some drivers, such as energy consumption, globalisation, urbanisation, trade openness (e.g. Refs. [10e13], or even more unexpected like obesity and overweight [14] or renewable energy consumption and out- door air pollution death rate [15].3. Data and methodology This section's main objective is to evidence, clearly and briefly, the data/variables, the group of countries, and the methodological approach that will be used in our experimental study.3.1. Data The necessity of an initial phase of mitigating and in a second one that achieves carbon neutrality had put pressure on govern- ments to act and citizens to behave in ways that limit CO2 emis- sions. For example, Bersalli et al. [16] support that global decarbonisation strategies are based on electricity generation development built on renewable energy technologies. This study focuses on the evolution of non-hydroelectrical renewable installed capacity, between 2001 and 2017, in 19 coun- tries from the Latin America and the Caribbean (LAC) region, such as Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Dominican Republic, Ecuador, El Salvador, Guatemala, Haiti, Honduras, Ja- maica, Mexico, Nicaragua, Panama, Paraguay, Peru, and Uruguay. Indeed, the choice of this group of countries is motivated by several reasons: (a) they possess exceptional natural conditions to produce energy from renewable sources; (b) they experienced rapid and steady growth in non-hydroelectrical installed capacity since the beginning of the new millennium; (c) data on the evolution of renewable installed capacity and its drivers are available for all these countries. One of the most used vehicles to stimulate investment in renewable energies have been public policies. Bersalli et al. [16] indicate that public policies to boost renewable energies sources began in 1980 in developed economies and spread to emerging countries in the 2000s. For example, these authors found that policy's particularities act to turn auction the elected instrument to promote de diffusion of renewable energies in LA countries. Bradshaw [17] analysed market-oriented regulations in Brazil and concluded that reforms in the power sector have begun in the 1990s. The author underlines that Brazil's political and institutional situation act to design a regulatory reform that went beyond the traditional approach centred on efficiency and competition. That include social and environmental concerns in the planning of the energy sector. Indeed, Brazilian regulators followed creative man- dates in promoting renewable energy use. Bradshaw also shows that the evolution of wind and solar shares in energy supply was conditioned by the dominance of hydropower infrastructure in Brazil.This paper aims to assess the evolution of new renewable en- ergy sources in the LAC countries. However, our analysis excluded hydroelectrical power because most of its capacity was installed in the last century before our dataset starts. We measure the growth of new renewable energy sources through the yearly change in Gigawatts (GW) of non-hydroelectrical installed renewable capac- ity (DRen) and use it as our main dependent variable. We also assess, alternatively, the determinants of the evolution of the yearly change in Gigawatts (GW) of the major components of non- hydroelectrical installed capacity, namely: wind (DWind), solar (DSolar), and bioenergy (DBio). All this data was retrieved from Ref. [22]. Our choice of measuring the evolution of installed ca- pacity through the yearly change in GW instead of using the growth rate may provide more accurate insights into the causal impact of factors that promote renewable energy sources. The latter option would produce unreasonably high growth rates in the first years of the sample because the initial installed capacity was fairly low in most countries. Indeed, when computing growth rates, we have very high values in the initial stages, but these values are only very high because they are relative (rate) values. For example, if the initial value is zero, any addition, however insignificant, produces an infinite rate of change. Our approach is, in these cases, more suitable than traditional approaches. Furthermore, several coun- tries have no non-hydroelectrical installed capacity at the begin- ning of the sample, rendering the computation of growth rates impossible. Fig. 1 below shows that installed capacity from non- hydroelectrical renewable sources increased more than tenfold between 2000 and 2017. Alvarado et al. [18] found that output causes renewable energy in LAC countries. This causality was independent of countries that were high-income or medium-income. Renewable energy growth was also seen as a current energy policy trend (e.g., Ref. [19]. Indeed, renewable energy analysis entails economic, social, politi- cal, and technical dimensions. Bersalli et al.[16] analyse the public policies promoting renewable energy technologies for electricity generation in a panel of LA countries and a European country panel. They conclude that LA countries have been proactive in their tentative to implement renewable energy. These authors emphasise promoting public policies to increase renewable energy investment. Another finding of these authors' research is that tax incentives are insufficient to deploy renewable energy technologies. LAC countries are mostly rich in conventional sources of energy. The transition of fossil sources to renewable ones are exposed to prices of exportation of fossil energy. The technology of extraction of non-renewable energy has evolved to allow the exploitation of these reserves abundantly. Cordano and Zellou [20] dissect the possible consequences of super-cycles in natural gas prices on LA countries' environment. They focus on the use of hydraulic frac- turing on underground water. They conclude that the mix of con- ditions prevalent in LA countries, such as political instability, weak institutions, governmental interventionism, and Brazilian pre-salt endowment, turn the impact of gas-price super-cycle uncertain. Indeed, to realize the following investigation, we consider several explanatory variables that are expected to have an impact on renewable energy investments, such as:  Renewable Energy Finance Flows: IRENA [22] provides a database of renewable energy projects that received interna- tional financial institutions' support. This database includes in- formation about the projects' recipient country, supporting institution, technology (hydropower, geothermal energy, bio- energy, solar energy, wind energy, marine energy, multiple re- newables, and other renewable energy), and instrument type Viviescas et al. [21] focused their analyses on the impact of renewable energy on energy security in LA countries. They look to the intermittency of renewable energy sources and inquire how to mitigate it by exploring energy sources' complementarities. Their study also probes the impact of climate changes on the interaction N. Silva, J.A. Fuinhas and M. Koengkan Energy 237 (2021) 121611  Carbon dioxide emissions: This variable was retrieved from the World Bank Open Data [27] and provided a database of the amount of CO2 emissions in Kilotons per capita.Indeed, CO2 is correlated with other pollutants' emissions, and it may be used as a proxy for air pollution. Thus, higher values of this variable are expected to raise public awareness about the environment and increase the political pressure to promote the deployment of renewable energy sources. Moreover, in this investigation, we renamed the variable as (CO2).table_1 Gross Domestic Product (GDP) (cyclical component): The gross domestic product in constant local currency units were retrieved from the World Bank Open Data [27]. Then we applied the Hamilton [28] filter, with four lags and the differentiation parameter equal to 2, following the author's recommendation, and computed the cyclical component of GDP. An above-trend growth may generate a positive investor sentiment that raises the investment in renewable sources. On the other hand, it could increase the expectation of a reversal in the growth trend that causes renewable energy investors to postpone their projects. Thus, the expected effect of this variable on renewable energy installed capacity is ambiguous. Moreover, in this empirical investigation, we renamed the variable as (GDP_Cyc).table_3table_2(grant, loan, insurance, credit line, private development finance, equity investment, concessional loan, guarantee, and other official flows). Indeed, we define, as our explanatory variable, in the main regression whose purpose is to assess the evolution of all the non-hydroelectrical installed capacity, the total amount of finance flows, in constant 2010 United States Dollar (USD), directed at the promotion of renewable energy, except hydro- electrical, regardless of the instrument type. In this study, we named this variable as (REFF). We also build three other vari- ables that include the finance flows directed, specifically, at the promotion of wind energy (WFF), solar energy (SFF), and bio- energy (BFF). International finance flows render renewable in- vestment more affordable. They may foster their development, particularly in LAC countries, most of which do not have a suf- ficiently robust economy to support these investments on their own. In this investigation, we converted the variables DRen, DWind, DSolar, DBio, REFF, WFF, SFF, and BFF into per capita values through their division by the population of each country. Indeed, per capita values allow us to reduce the disparities between the variables caused by population growth over time. Certainly, the descriptive statistics of all variables that will be used in this empirical investigation are shown in Table 1 below. Indeed, before the execution of the main regression model, we applied the Shapiro-Wilk test to verify the presence of normality in the model. The null hypothesis of this test is that the variables follow a normal distribution. Table 2 below displays the results from the Shapiro-Wilk test for normality test. As can be observed in the table above, the p-values of this test lead to a clear rejection of the null hypothesis that the variables follow a normal distribution in all cases. Thus, traditional panel regression methods which rely on the normality of the variables are inadequate and may generate misleading conclusions. In this sub- section, we approached the group of countries and the variables that will be used in our empirical study. In the next section, we briefly show the methodological approach that this investigation will use. Financial Development Index: This variable was retrieved from the International Monetary Fund (IMF) Financial Development Index [23] (2020) and provides a development index of financial institutions and financial services, and efficiency (the ability of institutions to provide financial services at low cost and with sustainable revenues and the level of activity of capital markets. Indeed, as we already know, electricity production from renewable energy sources, such as wind and solar, requires a large upfront investment but has lower running costs compared to fossil fuels. Thus, an accessible and efficient financial system is fundamental to support these projects and make them prof- itable for the investor. We measure the LAC countries' financial systems' soundness through the IMF Financial Development Index [24] and it is expected a positive relationship between this indicator and the dependent variable. Moreover, in this study, we renamed the variable as (FDI).3.2. Methodology As mentioned before, this subsection will approach theTable 1Table 1 Descriptive statistics of variables. Table 1 Descriptive statistics of variables. Variables Descriptive statistics Obs Mean Std. dev Min Max DRen 323 4.8966 15.5919 2.0529 131.6383 DWind 323 2.7228 12.4824 0.2345 124.0724 DSolar 323 1.0456 4.8523 0.8106 44.9566 DBio 323 1.3422 5.2247 25.5991 51.9247 REFF 323 3.5667 10.2401 0.000 70.6254 FDI 323 24.6110 12.2706 6.2964 63.3827 CO2 323 1.9575 1.2129 0.1584 4.6916 EGI 323 50.6871 12.8314 22.0174 83.4947 GDP_Cyc 323 1.0456 4.4273 19.6317 13.8450 Notes: Obs denotes the number of observations; Std. dev denotes standard devia- tion; Min denotes minimum, and Max denotes maximum. This table was built using the Stata function summarize.  Economic Globalisation Index De facto: This variable was retrieved from KOF Swiss Economics Institute [25] and was developed by Gygli et al. [26]. The index contemplates two sub- dimensions: (i) trade globalisation, which assesses the impor- tance of a country's trade in goods and services relative to GDP, and the diversity of trading partners; (ii) financial globalisation, which is measured by capital flows and stocks of foreign assets and liabilities. As already know, an economically globalised country may find it easier to import state-of-the-art technology designed to produce energy from new renewable sources. Thus, we expect economic globalisation to have a positive effect on the dependent variable. Furthermore, in this analysis, we renamed the variable as (EG). N. Silva, J.A. Fuinhas and M. Koengkan Energy 237 (2021) 121611 (2) Estimate the covariates' coefficients through the solution of the following problem, Table 2 Shapiro-Wilk test for normality. Variables Obs W V Z Prob > z DRen 323 0.3327 151.794 11.831 0.00000 DWind 323 0.3090 157.192 11.914 0.00000 DSolar 323 0.2717 165.685 12.038 0.00000 DBio 323 0.4148 133.142 11.523 0.00000 REFF 323 0.4215 131.600 11.495 0.00000 FDI 323 0.9175 18.770 6.907 0.00000 CO2 323 0.9059 21.404 7.217 0.00000 EGI 323 0.9752 5.649 4.079 0.00002 GDP_Cyc 323 0.9782 4.960 3.772 0.00008 Notes: This table was built using the Stata function swilk.Table 2 where d DReni;t≡Reni;t  bai, T and n denote the number of years and countries included in the estimation, respectively, and rt is the check function for quantile t. Canay [31] shows that the resulting estimates are consistent and asymptotically normal. We also evaluate the determinants of the evolution of wind, solar, and bioenergy installed capacities, with the dual purpose of testing the robustness of our results regarding the main dependent variable (DRen) and assessing the possible differential impact of the covariates on the individual components of renewable energy installed capacity. Specifically, we model the relation between the covariates and the various sources of renewable through the following equations. methodology that will be used in this empirical investigation. Indeed, to assess the effect of the explanatory variables on the change of non-hydroelectrical renewable installed capacity, this investigation will use the panel quantile regression model. This method presents several advantages relative to traditional least- square methods. It provides a complete picture of the indepen- dent variables' impact over all the dependent variable distribution, whereas least-square methods focus only on its conditional mean. Furthermore, quantile regression is broadly insensitive to outliers and, unlike least-square methods, does not require data to be normally distributed [29,30]. This characteristic is especially important for this study, given that our data is strongly non-normal (see Table 2 above). variable (DRen) and assessing the possible differential impact of the covariates on the individual components of renewable energy installed capacity. Specifically, we model the relation between the covariates and the various sources of renewable through the following equations.We model the relation between the change in renewable energy installed capacity and the explanatory variables through the following Equation, where i ¼ 1, …, 19 identifies the country, t ¼ 2001, …2017, corre- sponds to the observation year, XW0 i;t1 ¼ ð1; WFFi;t1; FDIi;t1; CO2i;t1; EGIi;t1; GDP Cyci;t1; TREND Þ, XS0 i;t1 ¼ ð1; SFFi;t1; FDIi;t1; CO2i;t1; EGIi;t1; GDP Cyci;t1; TREND Þ, XB0 i;t1 ¼ ð1; BFFi;t1; FDIi;t1; CO2i;t1; EGIi;t1; GDP Cyci;t1; TREND Þ, are the explanatory vari- ables vectors for the wind, solar and bioenergy equations, bWðUW i;t Þ ¼ ðbW 0 ; …; bW 6 Þ, bSðUS i;tÞ ¼ ðbS 0; …; bS 6Þ, bBðUB i;tÞ ¼ ðbB 0; …; bB 6Þ are the corresponding coefficients vectors, aW i , aS i , aB i are the unobserved fixed effects, and UW i;t  Uð0; 1Þ, US i;t  Uð0; 1Þ, UB i;t  Uð0; 1Þ are uniformly distributed random var- iables. Equations (4)e(6) are estimated through the same proced- ure described above for the main dependent variable, DRen. where i ¼ 1, …, 19 identifies the country, t ¼ 2001, …2017, corre- sponds to the observation year, X0 i;t1 ¼ ð1; REFFi;t1; FDIi;t1; CO2i;t1; EGIi;t1; GDP Cyci;t1; TREND Þ is the explanatory variables vector, bðUi;tÞ ¼ ðb0; …; b6Þ is the coefficients vector, ai is an unobserved fixed effect, Ui;t  Uð0; 1Þ, and TREND represents a time trend that may be seen as a proxy for the reduction in the levelized cost of renewable en- ergy over time. Indeed, in this investigation, we choose to lag the explanatory variables by one year because it takes a substantial amount of time for renewable energy investments to materialize in new installed capacity.where i ¼ 1, …, 19 identifies the country, t ¼ 2001, …2017, corre- sponds to the observation year, X0 i;t1 ¼ ð1; REFFi;t1; FDIi;t1; CO2i;t1; EGIi;t1; GDP Cyci;t1; TREND Þ is the explanatory variables vector, bðUi;tÞ ¼ ðb0; …; b6Þ is the coefficients vector, ai is an unobserved fixed effect, Ui;t  Uð0; 1Þ, and TREND represents a time trend that may be seen f h d i i h l li d f bl All the estimations were executed in the software R, using the QRPanel.R code provided by Canay. Fig. 2 shows the methodological steps we follow in this research. First, we present descriptive statistics for all the variables and perform the Shapiro-Wilk and Pesaran tests to check their normality and stationarity, respectively. Then, we assess the impact of the covariates on the dependent variables through the estima- tion of panel quantile regressions, using Canay's method. Finally, The fixed effect's unobservability renders the traditional quan- tile regression unable to identify the covariates' coefficients because ai can be arbitrarily related to the remaining variables. To circumvent this problem, we follow Canay [31]; who proposed a simple two-stage method that achieves identification, assuming that the fixed effect is a pure location shift, and ai and Ui;t are in- dependent. Let ui;t≡X0i;t½bðUi;tÞ  bm, and bm ¼ ½b0;m… b6;m , the conditional mean of bðUi;tÞ. Then, using Equation (1) we have, The estimator proposed by Canay [31]; for quantile t, involves two steps: (1) Obtain a consistent estimator1 of bm from Equation (2) and define bai≡½Reni;t  X0i;tbbm 1 We use the fixed effects method. N. Silva, J.A. Fuinhas and M. Koengkan Energy 237 (2021) 121611 Table 3 Panel unit root test (CIPS test).Variables Panel unit root test (CIPS) (Zt-bar) Without trend With trend Lags Zt-bar Zt-bar DRen 0 8.907 *** 8.131 *** DWind 0 6.237 *** 5.671 *** DSolar 0 2.815 *** 1.973 ** DBio 0 4.909 *** 4.460 *** REFF 0 4.480 *** 2.426 *** FDI 0 3.245 *** 2.964 *** CO2 0 0.842 ** 0.633 EGI 0 0.637 1.257 GDP_Cyc 0 4.521 *** 2.871 *** Notes: This table was built using the Stata function multipurt. The null hypothesis for CIPS stipulates that the series have a unit root. ***, **, and * denote a statistically significant rejection of the null hypothesis at the 1 %, 5 %, and 10 % levels, respectively. crucial role in promoting renewable energy in LAC countries, particularly in large projects. The estimates for CO2 emissions are positive, sizable, and broadly similar for all the quantiles. This ev- idence is consistent with the hypothesis that CO2 emissions generate political pressure that fosters renewable energy devel- opment. The dependent variable also bears a significant positive relation with economic globalisation for all the quantiles. Thus, countries with an economically open economy are more prone to develop new renewable energy sources. The time trend, which may be interpreted as a proxy for reducing the levelized cost of renewable energy over time, also positively affects the dependent variable in the lowest and highest quantiles. Finally, neither the financial development index nor cyclical GDP seems to be associ- ated with renewable energy development.Notes: This table was built using the Stata function multipurt. The null hypothesis for CIPS stipulates that the series have a unit root. ***, **, and * denote a statistically significant rejection of the null hypothesis at the 1 %, 5 %, and 10 % levels, respectively. The results for the wind installed capacity determinants' (Table 5) is strongly consistent with the main estimation (DRen) about the impact of finance flows, CO2 emissions, and the economic globalisation index: all these covariates exert a positive effect on installed capacity and the impact of financial flows is particularly prevalent at the highest quantiles. However, unlike in the (DRen) estimation, the financial development index negatively affects wind installed capacity.we check the robustness of our results to the presence of outliers using dummy variables. Table 6, which displays the estimated coefficients for the solar energy equation, corroborates our previous findings regarding the importance of finance flows, CO2 emissions, and time for the pro- motion of renewable energy sources, but it shows that the financial development index and the economic globalisation index may hinder the increase in solar installed capacity.4. Empirical results As mentioned previously in the introduction of this investiga- tion, this section will present the quantile regressions' estimation results, using the Canay [31] method. In this investigation, we choose to estimate the effect of the covariates on the change in non- hydroelectrical renewable installed capacity for the 25 %, 50 %, and 75 % quantiles and the 90 % quantile because we want to find the main drivers of the substantial increases in renewable installed capacity in the recent years. The estimation for bioenergy (Table 7) reveals that CO2 emis- sions, the economic globalisation index, the financial development index, and finance flows (only in the 75th quantile) foster the in- crease in bioenergy installed capacity. However, unlike in the pre- vious estimations for other renewable sources, time appears to have a detrimental effect on bioenergy. Before estimating the quantile regressions, we perform the Pesaran [32] panel unit root test to assess the series stationarity. This test, whose null hypothesis assumes that the variable is non- stationary, follows a non-standard distribution, even when the number of members in the sample is large. It is based on the averaging of panel-member specific Dickey-Fuller type regressions' coefficients and is robust in the presence of cross-section depen- dence. Indeed, Table 3 below shows that the null non-stationarity hypothesis is strongly rejected for all the dependent variables in both specifications. This hypothesis is also rejected for most explanatory variables, except CO2 emissions, in the test with a time trend and the economic globalisation Index in both tests. It is well known that in a relatively short panel, such as ours, the estimation results may be driven by a few extreme observations, which distort the causal link between the covariates and the dependent variable and leads the researcher to draw wrong con- clusions. To check our results' robustness to the presence of out- liers, we identified as outliers those observations whose residuals are more than three standard deviations away from the mean. Then, we created a dummy variable for each one of these obser- vations and re-estimated all the models.These exercise results are presented in Table 8, 9, 10, and 11 below. The comparison of Tables 4 and 8 reveals that the results remain broadly unchanged when dummies are included in the estimation of the evolution of non-hydroelectrical installed capacity: finance flows, CO2 emissions, and the economic globalisation index remain its main drivers, while the time trend impacts it only in the 25th Table 4 below shows the estimated coefficients for the change in the non-hydroelectrical installed capacity equation. The effect of Finance Flows on renewable energy development is positive and statistically significant for all the quantiles, except the fiftieth. Its impact is increasing across the quantiles, suggesting that it plays a Table 4 Estimation results - DRen. Independent variables Dependent variable (DRen) Quantiles 25th 50th 75th 90th REFF 0.1557 *** 0.3520 1.3034 ** 1.8932 *** FDI 0.0330 0.0221 0.0235 0.0612 CO2 2.6207 *** 2.3734 *** 2.3609 *** 2.8066 *** EGI 0.1422 *** 0.1514 *** 0.1839 *** 0.2173 *** GDP_Cyc 0.0255 0.0557 0.0863 0.1689 TREND 0.0885 ** 0.0913 0.0789 0.2654 * Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; ***, **, and * denote a statistically significant rejection of the null hypothesis at the 1 %, 5 %, and 10 % levels, respectively.Table 4 using the R code QRPanel.R; ***, **, and * denote a statistically significant rejection of the null hypothesis at the 1 %, 5 %, and 10 % Notes: The estimations were performed using the R code QRPanel.R; ***, **, and * denote a statistically significant rejection of the evels, respectively. tions were performed using the R code QRPanel.R; ***, **, and * denote a statistically significant rejection of the null hypothesis aNotes: The estimations were performed using the R code QRPanel.R; ***, **, and * denote a statistically significant rejection of the null hypothesis at the 1 %, 5 %, and 10 % levels, respectively. Energy 237 (2021) 121611N. Silva, J.A. Fuinhas and M. Koengkan Table 5 Estimation results - DWind. Independent variables Dependent variable (DWind) Quantiles 25th 50th 75th 90th WFF 0.7788 *** 1.5451 *** 2.3467 *** 3.6648 *** FDI 0.0709 *** 0.0889 *** 0.0487 *** 0.0479 *** CO2 2.0195 *** 2.4668 *** 2.4109 *** 2.4716 *** EGI 0.0835 *** 0.0942 *** 0.1136 *** 0.1288 *** GDP_Cyc 0.0183 0.0055 0.0089 0.0349 * TREND 0.0052 0.0105 0.0259 0.2457 *** Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** and * denote a statistically significant rejection of the null hypothesis at the 1 % and 10 % levels, respectively.Table 5 25th 50th 75th 90th WFF 0.7788 *** 1.5451 *** 2.3467 *** 3.6648 *** FDI 0.0709 *** 0.0889 *** 0.0487 *** 0.0479 *** CO2 2.0195 *** 2.4668 *** 2.4109 *** 2.4716 *** EGI 0.0835 *** 0.0942 *** 0.1136 *** 0.1288 *** GDP_Cyc 0.0183 0.0055 0.0089 0.0349 * TREND 0.0052 0.0105 0.0259 0.2457 *** Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** and * denote a statistically significant rejection of the null hypothesis at the 1 % and 10 % levels, respectively. Table 6 Estimation results - DSolar. Independent variables Dependent variable (DSolar) Quantiles 25th 50th 75th 90th SFF 0.0354 *** 0.2111 *** 0.2727 1.4467 *** FDI 0.2598 *** 0.2477 *** 0.2453 *** 0.2339 *** CO2 0.7301 *** 0.7918 *** 0.8283 *** 0.8706 *** EGI 0.0320 *** 0.0207 *** 0.0132 0.0041 GDP_Cyc 0.0336 0.0404 * 0.0228 0.0432 TREND 0.1677 *** 0.1720 0.1662 *** 0.1554 *** Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** and * denote a statistically significant rejection of the null hypothesis at the 1 % and 10 % levels, respectively. Table 7 Estimation results - DBio. Independent variables Dependent variable (DBio) Quantiles 25th 50th 75th 90th BFF 0.0983 0.3685 0.8479 *** 0.9930 FDI 0.1421 *** 0.1473 *** 0.1406 *** 0.1406 *** CO2 1.9514 *** 1.9026 *** 1.8436 *** 1.4957 *** EGI 0.0697 *** 0.0795 *** 0.1057 *** 0.1114 *** GDP_Cyc 0.0027 0.0338 0.0487 * 0.0007 TREND 0.0844 *** 0.1030 *** 0.0858 0.0048 Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** and * denote a statistically significant rejection of the null hypothesis at the 1 % and 10 % levels, respectively.Table 6 Table 6 Estimation results - DSolar. Independent variables Dependent variable (DSolar) Quantiles 25th 50th 75th 90th SFF 0.0354 *** 0.2111 *** 0.2727 1.4467 *** FDI 0.2598 *** 0.2477 *** 0.2453 *** 0.2339 *** CO2 0.7301 *** 0.7918 *** 0.8283 *** 0.8706 *** EGI 0.0320 *** 0.0207 *** 0.0132 0.0041 GDP_Cyc 0.0336 0.0404 * 0.0228 0.0432 TREND 0.1677 *** 0.1720 0.1662 *** 0.1554 *** Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** and * denote a statistically significant rejection of the null hypothesis at the 1 % and 10 % levels, respectively. Table 7 Estimation results - DBio. Independent variables Dependent variable (DBio) Quantiles 25th 50th 75th 90th BFF 0.0983 0.3685 0.8479 *** 0.9930 FDI 0.1421 *** 0.1473 *** 0.1406 *** 0.1406 *** CO2 1.9514 *** 1.9026 *** 1.8436 *** 1.4957 *** EGI 0.0697 *** 0.0795 *** 0.1057 *** 0.1114 *** GDP_Cyc 0.0027 0.0338 0.0487 * 0.0007 TREND 0.0844 *** 0.1030 *** 0.0858 0.0048 Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** and * denote a statistically significant rejection of the null hypothesis at the 1 % and 10 % levels, respectively.Table 7 inclusion of dummies weakens the positive effect of the economic globalisation index and strengthens the time trend's impact. and 90th quantiles, in the baseline estimation, and the 75th quantile in the estimation with dummies. g g p Solar energy results seem to be the most sensitive to outliers (Tables 6 and 10). Even though the effect of finance flows remains Tables 5 and 9 confirm the positive influence of finance flows and CO2 emissions on wind energy installed capacity. However, theTable 8 E i iTable 8 Estimation results (with dummies) - DRen. Table 8 Estimation results (with dummies) - DRen. Independent variables Dependent variable (DRen) Quantiles 25th 50th 75th 90th REFF 0.1269 *** 0.3155 *** 1.4292 *** 2.3875 *** FDI 0.0107 0.0045 0.1329 *** 0.1383 *** CO2 3.3662 *** 3.1722 *** 2.7964 *** 4.0406 *** EGI 0.0416 *** 0.0527 *** 0.1251 *** 0.2379 *** GDP_Cyc 0.0006 0.0013 0.0802 0.0125 TREND 0.0565 0.0912 0.1783 *** 0.1692 Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** denotes a statistically significant rejection of the null hypothesis at the 1 % level. Energy 237 (2021) 121611N. Silva, J.A. Fuinhas and M. KoengkanTable 9 Estimation results (with dummies) - DWind.Table 9 Table 9 Estimation results (with dummies) - DWind. Independent variables Dependent variable (DWind) Quantiles 25th 50th 75th 90th WFF 0.3001 *** 0.8894 *** 2.6437 *** 4.0447 *** FDI 0.0927 *** 0.0781 *** 0.1436 *** 0.0479 *** CO2 2.0901 *** 1.9496 *** 0.6409 *** 0.9926 *** EGI 0.0045 0.0068 0.0316 *** 0.0997 *** GDP_Cyc 0.0241 0.0062 0.0271 0.0103 TREND 0.0450 *** 0.0291 ** 0.0861 *** 0.2313 ** Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** and ** denote a statistically significant rejection of the null hypothesis at the 1 % and 5 %, levels, respectively. Table 10 Estimation results (with dummies) - DSolar. Independent variables Dependent variable (DSolar) Quantiles 25th 50th 75th 90th SFF 0.0297 *** 0.2912 *** 0.4642 *** 2.5146 *** FDI 0.0521 *** 0.0707 *** 0.0570 *** 0.2140 *** CO2 0.3701 *** 0.2463 *** 0.3227 *** 0.9518 *** EGI 0.0085 * 0.0076 ** 0.0035 0.0090 GDP_Cyc 0.0101 0.0010 0.0090 0.0186 TREND 0.0573 *** 0.0618 *** 0.0566 *** 0.0055 *** Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; ***, **, and * denote a statistically significant rejection of the null hypothesis at the 1 %, 5 %, and 10 % levels, respectively.Table 10 Table 11 Estimation results (with dummies) - DBio. Independent variables Dependent variable (DBio) Quantiles 25th 50th 75th 90th BFF 0.0194 0.3471 ** 0.8316 *** 0.6750 *** FDI 0.0913 *** 0.1030 *** 0.1068 *** 0.0812 *** CO2 0.4595 *** 0.3726 *** 0.2794 *** 0.5431 ** EGI 0.0203 0.0066 0.0001 0.0078 GDP_Cyc 0.0290 0.0338 0.0074 0.0272 TREND 0.0473 *** 0.0322 ** 0.0196 0.0048 Obs 323 323 323 323 Notes: The estimations were performed using the R code QRPanel.R; *** and ** denote a statistically significant rejection of the null hypothesis at the 1 % and 5 % levels, respectively.Table 11Table 11 Estimation results (with dummies) - DBio.Estimation results (with dummies) - DBio. broadly unchanged, dummies lead to a total reversion in the co- efficients related to CO2 emissions. This instability is hardly sur- prising, given that solar energy was the newest energy source to be widely deployed in LAC countries. Thus, in the early years of our sample, most countries did not install new solar capacity, which effectively reduces our sample's information content for this en- ergy source. variables and are on the borderline between the I(0) and I(1) orders of integration (see Table 3, above). Indeed, both two results had been already confirmed by Koengkan et al. [33,34], that studied the LAC region and found the same variable characteristics in their investigations. This is pointing that the results from the preliminary tests that were found in this empirical study are in consonance with the literature review. gy Finally, Table 11 confirms the positive impact of finance flows, time, and CO2 emissions (except in the 90th quantile) on bioenergy found in Table 7. However, the inclusion of dummies leads to the disappearance of the statistical significance for the variable eco- nomic globalisation index. Moreover, the model estimations' results indicate that the in- dependent variables of renewable energy finance flows, economic globalisation index, and carbon dioxide emissions positively affect our dependent variable named change in non-hydroelectrical renewable installed capacity. Besides, Fig. 3 below summarises the effect of independent variables on the dependent one. This figure was based on the results of the Canay model estimation.5. Discussion Then, based on these empirical founds we elaborate on the following question - What are the explanations for the positive impact of these variables in our model? e Well, the positive effect of renewable energy finance flows could be related to the increase of capital inflow for finance the new renewable energy projects As mentioned before, this section will present the possible ex- planations which were found in this empirical investigation. Well, the results from the preliminary tests show the non-presence of normal distribution (see Table 2, above), the indication that the N. Silva, J.A. Fuinhas and M. Koengkan Energy 237 (2021) 121611 energy demand. However, to attend to the energy demand will be necessary more investments in alternative energy sources. More- over, this economic liberalisation process will boost the imports of renewable energy technologies to increase energy efficiency and reduce the consumption of non-renewable energy sources. Finally, the positive impact of CO2 emissions could be related to the increase of environmental, political, and social pressure about the increase of environmental degradation caused by these emis- sions' growth. Indeed, this pressure will encourage the develop- ment of renewable energy policies that will facilitate investments, development, and the consumption of renewable energy sources in the LAC countries. Another explanation is related to the capacity to consume new renewable energy sources to mitigate these CO2 emissions in the LAC region. This section showed the results from the main model and a brief explanation for this study's results. The next section will show the conclusions of this experimental investigation. caused by the restructure plans, for example, ‘Washington consensus' and ‘Brady Plan’, which were adopted to promote the restructuring of external debts and the macroeconomic adjust- ments and mentioned by Koengkan [35]; and Santiago et al. [36]. Indeed, this period of adjustments occurred between 1989 and 1992, in several LAC countries (e.g., Argentina, Brazil, Costa Rica, Mexico, Venezuela (RB), and Uruguay) adopted these strategies.The restructuring plans and macroeconomic adjustments allowed a deep financial, trade, and foreign investment liberalisation, as well as the privatisation of significant portions of the public sector and the reduction of import barriers in most countries from the LAC region. As a result, the capital inflows reappearance in the 1990s, at the middle of the decade (1994), the gross fixed capital stock per capita in the LAC region (Argentina, Brazil, Chile, Colombia, Ecuador, Mexico, and Venezuela) reached a mean value of 11,260 US$ (in 1980 international dollars) that compares with 54,089 US$ for the USA [37]. Indeed, the LAC region's capital inflows acceler- ated again between 2004 and 2014, caused by the “commodities boom” where the capital inflows registered a growth rate of 25 % during this period and reached a value of 1743.936 US$ in 2014 [35]. That is, the renewable energy finance flows followed the same trend, where the inflows of investments in new renewable energy sources (e.g., marine, wind, solar, geothermal, solid biofuels and waste and liquid biofuels) were a value of 1.6 US$ billion in 2014 and 2016 reached 8.7 US$ billion [35].6. Conclusion This analysis assessed the advancement of new renewable en- ergy sources in a group of nineteen countries from the LAC region between 2001 and 2017. This investigation is in the initial stages of maturation, where it will supply a solid foundation for second- generation researchers regarding this topic. This study is a kick- off about the advancement of new renewable energy sources in the LAC region. This empirical research has been based on eco- nomic principles to construct a model that provides an accurate explanation of the results that were found in this study. This empirical investigation utilised as a method the Canay model estimation to realize the following analysis. The results from the preliminary tests pointed to the non-presence of normal dis- tribution and that the variables are on the borderline between the I (0) and I(1) orders of integration. Indeed, both two results agree with the literature that approached this group of countries. The Canay model estimation results showed that the indepen- dent variables renewable energy finance flows, economic globali- sation index, and carbon dioxide emissions positively affect our dependent variable named change in non-hydroelectrical renew- able installed capacity. It is worth remembering that this empirical investigation opted to choose from the Canay model estimation because it can capture the variations between countries (i.e., cross- sections). That is, the results found by this empirical investigation can answer the research question that arose. Therefore, the increase of capital inflows caused by financial liberalisation will reduce the financing costs and consequently will encourage the development and investment in new renewable energy technologies. This explanation is confirmed by Narayan and Smyth [38]; and Koengkan et al. [33]; where the increase of capital inflows to invest in renewable energy projects will reduce their costs of financing, making the credit cheaper and makes the new energy sources more feasible. Moreover, another factor that could be related to this positive impact is the existence of efficient renewable energy financial policies that encourage public and private financing in new renewable energy projects.As mentioned before, the possible explanation for the positive impact of renewable energy finance flows on change in non- hydroelectrical renewable installed capacity could be related to the increase of capital inflow caused by financial liberalisation and that consequently reduced the financing costs and encouraged the development and investments in new renewable energy technol- ogies (e.g., marine, wind, solar, geothermal, solid biofuels and waste and liquid biofuels). That is, the reduction of financing cost made the new renewable energy technologies more attractive than the hydropower plants in the LAC region. Moreover, another factor could be related to this positive impact is the existence of efficient renewable energy financial policies that encouraged public and private financing in new renewable energy projects. Regarding the positive impact of economic globalisation index on change in non-hydroelectrical renewable installed capacity, also could be related to the process of deep financial, trade, and foreign investment liberalisation, as well as the privatisation of significant portions of the public sector, and the reduction of import barriers as was mentioned before. All these changes in the economic structure consequently affect the index of globalisation, where financial and trade liberalisation are the main components of this index. Indeed, as we already know, the process of financial and trade liberalisation will boost the capital inflows to realize a new investment in the productive sectors or speculative, the economic activity, and so the The positive impact of the economic globalisation index could be related to the process of deep financial, trade, and foreign in- vestment liberalisation, as well as the privatisation of significant portions of the public sector, and the reduction of import barriers caused by the restructuring plans that were adopted by some LAC countries that encouraged the economic activity and so the energy demand. However, to attend to the energy demand will be neces- sary to carry out new investments in alternative energy sources. N. Silva, J.A. Fuinhas and M. Koengkan Energy 237 (2021) 121611 developing countries. Ecol Indicat 2016;67:543e55. https://doi.org/10.1016/ j.econmod.2014.10.022.Moreover, this liberalisation process in Latin American economies also encouraged the import of new renewable energy technologies to increase productivity and reduce the consumption of non- renewable energy sources. [13] Kasman A, Duman YS. CO2 emissions, economic growth, energy consumption, trade and urbanisation in new EU member and candidate countries: a panel data analysis. Econ Modell 2015;44:97e103. https://doi.org/10.1016/ j.econmod.2014.10.022. Moreover, the positive impact of CO2 emissions on change in non-hydroelectrical renewable installed capacity could be related to the increase of environmental, political, and social pressure regarding the increase of environmental degradation caused by the growth in these emissions that encouraged the new investments in renewable energy technologies, as well as could be related to the own capacity of consumption of new renewable energy sources to mitigate these CO2 emissions in the LAC region. [14] Koengkan M, Fuinhas JA. Does the overweight epidemic cause energy con- sumption? A piece of empirical evidence from the European region. Energy 2021;216:119297. https://doi.org/10.1016/j.energy.2020.119297. [15] Koengkan M, Fuinhas JA, Silva N. Exploring the capacity of renewable energy consumption to reduce outdoor air pollution death rate in Latin America and the Caribbean region. Environ Sci Pollut Control Ser 2021;28:1656e74. https://doi.org/10.1007/s11356-020-10503-x. p // g/ / [16] Bersalli G, Menanteau P, El-Methni J. Renewable energy policy effectiveness: a panel data analysis across Europe and Latin America. Renew Sustain Energy Rev 2020;133:110351. https://doi.org/10.1016/j.rser.2020.110351. [17] Bradshaw A. Regulatory change and innovation in Latin America: the case of renewable energy in Brazil. Util Pol 2017;49:156e64. https://doi.org/10.1016/ j.jup.2017.01.006.Authors contributions [18] Alvarado R, Ponce P, Alvarado R, Ponce K, Huachizaca V, Toledo E. Sustainable and non-sustainable energy and output in Latin America: a cointegration and causality approach with panel data. Energy Strat Rev 2019;26:100369. https:// doi.org/10.1016/j.esr.2019.100369. Nuno Silva: Investigation, Methodology, Validation, Writing- Original draft preparation. Jose Alberto Fuinhas: Conceptualiza- tion, Supervision, Visualization, Writing-Reviewing and Editing. Matheus Koengkan: Data curation, Formal analysis, Investigation. [19] Xu X, Wei Z, Ji Q, Wang C, Gao G. Global renewable energy development: influencing factors, trend predictions and countermeasures. Resour Pol 2019;63:101470. https://doi.org/10.1016/j.resourpol.2019.101470.Declaration of competing interest [20] Vasquez Cordano AL, Zellou AM. Super cycles in natural gas prices and their impact on Latin American energy and environmental policies. Resour Pol 2020;65:101513. https://doi.org/10.1016/j.resourpol.2019.101513. 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. [21] Viviescas C, Lima L, Diuana FA, Vasquez E, Ludovique C, Silva GN, Huback V, Magalar L, Szklo A, Lucena AFP, Schaeffer R, Paredes JR. Contribution of Var- iable Renewable Energy to increase energy security in Latin America: complementarity and climate change impacts on wind and solar resources. Renew Sustain Energy Rev 2019;113:109232. https://doi.org/10.1016/ j.rser.2019.06.039.Acknowledgements [22] IRENA. International Renewable Energy Agency; 2020. [23] International Monetary Fund (IMF). Financial development index database. URL: https://data.imf.org/?sk¼f8032e80-b36c-43b1-ac26-493c5b1cd33b; 2020. CeBER R&D unit funded by national funds through FCT e Fundaç~ao para a Ci^encia e a Tecnologia, I.P., project UIDB/05037/ 2020. [24] Svirydzenka K. Introducing a new broad-based index of financial develop- ment. IMF Working Papers 2016;16(5):1. https://doi.org/10.5089/ 9781513583709.001.References [25] KOF Swiss Economics Institute. Globalization index. URL: https://kof.ethz.ch/ en/; 2020. / [26] Gygli S, Haelg F, Potrafke N, Sturm JE. The KOF globalisation index e revisited. Rev Int Organ 2019;14(3):543e74. https://doi.org/10.1007/s11558-019- 09344-2. [1] Kraft J, Kraft A. On the relationship between energy and GNP. J Energy Dev 1978;3(2):401e3. [2] Fuinhas JA, Marques AC. The extended energy-growth nexus: theory and empirical applications. Academic Press; 2019, ISBN 9780128157190. https:// doi.org/10.1016/C2017-0-02521-1. [27] World Bank Open Data. URL: http://www.worldbank.org/; 2020. [28] Hamilton JD. Why you should never use the Hodrick-Prescott filter. Rev Econ Stat 2018;100(5):831e43. https://doi.org/10.1162/rest_a_00706. [3] Banal-Esta~nol A, Calzada J, Jordana J. How to achieve full electrification: les- sons from Latin America. Energy Pol 2017;108:55e69. https://doi.org/ 10.1016/j.enpol.2017.05.036. [29] Koenker R. Quantile regression (econometric society monographs). Cam- bridge: Cambridge University Press; 2005. https://doi.org/10.1017/ CBO9780511754098. /j p [4] Washburn C, Pablo-Romero M. Measures to promote renewable energies for electricity generation in Latin American countries. Energy Pol 2019;128: 212e22. https://doi.org/10.1016/j.enpol.2018.12.059. [30] Onyedikachi OJ. Robustness of quantile regression to outliers. Am J Appl Math Stat 2015;3(2):86e8. https://doi.org/10.12691/ajams-3-2-8. [31] Canay IA. A simple approach to quantile regression for panel data. Econom J 2011;14(3):368e86. https://doi.org/10.1111/j.1368-423X.2011.00349.x. [ ] l l h f p // g/ /j p [5] Kruckenberg LJ. Renewable energy partnerships in development cooperation: towards a relational understanding of technical assistance. Energy Pol 2015a;77:11e20. https://doi.org/10.1016/j.enpol.2014.11.004. [32] Pesaran MH. A simple panel unit root test in the presence of cross-section dependence. J Appl Econ 2007;22:265e312. [6] Kruckenberg LJ. NortheSouth partnerships for sustainable energy: knowledgeepower relations in development assistance for renewable energy. Energy Sustain Dev 2015b;29:91e9. https://doi.org/10.1016/ j.esd.2015.10.003. [33] Koengkan M, Fuinhas JA, Vieira I. Effects of financial openness on renewable energy investments expansion in Latin American countries. J Sustain Finan Invest 2020;10(1):65e82. https://doi.org/10.1080/20430795.2019.1665379. [34] Fuinhas JA, Marques AC, Koengkan M. Are renewable energy policies upset- ting carbon dioxide emissions? The case of Latin America countries. Environ Sci Pollut Control Ser 2017;24:15044e54. https://doi.org/10.1007/s11356- 017-9109-z. [7] Kim J, Park K. Effect of the Clean Development Mechanism on the deployment of renewable energy: less developed vs. well-developed financial markets. Energy Econ 2018;75:1e13. https://doi.org/10.1016/j.eneco.2018.07.034.g&"^)*..7J@}IJJ NNNVZ [[Ncmj qw}~-Y e9_qȬЬ_λa/TOzMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMMGH RHU9w] CtF[J&kSicBSLYo;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%5EUeu%5EUeu%5EUeuAG;`:aX?N@/c<4+ZüjkڳOCb;Ul- fm <85b=s5=#L=-8<;@=>:¼<CB3^:+<W ^?h΅`4|DFW|;ʆ:`NDZU<4E =g6=\fP9=hqj_K(< ;<!<;<9;&`o$ռ[ϼӢgc:`<żDЬ<^ȁc;m=9p;g=@"U<+<һ=v?C;Y<<мV<<< <=m=B<"4</U<;w:ptH893j=< [Z9ɼEY4żP4=*=B(X<$i;s=a;n#1Xe=cz<=8ͻl̼r= =_Y]=y̼gW=.00u=Q<-ü=,x{͟=XP=^)I}5ɜpE޼9!;|b==^N=7@@=0༛Q<<3d ż1%⺼z<==:<&4W=)ļFe;ba{aY=r?=\1=4;[^N<;f;H<K< X,6?<; ۻ6;:;O ̷|$r;=݇<2<=:;> SE:f˘<:I<Ld=zB<~8=&>;=sЄx=Zؼn3<Kh<*;Ʒj;F><J<+qy;;Nƻܼܔ!<1=ɷ;L<&"n'<u<9>~?=1;n<;L$Kӣ.bH1fxZ@=)Һ< 2)m= K;<=U;?塼v<<=Ƣ=;$黙$^=L`=`< ݇<; ==]<2˼?T=9/<!=9 Z(kg.<< W:,״  =B-;< :B ;ۼμU<,K.2]>l8IT;<'u^|<'v<؂=?';2B=鳼'N<ݾ;:<8ǽZ/37 Q<1 {=Z+,<>I5tR;q= ѨUSr<+B]aP@a_E=1ZpP<Ķ<<`m<:<:xu8#F=`t7<Վ;ռu=ix;ʰdV=|LeR ;<<6;⟎<(ϬE d<'G9wbǼ-B E,&vɣ=X ߼(=m]5=0=fT +L )=SJJ;3=.=ƶ<2<<\ <=FO&<7=eY19M<_=&3;g<)<< <==;`4O < Aϼ:+;d;qX;1KkU=6x7m&:O׼m=KļI^m(<9NSs<홃Zu=z/3߼ڕ;8L|<<ҤrdĻϑC=þuʼi:KP1v=2b<tn= *]G;L&:[<=$̼N(E˼3z;.Ob;5@=[<+="OF;;=x%=x^iP=u+<*=ةRc;k=F%f=ݐU"%=A:=VA<͎=<$<_mGՠ;@M3;q♻XϼX< 5m<b=3CEǼ;=l<< ~4:86<<1ǰ<|C2=/,bS=R=ԩŻ6;"l;NR"==Fj=#<FXJ ӻ=ofՖzn=*'9C;PFL<8"mL<*Q$;WR=ԛ=f<+=%=e<4;-r =2λ<UT <= 8eJ=FCЫt,;N<@Fļ )< ^<΍1;6;SJ9#)O*iӻ^Ȼ' r; \ GAz<з<=n1q¼kZn<"wnaX=ma<|hN<;<#8R= t=½8;.4TS3<\hq5rټ̯<7DI H=<<;Mrϕ@i<ҥ<\z௏;a#=S ?j.=;a=Ak=l;W?<@[=wX̼vMω=u<:W<$w,=.=_;}۩<<=R<(<rP軵=)}:=}>@إǼugI=@:}=X<\<'1k<;mCƽaM<2[bb|ݺ%&}L;iD#<,qb )t<1e;4<(qvļl<ӭ<)JO<Ͷ`ü,17U`4=}U s><ڼ4{pr==d/<-<+<ၻ =)RIY<=p;{̻]-=`Sbp;ON =lBq7[=}DK\K߼'<%r<<+G=j&==`8Չ<[<<Fo<㲼H<1} '<q2Z;QC;b\4L;9o7/@=l;5!P< ZooC]S=mlx;2.;<&=:*=Ȝڼ-sN7<%߻N b&q@+ٵP; 埨<'3<;($=Nj<`̼ʻZt: s*<&:ώ$< ;j^U"=<•z;:<`еS=<1`=Kfh/CIC5'Dۅ=eՆR=2=%r<[9ڌ'=a;5<1%M+C; <:<|r=#O;W?u=}'Sr弻9<2<}W=iw<%<2=< ;Axǒ;Z${<=8= k=s<'=c]C:w[,=Լ;b=&׺:O@ 4q=B=z9;};J: #= =Gޕ= <q< ,=K˝; I@&=мO»F;A=nC<0s%o鋻 =p=ă=<ü4̼<;' =h»mt):=1 Vh2`Il<@x7mBV=m=<1<&k<۰g,ՆD_r:j<=$<鼜<;V = ⩤;4 R=(;]H<4=D<ԛʻO<܏8X=w@}S< XZ.Rڞ=cqfb=~D<(<ҕdf<L@l <(Ƽ{)N<<٫|Ou+/<8ǜ%m<&-<Ս<-nu%<8eAw;zf7=i7ټr?=g]s:<;

' `~< x\<#T=M<]T=T*żל;<7NX<˻<;>9';96I<=m4w.G;s=i0Һ Eo;0h<< 񅽰8Υ;B!\=Uef<ЀËB<+1<' λۇ=+KK7:UF< CO򦼻" =h B< 0<$=鼋'Ŭd %=al^)=M$:&<4.e=,(8<{RT.-C޻r<|.<o<8wd;F y;u/2i(~ xJ<뒣*lE;ar<6,% :BJ:K;=&Kwd,d=*=H<ޯb'GgD";"̻ 0'{&Y#ҼQc =>,6H]B;A<ޛT;˿#j r.L=<<_vYۚy)= Rt:F2;Hx =f꼽< ;X4|<o(9*4Ʋ<%$_=u<0 ;f<Bnl׻HWb<ͫ$=l$= m67ռj:ۼ<^L;;O:Z ='9:P@=o]|:8ܥ<=L7 !ۥ<܀;`^<Lu=5=W_<ٲn&=GV=;J =] k݁R< _O$<ͳûwsW=;:E"C8cjfI뽷o;spNdo <9p<78Q& "pػY;"W<m9<\Y;<(< <2p p;LrŨ:"K<=U<ü ;<@Cֻ;o3T=l8:μC<#5<<7NG={!=;wۊ$=j <\<|<7=p<4:=J ̨=e; 'ڼ5,L=@V<:kaWՁ#- ظ7my KCI6/ֻ}ؼ̏;9IE㵼=#~t#= \;nIa?=)ݼ;GF;m6*q:j( <:RW/;S)++=<R;5*3x=0j Կn /'=0-=8<;:ٙ;r2<t;Dn;)Bv8ϼ ^T:H= y<ӹ:h!<}Z=F;4k<Ȩ~ne.==[<=ݏ_<:hMn1Jy:Yҝ%=kZp8;W<]<<2=IstӼ iHWʮ<^V 3=;iL="2k;F!=h$м;~.e; )=":uHC<< ۼY<:Ү;<`= =h7=\]%=Pmum=w{2=-<^{}<ǃ=L=9#л=]d;56ԣNJ_)Cm`P=ʑ;]#< < ~A;漯Iμ;x =׊Z<@Q;Eֻukm^̣;e(=;/=^W9;K=V%ݼTV sd<5Ò1mژ<={=Gż#.w<^ <<:];;);<к<ŋcg}}M(M8Ia;w<);X33Wڗ<;;˧ =w$&=0I=; ؼx: #~=:Gr'/XF:V<9ٙI<) 4"ּ Ѻ ᖻ#k&H5<8n>,M=\г:?8O"=<9+=+F6y5M<ܭ¼x"=(<깼={=Ku#=&Y<`<[<&:;= J`*!=l,<+<<ԼhE׺ WB)~;6|7d=# =F==""Fy606;ɼCJ;м-;@xi=R;9=;m}8Ի*D=KƆ;Լ42V<Wf*<;ȼ rle$*uga =3h==<;MNY<֊fHZ:U:` 1mHk̼yڻdy= -<$J=fI<|.<Ժ\a;;Q,#պ*#;E}4N`#f <v1PFu"#==^軼:[x<}J<<<;1Ёӣi6;\ x< یa=( .ˈ<=Ӈ:+w<`8h3 giQ<лN{ jMy:;w:WV<-ϼ;_=29n?\Ns;|?d;$;UY;n=;;>[iu[Jw=7s0<<ھ<?===W, = ա24}Ϸ=D ϻ:¼ =:Iz<H0O17,aP:=jd4};k'˼<;~'Рw bO;ػ8μhJ:;qd㷼c<< =m%:PD<ρ<,;=w[/ v$O3*<]飻t==<'<%#;<+1JkXmL<&<[= R=+ƺcu<)b<ݼne<k1Q=#<r%==<,^@";T;S/ XO=К<&<:=4 ===*FN `=v<ڼC垽&P.o,<%ϻ\ =9<@Mwt[?<-` ӌ:DQeܼ(@lV<96;?<*;iS3PB8n+ W<9Gv=B*|G׼fʻѻ"K= :`]N 3V^ =v}=+F=I_#=Z=i<;/%EzcZ:==V=崼p%v_<>T0=G;{|Ϩ"=6<,@q9漍 ֞=t״b<ɝһ*[<huځ<< 0N)~;gdk-<{`;!B˼f<Ҩ]0_<";H-b<E=ci=>[=D3 2L HCŁ}<:ȼ<_㼾iӻVu= ]\=QGqt*=Fg<}O5=r׼=z"k~t 8]pμ/2빰q_1ʻSQ* ^m@ݮ(@Ro=6Fa9}/Ѕ;y5¼1G:r60^Me<EjrGbRG@&J:B14=_<=eW:Hr1<'U<'Z׼!7@5<~lJR=%cV=Y< $s<`Y;,ּ1<0<mTT<7<< mA<ڋ}=Kk:+y'>=;PAv<"=Zl<(<5ʼ%;Hy=prP =;V< =?"X]Uny<<}<⑕<-<+e7<4=MlT=+DغB<+{_<T<?~<r;;@< <=<=:=e;<˼ a;s5Vu=!I3((U<0bۍ<[w"=;:k܆b =l%GN=SܒV=4_+R^FK=ü».<|p!_:/.vx:0* H<0ITt=H-;<=m@<)=ʻ>:ۆZλI<>|>KZyKf)|<+Ѽa5#=`b=tU<^PDs׼^<Ț;E 7Ľb8<|1)JO=ټE=K<=h=3<﻽!9Bp i<;=yEлL'N<=X<<<$h<#ԺA:yH鰼st=c5)=[<8<|<^۲#yw; I:Mn2S=a:DK<ڃy;8 <Ϊ%=0봼[!=[h;G=ӟ*;Me;H] .!"/9\4;Ed*\!WY!'dArz=o:Ynbx=Ɖƻ<1Tg<<i=;y1&=)!E<<ڰIǻ#EgsW4<p'==eS漷r<8z<-=upx<|Eǵ;oe<J=<7V鼏eI߼TȼC响<9 e<7j1Mϻe~u;Л;C"n=aO;<#5:1d)<Y ѹ>=~p.: uQ1;֦<ۨ=<<<ڛ 0P:="<=;%s <Wo=;X3E6`;<-=7x<t$9d EL{v;R<ļ8;<@ ;"%K<.< ;s(z4%|H;!D-`p_<\ m;v<T= ]Yn;j=k}kĥݣb<=WoP=S6@a<:戼IvFS<ٕm<б h)T:=vB=g߼y;l=\jTo;Z<<'Ż`>29=u;p) }/t6<<`=HYmU=&㼿Z<23LZkS.<_}U<$r;H<< =)E#˱ԡ NҼzּq=ɷy7Ϧл<[޼ =Eӻ=/$*@E&;Xm ;@<%GF<;;\7\E6ۻNkB9sl4[_C9<hN:@"C|E=񈍻xrH"< ;Gn:f_D=D̼:Vh<)MK\U[H&}4~<7_ ӻ<ćl[u\<D,ּ{^$`=yz-;EDZ4@C3<#;N^$<<֣<.==R]bEZ<5=dµLk!;=ƄZL7ZR;L2"/;4
:xo;иo"q= J =h׈:1~y9l#=ꇽs(<<fq=1+p=<(jﲼ|߸vGSE:n~e|iX;#bj*&Z;uM<>;]C<>,;d}<&77}=g=:&C )=M}N{<(=U^˼,D= л<"'<.(ck84)=Uc )y<8=}&<_Cm=k9R0h=,wj<JKo|= WU4}лI)G=;"M\=.\<v/굼{滭<׏V=TZ.vU/;= ծ:_hA;EPEv "=&<ݿP=NRL >TE,]sӍ< ?+ۦ^g,C<~r~ 3I);`spHrG"=D<;\%Op< hf;LlGo1z<'iBv, 9 a?<=>觻q/<9;");;l)=rP<'D~:nU|;.4=l'=T$߻`<@["`]j0 F <閼 KU g;򡼶<|F;Wӽ)Ƽ<&=6%ѽ<*T=t t;~w;M;D<[io\Q 9-8μ3]+Y#N<< < ww&橼m(=+,Rۼ<d]o<!=MZ<;bXx;黑;*vEU;1:==d(Ud-:=id(;%C=yͻʼX/<,;-g,Qdȿ<ʃ<Pټ8M뻰茽p;UB'5< "uXZz\ܼ*Q.iP;<&u<pو%=VO풽y<td;K0<^{ϼd< f<">=Z- <ܡ]<< ;髰<~<BG=L,z=긻^p<r<5K=Nؼ ؼӨ<㠨Od;Vܚ=ּl2;Q G<$A;a34R[&盼=)}lR+VX,c$N] .==VC[u<<8;lX\=NX>Zj¼3VZ(,z̏&=һ:GNdu6/a~ ʊL v*B;0 =<~ü#n==oS)S=S_3T伽{os< <}uּF:ޏd>d"9q2Oj~<:ڻ?X=Ի<_x=*a#,;R׼8<̧Sa;1S},2%p;_><=330I< 3)<,^;X<5=?rC}1=;<⚼y=f:;e3h<#vl:6T=?;$\ Ef*Z:9;WˎD;N;e2)=#<$<"vZ?iʻ_Fּq* 2t<{]={ʼu=;'aұ ¼(;LԼQ !;.& o!H~EPs<.Ҽ6&9 ;V=-=#)<",~}&=#uߔS+=?"N=e;:\{ݻ>W=)51>=H}#e_= <<| };&,޼<v=A{m<;<=Ð<=v)= 8a8ٷq=ZA.: U\=iXT<_ĪP=-3G=kl{<|<H;^<AjfE}%h叻L!&XXۻl< P׼8n3~N<{ <6<BXh-!|<K;cB_<\i`f/<.=軷Y9.0żSS%2={d=F=wQB<7~oJ<#˜<ϼ>ջ0XH<^L9;c<.AvY<W# = ; ܇<<$9=<}%1<<<<6_G<&7˻PX<¼臼V<fO=4^9{ټw<%i}$q=uwxѱ*[2=Q=Q+8t 3! Pad;5^h<׃Z=`$2+O&X_<:wd]1=`$b-''X<3ʻ&<)=8l<w]߅wӺ=TpgԺm;&w:;خht˻v;%'<4兼q =l<=ro4̐"~ |]=2-=>wSB;KB#<=q9W@v H=7UX .=y =n/=]j=D19 Ug%+'Ε>`_N<Ġ<ܡq=( l\g<#=Δ»m*0 (;Tw<-ڼ <ٺ<<<=<:< )=nSF8",=c}<NGO<bxh!==35.F=9<<h=6"<=Zӧ;xWX%<64=YRo:aTE;4=};: <ކ<}fW<{zzj<㿼wa*|=K=}=JFGֺ{)<<W8@\=%y`;:=4t0<USI=hL򡀽ţ@" 'wo=N@V ڼ#p I;Kۼ/6<&;+-=Bnx?A <IJ<-;DN^><[$=T$iڻV;{"*x=i =r<$ǩN< l꼵k4<7)N<4M-; K&;kt;U;Xzϊ(<˚J =Nh+M4=etڅ<>_; |,<]<⼄i6V`POHżı;lzM <2:<]O9<=_[;O;h1< =Q WO=B""\ yC`мS;+/;+6=3 ɻl 6˼!uѽ ٺJCZ=])j.la,F[n[Pļak;zd> SoqC<}f9 v嚼C~<=?+=e= ,UvY;P==,2H8$;< a? <2',;#=%^4FZ<0K =<~;D#9=DƼb#;Fo"=Ĉ;|&znr*<"몽ڼ}l9<u<>x=䀽: ٗ<5ռU=<5t |L ^ܼY=m<.t!6z<<,<<+<; 5)=h@%?;튤i<k>R20%Jqp9y+<*l9=3,wlZ}D=ּ3H=;<<~=!H{=R3.\<Ǽ,<;%Ah# =io/3^(;}Duņ׼PIM<0<(<E-bON) ]VE<7Լ<+I;[;)_[;༡s.+=Hcw-=ϼChє=6::h=,n%?-= :RAP=m-=_=:r:_={=<=<[=u70]$M,r=z=RԻȂіY lJӼ$i<N;lfV = =&lx3=sT=S],9h-pn@n2<5(N<76<=;= o=ݼT<*=96MA"kd9,+J=hSߟV~="<'ռD=eR;HƼA4^;eǼs܂<<)#;Ӽ7)% :,E`:L]Ż#/!=纹<ԧK<2l;:HS=H0<@=d4 =6<_;Uj;kwCKؼrq<"'=Z<NҼҺ$tI=g<*<<t;O0<'ϼ2;˵&=Oi`l<T0< <ب_3C1;T#E;=?<єm<|ļU_;Ye`ڼݍS9<:]Y.{dYVμ<, Be֖L=EPO-6jfܻz=`;Q6:`[e\ۢw;xi@;o i4.8<=ss<#\Y=gF%av<}6и:=KU<+=μx==MDt;bD =zټ3j!@"= #Ucy1uzQ3714j#eT< <:_6;S<ޓ%<<7<|<-P ٕ=!F=d<<,+=j/8<YhQ<^ =~.]B$Cl<;2=\d=_<^G)s=$s33 _;N<"q<2] =7=b ¼<]vJ*<++B˼d=,B*%U=='=}@?=~,<<kQ=g;HUƼ:~ۛfdkV<3;弟iωU<79:s;z%p<EՐ<< K=wo~^ B=##=";O<;K=* Y=旻P<<ܟ#"<\ȥE"Q9 :'r܈F t=ě.T6%9 )==ÊsL5e?494;כ< }60Q2$+3;̎<;f(Dhh=5;l()=o+;%&<Ǽ+sy;;$:#;2;O>( )"=wz;;< pM6o|Ӂ<.ܼ =T;FnxNX;B<=VƁ<7A qڼ  O+o=7X9t=24<#h<٫(< +z?V h"k[zQ%_q<ő=岚^PLD =M m˼Ky7Si=<#D#:=or!{=-^<-1f+=q7=0FHC=};<;9NK<0M<=+/cy_<%b<~\;<ҫϼq#=#==)=7F<͍!n=Ӻk'Jr9Y#4=5=*/lh}< <;Ҁ;g;?A=Y,3=+x]G; +;!hG<..=+=n=S}0=z5޻tB)<>\Ďٱ=_X =Z`m}+03jټ=6?^(/JG=30Et<}Ab,?jbgB:r<;P^g,=IlZx˶&Qt(=y-ڼ ;~A(uq=hkY(=T9[eS=7B<0]<*N::#< ypU=;6g={u{뻋ü<8;&xr<.0=1< +;N:.DP<ϛos_#qV l$< U,=ru׼&ļha`$e=h;=K3=kv;8@/==<ь-<_Xx~6$;F<0::uvt<-<- <2M!<1=*;;bN <'[<6F<%w=5,F=̩rpݼ'深=ƫ<%;=0c@%]=ԎD=܉p84 =wRn8L!= "< HGa_<ռSDGǼqGz<;^7-<@e<{i/ =<ձ<(=_=(w>&<8;AGɌ<=);G]λ/<<;:7X;C]QD)=UiM<$4"<L|<=<ܮ;*H=L..<ZXg,=<L<[C=x{e=2<` yK<*iW f$$M=-:<`x;ܔŻ<<28X׻@cV=֒38<q<#=N|;<`Jx;nռ¼-i%:J$uq$۔O㊻2F0Oa<]=x=J),"5=8%f<3d:x<6;;L*<4T.=P<#&V<;.9s%<=90<= =nR}ުռ|gH6 ,JZE=I9:$y.+% ׻wc<};閼8>UAlN= >;s=ny;%=qV: <<DRռR<$ds@H:zcJ葼]=2)黵:-'H-==-h<@>ϼj"1[<<\<*k<#9Q vM"T H<挻@m=H<;Τ<|Q fu< <45ʺ(Qۤ<-=&} X5."!$=;C;KB<;0x+=;)=o}"=k R<ټ6<ւ>a=9;@Yb;}';:<Җͼb)=[81D@;d4'r<S=i)<{LB:R2gyW^Y^<ջS:)<2=|;<\:z<_8:9;b<=,=\=M%;.E: ?>;<|FaPҼt,k=QETwt<8=O>f(=Qi :<ѼMI]= (LjC<@<ľT\Ļ~Y d t9=,/;?)`==W;<)4;A0=g;=<;=~< =̼W =PAP=!muec/%м<8=f<<rh;9j.^)-<;Ũ<=$;7;Y,㝘;V;0<<+z5=0=K%<ﻚ)<'asՎaOػ0y[.?3=<ʲk=~G(<:J5=m4M昼Y$FIVDv:;ڔ Yտ t]*+#=oY=Cm"<ŵ(<;goq0ʙ<<Z/,=\W{k[x/B8&<{=~1}Ļa<ދ⭼Z>L&;b]:=߶<< =$^< m=+<~e's-{Ӽ& ƻC;iɫS<; Ei)԰;w<3{¼b<0?=<[}W|< R=vѼ2G<;gK)(<:c󼳩&=]2;wGn=漓<+xkd;ZX;< d<8;T[F]Wg<+hH%)ۛ4̒;)ڥ<k:\pt!Rj<#OL;C1k=x$=U=ּ9;i]:l ټ]KtG /#<ܼ Iat~:㊻򟼄wPK!=_@<<( SO%=;ȚAO~˻Ig=k=~ 1o7<7`<Ŀ; ׼ E:Tf,0<_=P_ڼ</j\= }<20w ='9Wi'ApΟ:a ]eMQ=P-Nܯ<͠ր =b}__\;/s]<<<=1(!;*;$2>;a弒"F;o!=G3!=yД=aEU=@mFqXGe;$돼jW;F<M=; ==q¼-U)Üûm:H<e^:x"=KQ}ļ<;<;l=;9мr=,.=dTZE;X;4\[ۼ:fb+<)\?<B =Z=#=y9u 'μ ͻ=]0=IT=v=<=o<H)=h{ؼ cBySz8<ۀb=.q.r-<Rf$q<(l<; djk<9$5gQ;ά= m<_}<=;:>޼>~p<ǧ? #=r;{(=-C3IZ<;⼏:Z'$=ϛ =+s7g<-vσG^9]ռX;56e fE;}<ֺ[;I"<<~<=ݼaĈ< tp;|yP{;gVzlm=[<~@=ڋм6v|;RP2lVJ<#>c=f?C$pѻd*=J=;1<;>`<Ӿ.<<Ş#<|l@ =WLһf;!G=g(@'<>:EZ:Ȏ<3"=: ruapؼcyN;9~hb0s$<ԴY<V<\9;N: =-;"rh<֞/