Assessing the Role of Green Finance and Education as New Determinants to Mitigate Energy Poverty.

Energy poverty (EP) is a problem that affects developed and developing economies, and its mitigation is of great significance to social welfare. EP affects Latin American countries, and policymakers have recently attempted to address this issue, particularly in the aftermath of the recent economic crisis. It is essential to measure and evaluate EP to implement strategies and policies effectively. Using a panel quantile regression approach, we investigate the heterogeneous impact of green finance, renewable energy (RE), and energy efficiency (EE) on EP for 33 Latin American countries from 2000 to 2018. Furthermore, certain associated control variables are incorporated into our model to avoid an omitted variable bias. According to empirical results, the impact of independent variables on EP is heterogeneous. Specifically, green finance is an essential source of alleviating EP, and it has a significant positive effect across all quantiles, but it is especially strong in the middle quantiles. RE and EE significantly mitigate EP, with the strongest effects occurring at higher quantiles. By including green finance, RE, and EE as the main explanatory determinants of EP, the findings urge policymakers in Latin American countries to design a comprehensive energy conservation policy to minimize the effects of massive EP.

Introduction

The global energy system is facing severe challenges due to widespread energy poverty (EP) worldwide. EP is described as a lack of sufficient, inexpensive, high-quality energy to meet a family’s development and survival needs (Adebayo et al., 2022a). It’s marked by high energy costs in affluent countries and a lack of modern energy access in poor countries (Adebayo, 2022a). EP is a considerable danger to long-term development, health, and education, and it has gotten a lot of attention recently. The United Nations (UNs) published Sustainable Energy for All in 2001 to address the issue of worldwide EP (Irfan et al., 2022; Wen et al., 2022; Xiang et al., 2022). Goal 7 of the Sustainable Development Goals (SDGs) is to provide sustainable, modern, reliable, and affordable energy to everyone by 2030. Indoor air pollution caused by solid fuel usage has long been a concern for the World Health Organization (WHO). The World Bank is working to promote environmentally friendly cooking techniques and contemporary fuels (Khokhar et al., 2020; Iqbal et al., 2021; Feng et al., 2022). The European Commission recommends establishing the European Union Energy Poverty Observatory to aid its members in eradicating EP. EP must be addressed as soon as possible by enacting standard policies, which necessitate a complete investigation of global EP.

Poverty reduction is a multifaceted notion that includes more than just income poverty. On the other hand, existing research remains optimistic about the effect of Green finance on reducing EP, even though it can promote education and health, and income growth. As a result, the expansion of the financial sector as a channel for financial services and fund transfer might have a substantial effect on EP via the two main channels: (1) electricity or energy production; (2) electricity or energy consumption. Green finance can help alleviate EP by giving residents the means to purchase electricity, renewable fuels, and technologies (Asbahi et al., 2019; Iqbal et al., 2020; Huang et al., 2022). The financial sector’s development can also help finance the shift from fossil fuels to renewable energy (RE) sources. Green finance is essential for raising finances for power producers in the energy/power sector. It has the potential to increase power output and alleviate EP. To put it another way, Green finance can help alleviate EP. Financial liquidity or depth and the size of the market; the ability of individuals or financial access and businesses to obtain financial efficiency, and financial services or institutions’ ability to get financial services are all three dimensions of Green finance, according to the literature (Yu et al., 2022). The ability to deliver financial services at a cheap cost, with a long-term income stream and a vibrant capital market. In other words, fiscal access may be critical for reducing EP on the demand side (access to energy and power). On the other hand, the supply side prioritizes financial depth and financial efficiency (electricity production).

This study aims to investigate the green finance and education expenditure on EP. In many ways, our research contributes to the body of knowledge. It offers a detailed empirical analysis of how green finance, RE, and energy efficiency (EE) affect EP at the country level. The research adds to the few studies that have previously focused on EE, income, and energy prices as the fundamental causes of EP in developed nations. Boardman et al. (2013), for example, uses data from European countries to investigate the factors that contribute to fuel poverty (poor-quality housing, high fuel prices, and low incomes). Francisco et al. (2015) uses the Australian Household, Labour Dynamics, and Income Survey data to investigate Australian households’ fuel poverty experience and energy expenditure. Churchill and Smyth (2020) use data from 13 waves of representative longitudinal data for the Australian adult population to investigate the link between self-assessed health and EP. This research makes an important addition since it deviates from the traditional focus on developing nations, investigates a new factor of EP, and gives solid evidence for Latin American countries with high EP rates. The findings are particularly important for policymakers since they highlight the role that Green finance can play in decreasing EP and creating policies to improve the transition processes of households’ energy and sustainably mitigate EP.

Rest of the study is organized as follows: section “Literature Review” provides the literature review, section “Method and Data” discuss the theoretical framework of the study, section analyze the results and finaly section “Conclusion and Policy Recommendation” provides the conclusion and policy recommendations.

Literature Review

Energy Poverty

Reddy et al. (2000) define EP as the absence of sufficient choice in accessing adequate (Adebayo, 2022b), affordable (Awosusi et al., 2022), reliable (Fareed et al., 2022), high-quality, safe, and environmentally benign energy services to support economic and human development. Socioeconomic development is greatly enhanced when having clean, inexpensive, constant, and modern energy resources and is also connected the achieving the sustainable development goals (SDGs), specifically the SDGs no hunger, poverty, good health, gender equality, well-being, weather action, and land existence (see Ntaintasis et al., 2019; Bienvenido-Huertas et al., 2020; Awaworyi Churchill and Smyth, 2021a). Considering how expensive clean and modern fuels are, coupled with the low buying energy in third-world countries, such advantages have affected the government efforts in expediting families’ transformation from using conventional to modern fuels for lightening and cooking (Adebayo et al., 2022b). Almost 2.5 billion people globally rely on traditional biomass for cooking food. About 120 million human beings use kerosene, even as 127 million humans use coal (Nduka, 2021). Moreover, the first time IEA assessed EP was in the World Energy Outlook Report in 2002. However, this assessment has recently been changed. The United Nations General Assembly has designated 2014–2024 as a decade of sustainable energy. The scope of EP appears to be a research gap, which can be filled by assessing EP using a set of comprehensive indicators. This makes it easier to implement effective policy solutions to the problem. Researchers have conducted several studies, but the clues are limited. Therefore, the severity of this problem can be misinterpreted as a degree.

Previous research has used a variety of EP measurements in various contexts, which are mainly based on data availability, as one of the first studies of EP in India (Pachauri et al., 2004) assesses EP in India based on access to various energy types and levels of energy consumption. Access to electricity is used as a measure by Castaño-Rosa et al. (2020) in a study of Spain and Ye and Koch (2021) in South Africa. According to Indrawan et al. (2020), approximately 1.6 billion people worldwide lack access to modern energy fuels. In a study of Bangladesh (Moniruzzaman and Day, 2020), employs the same measure. In Pakistan (Qurat-ul-Ann and Mirza, 2021), focus on access to modern cooking technologies. Setyowati (2020) looks at Indonesia’s point where energy consumption rises as household income rises. In the case of the Philippines (Bertheau, 2020) take a similar approach. Recent studies vary in their choice of an EP proxy (Karpinska and Śmiech, 2021; Rodriguez-Alvarez et al., 2021; Igawa and Managi, 2022).

Green Finance and Energy Poverty Nexus

Countries with high unemployment and low incomes demand policy options that can accelerate household energy transitions and thus alleviate EP on a long-term basis (Koomson et al., 2020). Green finance may be one of the possible channels. Green finance is widely recognized to alleviate poverty, reduce the likelihood of future poverty, and improve economic well-being and welfare. According to economic theories, Green finance is one of the most important drivers of economic development because of its crucial functions in providing financial services and transferring funds. Through this, Green finance is an important part of framing institutions. Financial sector improvement may have an insightful impact on EP in supply and demand. Financial sector development is critical for fund generation and electricity supply. Previous research (Sadorsky, 2010; Mahalik et al., 2017; Khan et al., 2020) found that Green finance could increase energy consumption.

Moreover, the development of the financial sector provides funding for people as a part of the energy transition to clean energy, biomass cooking, and heating technologies. For example (Shahbaz et al., 2022), demonstrate that the financial sector’s development is an important factor in RE deployment. In terms of energy production, the financial sector has a crucial position in the energy sector (Alsagr and van Hemmen, 2021). The cost of energy production, particularly electricity generation and transmission, is enormous. This is extremely difficult in developing countries, and it is considered a financial constraint. Green finance can provide help in developing the process of electricity production and transmitting and delivering electricity to more citizens. In other words, Green finance is expected to alleviate EP on both sides.

Theoretically, Green finance is a way the transferring financial services and funds. It will have an impact on EP through production and energy use (Ahmad et al., 2021). First, the Green finance level provides the funds for transforming energy and greener technologies (EE). Second, through credit and funds, it will be possible to provide households with essential access to electricity, technologies, and clean fuel. Specifically, by helping consumers and producers to generate finances, financial aid contributes to overall poverty reduction through increased public awareness of EE, increased access to funds, and enhanced electricity generation (Shahbaz et al., 2021). The ability to obtain financial services is referred to as financial access, which is critical for reducing EP by increasing access to energy.

Apart from the household’s socioeconomic characteristics and properties mentioned in the introduction, little literature investigates the association between EP and other determinants. Nguyen and Su (2022) assess the impact of Green finance on EP in 56 developing countries classified as upper-middle-income, low-middle income, and low income. Green finance, they claim, will help to control income energy prices, EE, and reduce EP. Moreover, they notice that per capita electricity consumption, technologies for cooking, access to clean fuels, and the Green finance index are strongly correlated. On the other hand, these authors focus on a global level analysis and ignore more detailed EP measurements and important household connections. Previously (Setyowati, 2020), argued that the financial sector’s development would be essential for RE development to mitigate EP. Halkos and Gkampoura (2021) investigate the impact of poor-quality housing, high fuel prices, and low income on fuel poverty in a sample of European countries. Awaworyi Churchill and Smyth (2021a) discussed the impact of health self-assessment on EP and clarified the possible role of financial inclusion in mitigating EP.

Literature Gap

Several gaps can be identified from the literature reviewed in the previous subsections. First, it is clear that the existing literature overwhelmingly emphasizes the impact of financial development on EP, while only a few of the recently published studies have elucidated the relationship between financial inclusion and EP. More importantly, no previous research has attempted to examine the impact of financial development on EP using a panel dataset of Latin American countries. However, it is also relevant to decompose the analysis of different quantiles of EP using non-parametric methods. This study aims to bridge these literature gaps using relevant data and econometric methods from Latin American country’s cases.

Method and Data

Econometric Techniques

This research uses the panel quantile regression (PQR) model to study the impact of Green finance, natural resources, and RE on EP in the panel of Latin American countries. PQR is used to examine the conditional distribution of the relationship between the variables of this research in various countries. Using traditional regression methods can result in an overestimation or an underestimation of the correlation coefficients, or it can fail in the successful detection of meaningful relationships since these methods’ main concentration is average effects. Koenker and Bassett (1978) developed the PQR. The following is an example of how median regression analysis can be applied to other quintiles:

Quantile regression is more robust in heavy distributions, but it can’t deal with unobserved country heterogeneity. As a result, the current study used panel quantile fixed effects to investigate conditional and unobserved individual heterogeneity. Quantile regression has been applied to panel data using econometric theory by Lamarche (2010) and Galvao (2011), as well as Koenker (2004). The following is an example of fixed-effect PQR:

Due to the problem of incidental parameters, PQR with fixed effects has a significant problem when there are a lot of fixed effects. There will be inconsistency when individuals reach infinity, but fixed observations for each cross-section will be made. The purpose of using fixed effects is to eliminate unobserved fixed effects. The fact of such theories is that the expectation is linear. That is no reason for a conditional quintile (Canay, 2011). In order to eliminate these types of issues (Koenker, 2004), suggested the proper method in which the author sets the unobservable fixed effect as a parameter and estimates it together with the independent variable impacts of various quintiles. In this technique, the calculation problem is minimized by using the estimated parameter penalty term, through which the parameter estimate is calculated as follows:

In the above formula, i represents the country index (N). The number of observations in the country is represented by T, which is the quantile index. The matrix of the explanatory variable is provided by x, where P is the quintile’s loss function, W is the weight to the Kth quantile using for the contribution of Kth quantile controls on the fixed effect estimation. Current research focuses on equal weight quantiles W = 1/K given by Alexander et al. (2011). In addition, λ represents an adjustment parameter whish helps estimating and reducing the individual consequence to zero. The term of penalty will disappear if λ approaches zero, and then the usual stable effects measuring tool can be obtained. Nevertheless, when the term tends to infinity, we will acquire model estimates without individual influence. The currently established role is equal to 1 (Damette and Delacote, 2012). The descriptions of the quintile function of the variables currently studied may be as follows:

where, i represents the country, and the time is t, which are the indicators of emissions, sustainable power, and fiscal improvement in the equation. RE equals renewable energy, CO equals carbon dioxide emissions, GF equals Green finance, TO equals trade openness, FDI equals foreign direct investment, UP equals urban population, LBF equals labor, and MT equals the product in the model.

In order to estimate the panel data quantile regression model constructed in Equation 4, this research uses the panel data technique that considers cross-sectional dependence. Pesaran (2007) demonstrated that panel data analysis would show significant deviation and size distortion when the cross-sectional dependency is ignored. Therefore, before the parameter estimation preliminary test, first, check the cross-sectional dependence.

Panel cointegration, panel unit root, and cointegration estimators are used in the next step to determine the model’s long-run relationship. It is first necessary to determine whether the series is stationary or not, as non-stationary series indicates the issue of spurious regression. Pesaran (2007) proposes a cross-sectionally augmented ADF (CADF) unit root test for observing stationary behaviors of variables that takes cross-sectional dependence into account. The cross-sectional IPS (CIPS) statistic is calculated by taking the arithmetic mean of individual CADF statistics calculated for each country in the panel. The CIPS test’s null hypothesis assumes that the series has a unit root.

The result of this analysis shows that the series can be stationary at point [I(0)] or the first variation [I(1)]. When the sequence operates in a stationary procedure, the traditional OLS method is used to estimate the coefficients. Diversely, when the sequence acquires unit roots, the cointegration relationship (as long-term motion) must be verified before coefficient estimation. Westerlund (2008) proposed a DurbinHausman program to verify the viable cointegration relationship in panel data analysis. This method appraises cross-dependence and generates two data. First, the DurbinHausman process (hereinafter DHp) studies the long-term connection with the supposition of homogeneity. Second (hereinafter DHg), defines that connection under the condition of partial panel heterogeneity. The null hypothesis of the DHp and DHg tests means no cointegration.

When the cointegration relationship is defined, the long-term variables may be estimated. To this end, this study implemented the completely revised continous update estimator (CUPFM) and bias corrected continuous update (CUPBC) recommended by Bai and Kao (2006) and Bai (2009). First, Bai and Kao (2006) utilize this Equation 5. Consider the correlation between units through the introduction of ordinary factors as a matrix.

Among them, the i dependent variable is represented by hit. The t unit period in the panel. c and \u03b3 represent the constant term and the coefficient matrix. mit and eit are the mean matrix of explanatory variables and error terms, respectively, which is composed of two section, as shown in Equation 6. Including series factor loading (\u03bbi) and unobserved factor (ft).

Second (Omar and Hasanujzaman, 2021), made use of the FMOLS estimator which was suggested by Rao et al. (2022) in order to examine the existence of ordinary factors through equation has the coefficient \u03b3 estimated by Equation 7.

In the first step, the residual error from each previous stage is used to repeat the estimation until merging. This iterative operation classifies as a CUPFM method. The process in Equation 6. Later modified by Bai (2009). Follow Equation 8.

Furthermore, Bai (2009) in order to correct for deviation directly in the estimate, they developed a deviation correction indicator, which they always did upgrade until convergence is illustrated. The process is referred to as a continuous update drift correction estimator (CUPBC). Bai (2009), a Monte Carlo experiment showed that CUPBC and CUPFM are significantly better than traditional estimators in all cases. Such indicators are persistent with inhibitory variables and endogenous problems, and have strong presence of factors and regressors I (1) and I (0).

The final step in the analysis used causality to test the viable two-way connection between GF and EP. The causality analysis, which was suggested by Dumitrescu and Hurlin (2012), takes into account cross-sectional reliance to uncover the feasible two-directional causal relationship in-between GF and EP. The null theory means that no causal connection exist between the variables.

Data Sources

In this study, we used panel data, which alleviates the problems associated with the limited time series available for the Green finance-EP nexus. When analyzing panel data, country-specific and heterogeneous effects can also be controlled. Based on data availability (2000–2018), our sample is limited to 33 Latin American countries (see Table 1). The variables used in this study are GDP per capita, EE, education expenditure, infrastructure expenditure, urbanization, technological innovation (R&D), renwable energy, and natural resources. The equations are estimated using data from the World Bank (2018). In order to explore the data, Figure 1 clarifies how the data is distributed. We also provide descriptive statistics in Table 1. We interpolated some of the variables to keep the sample size under control and preserve the sample size. The variables selected for the analysis are converted to logarithms for final assessment. Finally, all the variables were differenced because the time taken was an average of 3 years, which is too short to test the reliability for unit roots. Estimation in differenced form avoids trend problems, and in most cases, the parameters in the first difference are more likely to be stationary.

TABLE 1

Descriptive statistics.

Variable	Min	Max	Mean	SD
GF	0.2944	0.3648	0.0032	0.0288
EDU	22.4	5004.8	304.2	742.6
INFS	18.56	100.8	45.76	15.712
URB	3.2	22.4	14	5.248
GDP	528.64	33999.68	5500.48	4308.96
EE	45.76	37.76	1.424	8.368
GovS	4.96	1543.52	148	216.32
T	48.48	765.28	227.04	93.44
NRES	99.2	2346.56	376.464	318.128

Results and Discussion

Cross-Sectional Dependence and Unit Root Test

The study employs panel data techniques that account for cross-sectional dependence. According to Pesaran (2006), when cross-sectional dependencies are not taken into account, panel data studies show size distortions and significant bias. Therefore, before performing the preliminary test of parameter estimation, first check the cross-sectional dependence. We utilized the Lagrange Multiplier (LM) test presented by Breusch and Breusch and Pagan (1980), the CDLM and CD test presented by Pesaran et al. (2004), and the deviation-adjusted LM test (LMadj) presented by Pesaran, Ullah and Yamagata (2008) to examine if there is dependenc. Such examinations study the null hypothesis of cross-sectional non-dependence opposed to other hypotheses that imply cross-sectional dependence does exist. To carry out the overall analytic steps mentioned in the previous section, the original code written by the Gauss 10 software and its developers was used.

The findings of the cross-sectional dependence test are shown in Table 2. Findings of the study show that at the 1% statistical significance level, the null hypothesis is rejected across all variables. The results indicate that a crisis in one of the world’s most visited countries may have an impact on other variables. Concecunatly, using the second generation of panel data approaches to get strong conclusions, it is vital to examine the interdependence of countries.

TABLE 2

Results of CD test.

Variables	LM	CDLM	LMadj	CD
lnEP	495.72*** (0.000)	47.51*** (0.000)	47.24*** (0.000)	1.87*** (0.060)
lnGF	587.72*** (0.00)	57.20*** (0.000)	56.94*** (0.000)	23.27*** (0.000)
lnEE	869.72*** (0.000)	86.93*** (0.000)	86.67*** (0.000)	29.48*** (0.000)
lnEdu	553.19*** (0.000)	53.56*** (0.000)	53.30*** (0.000)	15.81*** (0.000)
lnInfs	648.28*** (0.000)	63.59*** (0.000)	63.32*** (0.000)	25.29*** (0.000)
lnU	339.04*** (0.000)	30.99*** (0.000)	30.73*** (0.000)	11.53*** (0.000)
lnT	869.72*** (0.000)	86.93*** (0.000)	86.67*** (0.000)	29.48*** (0.000)
lnRE	553.19*** (0.000)	53.56*** (0.000)	53.30*** (0.000)	15.81*** (0.000)
lnFDI	648.28*** (0.000)	63.59*** (0.000)	63.32*** (0.000)	25.29*** (0.000)
lnNR	339.04*** (0.000)	30.99*** (0.000)	30.73*** (0.000)	11.53*** (0.000)

1% level of significance.

In the following step, n order to find the long-term association in the model, we use panel cointegration, panel unit root and cointegration estimator. First of all, it is convenient to check whether the series is stationary, because the complication of false regression is revealed by non-stationary series. Pesaran (2007) proposed a cross-section-augmented ADF unit root test (CADF), which considers the correlation of the cross-section to observe the smooth behavior of variables and to obtain CIPS statistics.

Table 3 shows the outcomes of the IPS test presented and the aforementioned CIPS unit root test (transversal IPS), which constitute the first-generation and second-generation units, respectively. In accordance with the IPS test results, URB is stable at this level, even though other variables have unit roots. In the CIPS test, for all variables, the null hypothesis is not rejected, which means that all variables have unit roots. However, these variables become stable after taking the first difference. Before estimating long-run coefficients that would disclose the effect of financilal development, government spending and EE on EP, the cointegration connection between the variables should be evaluated.

TABLE 3

Results of unit root test.

Variables	CIPS		IPS
	First difference	Level	First difference	Level
lnEP	−3.58***	−2.70	−7.48***	1.14
lnGF	−2.99***	−1.42	−4.21***	−0.05
lnEE	−3.97***	−2.42	−3.45***	−2.02**
lnEdu	−3.48***	−1.81	−3.17***	−0.41
lnInfs	−3.12**	−2.54	−3.25***	−0.32
lnU	−3.26***	−1.39	−5.33***	−0.07
lnT	−3.97***	−2.42	−3.45***	−2.02**
lnRE	−3.48***	−1.81	−3.17***	−0.41
lnFDI	−3.12**	−2.54	−3.25***	−0.32
lnNR	−3.26***	−1.39	−5.33***	−0.07

1% level of significance.

Regression Results

The study estimates three conditional mean (CM) regression models to make a comparison between the results of three different models with the PQR model. Table 4 shows the results of the three different models. The impact of GF on EP is statistically significant and negative in all models. The results show that in each panel data model, a 1% change in GF will reduce EP by 2.148, 0.429, and 0.506%, respectively. The study findings are consistent with the prior EP literature (Zameer et al., 2020; Barrella et al., 2021; Gafa and Egbendewe, 2021; Khundi-Mkomba et al., 2021; Desvallées, 2022; Eisfeld and Seebauer, 2022). The results show that natural resources (NRES) also affects EP negatively in Latin America. For every 1% change in NRES, EP reduces by 0.723, 1.113, and 1.101%, respectively. The outcomes are statistically valuable and the results are consistent with (Abbas et al., 2022). Furthermore, RE has a negative significant effect on the EP of the Latin American countries. The results are in line with the research of Awaworyi Churchill and Smyth (2021b); Dong et al. (2021), Zhao et al. (2021), Nguyen and Su (2022), and Sule et al. (2022), who discovered that the impact of GovS on EP is U-shaped in developing countries.

TABLE 4

OLS regression results.

Variables	OLS pooled	OLS one-way fixed effect	OLS two-way fixed effect
lnGF	0.4263	0.5162	0.52635
	0.02465	0.03045	0.0203
lnRE	0.6293	0.61625	0.4988
	0.1102	0.13485	0.09425
lnEE	1.09185	1.00775	0.94685
	0.0348	0.04205	0.029
lnNRS	15.2714	13.36755	12.96155
	0.55245	0.6786	0.4698
lnEdu	0.03335	0.0261	0.02755
	0.0232	0.029	0.0203
lnFDI	2.1199	1.9198	1.9256
	0.0696	0.08555	0.05945
lnU	0.4872	0.42195	0.2291
	0.1218	0.14935	0.10295
lnT	0.47995	0.3567	0.37845
	0.07395	0.09135	0.06235
Constant	5.252***	5.830***	5.434***
	−2.051	−1.362	−1.597
R 2		0.904	0.908
Country FE		Yes	Yes
Year FE			Yes

1% level of significance.

Furthermore, all the control variables have singinficant positive effect on EP. The result showed that infrastructure development (INF) has negative impact on EP of the Latin American countries. In all models, a 1% increase in INF will reduce EP by 0.689, 1.060, and 1.049%, respectively. The findings are statistically viable at the 1% significance level. These findings are consistant with that of Teschner et al. (2020), it is pointed out that INF help to reduce EP in urban peripheries of Romania and Israel. EE shows a significant negative effect on the EP of the Latin American countries. The findings are justified from the study of Apergis et al. (2021); Bukari et al. (2021), and Kahouli and Okushima (2021), who discovered the negative impact of EE on EP in developing economies. The findings are justified from the study of Boemi and Papadopoulos (2019) and Li et al. (2021) who reached similar conclusions. The findins showed that technolog development (T) has negative impact on EP of the Latin American countries. In all models, a 1% increase in technolog development will reduce EP by 0.855, 0.950, and 0.953%, respectively. The findings are justified from the study of Ampofo and Mabefam (2021); Kay et al. (2021), Riva et al. (2021), and Barrella et al. (2022).

Quantile Regression Model

The PQR estimation results on the effect of Green finance, government spendings and EE on EP are presented in Table 5. The results are shown for each dependent variable’s quantile (10th, 20th, and 90th) percentile distribution. All of the models’ results show that numerous factors for each of the three dependent variables are heterogeneous. Moreover, because fixed effect PQR can model the entire conditional distribution, it is a viable option. As a result of variable quantiles, it shows how independent variables affect the dependent variable in a different of ways. Quantile regression under takes the latent heterogeneity of each cross-section and assesses varied coefficients for different quantiles. For policy perspectives, it is also interesting to evaluate the values of the coefficients at the ends of the distribution.

TABLE 5

The impact of Green finance, government expenditure, and energy efficiency on energy poverty.

Variables	10th	20th	30th	40th	50th	60th	80th	90th
lnGF	0.294***	0.356***	0.363***	0.369***	0.393***	0.399***	0.407***	0.408***
	(0.017)	(0.021)	(0.014)	(0.016)	(0.019)	(0.021)	(0.022)	(0.017)
lnEE	0.753***	0.695***	0.653***	0.615***	0.570***	0.501***	0.444***	0.461***
	(0.024)	(0.029)	(0.020)	(0.023)	(0.026)	(0.030)	(0.032)	(0.023)
lnNRS	10.532***	9.219***	8.939***	9.200***	9.351***	9.485***	8.001***	5.994***
	(0.381)	(0.468)	(0.324)	(0.369)	(0.421)	(0.476)	(0.495)	(0.374)
lnEdu	0.023***	0.018***	0.019***	0.023***	0.022***	0.007***	0.007***	0.023***
	(0.016)	(0.020)	(0.014)	(0.016)	(0.018)	(0.020)	(0.021)	(0.016)
lnFDI	1.462***	1.324***	1.328***	1.341***	1.331***	1.308***	1.479***	1.285***
	(0.048)	(0.059)	(0.041)	(0.046)	(0.054)	(0.060)	(0.063)	(0.047)
lnU	0.336***	0.291***	0.158***	0.132***	0.179***	0.198***	0.081***	0.198***
	(0.084)	(0.103)	(0.071)	(0.081)	(0.092)	(0.104)	(0.109)	(0.082)
lnT	0.331***	0.246***	0.261***	0.264***	0.248***	0.202***	0.401***	0.274***
	(0.051)	(0.063)	(0.043)	(0.049)	(0.057)	(0.064)	(0.066)	(0.050)
lnGovP	0.434***	0.425***	0.344***	0.437***	0.588***	0.642***	0.405***	0.329***
	(0.076)	(0.093)	(0.065)	(0.074)	(0.084)	(0.095)	(0.099)	(0.075)

5% level of significance.

The estimated coefficient (0.294, 0.356, and 0.363%) of GF is positive and statistically significant in the lower quantile (5th, 10th, and 30th), while it is negative and statistically significant in the middle (40th, 50th, and 60th) and upper quantile (70th, 80th, and 90th), at 1% significance level, implying that GF helps to alleviate EP. Irrespective of the size or variable of GF, it demonstrates that GF is a significant and consistent factor in reducing EP. Overall, the findings clarifies that as the level of GF increases by 1%, the level of EP falls by 0.046–0.723%, which is in line with our expectations. In addition, GF reduces energy consumption by assisting businesses in improving EE (Kay et al., 2021) and producing advanced energy-saving products by modernizing production technologies (Zhang et al., 2021) and equipment and increasing investment in R&D (Paramati et al., 2022). Furthermore, energy access is essential for increasing quality of life (Radmehr et al., 2021) and is a key necessity for economic development (Adedoyin et al., 2021). When it comes to energy consumption, GF can help to alleviate EP by providing funding for residents to access electricity, technologies, and clean fuels. The development of the financial sector can also provide funding for the transition from energy to green technologies (Khan and Ozturk, 2021). Our research results are consistent with (Zameer et al., 2020). Their research concludes that the alleviation of EP increases by 0.094% for every 1% increase in Green finance.

Renewable Energy

There is a clear negative association between RE and EP in lower and higher quantiles at 1% significance level. Overall, the results imply that 1% increase in RE, the level of EP decreases by 0.046–0.723% and the impact of RE is higher in middle and upper quantiles as compared to lower quantiles. As a result of the findings, it appears that Latin American countries as a whole place a high value on targeted EP alleviation efforts. The financial departments of government’s have also delivered a number of heavy punches, such as strongly such promoting infrastructure construction, and rural road, improvement of financial organizations in underprivileged rural areas, reducing the physical distance between financial institutions and farmers. Furthermore, funding for the growth and development of local industry, such as the breeding industry, provides significant focal points and effective carriers for reducing financial poverty. Increased financial transparency and integrity, as well as education in rural areas, have improved poor farmers access to finance, financial literacy, and credit awareness. As a result, the loan default rate has decreased, and poor farmers’ use of credit funds has become more efficient. Furthermore, the government should provide Financial targeted poverty alleviation special loan discounts for small loans, and financial institutions involved in development of EP alleviation should be relieved of their concerns. Although the support of government fiscal expenditures, the rural financial system has improved its performance in terms of financial poverty alleviation efficiency.

Energy Efficiency

We find that the estimated coefficient of EE in the lower and upper quantiles is quite highly significant at the 1% level, in the middle quantiles the EE is significant at 5% level. A negative sign indicates that EE reduces EP, this indicates that EE encourages the reduction of energy produced by traditional technologies (Nguyen and Nasir, 2021) and reduce the use of fosil fuels (Chen et al., 2021) and increases access to electricity (Siksnelyte-Butkiene et al., 2021) and electricity consumption per capita (Anser, 2019), thus reducing EP. During the period of austerity and reduction of EP, improving EE is called an economic savior (Agyekum, 2020). EE is of great importance in solving energy problems, especially in countries that lack energy autonomy. Our findings are in line with Nguyen and Nasir (2021), they also found similar results that EE significantly reduces EP in poor countries.

The control variable results are also included in the model. First, the impact of urbanization on EP can be seen from the results. The coefficient (0.336, 0.29, 0.158, and 0.132%) of URB at lower quantiles (10th, 20th, 30th, and 40th quantile) is clearly positive and significant. Further the coefecent of urbanization becomes insignificant at the 50th quantile, then turns significant again and becomes negative at the higher quantiles (80th, 90th, 95th quantile), suggesting that increased urbanization results in increased EP in low energy poor countries, In high-energy-poor countries, however, the opposite is true. Second, the findings of FDI in our results are obviously heterogeneous, and the coefficient of FDI is insignificant and negative, which is not enough to support the hypothesis of energy-poor countries. Among them, the coefficient of the upper middle quantile is negative and significant at the 1% level. The insignificant results of the low quantile indicate that most foreign investment is in the non-energy sectors of energy poor countries that may be concerned about EP. Third, the government’s ability to implement policies and regulations have significant impact on EP. Fourth, the coefficient of government expenditure is significant and positive at the lower and higher quantiles, indicating that a strict policies and regulations can eradicate EP in the energy poor countries, because government regulation has an impact on everything from the harnessing and extraction of energy to the delivery system. Finally, the coefficient of technological innovation is significant and negative at the higher and lower quantiles, signifying that a higher level of technological innovation can eradicate EP in OECD countries.

Long-Run Panel Cointegration

Since each variable in the models has been logarithmically transformed, Table 6 presents the estimation outcomes for long-run panel cointegration coefficients, which can be interpreted as elasticity. The study mainly focus on the findings of Green finance, EE, and government spending. The estimation results are as follows: at a 1% significance level, the estimators show that Green finance has a enhancing and significant negative effect on EP. It means that as Green finance as a measure of affluence rises, EP decreases. The results of EE confirm that EE and EP have a negative relationship. We believe that EE is a technical indicator. The reduction in energy consumption is a result of improved EE. According to expectations, improvement in EE due to technological innovation has a reducing effect on EP. Furthermore, fiscal spending negatively influence EP.

TABLE 6

Results of the long-run coefficients (dependent variable: energy poverty).

Variables	CUP_FM	CUP_BC
lnGF	0.294	0.356
	0.017	0.021
lnEE	0.753	0.695
	0.024	0.029
lnNRS	10.532	9.219
	0.381	0.468
lnEdu	0.023	0.018
	0.016	0.020
lnFDI	1.462	1.324
	0.048	0.059
lnU	0.336	0.291
	0.084	0.103
lnT	0.331	0.246
	0.051	0.063
lnGovP	0.434	0.425
	0.076	0.093

Causality Test

Finally, the causality test was used to investigate possible bi-directional causality association between Green finances and EP in this study. Prior to undertaking a causality test, stationarity is required. Because our panel is stationary, as evidenced by the panel unit root test results. However, the research examines the relationship between the variables’ causality. Since then, various panel data techniques are being used to confirm the variables’ relationship. The causality test is now required to determine the direction of the relationship. To determine whether the observed variables are causal, the DH granger causality test is used. The study used the bootstrap method to conduct this test because the results of the slope heterogeneity and CSD tests indicated the presence of heterogeneity and CSD in the panel. The results of the causality test are shown in Table 7. The existence of a bidirectional causality relationship between Green finances and EP can be verified. In other words, Green finances have an impact on EP, while changes in EP effect Green finances. The results of the causality tests also show that EE and EP have a one-way relationship, whereas GDP per capita, urbanization, and EP have a two-way relationship.

TABLE 7

Causality test.

Null hypothesis	Wald stattictics	p-Value	Decision
lnGF does not cause lnEP	5.876	0.00	lnGF ↔ lnEP (bidirectional causality)
lnEP does not cause lnGF	4.225	0.05
lnGovS does not cause lnEP	4.694	0.00	lnGovS ↔ lnEP (bidirectional causality)
lnEP does not cause lnGovS	2.687	0.03
lnEE does not cause lnEP	5.201	0.00	lnEE → lnEP (unidirectional causality)
lnEP does not cause lnEE	2.474	0.49
lnU does not cause lnEP	9.874	0.00	lnurban → lnEP (unidirectional causality)
lnEP does not cause lnU	2.376	0.74
lnFDI does not cause lnEP	4.740	0.00	lnFDI ↔ lnEP (bidirectional causality)
lnEP does not cause lnFDI	4.523	0.00

Robustness Test

In this section, we perform robustness test to ensure the validity of our findings. Various values for λ are considered in the robustness test. The results of the study were tested to see if they were robust to different input values of λ. We conduct test with various values of λ varying from 0.1 to 1.5. To check robustness test, we only show the most important variables interest. Table 8 summarizes the findings. The results are nearly identical to those of the PQR with = 1. These findings corroborate our previous findings, which are listed in Table 6. In short, the sensitivity test confirms that GF, GovS, and EE have a positive and consistent impact on EP in Latin American countries. As a result, our findings are robust when we use different proxies and change the regression approach.

TABLE 8

Robustness test.

Variables		5th	10th	20th	30th	40th	50th	60th	80th	90th
lnGF	λ = 0.1	0.297	0.306	0.370	0.378	0.384	0.409	0.415	0.423	0.424
		0.017	0.018	0.022	0.015	0.017	0.020	0.022	0.023	0.018
lnGovS		0.438	0.451	0.442	0.358	0.454	0.612	0.668	0.421	0.342
		0.077	0.079	0.097	0.068	0.077	0.087	0.099	0.103	0.078
lnEE		0.760	0.783	0.723	0.679	0.640	0.593	0.521	0.462	0.479
		0.024	0.025	0.030	0.021	0.024	0.027	0.031	0.033	0.024
lnGF	λ = 0.50	0.317	0.327	0.396	0.404	0.411	0.437	0.444	0.453	0.454
		0.018	0.019	0.023	0.016	0.018	0.021	0.023	0.024	0.019
lnGovS		0.468	0.483	0.473	0.383	0.486	0.654	0.714	0.451	0.366
		0.082	0.085	0.103	0.072	0.082	0.093	0.106	0.110	0.083
lnEE		0.813	0.838	0.773	0.727	0.684	0.634	0.558	0.494	0.513
		0.026	0.027	0.032	0.022	0.026	0.029	0.033	0.036	0.026
lnGF	λ = 1.0	0.374	0.386	0.467	0.477	0.485	0.516	0.524	0.534	0.536
		0.022	0.022	0.028	0.018	0.021	0.025	0.028	0.029	0.022
lnGovS		0.553	0.570	0.558	0.452	0.574	0.772	0.843	0.532	0.432
		0.097	0.100	0.122	0.085	0.097	0.110	0.125	0.130	0.098
lnEE		0.959	0.989	0.913	0.857	0.808	0.748	0.658	0.583	0.605
		0.031	0.032	0.038	0.026	0.030	0.034	0.039	0.042	0.030
lnGF	λ = 1.5	0.431	0.444	0.538	0.548	0.557	0.593	0.603	0.615	0.616
		0.025	0.026	0.032	0.021	0.024	0.029	0.032	0.033	0.026
lnGovS		0.636	0.655	0.642	0.519	0.660	0.888	0.969	0.612	0.497
		0.111	0.115	0.140	0.098	0.112	0.127	0.143	0.149	0.113
lnEE		1.103	1.137	1.049	0.986	0.929	0.861	0.757	0.670	0.696
		0.035	0.036	0.044	0.030	0.035	0.039	0.045	0.048	0.035

Standard errors in parentheses ***p < 0.01, **p < 0.05, *p < 0.1. *,**, and *** means 10, 5, and 1% significance level.

Conclusion and Policy Recommendation

Poverty alleviation is a serious practical and social issue that people all around the world are dealing with. Similarly, it must be addressed as part of the process of establishing a prosperous society. Despite numerous attempts to reduce global EP, the problem persists, with the incidence of EP being higher in developing countries. Despite its potential to aid in the fight against EP, the Green finance, and EP nexus has received little attention among the many policy options being considered in the literary space. Thefore the major aim of this research is to find out how Green finance, government spending, and EE affects EP. The PQR approach is used to achieve the objective of the study. This technique considers unobserved individual heterogeneity as well as distributional heterogeneity. In addition, several related control variables are included in the model to avoid an omitted-variable bias. We believe that PQR models can provide a more thorough view of the indicators that influence EP than OLS mean regression. This study examines the annual sample period in the Latin American countries from 2005 to 2018. Overall, the findings showed that Green finance has a positive impact on EP, which is supported by alternative quasi-experimental methods. Other sensitivity checks, such as alternative weighting systems for Green finance and multidimensional EP, as well as alternative EP cut-offs, show that it is robust. Moreover, the government spending and energy efficeicny has a signifincat effect on EP. We conclude that Green finance, government spending, and EE significantly alliviate EP in the 33 Latin American countries.

Finaly, these findings may be useful to policymakers and governments because they can assess how EP has evolved in Latin American countries over the last 16 years, particularly in light of the financial crisis, which countries were the most affected, and which parameters can be considered EP determinants. As a result, plans and initiatives will be implemented more effectively and directly, potentially leading to the successful mitigation and eradication of EP in Latin American countries. These findings will also aid in the successful prevention and the development of resilience of an increase in EP levels as a result of a future economic crisis. EP drivers must be identified in order to successfully reduce negative outcomes through targeted policies, especially in light of the recent COVID-19 pandemic, which had direct and severe socioeconomic impacts on industrial sector, potentially worsening EP conditions.

Policy Recommendation

We put forward some policy recommendations based on the outcomes of this study.

The impact of Green finance in reducing EP has been found to be quite beneficial. As a result, it is recommended that policymakers focus and develop precise plans for delivering financial assistance to those in need. Furthermore, concentrating on the construction and improvement of government funding risk-sharing and guarantees arrangements, as well as utilizing financing platforms of government such as small loans with greater flexibility. For financial poverty alleviation, policymakers must implement local financial support programs on a serious basis, after that, it may concentrate on developing a modern, inclusive rural finance system. The rural revitalization strategy should be implemented as soon as possible. Poverty alleviation is the primary goal of creating a prosperous society in general. As a result only a strong financial support for rural revival policies can encourage the rural sector to experience fresh growth.

Technological innovation is critical for the alleviation of EP; therefore, it is recommended that technological innovation determination to alleviate poverty be strengthened, and to alleviate poverty technological innovation be promoted through market mechanisms. Incentives should be strengthened, and existing innovative talent should be utilized. Make good use of its counter-aid policy, keep looking for new ways to help with technology, and improve scientific research institutions’ innovation capabilities. With the right resource endowment, the government can provide new technologies. The agriculture sector and rural poverty are two areas where technological innovation can be focused. Adopting the technological production capacity of natural environment, innovation, social, and economic conditions can all play a key role. The importance of technological advancements is also demonstrated by the fourth industrial revolution; thus, focusing on this point in policy making will benefit people and the economy in the long run.

Energy poverty has a direct impact on the country’s ability to meet its long-term sustainable development goals, according to the findings of this study (SDGs). Transitioning from fossil fuels to solar and wind power is a good thing, because it can help the country better withstand EP shocks (but only partially). Countries should not rely solely on green energy policy as a panacea for all of their development challenges, even if this policy shift is beneficial. In addition, countries that are adopting RE should be aware of the potential risks and devise measures to minimize these risks.

Limitations

Long-term unavailability of data was a major limitation faced in conducting this study. Therefore, a country-specific analysis could not be performed. As part of the scope of future research, this study can be replicated to other groups of economies to examine the overall generalizability of the findings.

Data Availability Statement

The original contributions presented in this study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

Author Contributions

All authors listed have made a substantial, direct, and intellectual contribution to the work, and approved it for publication.

Conflict of Interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Publisher’s Note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.