Biomass energy consumption and financial development: evidence from some developing countries

ABSTRACT

The importance of renewable energy in the economic structures of countries is increasing and biomass is one of the most widely used renewable energy types. In this paper, the nexus between biomass energy consumption (BEC) and financial development (FD) for developing countries in the period between 1990 and 2018 has been investigated with Konya (2006) bootstrap panel causality tests, Westerlund (2006) panel cointegration with multiple breaks, CCE and AMG panel cointegration estimator are suitable for cross-section dependency and heterogenous panels. According to empirical findings, BEC and FD have cointegrated in the long-run and this relationship is positive. Moreover, the growth hypothesis is valid for Argentina, Bolivia, Mexico, Uruguay and El-Salvador; neutrality hypothesis is valid for Chile, Colombian, Peru, Turkey and Panama.

KEYWORDS:

• Biomass energy consumption
• financial development
• developing countries
• panel cointegration
• panel causality

JEL CODES:

• C33
• Q43
• P28

1. Introduction

Today, with the increase in the use of technology and globalisation, energy has started to be needed and used in all areas. It has also become one of the most basic needs of people. As the population increases, energy use is steadily increasing, and as a result, energy resources are constantly declining. Energy consumption (EC) has great importance in the world economy (Aslan 2016) and remains one of the significant triggers in ensuring economic growth (Destek 2017). Especially after the 1970 oil crisis, the importance and consumption of energy resources with renewable and sustainable structures have increased gradually (Bildirici and Özaksoy 2013). The main reasons for this usage are fluctuations in oil prices, dependence on fossil energy sources, climate change, global warming, and environmental problems. Besides, the distribution of non-renewable energy resources in the world is unbalanced as well as being limited in terms of both quantity and diversity. In parallel with this situation, with the Kyoto Protocol in 1997, the importance of environmental cleanliness and sustainability was recognised by both developing and developed countries (Öztürk and Bilgili 2015). Moreover, the more costly use of renewable energy sources stands out as one of the reasons for the use of non-renewable energy resources. According to the Environmental Kuznets Curve hypothesis, the demand for non-renewable energy will decrease, the demand for renewable energy sources will increase in the long term, and the share of environmentally friendly energy sources in total energy consumption is expected to increase with the increase in per capita income.

According to the research conducted in the literature, Sadorsky (2011) has the first study to analyse the relation between EC and FD and states that this relationship may be due to several possible reasons. First, the prices and consumption costs of vehicles that provide EC such as home, automobile, washing machine, and the refrigerator can affect the money needs of people. In this context, it is possible to explain how EC positively affects FD indicators. Considering that biomass energy is a source that can be converted into electricity (Bildirici 2013), it is possible to investigate the relationship between BEC and FD. Second, a reverse relationship can be explained as follows; it is a possible result that people or countries with economically developing economies use their energy resources more economically and thus decrease their EC costs. Also, increasing economic prosperity due to increased FD may lead to the incentive to spend money comfortably and an increase in EC (Zeren and Koç 2014). In order to meet this increasing EC, all countries in the world will be directed to renewable energy types. Renewable energy types are divided into five groups: hydroelectric, geothermal, wind, solar, and biomass energy. In this context, biomass, which is one of the renewable energy sources in today’s world, has been of serious importance. Biomass, which emerges as the origin of plants and living organisms, is often referred to as vegetative organisms that store solar energy with the help of photosynthesis. Through this photosynthesis, the oxygen needs of living beings in nature are met and this reveals the environmental structure of biomass energy. The total mass of living organisms belonging to a species or a community of various species at a given time can define as Biomass (EIA, Energy Information Administration). Biomass energy, an important renewable resource from the formation of electricity and other types of energy, contributes to poverty reduction and increases rural employment in underdeveloped and developing countries (Bilgili and Ozturk 2015). In this regard, biomass energy is of great importance in the provision of economic and FD especially in developing countries (Shahbaz et al. 2016).

As of 2016, according to a report of World Bio Energy (WBA 2018), BEC constitutes 70% of the total renewable EC. In fact, biomass is the oldest energy source in world history and consists of three different types: waste, wood, and alcohol fuels (Payne 2011). Biomass energy is consumed directly/traditionally and indirectly/modern (Bildirici and Özaksoy 2013). Direct BEC takes place in three different ways: cooking, heating, and industrial processes (Bildirici 2014). Indirect consumption is realised by converting secondary energy sources to biomass energy (Bildirici 2013).

When the studies in the literature are analysed in general, four basic hypotheses examine the relationship between growth and EC. The first is the Growth Hypothesis that expresses when EC is the cause of growth. The second hypothesis is the Conservation Hypothesis that will be valid if causality from growth to EC is detected. Third, the feedback hypothesis is explained as a two-way causality nexus between the two variables. The neutrality Hypothesis is the last. For this, there should be no relation between growth and EC. Since biomass is an energy type and FD is a growth type, the findings obtained from the study will be evaluated within the framework of these hypotheses.

This study aims to examine the relationship between biomass energy consumption and financial development. The most reasonable solution to ensure economic and financial growth for countries is to ensure sustainable production. However, it does not seem possible for countries that are dependent on foreign countries to provide this sustainability. Accordingly, it is important to determine the type of energy that developing countries should concentrate on to ensure production and financial growth. At this point, it will be a serious step for developing countries to provide this issue. In this respect, it was deemed appropriate to select the sample of developing countries, which are particularly affected by global economic developments, depend on foreign energy, and need economically sustainable growth. In this context, researching the relationship between Biomass which is one of the most produced and used renewable energy types, and economic and financial growth will add originality to the study. Being the first study to be done in this direction constitutes the motivation of the study.

In the second part, the findings of the studies examined in the literature are presented. The data and econometric method used in the third part of the study are mentioned, and empirical findings are presented in the fourth part. In the last part, the results and policy suggestions are presented.

2. Literature review

The relation between economic growth and BEC has been proven by many research, with Payne’s (2011) work leading the research. As a result of the study using the Toda-Yamamoto causality test, one-way causality has been determined from BEC, which supports the growth hypothesis, to economic growth. Following Payne’s (2011) work, three different studies have been published by the Bildirici that examined different countries and used different methods. In the first of these studies, Bildirici and Özaksoy (2013) examined 10 European countries and found different findings on the direction of causality for these countries. In the second study of Bildirici (2013), which examined 10 countries from the South American continent, the Bildirici encountered long-term evidence of cointegration for 9 of these countries. In another study, Bildirici (2014) aimed to estimate the cointegration coefficients between BEC and economic growth by using the Panel ARDL approach. According to the results obtained, BEC has been proven to have a positive effect on economic growth.

It is possible to come across two studies of Öztürk and Bilgili about the mentioned relationship in 2015. In the first of these studies, 51 sub-Saharan African countries have been examined by using dynamic panel data analysis. The results obtained are in support of the findings of Bildirici (2014) and BEC has a positive effect on economic growth. In the other study of Bilgili and Ozturk (2015), G7 countries have been examined and similar findings have been found.

In another study that examined the United States, Aslan (2016) found causality from EC to economic growth supporting previous studies and confirmed that BEC contributed positively to economic growth through ARDL testing. The other works belong to Shahbaz et al. (2016) and Destek (2017), and both studies have determined that the feedback effect hypothesis exists between BEC and economic growth. However, in the study of Destek (2017) has been found that the growth hypothesis and the neutrality hypothesis are valid for some countries.

Bildirici and Özaksoy (2018) have been studied for transition countries. As time-series analysis results of the study, it is not possible to reach a general opinion, conversation hypothesis is valid for some countries such as the Czech Republic, Hungary and growth hypothesis is valid for some countries like as Slovenia, Estonia. Also, according to panel data analysis results, the feedback hypothesis is valid in all countries.

The studies about the investigation of the relationship between biomass and economic growth also exist in 2019 and 2020. In one of the studies conducted in this direction, Solarin and Bello (2019) discussed the Brazilian sample. As a result of the analysis, it has been determined that Brazilian policymakers can achieve sustainable economic growth by consuming less non-renewable energy and more BEC. The Indonesian sample was taken in the study of Sinaga et al. (2019) and stated that increasing the BEC contributed significantly to both economic and environmental development.

Aydin (2019) analysed the relationship between BEC and economic growth and find that the growth hypothesis for Brazil and India, the conversation hypothesis for China and South Korea, and finally the feedback hypothesis for Russia. Destek and Okumuş (2019) discussed this relationship for the G-20 countries and found that the growth hypothesis was valid for these countries. In one of the recent studies in the literature, Ajmi and Inglesi-Lotz (2020) examined this relationship for 26 OECD countries. As a result of the study, the feedback hypothesis was found to be valid, it was confirmed that biomass is an important renewable energy source in improving economic growth.

When the studies carried out in the literature are analysed, besides, the relation between the BEC and economic growth, many works in EC and economic growth, EC and FD matches are found. On the other hand, the relationship between BEC and FD is seen as a virgin area suitable for research. This study reveals its originality in terms of being the first study that examines the relation between BEC and FD. This relationship has been explained by panel cointegration and causality analysis method.

The data and econometric method used in the third part of the study are mentioned, and empirical findings are presented in the fourth part. In the last part, the results and policy suggestions are presented.

3. Data and methodology

In this paper, annual data has been used in the study and covers the period of 1990–2018. Data of BEC and FD has been obtained from the official website of U.S. Energy Information Administration and the official website of the World Bank respectively. Information about the data set of the study based on the research in the literature is presented in Table 1.

Table 1. Information on the data set of the study.

Abbreviation of variables	Variables used in the study	Unit of the variables	Researches using the variables
BEC	Biomass energy consumption	Used Extraction of Biomass (ktoe: kilotons of oil equivalent)	Bilgili and Ozturk (2015), Öztürk and Bilgili (2015), Shahbaz et al. (2016), Bildirici and Özaksoy (2018), Destek (2017), and Aydin (2019)
FD	Financial development index	Index Consisting of Banking Sector Development, Bond Market Development, Stock Market Development	Tang and Tan (2014), Shahbaz et al. (2016), Topcu and Payne (2017), and Destek (2017)

FD used in the study is a variable created by combining three separate data. FD is analysed according to three sub-market indices and FD indices. The financial development index consists of the indices of these three sub-markets which are the bond market, stock market, and banking sector. The banking sector index is formed using deposit money bank assets to GDP, financial system deposits to GDP, liquid debt to GDP, and special loans to GDP of deposit money banks. The stock market index covers the stock market capital to GDP, the stock market turnover ratio and the total value of the stock market traded to GDP. Domestic private debt securities outstanding to GDP, domestic public debt securities outstanding to GDP, international private debt securities outstanding to GDP, and international public debt securities outstanding to GDP are the composing of the bond market index (Tang and Tan 2014, Shahbaz et al. 2016, Topcu and Payne 2017, Destek 2017).

As can be understood from the variables related to FD, each of them has been used as the rate of GDP. In this way, GDP, which is one of the most important indicators of the economic growth of countries, is the control variable of the study. Accordingly, the model of the study has been established as follows; $BEC=\propto +{\beta }_1\left({\displaystyle \frac{FD}{GDP}}\right)+u.$ $\left({\displaystyle \frac{FD}{GDP}}\right)=\propto +{\beta }_{1}BEC+u.$In the study, following the slope homogeneity and cross-section dependency tests, Fourier Panel ADF test developed by Lee, Wu, and Yang (2015) has been applied in accordance with the results obtained from them. The expansion of the periodic y_t function in the form of an infinite cosine and sine sum is expressed as the Fourier series. The Fourier ADF test uses trigonometric functions to capture large variations from the mean of the dependent variable and it is developed by Enders and Lee (2012). The most important advantage of this test is that it can find structural breaks with a smooth transition that traditional refraction tests cannot find. The econometric equation of this test can be described as following: $y_t={\lambda }_0+{\lambda }_1\sin \left({\displaystyle \frac{2\pi kt}{T}}\right)+{\lambda }_2\cos \left({\displaystyle \frac{2\pi kt}{T}}\right)+v_t$where λ₁ and λ₂ denotes Fourier coefficients, π refers 3.1416, the means of k is the frequency used to find the optimal value that makes the sum of residual squares the least and T refers to the sample size. This method is a panel-applied form of the Fourier ADF test developed by Enders and Lee (2012). According to these tests, the null hypothesis (H₀)point outs the presence of the unit root in the panel, and the alternative hypothesis (H₁) point outs that the series is stationary.

Another method used in the study is Westerlund’s (2006) panel co-integration test. This test permits multiple structural breaks, as well as determines the results with and without cross-section dependency through two separate test statistics. This method, which has a structure similar to the Maki multiple break cointegration test in the time series, determines the number of structural breaks endogenously by the model, not exogenously, unlike traditional one or two structural breaks models.

This method is referred to as multiple structural breaks, not any number of structural breaks. It is not possible to forecast the date and number of structural breaks. For this reason, the number of structural breaks will be determined in accord with the structure of the data. Westerlund (2006) multiple structural fracture panel cointegration test is implemented with the help of the following model: $Z(M)=1/N\sum _{(i=1)}^N\sum _{(j=1)}^{(M_{?}+1)}\sum _{(t=T_{(ij-1)}+1)}^{(T_{?}j)}\left({\displaystyle \frac{S_i{t}^2}{{(T_{i}j-T_{(ij-1)})}^2{\sigma }_I^2}}\right)$

The symbol S can be defined as ${\sum }_{s=T_{ij-1}+1}^t{W}_{st}$. Also, like the Fully Modified Least Squares (FMOLS) method, the vector of residues obtained from an effective estimator is expressed as S. The H₀ of this test indicates that there is a cointegration relationship in the panels, while the (H₁) indicates that there is no co-integration relationship in the panels. The last method used in the study is the Bootstrap panel causality test developed by Kónya (2006) and it works accordingly Seemingly Unrelated Regression (SUR). This estimation with the SUR method takes place as follows: $y_{1,t}={\alpha }_{1,1}+\sum _{l=1}^{ml{y}_1}{\beta }_{1,1,l}{y}_{1,t-1}+\sum _{l=1}^{ml{x}_1}{\gamma }_{1,1,t}{x}_{1,t-1}+{\epsilon }_{1,1,t}$ $\begin{array}{rl}{y}_{2,t}& ={\alpha }_{1,2}+\sum _{l=1}^{ml{y}_1}{\beta }_{1,2,l}{y}_{2,t-1}+\sum _{l=1}^{ml{x}_1}{\gamma }_{1,2,t}{x}_{2,t-1}+{\epsilon }_{1,2,t}\\ & ={\alpha }_{1,N}+\sum _{l=1}^{ml{y}_1}{\beta }_{1N,l}{y}_{N,t-1}+\sum _{l=1}^{ml{x}_1}{\gamma }_{1,N,t}{x}_{N,t-1}+{\epsilon }_{1,N,t}\end{array}$and $x_{1,t}={\alpha }_{2,1}+\sum _{l=1}^{ml{y}_1}{\beta }_{2,1,l}{y}_{1,t-1}+\sum _{l=1}^{ml{x}_1}{\gamma }_{2,1,l}{x}_{1,t-1}+{\epsilon }_{2,1,t}$ $x_{2,t}={\alpha }_{2,2}+\sum _{l=1}^{ml{y}_1}{\beta }_{2,2,l}{y}_{2,t-1}+\sum _{l=1}^{ml{x}_1}{\gamma }_{2,2,l}{x}_{2,t-1}+{\epsilon }_{2,2,t}$ $x_{N,t}={\alpha }_{2,N}+\sum _{l=1}^{ml{y}_1}{\beta }_{2N,l}{y}_{N,t-1}+\sum _{l=1}^{ml{x}_1}{\gamma }_{2,N,l}{x}_{N,t-1}+{\epsilon }_{2,N,t}$

Here the symbol l denotes the number of lags, and y and x represent the variables BEC and FD, respectively. The model’s (H₀) indicates that there is no causality between the panels, and the (H₁) indicates that there is causality in the panels. Empirical application of these methods, which are explained mathematically in the following sections of the study, has been done and the results have been presented.

4. Empirical findings

The investigation of the relationships between BEC and FD has been performed through panel data analysis. In this regard, slope homogeneity, cross-section dependency and unit root test are some preliminary tests and required to detect these relationships. The first step to be done in panel data analysis is to determine the homogeneity of the panels.

According to the Pesaran and Yamagata (2008) results presented in Table 2, both panels have a heterogeneous structure. Because both delta (Δ) and corrected delta (Δ_adj) values are econometrically significant at 1% level. So, these results are correct with 99% reliability.

Table 2. Slope homogeneity test.

	Δ	P-value	Δadj	P-value
BEC	7.144	0.00	7.596	0.00
FD	6.886	0.00	7.298	0.00

Another pre-test that needs to be done to determine which methods will determine the relationships between the panels is the cross-sectional dependency analysis. In the studies conducted until 2008 to examine the existence of cross-sectional dependence, The Breusch Pagan (1980) CDLM test has been used, which gives accurate results when individual means are zero, but not reliable when individual means are different from zero. Along with the work of Pesaran, Ullah, and Yamagata (2008), this deficiency has been detected by adding mean and variance to the test statistics.

Therefore, the adjusted CDLM test, which gives the new cross-sectional dependence test statistics, has been used in the study. According to the findings shown in Table 3, Breusch and Pagan (1980) CDLM test showed the presence of cross-sectional dependency with one of the panels with 99% reliability and the other with 90% reliability, while the CDLM_adj method of Pesaran, Ullah, and Yamagata (2008) with the reliability of 99% reliability for both panels used in the study is indicating. Although the two tests mentioned offer nearly analogue results, the findings of the CDLM_adj test developed by Pesaran, Ullah, and Yamagata (2008) are stronger in terms of reliability.

Table 3. Cross-sectional dependency test results.

	CDLM	P-value	CDLMadj	P-value
BEC	88.03	0.00	9.53	0.00
FD	61.027	0.056	9.99	0.00

It is possible to decide which panel unit root test to use based on the findings of homogeneity and cross-section dependency test. In such a situation where there is a cross-sectional dependence between the panels and a heterogeneous structure, it is appropriate to use the Fourier Panel ADF unit root test developed by Lee, Wu, and Yang (2015). In addition to these advantages, it takes into account a lot of number of structural breaks with smooth transition.

When the panel unit root test results presented in Table 4 are examined, it is understood that the panels have a unit root in their level values, and when their first differences are taken, they become stationary. Because the test statistics, which are higher than the critical values in the level value, fall below the critical values after the difference is taken. These results are valid for all three Fourier frequencies.

Table 4. Lee, Wu, and Yang (2015) results of Panel ADF unit root test with Fourier function.

	Fourier frequency	Level	First difference
BEC	1	−3.27	−4.64***
	2	−2.36	−4.37***
	3	−2.17	−4.42***
FD	1	−1.89	−3.72***
	2	−0.97	−3.44***
	3	−0.76	−2.99***

Note: Critical values are taken as – 3.64 for 1 Fourier frequency, −3.37 for 2 Fourier frequency, −2.97 for 3 Fourier frequency obtained from this study and *** expresses with 99% reliability.

Since the panel unit root test is a model that takes into account the structural breaks, the following cointegration test will also be in accordance with this structural break model and it will be a correct approach to ensure unity in practice. In this direction, Westerlund panel cointegration test (2006), which performs cointegration analysis considering multiple structural breaks, has been used at this stage of the study. This test presents the results of both cross-section dependence and non-cross-section dependence with asymptotic and bootstrap p-value values. In the nonexistence of cross-sectional dependency, asymptotic p-value is used, in the other case, bootstrap p-value is used.

According to Table 3, panels forming the dataset of the study have cross-section dependence. For this reason, bootstrap p-value will be considered when evaluating panel co-integration test results. Accordingly, when the results in Table 5 are analysed, the H₀ that there is cointegration is accepted in the panels. From this point of view, if there is no horizontal cross-section dependency, the asymptotic p values are considered, and the H₁ is accepted and the panels indicate that there is no cointegration. On the other hand, when the breaks detected in this test are examined, it is understood that there are 2 or 3 structural breaks for the countries under consideration. Many of these structural breaks coincide with the 1997 Southeast Asian Crisis, the 2001 Dot-com Bubble, the 2001 Enron Scandal, and the 2008 Mortgage Crisis, and it can be considered as evidence of how strong the method used is to detect multiple breaks like these.

Table 5. Westerlund (2006) panel cointegration test with multiple breaks.

	LM Stat	Asymptotic P-value	Bootstrap P-value
BEC-FD	572704.446	0.000	0.113
Structural breaks dates
	First break	Second break	Third break
Argentina	1997	2002	–
Bolivia	1996	2001	2007
Chile	1997	2003	2009
Colombia	1997	2003	–
Mexico	1999	2004	2009
Peru	1999	2004	2009
Turkey	1998	2010	–
Uruguay	1996	2001	2007
Panama	2000	2005	–
El Salvador	1998	2007	–

After the findings obtained from Table 5, the analysis to be performed will be the determination of this cointegration direction. Accordingly, Pesaran’s (2006) CCE and Eberhardt and Bond’s (2009) AMG panel cointegration estimator, which takes into account the cross-sectional dependency and is possible if the panels are in a heterogeneous structure, is used. When the results presented in Table 6 are examined, it is understood that the cointegration relationship between BEC and FD is positive and the results of both panel cointegration estimators are consistent with each other. Accordingly, it is understood that increasing the BEC will have a positive contribution to the FD of the countries included in the study.

Table 6. CCE and AMG mean group estimation results.

	Coefficient	Standard deviation	T statistics
CCE	12.478	0.68	−3.69
AMG	12.469	0.70	−3.65

At this part of the study, bootstrap panel causality test developed by Kónya (2006), which takes into account heterogeneity and cross-sectional dependence, has been used. This test shows the Wald statistics obtained for each country in the panel and the results obtained are presented in Tables 7 and 8. According to these results, H₀ hypothesis is accepted and no causality has been found from FD to BEC for any country. On the other hand, unidirectional causality has been found from BEC to FD for Argentina, Bolivia, Mexico, Uruguay, and El-Salvador that means the growth hypothesis is valid for these countries. However, any causality has not been found for Panama, Chile, Colombia, Peru, and Turkey so, neutrality hypothesis is valid.

Table 7. From FD to BEC bootstrap panel causality test results.

	Wald statistics	Bootstrap p-value	Critical values
			%1	%5	%10
Argentina	1.452	0.37	5.225	3.559	3.065
Bolivia	0.282	0.88	14.188	10.234	6.911
Chile	0.185	0.94	6.323	6.117	5.042
Colombia	0.097	0.60	4.792	2.376	2.163
Mexico	0.087	0.76	6.977	3.567	2.478
Peru	1.379	0.64	7.234	5.731	5.233
Turkey	0.857	0.78	8.759	7.388	5.213
Uruguay	1.199	0.36	5.153	3.046	2.488
Panama	0.000	0.99	7.112	4.993	3.614
El Salvador	0.487	0.40	2.838	2.074	1.530
Panel Fisher	9.224		p-Value	0.98

Table 8. From BEC to FD bootstrap panel causality test results.

	Wald statistics	Bootstrap p-value	Critical values
			%1	%5	%10
Argentina	10.582***	0.00	9.094	5.813	4.077
Bolivia	12.941**	0.02	19.393	10.847	8.66
Chile	0.002	0.99	8.299	5.402	4.189
Colombia	0.796	0.20	4.167	2.740	1.530
Mexico	31.021***	0.01	15.032	9.526	7.733
Peru	0.285	0.99	18.797	15.424	13.764
Turkey	4.135	0.46	26.254	11.671	9.767
Uruguay	9.934***	0.01	5.592	3.545	2.742
Panama	0.781	0.43	11.956	5.197	3.828
El Salvador	2.257**	0.04	3.821	2.051	1.804
Panel Fisher	39.162		p-Value	0.006

Note: *** 99% and ** 95% represent level of significance with reliability.

5. Conclusion remarks

Biomass energy use is an important factor for both economic growth and FD. Biomass energy use can reduce dependence on oil, and if the country is dependent on imports of oil, its dependence on imports can decrease and affects foreign trade development in a positive way. At the same time, biomass energy can lead to rural employment, resulting in a reduction the unemployment rate and a positive contribution to economic growth. Along with the results of these effects, economic growth stability can be achieved, thus the country can be in a position suitable for foreign investment. With the foreign investment, it can be said that the country’s stock and banking markets will be affected, which will have a positive impact on FD.

The first paper to investigate the relationship between BEC and FD reveals the originality of the study and aims to be a pioneer in the literature by filling the gap in this subject. In this study, the relations between BEC and FD for the period between 1990 and 2018 have been examined with Westerlund (2006) panel cointegration with multiple breaks, Pesaran’s (2006) Common Correlated Effects (CCE) panel cointegration estimator, and Kónya (2006) bootstrap panel causality tests. The economic crisis periods in the data range of the study have been taken into account by means of these methods that take into account the structural breaks. According to the Panel Cointegration test results, it is understood that BEC and FD act together in the long term. Moreover, the relationship with the cointegration estimator has been found to be positive. On the other hand, causality from BEC to FD has been determined in most of the countries throughout the panel that means the growth hypothesis is valid. Also, it is understood that the neutrality hypothesis is valid for the countries that do not have any relation between them. The findings obtained as a result of the research are similar to the studies of Payne (2011), Bildirici (2013), Bildirici (2014), Bilgili and Ozturk (2015), Aslan (2016), and Destek (2017) in the literature.

According to the findings, BEC is an important trigger for economic growth and FD for developing countries. Therefore, energy policies to improve energy infrastructure and increase energy supply will be the right option for these countries. In this regard, it is aimed to enhance the consumption and production of modern biomass energy in order to tackle with climate change problems and global warming, which has serious negative effects on the global economic environment, to decrease greenhouse gas emissions and foreign dependency levels in energy, to improve the biomass energy infrastructure and to attract technology-intensive investments to the country. Sustainable economic growth and FD will be achieved by reducing the dependence on non-renewable EC by policymakers acting in this direction and increasing renewable EC. In this way, it seems possible to ensure sustainable development by minimising the poverty level in developing countries. In this context, meeting the renewable energy needs of countries due to the increase in production, reducing external dependence on energy imports, reducing costs arising from energy imports, and eliminating the volatility that may occur in energy supply may be possible thanks to investments in biomass energy. It will also ensure to create new employment areas and their long-term growth plans will be more effective in developing countries, in particular. In short, it can be said that it will contribute to the economic growth of countries. However, the installation of renewable energy sources such as Biomass is highly cost. In addition to this cost, non-renewable energy use continues for reasons such as continued subsidies for fossil fuels and the exclusion of the total cost of pollution in the cost of fossil fuels. Therefore, its preference for the use of renewable energy like biomass, and to increase investment in these resources should be supported by different economic policies, such as tax incentives. In this direction, it will be possible to reduce external dependence on energy.

In future studies, the classifications used by the global authorities such as the World Bank and OECD can be analysed by taking into account the income levels of the countries where the data subject to the research can be accessed. In addition, studies to be carried out by adding variables such as foreign direct investments, trade openness, CO₂ emission to the model of the study will reveal important findings in this regard. Moreover, studies that will reveal the relationship between renewable energy types such as solar, wind, hydropower, and geothermal with FD and economic growth will make important contributions to the literature in question.

Disclosure statement

No potential conflict of interest was reported by the author(s).