The impacts of governance on environmental pollution in some countries of Middle East and sub-Saharan Africa: the evidence from panel quantile regression and causality

Governance is one of the basic determinants of pollution levels through property rights, the effective judicial system, etc. it is accepted as that bad governance because of inefficient regulatory structures, government bureaucracy, weak law enforcement, etc. support environmental pollution. In this context, in some countries of the Middle East and sub-Saharan Africa, it will be studied the impacts of governance on environmental pollution over the period of 1996–2018 by the panel quantile and Granger causality methods. The countries were selected by considering two different measurements, EPI (2020) index and governance index (2020). According to EPI (2020), these countries have low scores in terms of environmental quality, and in the governance index (2020), they have bad governance scores. In this study, in which panel quantile regression model is used, control variables are included in the model to prevent omitted-variable bias. The results of the analysis determined that the effect of governance on carbon emissions is positive, as well as that the effects of independent variables on CO2 emission are heterogeneous across quantities. Panel quantile regression revealed the evidence of the relation among the environmental pollution, two parameters of governance, FDI, financial development, human development index, and trade openness used as the explanatory variable and determined that government has the greatest positive effect on CO2 emission. On the other hand, by using traditional Granger causality and Dumitrescu-Hurlin causality methods, it was found the evidence of causality among governance and environmental pollution in the context of two parameters of governance. Accordingly, it was determined the evidence of unidirectional causality relation from political governance to environmental pollution and besides from economic governance to environmental pollution. And it was determined the evidence of unidirectional causality from FDI and the other explanatory variables to environmental pollution.


Introduction
Environmental pollution under the influence of dangerous pollutants such as CO 2 , NO x , and GHG became a major threat worldwide. According to the IEA (2019), CO 2 emissions increased by 33.1 billion tons worldwide in 2018, an increase of about 145% compared to pre-industrial levels (International Energy Agency 2019). The reasons for the rise in CO 2 releases have been discussed in many papers that accent the impacts of various factors. In pursuit of Grossman and Krueger (1991) 1 , many papers accepted industrialization and economic growth as important factors that have impacts on environmental pollution. Although governance has an impact on both economic growth and environmental pollution (Danish et al. 2019), only a few environmental economists discussed that governance has an important role in environmental pollution. Whereas in the context of environmental damages, the governmental system and institutions' quality affect environmental quality both directly and indirectly. Good governance helps to decrease the level of pollution (e.g., Leonard 1988;Shandra et al. 2008;Clapp and Dauvergne 2005;Jorgenson 2009), and it contributes to the environmental improvement and regulation of emission level. In some papers, institutional quality and strong institution structure are accepted as the basic tools (Liu et al. 2020;Hassan et al. 2020) to decrease pollution levels (Hosseini and Kaneko 2013) through property rights, the effective judicial system, etc. And the faults in governance support corruption in the environmental quality (Leitao 2016). So, breaking the rules will eliminate the CO 2 reduction effect (Abid 2016;Liu et al. 2020). Moreover, institutional quality in the frame of governance can control the increasing impact of CO 2 emissions (Goelet al. 2013).
On the other hand, there is an important relation between governance, environmental pollution, and FDI inflow. In the less developed countries, FDI inflows have an adverse effect on environmental sustainability, and besides the relation between environmental sustainability and FDI depends on the role of institutions and governance. The nexus between FDI, environment, and sustainability was explored by WCED (1987). Some papers showed that energy consumption under the influence of FDI inflows is a significant reason for environmental pollution in the frame of the PHH (Pollution Haven Hypothesis) that was developed by Copeland and Taylor (1994). In several articles, the problem of deepening external debt in underdeveloped countries, which is caused by governments imposing exemptions on environmental regulations in order to protect the natural environment from damages caused by various sectoral activities, has been discussed (e.g., Leonard 1988;Shandra et al. 2008;Jorgenson 2009;Clapp and Dauvergne 2005). Cole and Fredriksson (2009) discussed these firms put pressure on policies of host countries for environmental regulations. Desbordes and Vauday (2007) stated that these firms provide regulatory and substantial tax advantages.
In this paper, the relation and causality among environmental pollution, governance, FDI inflow, financial development, human development index, trade openness, and real GDP for a sample of the countries consisting of Afghanistan, Angola, Bangladesh, Benin, Cameroon, Congo Dem., Congo, Chad, Central African Rep, Ethiopia, Ghana, Gambia, Eritrea, Iraq, Kenya, Lesotho, Madagascar, Mali, Mozambique, Myanmar, Niger, Pakistan, Syria, Uganda, and Zimbabwe by using panel quantile regression and panel causality tests and traditional and Dumitrescu-Hurlin will be analyzed during 1996-2018 period. This paper will fill the gap in the literature by analyzing the relation between these variables for Middle East and sub-Saharan Africa countries. While this article is considered to be a complement to previous articles that have an empirical approach in this area, it differs from the literature regarding the simultaneous estimation of the relationship between selected variables using quantile methods.
The countries in this paper were selected by depending on the global environment index (EPI 2020) and governance performance index (2020) of the countries. These countries have low scores of environmental quality. And the selected countries are categorized by inefficient regulatory structures, government bureaucracy, and weak law enforcement. In regard to methodology, firstly panel quantile regression (PQR) method was applied. PQR is robust to heavy distributions and outliers. PQR estimators will provide one result to each quantile. By employing PQR method, the determinants of CO 2 releases through the conditional distribution will be found. From a policy outlook, it is more exciting to see what happens at the extremes of distribution. So, PQR will give the long-run estimations to determine accurate economic policy suggestions that cover the crucial points in this paper. The results of PQR will be compared with one's of two-way OLS, FMOLS, and DOLS. Lastly, traditional Granger causality and Dumitrescu-Hurlin causality tests will determine the direction of causality that is very significant to determine appropriate economic policy suggestions. If these methods give the same results, the results will be considered as correct.
The design of this article was constructed in the following way. The "Literature" section is given in the following section. The third part presents the data and definitions of the variables. The methodology is explained in the "Econometric methodology" section. The "Empirical results" section supplies the econometric results. The discussion has taken place in the "Discussion" section. The conclusions are exhibited in the last part.

Literature
Dryzek (1987) discussed radical decentralization on environmental degradation. Desaii (1998) showed that in developing countries, corruption contributes to environmental problems. Fredriksson and Svensson (2003) found that environmental policies have the impacts on corruption and political instability and showed that governance and the level of corruption have an adverse impact on the stringency of environmental regulation. Fredriksson et.al. (2004), Pellegrini and Gerlagh (2004), and Damania et al. (2003) determined that corruption decreases the severity of environmental regulations. Bhattarai and Hammig (2004) found support for the EKC hypothesis in the context of control variables of institutional factors and the rule of law. Fredriksson et al. (2005) found that democracy encourages the administration to better characterize public preferences and that the governments should prefer more strict environmental policies. Welsch (2004) tested the direct and indirect impacts of corruption on environmental quality and determined that it hampers the application and formation of environmental regulations. Cole (2007) found that corruption positively affects both CO 2 and sulfur (SO 2 ) emissions for 94 countries in the period of 1987-2000. Dutt (2009) examined the relationship between the environment and income in the context of institution and governance. Gani (2012) who examined the relationship between CO 2 releases and five dimensions of governance in 99 developing countries found that good governance enables lower CO 2 releases. Halkos and Tzeremes (2013) found the importance of the relationship between CO 2 emission and governance by non-parametric estimator. Halkos et al. (2015), for the UK, Germany, and France, tested the impact of regional quality on the environment, and they determined that high regional quality will not enable environmental effectiveness. Sarkodie and Adams (2018) showed that corruption control, better governance, and political-institutional quality should be supported to mitigate pollution in South Africa. Abid (2016), in sub-Saharan Africa, explored institutional quality. Danish et al. (2019), for BRICS countries, showed the importance of governance on environmental damages. Liu et al. (2020), for five countries that have high CO 2 emission countries, showed that good governance is a way to protect and enhance environmental quality. Ozturk et al. (2019) analyzed the relation between the control of corruption and energy efficiency. Bernauer and Koubi (2009) and Wilson and Damania (2005) showed green parties' strength and regulatory systems have a nonnegative effect on environmental quality. Adedoyin et al. (2020a) tested climate change protests and their interconnectedness with CO2 emissions 41 countries. Adedoyin et al. (2020b) and Adedoyin et al. (2021) discussed the role of economic policy to control high levels of CO 2 emissions in the region.
If the literature of the relation between FDI inflow and governance is explored, it is seen that good governance is emphasized by some articles. Gani (2007), Globerman and Shapiro (2002) determined a positive correlation between FDI and good governance. Lehnert et al. (2013), Nguyen (2015), and Mengistu and Adhikary (2011) determined the importance to increase the attractiveness of FDI. Peres et al. (2018), Li et al. (2019), and Wang and Chen (2014) determined the impacts of FDI on the environmental damages of the host country are based upon the level of governance and showed that governance of the host country is affected by behaviors and policies of transnational firms.
In the less developed countries, it is accepted that FDI has detrimental impacts on the environment in the frame of exhaustible resources (Solow 1974;Stiglitz 1974;Bokpin et al. 2015). Several articles state that natural resources are the most important determinant of foreign direct investment flows to Africa. Bokpin et al. (2017) showed how FDI towards Africa contributes to the environmental degradation over the period of 1990-2013. The results determined that a rise in FDI inflows importantly rises environmental damages and affirmed that there is a need for strong governance and quality institutions to have non-negative effects on the environmental sustainability of FDI. Some papers focused on relations for the other countries. Omri and Hadj (2020), in 23 emerging countries over 1996-2014, tested how good governance complements FDI to diminish CO emissions. They found that increasing governance quality has negative effects on carbon emissions and that both institutional and political governance decrease the level of CO 2 releases.

Data and definitions
This paper aims to examine the effect of governance on environmental pollution using variables from the Middle East and sub-Saharan Africa countries. CO 2 emissions are used as a measurement of environmental pollution. CO 2 emissions as a dependent variable were measured in terms of metric tons. Annual data covering the period of 1996-2018 for Afghanistan, Angola, Bangladesh, Benin, Cameroon, Congo Dem., Congo, Chad, Central African Rep, Ethiopia, Ghana, Gambia,, Eritrea, Iraq, Kenya, Lesotho, Madagascar, Mali, Mozambique, Myanmar, Niger, Pakistan, Syria, Uganda, and Zimbabwe was employed. The variables used in the analysis are environmental pollution (co), FDI inflow (fdi), real GDP (y), energy consumption (c), human development index (hdi), trade openness (tr), financial development (fin), political governance (p), and government effectiveness (g) as a measurement of governance. Economic growth is represented by realGDP (y) in constant 2005 USD.
Because the relation between governance, energy consumption, CO 2 emissions, and economic growth can be affected by others variables, to avoid omitted variable bias, it was adapted a multivariate approach. As additional explanatory variables, the trade deficit, human development index, and financial development are included in the model. In measuring the trade deficit, the share of the trade deficit in the GDP is taken into account, and financial development is measured by the total value of domestic loans to the private sector as a share of GDP.
All variables for the analyzed countries were obtained from the World Bank (WB) and country statistics. Two governance indicators as economic governance (government effectiveness) and political governance (political stability and absence of violence/terrorism) were used. Government effectiveness and political governance are measured with percentile rank. They were taken from the WGI (Worldwide Governance Indicators) database. And the variables are transformed into logarithms, and they were converted as log(variables t ) In this paper, government effectiveness estimate and political stability and absence of violence/terrorism estimate variables were not used because of their negative values. But these variables and other variables of government effectiveness and political governance express the similar problems. Government effectiveness estimate and political stability and absence of violence/terrorism estimate were drawn to show significant effects of government and political effectiveness. Figures 1 and  2 show government effectiveness estimate and political stability and absence of violence/terrorism estimate According to Figs. 1 and 2, the selected countries have negative values for both political and economic governance variables by signing serious problems in the context of both political governance and government governance.
EPI as other index is in Table 1. According to EPI (2020), the countries in the African continent have low scores of environmental quality. Low scores show the need for domestic sustainability struggles. And global environmental index (EPI 2020) shows how far to establish the environmental policy goals of these countries.
The results of Table 1 show that the lowest ranking of Myanmar, Afghanistan, Cote d'Ivoire, Guinea, and Madagascar indicates the presence of more serious environmental damage in these countries.

Panel quantile method
The panel quantile regression (PQR) that was suggested by Koenker and Bassett (1978) has some advantages over the OLS regression. Firstly, it can be obtained more robust the result from PQR (Bera et al. 2016). And it is not needed to shape distributional assumptions by employing PQR (Sherwood and Wang 2016). Furthermore, this technique can seizure the characteristics of the full conditional distribution of the selected variables (Yu and Jones 1998;Chen et al. 2019). Koenker and Bassett (1978) proposed that quantile regression is useful in examining asymmetric features of variable distributions (Chen and Lei (2018)).
The conditional quantile of y i is given as follows: PQR is robust to heavy distributions and outliers. In this paper, the fixed effect PQR method was used: N is the number of observations, and t is the index of time. The parameter estimate is calculated as follows: To estimate the impacts of FDI (fdi), economic growth (y), political governance (p) and economic governance (g) on CO 2 emissions, it was used equally weighted quantiles wk= 1/K as Ang (2007) and Chen and Lei (2018) and set λ = 1 as Chen and Lei (2018). λ is a setting parameter used to reduce the individual effects to zero by improving the performance of the β estimate (Lamarche 2011;Alexander et al. 2011;Zhu et al. 2016). In this paper, it was setted as λ=1 (Damette and Delacote 2012;Zhu et al. 2016) The conditional quantile function is given as follows: Additional explanatory variables containing of FDI, hdi, fin, and tr are included.

Empirical results
In this paper, the empirical results were obtained in four steps 2 :

1-
The descriptive statistics, cross-sectional dependence tests, and panel unit root tests were given.

2-
The results of panel quantile regression were obtained, and two-way OLS, DOLS, and FMOLS tests determined the long-run coefficients. These tests were preferred to compare the results determined by quantile methods. 3- Kao (1999) and Westerlund (2007) tests to determine cointegration were applied. 4-And lastly, the direction of causal relation by Granger causality and Dumitrescu-Hurlin test for two different conditions was determined.

The unit root tests
Before conducting panel methods, it was tested if there is presence of cross-sectional dependence. To explore crosssectional dependence, the Breusch and Pagan (1980) LM test, Pesaran (2004) scaled LM test, and Baltagi et al. (2012) biascorrected scaled LM test were employed, and the findings were reported in Table 3. As shown in Table 3, all employed tests provide rejection of the null hypothesis of no cross-sectional dependence at the 1% level. To avoid inconsistency, IPS (Im et al. (2003) and CIPS (Pesaran (2007)) panel unit root tests were preferred and presented the results in Table 4.
These results show that for all variables at the 1% level, the H 0 hypothesis of a unit root is not rejected. However, at the first difference, it can be rejected at the 1% level.
And now, it can be examined the evidence of a longrun relation among these variables by employing the Westerlund and Kao tests. According to the results reported in Table 5, H 0 hypothesis can be rejected, so the evidence of the cointegration between the variables is determined.  The results of the model The results of Pedroni DOLS and FMOLS and quantile methods showed that the coefficients of governance have statistically significant by determining that they have effects on environmental pollution. The estimations of the FMOLS model disclosed that a 1% increase in political governance and government governance increases environmental pollution by 70% and 90% respectively, and a 1% rise in FDI increases environmental pollution by 7.8%. The quantile regression with fixed effects in Koenker (2004) is used to control for the distributional heterogeneity. In Table 6, according to our results, all variables have statistically significant impacts on environmental pollution.
According to the determined the results by the panel quantile regression, several significant points were determined. There are an important effects of FDI, HDI, energy consumption, tr, fin, and governance on CO 2 emissions. The coefficients of Δg and Δp are statistically significant and has a positive sign at all quantiles by increasing through the increase in the ΔCO 2 quantiles.
Furthermore, when compared with other variables, the political and economic governance have the greatest positive effect on CO 2 emissions across quantiles. The influential intensity of political and government governance on CO 2 emissions changes from 0.597 to 0.76 and from 0.785 to 0.88. The influential intensity of economic growth on CO 2 emissions changes from −0.325 to 0.224. According to results, good governance system supports environmental regulations and decreases CO 2 releases. The impacts of governance on environmental pollution is clearly homogenous. On the other hand, the impact of economic growth on environmental pollution is clearly heterogeneous. Different percentiles of the conditional distribution have some important differences. Although the statistically significant y coefficient initially cause to increase and then decrease in ΔCO 2 amounts, it has a positive sign in various quantiles except the 90th quantile.
FDI cause increasing carbon releases, ranging from 0.122 to 0.24. FDI leads to increase in carbon emissions in these countries due to the old-fashioned the technologies and due to insufficient environmental laws and regulations. The positive coefficient of FDI support to the pollution haven hypothesis.
If the other results for the control variables included in the model are analyze, it is found the impact of human development index on carbon emissions. Unlike other variables, some coefficients of the HDI variable is negative. The coefficient of Δhdi is clearly significant and positive at the 5th quantile and at the 90th quantile, but it turns negative at the 0.25th, 0.50th, and 0.75th, suggesting that hdi leads to decreasing carbon

Robustness analysis
To test the validity, robustness check was conducted. For λ, it was considered alternative model specification and different values. It was tested if the results are robust to different values of λ ranging from 0.5 to 1.5. In Table 7, the findings are stated. The results are more or less consistent with those from the panel quantile regression with λ=1. Zhu et al. (2016), Ang (2007), and Salahuddin et al (2018) showed that FDI indicators have an important effect on economic growth. They debated that in the long run, FDI is positively connected to economic growth. Some papers determined hdi is positively connected to economic growth. Many paper found that FDI, financial development, and trade openness are positively connected to energy consumption. Given the connection among these variables, it was tested another model. This model excludes only fdi, hdi, fin, and tr. The results are reported in Table 8.
The results are near to those from the model including FDI, HDI, tr, and fin variables. According to the results, for the energy consumption variable, there is a significant difference. In addition, for the economic growth, there is a significant difference between the 90th quantiles. For governance (g) variable, the coefficients are very small accordingly other model.  *In Westerlund test, in small sample size, the results can exhibit the sensitivity to the selection of parameters as the kernel width and lag-lead lengths. By taking into account suggestion of Persyn and Westerlund (2008), a shorter kernel window is used. **p-value is exhibited, but not z-value and robust p-value.

Causality results
As the cointegration among the variables is determined, it is expected that through a feedback mechanism, a change in a variable has an influence on the other variables.
In Table 9 and 10, the results of causality were exhibited. The results can be given as follows: 1-It was found that there are uni-directional causalities running from economic governance to CO 2 release and from political governance to environmental pollution. More clearly, bad economic and political governance is a Granger cause of environmental pollution. 2-The evidence of uni-directional causality from FDI to CO 2 releases was corrected. Production technology and insufficient environmental awareness are very important factors for increasing environmental pollution. 3-As approved by many papers, it was determined that there is uni-directional causality from Y, HDI, and energy consumption to CO 2 releases. 4-It was found the evidence of unidirectional causality from trade openness and financial development to CO 2 releases. 5-It was found the evidence of two-way causality between political governance and Y, between economic governance and Y, between political governance and economic governance, between political governance and FDI, between economic governance and FDI, between economic governance and financial development, between political governance and financial   development, between economic governance and trade openness, and between political governance and trade openness. 6-And it was found the evidence of bidirectional causality between political governance and HDI and the evidence of unidirectional causality from economic governance and HDI.
Accordingly both traditional and Dumitrescu-Hurlin causality tests, there is uni-directional causal nexus from economic growth, political and economic governance, energy consumption, HDI, trade openness, financial development, and FDI inflow to CO 2 emission.

Discussion
Quantile regression method takes the unobserved individual heterogeneity and distributional heterogeneity into consideration. And to avoid an omitted-variable bias, certain related control variables are included in the model. The results of panel quantile regression models provided to see the whole picture of these variables that have significant effect on carbon emissions. Governance increases CO 2 releases with the strongest effects observed at higher quantiles. And economic growth and energy consumption variables have a positive and significant effect on carbon emissions, but for economic growth except at the 90th quantile, the effect of economic growth is negative at the 90th quantile. This state determined that a higher level of economic growth can decrease the rise in CO 2 releases in high-emissions countries. The results are robust for the different values of λ.
When the results of regression and Granger causality are simultaneously evaluated, the whole picture can be obtained. The coefficients of economic and political governance, FDI, financial development, energy consumption, trade openness, and economic growth on CO 2 emissions have positive signs in various quantiles ( Table 6). The coefficients of economic and political governance in all the estimated models are positive and determine that bad economic governance and political governance lead to increasing CO2 emissions. And according to causality results, in these countries, the evidence of unidirectional causality from bad economic governance to CO 2 release and the evidence of unidirectional causality from political governance to environmental pollution are determined.
And according to the other results of this paper, FDI as other variable cause increasing carbon emissions in changing ranges from 0.122 to 0.24. According to causality results, there is evidence of unidirectional causality from FDI to CO 2 releases. In these countries, the FDI inflows provide an important contribution to economic growth because of the evidence of unidirectional causality from FDI to y, but it leads to an increase in environmental pollution. Rising pollution caused by FDI is not controllable up to some extent because of bad governance; moreover, these countries have problems of economic development, and in these countries, FDI is vital for economic development. In these countries, FDI cause to increase in carbon emissions due to old-fashioned technologies and insufficient environmental laws and regulations and bad governance. Bad governance is to prevent government policies from addressing environmental issues caused by FDI. In these countries, there is bidirectional causality between FDI and governance parameters. FDI is sensitive to bad governance because bad governance attracts FDI and FDI acts to make bad governance worse.
And HDI, financial development, and trade openness variables are Granger cause of CO 2 releases. Moreover, there is bidirectional causality between economic governance and financial development, between political governance and financial development, between political governance and trade openness, between economic governance and trade openness, and between political governance and HDI. And there is the evidence of unidirectional causality from economic governance to HDI. The causality results show that economic and political governance has significant explanatory power on financial development, trade openness, HDI, FDI inflow, CO 2 releases, and real GDP. In addition to the control of environmental pollution, to prevention and control of FDI's polluting technologies, and to solve economic development problems, good governance is important, because in these countries, bad governance does not allow policymakers to adhere to strict environmental policies and to control corruption. Mismanagement creates more environmental pollution, and good political and economic governance can decrease the environmental pollution. There is a need for the good governance of these countries because good governance covers the protection of the environment and sustainable natural sources.

Conclusions
This paper tested the relation between governance parameters, energy consumption, economic growth, and environmental pollution with additional variables such as FDI inflow, financial development, trade openness, and HDI in Afghanistan, Angola, Bangladesh, Benin, Cameroon, Congo Dem., Congo, Chad, Central African Rep, Ethiopia, Ghana, Gambia, Eritrea, Iraq, Kenya, Lesotho, Madagascar, Mali, Mozambique, Myanmar, Niger, Pakistan, Syria, Uganda, and Zimbabwe for the period of 1996-2018 by utilizing panel quantile test, panel cointegration test, and panel Granger causality tests. Panel cointegration test determined the presence of cointegration among the selected variables. If compared with the results of two-way OLS, FMOLS and DOLS methods, panel quantile regression models provided a more whole picture of the factors that affect CO 2 emissions. The panel quantile regression results found the effects of government and political governance, FDI and the other explanatory variables on environmental pollution. The coefficients of economic (government) and political governance in all the estimated models are positive and determine that an increase in bad economic and political governance will lead to increasing CO 2 emissions. Both traditional and Dumitrescu-Hurlin causality approaches were employed to determine the evidence of the causality between the analyzed variables. The results determined unidirectional causality from governance parameters to environmental pollution. And if bad governance can be improved, the governments can apply incentive policies to attract high-tech firms to further invest in their countries, and on the other hand, to limit the entry of polluting industries into the country, they must have strict environmental regulations. The government policies should look for the efficient and environmentally friendly source consumption of FDIs.
Despite of unsustainable development, the policymakers do not have the opportunity to choose the variables which have adverse effects on the environment. Bad governance prevents government policies from addressing environmental issues. Political stability helps to control CO 2 emissions and to recover environmental quality. Decreasing CO 2 emissions with a better political institution can be provided because stability is one of the most significant requirements of environmental quality.
Author contribution All paper was prepared by Melike Bildirici.
Availability of data and materials All data are received from official source.

Declarations
Ethics approval and consent to participate This paper was not previously published in any journal. It is currently not under consideration by another journal. Consent to participate is not applicable.

Consent for publication Not applicable
Competing interests The authors declare no competing interests.