FINANCIAL MARKET DEVELOPMENT AND CO2 EMISSIONS NEXUS IN NIGERIA: AN APPLICATION OF ARDL APPROACH

The paper investigated the impact of financial development on CO2 emissions in Nigeria from 1981 to 2019. In the process of investigating the impact, Augmented Dickey-Fuller and Philip Perron, as well as the Zivot-Andrew structural breaks, unit root tests were applied. Their results indicated that financial development, level of income, and CO2 emissions were stationary at the first difference and that of Zivot-Andrew structural breaks indicated a mixture of integration. Cointegration relationship among the variables was established through autoregressive distributed lag model bounds test. The autoregressive distributed lag model long-and-short run models results indicated that financial development and income level significantly negatively impact the CO2 emissions. The suggestion based on these results is that financial development and income level help in financing clean projects in the long-and-short runs. The Granger causality result revealed bidirectional causality from financial development to CO2 emissions, income level to CO2 emissions, and financial development to income level. The variance decomposition analysis indicates that financial development and income level have contributed less to CO2 emissions, and impulse response function results revealed that CO2 emissions respond negatively to shocks in financial development and income level. Therefore, we recommend expanding the Nigerian financial market in financing clean projects for a clean environment alongside checking income generation activities that bring about emissions of CO2, such as burning trees for charcoal production in the forest, among others.


Introduction
The evolution of the financial development concept started in the 6th century BC-15th century A.D., and the concept continued to widen up to the present days, passing through several periods and terminological modifications beginning with the basic as a financial market to current clarification of financial development due to purposes and results (Čižo, Lavrinenko, & Ignatjeva, 2020). One of the most excellent millennium development goals is providing a clean or fresh environment in the world to benefit from contemporary and clean technologies. If not, the technologies that promote pollution emissions would become harmful to the globe in pollution and fair provision of short-term welfare to the societies. For sustainability in the long-term, these technologies from clean sources are crucial in realizing a clean environment. Thus, for the advancement of clean technologies, the financial sector needs to categorize the loans. Otherwise, the emissions of pollution and financial market development will create unmaintainable development in the globe due to ignorance (Claessens & Feijen, 2006;Mahmood, 2020).
Accompanied by economic development and growing energy consumption, several additional forces within a nation substantially affect the environment. One among these forces is the impact of financial market development. As stated by Frankel & Romer (1999), financial markets might help in achieving a healthier development in the nation alongside a rising need for a cleaner environment due to more development. However, Zhang (2011) claimed that financial market development would increase investment and consumer credit. Due to financial market development, the growing economic events might emit pollution, which would convert to degrading the environment while regarding economic growth and financial market development. The actual positive environmental influence could be anticipated if the method and structural possessions are prevailing on the measure consequence, and the contrary environmental consequence may be anticipated then (Mahmood, Furqan, & Bagais, 2019). The positive environmental part of financial market development proved that it helps achieve clean environmental technologies and inventions with financial market development, raising the capacity of innovation and industrial size to invents effective technology, decreasing the nation's levels of pollution (Birdsall & Wheeler, 1993).
To the extent of this argument,  examined the connection between financial market development and environmental pollution in Malaysia's case. They reported a negative effect of financial market development, and therefore the study corroborates the argument of the positive environmental influence of financial market development. Another investigation revealed a positive effect of financial market development in achieving environment quality in the case of the leading emitter of CO 2 China and concluded that financial market development had decreased the level of pollution (Jalil & Feridun, 2011). Zhang (2011) reported that financial market development caused pollution in investigating the relationship between financial market development and pollution. Moreover, more recently, Mahmood (2020) reported that financial market development had an insignificant influence on the level of emissions in G.C.C. countries. Financial market development increases the growth of industries and urbanization, which is measured as necessary for the complete economic growth; however, it can possess some harmful effect in pollution emission when the environment's side is neglected. Therefore, it becomes significant to examine the impact of financial market development on environmental pollution to control the dirty industrial growth, which is financed to raise their business actions which consequently caused damage to the environmental health. In line with the above argument and the paper's background, our objective is to explore the impact of financial market development on CO 2 emissions. Testing the objective would aid in filling the literature gap in the Nigerian environment.
Therefore, based on this background, the paper is structured as follows. This section is section two that provides the review of related literature on the relationship. Section three offered data description and research methodology. Section four provides the empirical results and discussions of findings. Section five concentrated on the conclusion and policy recommendation.

Literature Review
The review of related literature in this paper covered studies on financial market development and CO 2 emissions relationship together with those studies on CO 2 emissions and other related variables both within and outside the country of study.
In their investigation for the possible existence of the EKC hypothesis for 1971 to 2014 using the case of Saudi Arabia, Mahmood et al. (2018) analyzed the asymmetric influence of financial market development and energy consumption on CO 2 emissions. The outcome indicated the existence of EKC and long-run asymmetric influence of financial market development together with short-run and long-run asymmetric effect of energy consumption on CO 2 emissions. Again, financial market development is accountable for environmental degradation, and reducing energy consumption is found to support CO 2 emissions. However, in the Turkish economy, Çetin & Ecevit (2017) studied the effect of financial development on CO 2 emissions from 1960 to 2011. The data was analyzed using ARDL and VECM Granger causality test. The results revealed that financial development, trade openness, economic growth affected CO 2 emissions in the long-run, and long-run causality exists from financial development, trade openness, and economic growth to CO 2 emissions. The results imply that the EKC hypothesis exists and that financial and economic development has arisen due to the environment's degradation.
Using the case of 12 MENA nations and applied the G.M.M. technique in the data estimation for the 1990-2011 period, Omri et al. (2015) examined the connection among financial development, trade openness, CO 2 emissions, and economic growth. The result indicated a negative connection between financial development and CO 2 emissions in Jordan, and again in Qatar, CO 2 emissions have a positive correlation with financial development. Shahzad et al. (2014) investigated the link between economic growth, energy consumption, financial development, trade openness, and CO 2 emissions for the sample duration of 1973 to 2011 in Pakistan. The sample period data was analyzed using ARDL bounds for cointegration, fully modified, and dynamic ordinary least squares. The result revealed that there exists a bi-directional long-run causal relationship between financial development and CO 2 emissions.
In another development,  studied the relationship between financial development and CO 2 emissions with the help of energy consumption, economic growth, and CO 2 emissions in the case of Malaysia between the period of 1971 to 2011 and the data for the sample period was analyzed using ARDL and VECM approaches. The long-run result indicated that financial development reduces CO 2 emissions, and bidirectional causality exists between financial development and CO 2 emissions. However, in the South African economy, Shahbaz, Kumar Tiwari, & Nasir (2013) examined the influence of coal consumption, financial development, trade openness, economic growth on environmental pollution for the sample period of 1965 to 2008 and applied ARDL VECM procedures. The analyses revealed that financial development has a long-run negative impact on economic growth and that unidirectional causality exists from CO 2 emissions to financial development. Mahmood et al. (2018) investigated the environmental impact of financial market development, foreign direct investment, trade openness on CO 2 emissions together with the EKC hypothesis in the six East Asian nations for the 1991-2014 period using the technique of spatial econometrics to reflect the neighboring nations' spillover effect. The result indicated a supportive spillover effect from the neighboring nations' CO 2 emissions, financial market development, foreign direct investment, trade openness, and energy intensity. They were accountable for the resident environmental degradation and neighboring nations' financial market development's significant positive effect on CO 2 emissions. In their determination for the influence of economic growth and energy consumption on environmental pollution, Maijama'a & Musa (2020) applied the ARDL procedure on the time series data spanning the 1981-2014 period of the Nigerian economy. Their result from the ARDL procedure indicated that all the series are cointegrated. Energy consumption and economic growth have significantly signed positively with environmental pollution, whereas negative and significant signed was witnessed between crude oil price and environmental pollution in the long-and-short runs.
In another development using the same case study and applied ARDL approach, Maijama'a & Musa (2020) analyzed the effect of crude oil price and urbanization on the level of environmental pollution for the 1981-2016 periods. The analysis revealed that all the series were cointegrated and that crude oil price and foreign direct investment have significantly influenced environmental pollution negatively in both long-and-short run periods. However, a significant favorable influence was witnessed between urbanization and environmental pollution. Ma & Fu (2020) document that financial development positively affects energy use through financial institutions and financial markets in developing countries. However, they do not identify this effect in the case of developed countries. The authors employ the G.M.M. approach and investigate different periods of 1991-2014, 1981-2014, 1970-2014, and 1960-2014 as the whole sample covered 120 countries.
On the other hand, in modeling the relationship between economic growth and financial development, most studies reported a positive connection between the two variables implying that financial development promotes economic growth. Among the studies that reported the existence of the positive relationship between financial development and economic growth using different methodologies in different case studies to include Alrifai et al. Therefore, based on the sufficient literature reviewed, the general studies present conflicting findings regarding the exact relationship between financial development and CO 2 emissions. The studies that specifically looked at the relationship between financial development and CO 2 emissions in the Nigerian economy are relatively scarce. Hence, there is a sufficient gap in the prevailing literature in Nigerian studies to investigate the impact of financial development on CO 2 emissions.

Data and Research Methods
The data and methodology in this section are organized in sections. Section one focused on the source of data, justifications, and variables' measurements. Section two presented the general paper methodology.

Data Source and Variables Measurement
The variables included in this study are financial market development, CO 2 emissions, and level of income for the 1981-2018 periods. The study period's choice was highly influenced by the data available on all these variables from two sources. The remaining required information concerning the variables' measurement, justification, and extracted sources are offered in Table 1.

The World
Bank (2020)

Methodology
The study's primary focus is to determine the effect of financial development on CO 2 emissions in Nigeria within the sample period of 1981 to 2018. A theoretical framework for this research is the environmental Kuznets theory. According to the environmental Kuznets theory, industrial advancement initially causes environmental destruction, but after a certain degree of economic growth, a society's interaction with the environment improves, and environmental degradation levels decrease. Therefore, to derive the empirical model connecting the variables of interest, line with previous studies such as Ang (2008), Halicioglu (2009) and later on, Shahbaz & Lean (2012) utilized a single equation model to study the relationship between economic growth, energy consumption, and CO 2 emissions. Later on, Tamazian, Chousa, & Vadlamannati (2009) and Jalil & Feridun (2011) augmented the single equation model by including financial development as a potential determinant of CO 2 emissions. Therefore, following these studies, we use financial development and income levels within a multivariate framework in Nigeria. To derived the model equation, we have followed the empirical economic growth model given as Q = F (K, L). In our modification of the model, Q is replaced with CO 2 , F.D. replaced K and I.C. replaced L, and the final modified estimable model is given (1) CO 2t represents the carbon emissions; FD t represents financial development at time t; IC t represents the level of income at time t; t represents 1981-2018 periods.
Ln is the natural logarithm; 0 θ represents the drift parameter; 1 2 & θ θ represents the coefficients of explanatory variables to be estimated; F.D.A. represents financial development; I.C.T. represents the level of income; t ε represents the disturbance term.

Unit Root Tests
When the given variables possessed unit root, it is indicated that the variable is non-stationary, and when non-stationary variables are regressed on other non-stationary series, the generation of spurious or non-sense regression results is guaranteed. Therefore, to dodge all possibilities of generating spurious results, this paper utilized the convectional Augmented Dickey-Fuller (Dickey & Fuller, 1981). Besides, Phillips & Perron (1988) unit root tests and Zivot & Andrews (1992) structural breaks unit root test is also engaged as a robustness check to the result A.D.F. and P.P. unit root tests. The A.D.F. modeling is presented in Equation 3.
We test the null hypothesis, H.N.: 0 γ = for non-stationary as against the alternate hypothesis H.A.: 0 γ ≤ for the stationary series.
After this stage for unit root, we continued to test for cointegration association among our variables. For this reason, we are following the Autoregressive Distributive Lag (ARDL) bound test advanced by Pesaran et al. (2001) in a subsequent system: We test for the existence of a long-run relationship using the null hypothesis that says no cointegration relationship among the series given as H.N.: θ 1 = θ 2 = θ 3 = 0. The alternative hypothesis says there is a cointegration relationship among the variables given as H.A.: θ 1 ≠ θ 2 ≠ θ 3 ≠ 0. According to Pesaran et al. (2001) a cointegration relationship among the variables of interest exists when the calculated F-statistics is greater than the upper bonds critical values only. Therefore, we reject the null hypothesis. However, when the estimated F-statistics is less than the lower bond critical values, then there is no cointegration relationship among the variable, and we accept the null hypothesis. On the other hand, if the calculated F-statistics is greater than the lower bond critical values but is less than the upper bond critical values, the result is inconclusive.
Following the cointegration relationship among our variables, we proceeded to test for the long-run and short-run coefficients together with the error correction coefficient that determined the speed of adjustment back to the equilibrium position. However, the long-run and short-run coefficients' generation depends mainly on the pie (π) being negative, less than one, and statistically significant. Also, as the coefficients of the pie (π) have satisfied these three conditions, it is said to substantiate the existing long-run association in the model as given in Equation 5.

Granger Causality Technique
After establishing the series's long-run connection, the paper continued examining the causal relationship between the series with the Granger causality test's help. Granger (1969) supported this test such that series Y Granger caused X if series Y can be projected with the help of better certainty using previous values of series X and ceteris paribus. We have utilized the following V.A.R. model given in Equations 6, 7, and 8, respectively.
Where series X, Y, and Z represent the CO 2 emissions, financial development, and income level, respectively. From Equations 6, 7, and 8, we test for 0 The decision regarding the acceptance or rejection of null and alternative hypotheses is based on the following two situations: 1. We reject each of the H 0 given above only if the calculated F-statistic is larger than the critical values at a certain level of significance. In other words, we do not reject H 0 , and rejecting the H 0 in Equations 6, 7, and 8 suggested that financial development and level of income Granger caused CO 2 emissions. The previous values of financial development and level of income significantly project CO 2 emissions. 2. Similarly, rejecting H 0 in Equations 6, 7, and 8 also suggested that CO 2 emissions Granger caused financial development and income level. Previous values of CO 2 emissions could be employed to project financial development and income levels in question.
Incorporating our three variables that include CO 2 emissions, financial development, and level of income into the model, Granger causality equations are given in Equations 9, 10, and 11, respectively, based on the vector autoregressive (V.A.R.) system of the equation as follows:

Finding and Discussion
This section offered the empirical results and discussions in graphical presentation of the series, descriptive, and correlation analyses. Also, the unit root test results' presentation using the traditional unit root tests and structural breaks unit root to determine the order integration of the series, the optimum lag selection test, ARDL bounds test, Granger causality test result, and diagnostic tests.
The graphical presentation of CO 2 emissions, financial development, and income level are given in Figure 1. The trends of CO 2 emissions show a decreasing fluctuating trend for the period under study, whereas financial development indicated an increasing fluctuating trend throughout the study period. However, the level of income trend possessed some form of fluctuations from 1985 to 2000. From there, the trend maintained an upward movement for the rest of the period. The rest of the interpretation regarding the study period is presented in Figure 1.  (2020) After the graphical presentation of financial development, level of income, and CO 2 emissions for the period under investigation as offered in Figure 1, here comes the presentation of descriptive and correlation analyses given in Table 2. The areas described include the average values given as the mean, the variable variation of variables more formally known as deviation, the measure of how positive or negative the variables are in terms of skewness, the measure of variables normality given as the Jarque-Bera values and their probability values given in bracket and lastly the total number of observations after the descriptive analysis followed by the correlation analysis offered in the lower part of Table 2. The result indicated a negative correlation running between the independent variables and the dependent variable, as shown by the negative values of -0.868 and 0.896 for the financial development and income level. Therefore, these negative values entail that an increase in financial development and income level would reduce the level of CO 2 emissions for the period under study. The rest of the analysis result is presented in Table 2.

Unit Root Tests Results
The Augmented Dickey-Fuller (A.D.F.) and Philip Perron (P.P.) unit roots tests were engaged in ascertaining the order of integration of the series and their estimation results offered in Table 3. The results revealed that under A.D.F. and P.P. unit root tests, all the series were not stationary at the level. This result makes it impossible to reject the null hypothesis of the non-stationary series. However, these variables became stationary after first differencing, and this makes it possible to reject the null hypothesis of the non-stationarity series. Therefore, all the series are integrated of order one or popularly known as I(1) variables. The two tests' results indicated that the ARDL approach is more appropriate to be conducted. The bounds test for cointegration is efficient to handle the cointegration relationship between the series. At 1st Difference lnCO 2t -6.382 (0.000) *** -6.287 (0.000) *** -8.526 (0.000) *** -8.391 (0.000) *** lnIC t -3.825 (0.006) *** -3.741 (0.032) ** -3.825 (0.006) *** -3.741 (0.032) *** lnFD t -5.430 (0.000) *** -5.317 (0.000) *** -7.890 (0.000) *** -7.753 (0.000) *** Note: *** stands for the 1% level of significance, respectively ** stands for the 5% level of significance, respectively Source: Author Calculation Knowing the integration of the series using the traditional A.D.F. and P.P. unit root tests does not guarantee that the result is free from structural breaks. To handle the stationarity test with the possible existence of structural breaks in the series, we have employed the Zivot-Andrew structural breaks unit root test. The result of utilizing the test is given in Table 4. The result of the test indicated that CO 2 emissions and financial development were stationary at the level. Therefore they are said to be integrated of order zero or I(0). In contrast, the level of income is the only variable not stationary at level but became stationary after first differencing and is said to I(1). The remaining interpretation of the result can be obtained in Table 4.
The result of the Zivot-Andrew structural breaks unit root test also supported the application of the ARDL bounds approach.  430 (2) ** 1990 Note: * stands for the 10% level of significance, respectively ** stands for the 5% level of significance, respectively *** stands for the 1% level of significance, respectively Source: Author Calculation

Figure 2: ARDL Model Selection Criteria Graph
After knowing the optimum lags combination of the ARDL to be 2,0,0 as given in Figure  2, here comes the estimation of the long-run equilibrium relationship among the variables, and the result of the estimation is given in Table 5. The estimation results revealed that all the variables have a strong cointegration relationship as indicated by the estimated F-statistic value of 6.143, which appears to be higher than lower and upper bounds critical values at 1 percent, regarded as more stringent. Therefore, all the series are highly cointegrated, and they move together in the long-run.

ARDL Shot-and-Long Run Estimated Results
The existence of a cointegration relationship among the series given by the result of the bounds test in Table 5 necessitated estimating short-run, long-run, and error correction coefficients, respectively. The result indicated that financial development and income level have a significant negative impact on CO 2 emissions at a 1 percent level of significance. Precisely, a percentage change in financial development and level of income are associated with 0.136 and 0.502 percent decrease in CO 2 emissions in the short-run period. There is a strong influence of financial development and income level on CO 2 emissions in the long-run period, where financial development reduced CO 2 emissions by 0.150 percent and level of income reduced CO 2 emissions by 0.553 percent, respectively. These findings are regarded as a strong indication that financial development and income help clean project financing. The findings corroborate that of Mahmood (2020) for the United Arab Emirate, Shahbaz, Solarin, et al. (2013) for Malaysia, Jalil & Feridun (2011) for the Chinese economy, and Tamazian et al. (2009) for BRIC nations.
The coefficient of error correction term as reported in the lower part of Table 6 has satisfied all the three econometric conditions of being negative, less than one, and statistically significant. Therefore, these conditions' achievement is also a confirmation of a long-run equilibrium relationship among the series. Precisely, the error correction term coefficient of -0.908 signifies that there is a solid and fast speed of convergence in case of dynamic shortrun disequilibrium back to the equilibrium position. The speed of convergence back to equilibrium position is approximately 91 percent every year within the sample study period. The coefficient of R-square in the lower part of Table 6 measured the number of variations or changes in the explanatory variable that is jointly explained by the explanatory variables in the model. In the ouR-square coefficient, 93 percent change in CO 2 emissions is explained jointly by the financial development and income level, implying that only 7 percent is captured by the error term. The F-statistic value measured all the independent variables' joint significance in explaining the dependent variable in an income level model. The estimated F-statistic is statistically significant, which indicated that financial development and income level are jointly significant in explaining changes in CO 2 emissions. The estimated models' coefficients reported in Table 6 are not reliable for policymaking if the coefficients are not subjected to diagnostic tests. To ascertain the reliability of these coefficients, we have engaged the serial correlation test, heteroscedasticity test, functional form test, normality test, and stability test, respectively. Their results are reported in Table 7. The result indicated that the null hypothesis of no serial correlation could not be rejected for the serial correlation test since the test's p-value is not significant. For heteroscedasticity, the null hypothesis of no heteroscedasticity could be rejected given that the test p-value is not significant. In the case of errors specification or misspecification in the model, the functional form test in the form of the RAMSEY Reset test revealed that the null hypothesis says errors are specified accepted since the p-value of the test is insignificant. Again, errors in the model are normally distributed, as shown by the insignificance of the normality test p-value. Lastly, the stability test via the CUSUM and CUSUM of squares indicated that errors are stable since the CUSUM and CUSUM of straight lines are within the 5 percent significance boundary. Therefore, in summary, our estimated model has passed the serial correlation, heteroscedasticity, functional form, normality test, and stability tests, respectively. The estimated coefficients are now safe and can be relied on for policymaking and other statistical inferences.

Robustness Checks Results
For robustness checks on the calculated coefficients of the long-run ARDL model, we have utilized the dynamic ordinary most minor (DOLS), fully modified ordinary least square (FMOLS), and the canonical cointegration regression (C.C.R.), respectively. The results are reported in Table 8. The three estimators' results indicated that the independent variables' coefficients are in line with their ARDL long-run coefficients. This condition is because the financial development and level of income coefficients were negative and statistically significant at a 1 percent level of significance.

Granger Causality Test Result
The Testing of the direction of causality among CO 2 emissions, financial development, and income level was highly necessitated by the existence of a cointegration association between CO 2 emissions, financial development, and level of income, respectively, as shown by the bounds test result reported in Table 5. The result of the Granger causality test reported in Table 9 indicated that there exists bidirectional causality between financial development and CO 2 emissions. Another bidirectional causality is running from the level of income to CO 2 emissions. Similarly, there exists bidirectional causality running from financial development to level of income, respectively. 3.103 (0.087) * lnIC t ← lnFD t Note: *** stands for the 1% level of significance, respectively ** stands for the 5% level of significance, respectively * stands for the 10% level of significance, respectively Source: Author Calculation

Variance Decomposition and Impulse Response Results
This section focused on estimating the percentage change or variation in the dependent variable caused by the shock in the independent variables and one variable's response to shock in other variables. Test results are reported in Table 10 and Figure 4, respectively. The result from Table 10 showed that in period 1, CO 2 emissions response 100 percent to their shock or innovation. However, in period 5, the response percentage decreased to 87 percent approximately while financial development and level of income contributions were approximately 6 percent each. Toward the long-run in periods 9 and 10, own response decreases from 84 percent to 83 percent approximately. In contrast, the financial development response was approximately 6 percent for the two periods while that of the income level increases from 10 percent to 11 percent in periods 9 and 10.
Moreover, the financial development percentage response to its shock was 91 percent approximately, and CO 2 emissions contributed the remaining 9 percent. Again, in period 5, the own response decreases to 53 percent approximately where CO 2 emissions provide 37 percent and the level of income provide 10 percent approximately. In periods 9 and 10, the own response decreased where it accounted for 50 percent and 49 percent only. In comparison, CO 2 emissions and income level accounted for the constant approximated values of 35 percent and 15 percent for the two periods.
Similarly, the response of income level to shocks in CO 2 emissions and financial development was 98 percent, 87 percent, and 81 percent for the 1st, 5th, 9th and 10th periods, respectively. Simultaneously, CO 2 emissions are responsible for the 1 percent, 0.2 percent, 0.2 percent, and 0.3 percent, respectively, for the same periods. However, an approximated value of 1, 12, 19, and 19 percentage of the contributions were accounted for the financial development, especially 1st, 5th, 9th and 10th periods.
In summary, the analysis of variance decomposition revealed that apart from own shock response, level of income contributed more than the financial development to CO 2 emissions. However, when CO 2 emissions were the dependent variable, CO 2 emissions contributions were more significant than that of the income level when financial development was the dependent variable. Lastly, financial development contributed more than CO 2 emissions when the level of income was the explained variable. Response of LIC to LIC Response to Cholesky One S.D. Innovations ± 2 S.E.

Conclusion and Policy Recommendations
This paper investigated the effect of financial development on CO 2 emissions in Nigeria for the sample period of 1981 to 2018 using the ARDL approach analysis. The result of the unit root test from A.D.F. and P.P. revealed that all the series were stationary at the first difference, and therefore, they are all I(1) variables. In contrast, the Zivot-Andrew structural break test result showed a combination of two I(0) variables and one I(1) variable, respectively. These unit root test results guaranteed the ARDL bounds test application, and the result indicated a strong cointegration relationship among the series at a 1 percent level of significance.
The long-run and short-run ARDL results indicated that financial development and level of income have a negative and significant influence on the level of CO 2 emissions within the study period. The long-run ARDL estimates were subjected to some robustness checks using DOLS, FMOLS, and C.C.R., and their results conformed with that of the long-run ARDL coefficients. The ARDL analysis was subjected to some reliability tests, and their result indicated that the model is free and reliable for policymaking.
The Granger causality test results indicated bidirectional causality running from financial development to CO 2 emissions, level of income to CO 2 emissions, and financial development to level of income, respectively. The variance decomposition results indicated that shocks in financial development and income level contributed some quota to changes in the level of CO 2 emissions. The impulse response function result showed that a negative response was observed from financial development and income level due to shocks in CO 2 emissions.
Based on the above and the conclusion drawn from this paper's empirical findings, we recommended that there is a need for the expansion of the Nigerian financial market. This because in order to continue to supports financing clean environment alongside checking human income generation activities that bring about the emissions of CO 2 such as cutting down of trees in the forest for charcoal production, rapid and unnecessary bush burning, among others.