The Impact of Financial Development and Economic Growth on Environmental Quality of Kenya

Of recent times many countries are suffering from environmental problems such as global warming and emission of greenhouse gases. Emissions of carbon dioxide as been recognized as the major contributor to global warming and climate change. This paper examines the long run relationship among the variables environmental quality, financial development and economic growth in Kenya using time series for the period 1970-2019. Autoregressive distributed lag bounds test is used to investigate long run relationship and Granger causality method is used to test for causality among the variables. Empirical results indicate that there is long-run relationship among the variables. Long run results suggest that increases in financial development, lagged CO2, energy consumption, population growth, and trade openness significantly worsens environmental quality in Kenya. Natural resources significantly improve environmental quality in Kenya. According to the results the relationship between CO2 and financial development in Kenya is non-linear suggesting presence of EKC between CO2 and financial development. The empirical results confirm that the Environmental Kuznets curve does not exist between CO2 and economic growth in Kenya in the long run. Short-run results also show that financial development, lagged CO2, FDI, population growth, and trade openness increase CO2 emissions while natural resources reduce it. Causality results show unidirectional causality running from financial development to environmental quality and from CO2 to GDP. According to the findings, there is evidence of neutrality hypothesis between financial development in Kenya and economic growth. Existence of long run relationship suggests that the government of Kenya needs to implement appropriate environmental policies that reduce pollution during economic growth. The government should set policies and guidelines to the financial sector so that the sector offers credit to firms that reduce air pollution.

Many studies have investigated the relationship between economic growths and environmental quality, little attention has been given to the impact of financial development. In Kenya there are very few studies that have looked on the impact of financial development on environmental quality. This study fills this gap by investigating the role of financial development on CO2 emissions within a multivariate framework.
The relationship between air pollution and economic growth theoretically according to Grossman and Krueger (1995) is non linear and graphically looks as an inverted U-shaped curve referred as Environmental Kuznets curve hypothesis. Many studies have been done arriving at mixed results. Several studies confirm the presence of EKC ( Mohammadi, 2016;Omri et al ,2015;Ozatac and Taspinar, 2017;AC et al 2018, Asumadu-Sarkodie andOwusu ,2017). If this is true then in the long run economies can improve the environment otherwise the environment will continue to deteriorate and Kyoto targets cannot be achieved. Other research findings show that EKC does not exist (Aye and Edoja, 2017;Simiyu, 2017;Mikayilov et al ,2018) .
It is also possible as from a few studies that have been that financial sector can affect air pollution .Financial sector includes commercial banks, capital markets insurance companies and other financial intermediaries. A developed financial sector is one which is large enough and plays the following roles. Financial sector mobilizes and pools savings, allocates savings to the most productive use, provides information on possible investment, allocates capital , facilitates both domestic and international trade, manages risk, monitors firm`s activities and exerts corporate governance ( Mesagan and Nwachukwu ,2018). Development of the financial sector can accelerate economic growth and this will holding other factors constant affect environmental quality according to Environmental Kuznets curve (EKC) hypothesis. The industrial sector whose growth depends much on financing from the financial sector consumes a lot of energy in their industrial activities. Developing countries, Kenya inclusive mostly purchase energy intensive inputs which pollute the environment hence affecting the environment negatively ( Hoffmann, 2011). Financial intermediaries allocate credit to the private sector for investment and this can impact on the environmental quality negatively or positively (Mesagan et al, 2019). Through loans given to industry and factory activities and their energy consumption may increase pollution ( Sadorsky, 2010;Tamazian et al , 2009). Financial intermediaries can deny credit to those firms that are high in CO2 emissions and give more loans to those whose production processes emit less carbon dioxide and this can reduce CO2 emissions. It is also argued that financial intermediaries can access environmental friendly technologies which can reduce emissions of carbon dioxide ( Tamazian et al ,2009). Financial intermediaries also supply loans to consumers who purchase household appliances, refrigerators, houses among others. Household also buy air conditions and automobiles that increase air pollution ( Zhang , 2011). A developed financial facilitates investment with low cost including environmental friendly projects. A developed financial sector can reduce borrowing costs, promote investment in the energy sector and this will likely reduce pollution ( Tamazian et al ,2009, Tamazian andRao, 2010). A developed financial sector attracts foreign direct investment. Foreign direct investment impact on a country`s economic growth which on the concept of EKC hypothesis affects the environment ( Tamazian and Rao, 2010).
This study adds to the current literature by investigating the role of financial development and economic growth on environmental quality in Kenya using data for the period 1970-2019. There are very few studies that have investigated the impact of financial development on CO2 in Kenya. This study will be beneficial to environment policy makers and the governments in making policies that are friendly to the environment.
The paper is organized as follows. Section 1.0 gives the introduction to the paper, section 2.0 presents literature review, section 3.0 discusses methodology, section 4.0 presents empirical findings while section 5.0 presents conclusions and policy implications of the paper.

Literature Review
Most of the studies have investigated the EKC hypothesis, which shows the relationship between economic growth and carbon dioxide emissions. The hypothesis argues that the relationship between CO2 and economic growth is quadratic in the long run (Grossman and Krueger, 1995). This graphically looks like an inverted U shape curve. Empirical studies that have been done over a few decades, present mixed results. Ozatac and Taspinar (2017) investigated EKC for Turkey using the data for the period 1960-2013. The study used ARDL to test for presence of long run relationship and Granger and ECM to estimate short run relationship among the variables. The study found that EKC is applicable for Turkey and there is long run relationship among the variables included in the model.  Vol.11, No.12, 2020Simuyu (2017 investigated the presence of EKC and the impact of trade openness in Kenya. The research findings indicated that EKC hypothesis is not valid for the case of Kenya and that real GDP percapita has a negative impact on environmental quality. Ogundipe and Oshin (2015), investigated the validity of EKC theory for West Africa countries. The study used static panel data approach for 15 West Africa counties for the period 1980-2012. The findings confirm the existence of EKC theory for the West Africa countries.  investigated whether financial development reduce CO2 in Malaysia . The study time series data for the period 1971-2008 and applied ARDL bounds to test for cointegration among the variables. The results indicated that financial development reduces CO2 emissions in Malaysia and that energy consumption and economic growth worsen environmental degradation.  studied the effects of financial development, economic growth, coal consumption and trade openness on environmental performance using time series data for the period 1965-2008 for Sub-Saharan Africa (SSA) countries. The study used ARDL bounds method to test for existence of long run relationship among the variables and ECM method to investigate the short run impact of the explanatory variables on quality of the environment. The findings confirm existence of cointegration among the variables and that economic growth and coal consumption increase energy emissions while financial development improve the quality of the environment. The study also confirmed existence of EKC in SSA . Mohammadi (2016) investigated the impact of the major socio-economic variables as well as financial development on environmental quality using time series data over the period 1970-2013 for 16 middle income countries. The study used PSTR method. The findings of this study provided evidence that output and energy consumption have positive significant effect on environmental pollution but their effect at higher levels of financial development decrease and increase respectively. Population also affected CO2 emissions positively. Financial development has a significant positive effect on pollution with a threshold level of 34 % for financial development index. Omri et al (2015), studied the relationship between financial development, CO2 emissions, trade openness and economic growth using simultaneous equation panel data models for a panel of 12 MENA countries over the period 1990-2011.The study came up with several findings: the EKC hypothesis was confirmed for the 12 MENA countries, there is bidirectional causality between CO2 and economic growth, and between economic growth and trade openness. Mesagan and Nwachukwu (2018) studied the determinants of environmental quality for Nigeria using data for the period 1981-2016. The study used ARDL bounds test for cointegration analysis and Granger causality for the causal relationship among the variables. The study found that income, financial development energy consumption and trade positively and significantly determine air pollution.
Tsauri(2019) on the study on the role of financial development on carbon dioxide emissions for west Africa countries found that financial development significantly increase carbon dioxide emissions in West Africa countries. The study used time series data for the period 2003-2014 and applied pooled ordinary least squares method, fixed and random effects methods.
From the theoretical arguments and empirical findings, the impact and the relationship between carbon dioxide, economic growth and financial development is not conclusive. Theoretically financial development can worsen environmental quality and it can also improve it. The relationship between CO2 and financial development can be linear or non-linear. Most of the studies have found that economic growth increase pollution of the country , however the hypothesis of EKC is present in some studies while some studies invalidate this hypothesis.
This study contribute to the present knowledge by studying the impact of financial development and economic growth on carbon dioxide emissions using ARDL bounds test method to test for the presence of long run relationship among the variables CO2, energy consumption, economic growth, financial development, foreign direct investment, natural resource, population and trade .The short run coefficients are estimated by use of ECM and causality analysis is done by use of Granger causality method. The study uses Kenyan time series data for the period 1970-2019.

Methodology
This paper uses econometrics analysis to investigate long run relationship and causality among the variables financial development, economic growth energy consumption, natural resources, foreign direct investment, population growth, trade openness, and environmental quality in Kenya using time series data for the period 1970-2019.The paper uses ARDL to investigate long run relationship while Granger causality test is used to test for causality among variables. Data for these variables are obtained from World Bank development indicators, Kenya economic survey and Kenya statistical abstract.
In this paper the level of carbon dioxide (CO2) emissions in Kenya over years is taken as a proxy of environmental quality. CO2 emissions function is modeled as: Journal of Economics and Sustainable Development www.iiste.org ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online) Vol.11, No.12, 2020 CO2t=f (FDt, Yt, Et) (1) Where CO2 is carbon dioxide emissions (metric tons per capita), FD represents financial development ( proxied by credit to the private sector), and Y economic growth of Kenya while E is energy consumption per capita. The square of financial development was also included in the model to find out whether the relationship between CO2 and financial development is non -linear or not. The logic behind this is that the financial sector cares less for the environment in the earlier stages of economic growth. When the economy matures, the developed financial sector`s activities will improve the quality of the environment by issuing loans to environmentally friendly projects (Shahbaz, Solarin, Mahmood ,and Arouri, 2013). In this model foreign direct investment is also included as an explanatory variable to measure the impact of foreign capital on CO2 emissions. Kenya`s foreign direct investment has been growing over time and with a developed financial sector foreign direct investment my invest in environmental friendly projects that may reduce greenhouse gases emissions (Talukdar, and Meisner, 2001). Other variables that have been identified in the literature include population growth (POP) ,natural resources (NR) and trade openness (TR). As the population of a country increases, it engages in deforestation and increases its demand for energy for their various activities. This will worsen global warming of the country as the emissions of carbon dioxide increase (Tsaurai, 2019). Exploitation of natural resources such as drilling of oil needs usage of heavy machinery which consumes a lot of energy impacting positively to air pollution . Trade openness affects environmental quality, as countries engage in international trade they increase imports of industrial inputs which are high energy consuming and this may increase air pollution in the country. Therefore to include all these variables the above model becomes Since quite a number of studies have validated the presence of EKC between CO2 and income, the above model (2) incorporated the square of income to find out if it is present in Kenyan context. Therefore model (2) was also estimated with Y squared without the financial development squared as: In the model the indicator of environment quality is CO2 which represents total carbon dioxide emissions in Kenya measured in metric tons percapita. FD is financial development which is proxied by total domestic credit to the private sector expressed as % of GDP, FD 2 is financial development squared, Y represents economic growth in Kenya, Y 2 this is economic growth squared, E represents energy consumption percapita, FDI represents foreign direct investment expressed as % of GDP, NR is natural resources measured by natural rent expressed as a percentage of GDP, POP is population growth expressed as a percentage of total population TR is trade openness which are total exports and imports expressed as % of GDP. All variables are converted to natural logarithms. The empirical equation is expressed as: The coefficient of FD βfd as from previous studies it can be positive or negative. In case of presence of EKC between CO2 and FD, then the coefficient of FD 2 is supposed to be negative. The coefficient of economic growth βy is supposed to be positive, as economic activities increase pollution is supposed to increase. The coefficient of the square of income βy2 is supposed to be negative if EKC is valid as for the relationship between CO2 and income. That is after a certain threshold the quality of the environment is supposed to start improving as according to EKC hypothesis that is to produce an inverted U shaped curve. The coefficient of energy consumption is supposed to be positive. As counties develop they increase consumption of energy, this increases emissions of CO2 (Jalil and Feridun, 2011) holding other factors constant. The coefficient of FDI is expected to be positive as investment increases economic growth increases and this leads to more pollution to the environment (Tamazian and Rao, 2010) . The coefficient of natural resources βNR is expected to be positive. The logic behind this is that as the country exploits more natural resources, the demand for energy increases this worsens environmental degradation as more GHS gases are emitted to the environment . At the same time during natural resource exploitation heavy machinery are used which consume more energy there increasing air pollution. As the population of the country increase the demand for goods and services increase including demand for automobiles. High population also engages in deforestation. The increase in demand for automobiles and increase of deforestation impacts positively to air pollution therefore the coefficient of pop βpop is expected to be positive. Lastly the coefficient of trade openness is expected to be positive or negative. As the country engages more in international trade, more inputs to the industrial sector are imported which emit more carbon dioxide hence Journal of Economics and Sustainable Development www.iiste.org ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online) Vol.11, No.12, 2020 worsening environment quality. It is also possible that with expansion of trade countries can access environmental friendly inputs in the international market that may reduce CO2 emission (Grossman and Krueger, 1991).
To test for existence of long run relationship among the variables this paper uses ARDL bounds test for cointegration. This method is applicable for small samples and cases where the variables are I(0) or I(1) or a mix of the two levels of integration (Nkoro &Uko, 2016;Pesaran, Shin &Smith, 2001).
Testing for long run relationship, the ARDL bounds testing approach for cointegration involves two steps. The first step is to investigate existence of long run relationship among all the variables in the model. This involves estimating a dynamic error correction model computed from ARDL through a simple linear transformation. The unrestricted form of the ECM of the ARDL model is expressed as:  (6) was also expressed with Y 2 instead of FD 2 to test for EKC. ∆ is the first difference operator βi ,δi , αi , ci di , γi ,πi ri, ei,ki, are coefficients of short run dynamic model , σi are long run parameters of the model and ut are white noise errors. To perform the bounds test the following hypothesis is tested: H0: σ1= σ2= σ 3= σ 4 = σ 5 = σ 6= σ 7 σ 8 σ 9=0 (long run relationship does not exist) , H1: σ 1≠ σ 2≠ σ 3≠ σ 4≠ σ 5 ≠ σ 6 ≠ σ 6≠ σ 6≠ σ 6≠ 0 ( long run relationship exists). Once the test is done, if one fails to reject the null hypothesis then we conclude that there is no cointegration in the series, however if the null hypothesis is rejected then we conclude that there is cointegration in the series.
The decision is based on the calculated F statistic which is derived from the Wald test is compared with the critical values of Pesaran et al.(2001). Pesaran et al (2001) Where βij , are short term parameters , while λ is the error correction coefficient which shows the speed of adjustment of the dependent variable to its equilibrium value when a shock occurs in the system while ECT are the residuals obtained from the estimated equation (6).

Descriptive Statistics
The descriptive statistics for the variables included in the empirical model is presented in Table I in the appendix

Unit Root Tests
To test for unit root in this paper, Augmented Dickey-Fuller (ADF) and PP test of Phillips and Perron (1988) are used. This is done to confirm that all variables were I(0) and I(1) only since ARDL is applicable for the cases of variables which are I(0) and I(1) only and not I(2).The unit results are shown in Table II in the appendix.
To obtain lag length of the variables the Akaike Information Criteria (AIC) was used. According to the results in Table2, foreign direct investment and economic growth were found to be stationary at levels while carbon Journal of Economics and Sustainable Development www.iiste.org ISSN 2222-1700 (Paper) ISSN 2222-2855 (Online) Vol.11, No.12, 2020 dioxide emission, financial development, energy consumption, natural resources, population growth and trade were found to be stationary at their first differences. Since the variables were a mix of I(0) and I(1), the ARDL bounds test was therefore found to be applicable. The results for ARDL bounds test are presented in Table III in the appendix.
The ARDL bounds test was done to investigate the existence of long-run relationship among the variables in the model. The F statistic tests the joint null hypothesis which assumes that the coefficients of all lagged variables at levels is equal to zero meaning that there is no long run relationship among the variables. In testing for this hypothesis, each variable was assumed to be a dependent variable. The results shown in Table 3 shows that there is contegration when, financial development, foreign direct investment, economic growth and natural rent are dependent variables. Financial development, foreign direct investment and economic growth were significant at one percent as their respective F -statistic values (5.6, 6.3, 4.7) exceeded the Pesaran et al (2001) upper critical values while natural rent function with an F-statistic value of 3.38 was significant at five percent level. With these ARDL bounds test result for the four functions, the null hypothesis which states that there is no cointegration was rejected. The results in Table3  The trace statistic suggest presence of five cointegrating equations at the 0.05 level while maximum eigen value statistic shows presence of two cointegrating equations. Following confirmation of existence of cointegration the long run equation was estimated which gave the results in Table V. Two models are estimated , the first one included financial development squared to find out whether the relationship between financial development and CO2 is inverted U-shaped or not and the second model had economic growth squared also to check if EKC exists between CO2 and income. All variables were expressed in natural logarithm form. Financial development has a positive effect in model one while in model two it has a negative impact. In model one if financial development increases by one percent, CO2 increase by 4.12 percent. This result was statistically significant at 1 percent. The square of financial development shows a significant negative impact on CO2 .The significant negative impact of the square of financial development on environmental quality, shows that the relationship between environmental quality and financial development is non-linear. The environmental Kuznets curve exists between environmental quality and financial development. , found existence of EKC on the relationship between CO2 and financial development.
In both models the coefficients of CO2 lagged, energy consumption and natural rent were statistically significant at one percent. The results suggest that if lagged carbon dioxide emissions increase by 1 percent, carbon dioxide emissions will increase by about 0.3 per cent in model one and around 0.4 percent in model two. This result was statistically significant.
Energy consumption in both models has a positive impact on environmental quality. Increase of energy consumption by one percent worsens environmental quality by 2.54 percent in model one and 1.96 percent in model 2, respectively. Both coefficients were statistically significant at 1 percent. Increases foreign direct investment worsens environmental quality in Kenya. An increase of foreign direct investment by one percent increases CO2 by 0.017 percent in model one. An increase in natural rent in both models improves environmental quality. An increase of natural rent by one percent reduces carbon dioxide emissions by 0.18 percent in model one, and by 0.145 in model two. Both coefficients were statistically significant at 1 percent. Growth in population and trade openness worsens environmental quality in Kenya during the sample period. Growth in Kenya`s population by 1 percent increases CO2 emissions by 0.21 per cent while growth in trade openness by 1 percent increases CO2 emissions by 0.19 percent. The population growth coefficient was statistically significant at five percent while the trade openness coefficient was statistically significant at one percent in model one. From the long run results EKC theory, for the relationship between CO2 and economic growth does not exist for the case of Kenya. This result was also found by (Simiyu, 2017).The diagnostic tests done for the error term shows that the two models estimated was sound. The results in Table VI shows short run estimated coefficients of the model. Figure II, shows the plot of the CUSUM of squares.
According to the results in Table VI, increase in financial development, energy consumption, foreign direct investment, population growth and trade significantly worsens environmental quality in Kenya in the short run during the study period. However, increases in natural rent holding after factors constant, improves environmental quality in nature as its coefficient (-0.011) is statistically significant at one percent. The error correction coefficient has a negative significant coefficient (-0.70). The results suggest that short run deviations of CO2 from its long run value were corrected about 70 percent each year. The significant ECT coefficient also confirms existence of long run relationship among variables in the model. The model passed all the diagnostic tests of the error term .The plot of the CUSUM of squares in Figure I, shows that the estimated coefficients are stable in the entire sample Journal of Economics and Sustainable Development www.iiste.org ISSN 2222-1700(Paper) ISSN 2222-2855(Online) Vol.11, No.12, 2020 period. Table VII, shows Granger causality analysis results.
From the Granger causality results there is unidirectional causality running from financial development to environmental quality. The F -statistic 3.59 is statistically significant at ten percent. There is also unidirectional causality running from CO2 to GDP. CO2 emissions Granger causes GDP. The F-statistics 3.39 is statistically significant at ten per cent. Neutrality hypothesis is confirmed between financial development and GDP in the sample period.

Conclusion and policy recommendations
This paper investigated the relationship among environmental quality, financial development and economic growth in Kenya. The paper used ARDL bounds test to test for long run relationship among the variables. Time series data for the period 1970-2019 was used. The paper found that there is long run relationship when financial development, foreign direct investment, economic growth and natural rent are dependent variables. The paper found that financial development significantly worsens environmental quality in model one and that there is existence of EKC between environmental quality and financial development. Long run results show that in both models increases in lagged CO2, and energy consumption worsen environmental quality while increases in natural rent improve environmental quality. Other variables that significantly worsen environmental quality are foreign direct investment, population growth and trade openness .As the country increases its international trade, and foreign direct investment, CO2 emissions increase. Growth in population as it increases its energy consumption, destroys forests and its increase in demand for industrial products, emission of greenhouse gases increase all these degrades the environment. From the results EKC does not exist between CO2 and economic growth in Kenya during the sample period Short-run results confirm that increases in financial development energy consumption foreign direct investment trade and population growth worsen environmental quality while natural rent improve CO2.
Causality results indicate that there is unidirectional causality running from financial development to CO2 emissions and from CO2 emissions to GDP.
The conclusions of this paper are that financial development, energy consumption, population growth, trade openness, and foreign direct investment impact on environmental quality negatively while natural resources improve air pollution in Kenya. The hypothesis of EKC for the relationship between income and air pollution is not valid in Kenya, however it is applicable for the relationship between air pollution and financial development.
The policy implication of the paper findings suggest that the government of Kenya should come up with policies that make sure that the loans given by financial sector are investment in environment friendly projects in the short run however in the long run as the financial sector develops pollution will go down without much effort by the government. Appropriate policies should be implemented to control population growth as the results show that high population worsens CO2 emissions. The government and all stake holders should increase investment in clean energy as this is known to pollute less the environment. Foreign direct investment flows to be invested more on environment friendly projects. Lastly as the country engages more in foreign trade, the government should put in place policies that inhibit importation of inputs that increase pollution.  -1.61 -7.69*** -1.53 -7.82*** I(1) (***), (**), means the series is stationary at 1% , and 5% respectively, all tests were done with intercepts only