Environ Sci Pollut Res DOI 10.1007/s11356-016-8326-1
RESEARCH ARTICLE
Financial development and sectoral CO2 emissions in Malaysia Ibrahim Kabiru Maji 1,2 & Muzafar Shah Habibullah 1 & Mohd Yusof Saari 1
Received: 3 October 2016 / Accepted: 21 December 2016 # Springer-Verlag Berlin Heidelberg 2017
Abstract The paper examines the impacts of financial development on sectoral carbon emissions (CO2) for environmental quality in Malaysia. Since the financial sector is considered as one of the sectors that will contribute to Malaysian economy to become a developed country by 2020, we utilize a cointegration method to investigate how financial development affects sectoral CO2 emissions. The long-run results reveal that financial development increases CO2 emissions from the transportation and oil and gas sector and reduces CO2 emissions from manufacturing and construction sectors. However, the elasticity of financial development is not significant in explaining CO2 emissions from the agricultural sector. The results for short-run elasticities were also consistent with the long-run results. We conclude that generally, financial development increases CO2 emissions and reduces environmental quality in Malaysia.
Keywords Financial development . Sectoral carbon emissions . Cointegration method . Malaysia
Responsible editor: Philippe Garrigues Electronic supplementary material The online version of this article (doi:10.1007/s11356-016-8326-1) contains supplementary material, which is available to authorized users. * Ibrahim Kabiru Maji
[email protected]
1
Faculty of Economics and Management, Universiti Putra Malaysia, Serdang, Malaysia
2
Department of Economics, Bauchi State University, Gadau, Nigeria
Introduction Financial system plays a key role in ensuring financial intermediation by facilitating the flow of funds from the surplus spending unit to the productive sector of the economy. This ensures that financial resources are allocated efficiently in promoting economic growth and development. Financial stability on the other hand describes the situation in which the financial intermediaries function effectively and ensure the working of key financial institutions within the economy (BNM 2016). In Malaysia, the Financial Sector Master Plan (FSMP), which was set up for the period 2001–2010, provides the foundation for the development of the financial sector. The Master Plan helps in strengthening the domestic financial intermediaries and their infrastructure. This was later supported by the Second Master Plan called the Financial Sector Blueprint to provide guide to financial institutions on increasing the financial inclusion for high value added and ensure that the Malaysian economy is among the developed countries in the year 2020. As such, the Financial Sector Blue Print for the period of 2011–2020 builds on the achievement of FSMP. In the next decade, the financial sector in Malaysia is projected to grow beyond its role as an enabler of growth and development to be a key driver of economic growth, with growth in the financial sector strongly anchored to growth in the real sector. Based on the projected growth rate for the next decade, it is envisaged that the financial sector will grow at an annual rate of 8–11%, increasing the depth of the financial system to six times of gross domestic product (GDP) in 2020. Moreover, the contribution of the financial sector to nominal GDP is expected to grow from 8.6% of nominal GDP in 2010 to between 10 and 12% by 2020 (BNM 2016). To further ensure financial deepening, the financial sector through bank Negara had strengthened the management of domestic liquidity, exchange rates and managing external
Environ Sci Pollut Res
reserves to safeguard its value and optimize its returns through monetary policy decisions. To sustainably achieve these plans, financial institutions must live by example through ensuring friendly environment and minimizing negative impact on environment. This requires the management of financial institutions’ environmental footprint through efficient use of resources such as energy and waste management in physical operations and supply chains. Thus, it requires the development of an environmental management program that addresses climate change and greenhouse gas emission reduction. For instance, a Bank can support energy efficiency by promoting renewable power generation and adopting green building standards and practices in their operations. Financial development and environmental quality nexus have also been documented in the literature. Recent literature in this respect emphasizes on the role of financial sector in increasing energy consumption when it provides services to other sectors that subsequently lead to CO 2 emissions (Shahbaz et al. 2013a). As such, greater economic growth and enhanced financial development could be associated with environmental pollution. Hence, improvement in financial development such as providing credit facility to private sectors may cause the benefiting sectors to demand more intermediate input such as energy and other material inputs that could pollute the environment. In January 2016, the net financing to the private sector in Malaysia grew by 8.3% and between January 2014 and January 2016; the gross private sector financing has increased to about RM140 billion, representing 12.7% of 2014 gross domestic product (GDP). Similarly, the loan disbursement to private businesses and household has also increased to about RM110 billion from 2014 to 2016. This could increase the purchase of energy-consuming materials like production machines, vehicles, and equipment that further increase carbon emissions. On the other hand, financial development has also been perceived as one of the ways that can help in achieving sustainable and quality environment (Sadorsky 2010). For instance, monitoring the adherence to cleaner environmental policies of private sectors by financial sector before providing the private sector with any service could assist in mitigating CO 2 emissions, an indicator for environmental quality (Sadorsky 2011). The financial sector can also penalize businesses that do not adhere to energy conservation policy by not providing them with credit facilities. After the work of Grossman and Krueger (1995) which shows that economic growth will continue when carbon emissions decrease (i.e., inverted U-shape relationship between economic growth and CO2 emissions), empirical evidences from the work of Ozturk and Acaravci (2013) and Omri et al. (2015) have verified this finding. On the other hand, Wang et al. (2016) have shown a positive causality running from economic growth to CO2 emissions in China. More empirical findings from testing the Environmental Kuznets curve
(EKC) theory shows that in the cases of Brazil, China, and Indonesia, CO2 emissions will decrease over time when income increases (Alam et al. 2016). Evidence for the case of Malaysia, has shown that petroleum and natural gas energy consumption were estimated to account for 40 and 36% of total energy consumed (EIA 2015). Thus, Malaysia’s CO2 emissions per capita share of total emissions is about 0.77%. This is greater than the per capita emissions of some oil-exporting countries like Iraq and Venezuela that emit about 0.37 and 0.67%, respectively. In 2011, total greenhouse gas (GHG) emission in Malaysia is about 0.6% of global emissions. However, Malaysia intends to reduce its GHG emission intensity of GDP by 45% by 2030, relative to the emission intensity of GDP in 2005. This consists of 35% on an unconditional basis and a further 10% is conditioned upon receipt of climate finance, technology transfer, and capacity building from developed countries (UNFCCC 2015). Furthermore, related studies on the relationship between financial development and energy and environmental quality in Malaysia include the following: Shahbaz et al. (2013a), Alam et al. (2015). On the other hand, Shahbaz et al. (2013a) reveals that financial development reduces CO2 emissions; Alam et al. (2015) found a positive relationship between indicators of financial development and environmental quality. The difference between our study and these studies is that we have considered sectoral CO2 emissions in our analysis. The knowledge of sectoral CO2 emissions approach help to show which sector’s emissions is due to financial development and thus guide policy maker in decision making regarding sectoral CO2 mitigation measures. As such, disaggregated CO2 emissions can help in the efficient allocation of a country’s scarce resource to cub environmental problems of targeted sectors. This paper investigates the impacts of financial development on sectoral CO2 emissions in Malaysia. Investigating sectoral carbon emissions with respect to financial development is not common in literature, particularly in an augmented theory that recognized energy as an important production input beside labor and capital. Figure 1 shows stylized facts on the trend of variables that include per capita and sectoral CO2 emissions with other control variables. The BLiterature review^ section of this paper presents related literatures on financial development and environmental quality. The BMethodology^ section detailed the methodological framework. The BResults and discussions^ section presents the results and major findings while the BConclusion^ section focused on conclusion and policy implications.
Literature review Schumpeter (1911) was one of the pioneer scholars that postulates the relationship between financial development and
Environ Sci Pollut Res CO2 emissions from manufacturing and construction
CO2 emissions per capita 2.4
-12.8
2.0
-13.2
1.6 -13.6 1.2 -14.0
0.8 0.4 1980
1985
1990
1995
2000
2005
2010
-14.4 1980
1985
CO2 emissions from agriculture -13.0
-7.4
-13.2
-7.5
-13.4
-7.6
-13.6
-7.7
-13.8
-7.8
-14.0
1985
1990
1995
2000
2005
1995
2000
2005
2010
CO2 emissions from transportation
-7.3
-7.9 1980
1990
2010
-14.2 1980
1985
1990
CO2 emissions from oil and gas
1995
2000
2005
2010
2005
2010
2005
2010
2005
2010
Income
-12.0
9.2
-12.4
8.8
-12.8 8.4 -13.2 8.0
-13.6 -14.0 1980
1985
1990
1995
2000
2005
2010
7.6 1980
1985
1990
Energy consumption 8.0
-11.8
7.6
-12.0
7.2
-12.2
6.8
-12.4
6.4 1980
1985
1990
1995
2000
1995
2000
Financial development
2005
2010
-12.6 1980
1985
1990
Capital
1995
2000
Trade openness
-12.8
-11.4 -11.6
-13.2
-11.8 -13.6 -12.0 -14.0
-14.4 1980
-12.2
1985
1990
1995
2000
2005
2010
-12.4 1980
1985
1990
1995
2000
Fig. 1 Financial development, sectoral CO2 emissions, and other determinants (source: authors, using data from WDI 2015)
economic growth by emphasizing the importance of the financial sector in the process of economic growth. He argues that the financial system plays an important role in mobilizing
savings and allocates them to productive activities. Goldsmith (1969), McKinnon (1973), and King and Levine (1993) are also of the view that sound financial system is a
Environ Sci Pollut Res
catalyst for economic growth due to its capacity to reallocate excess fund from the surplus spending unit to the deficit spending unit of an economy. The surplus funds can be mobilized from household and government savings. Thus, the financial system helps in reducing transaction cost, information cost, and monitoring cost of firms and increases productivity (Shahbaz 2013). The use of financial system facilities can also reduce the risk associated in dealing with cash during transaction process. Recently, Sadorsky (2010) argues that financial development is an engine of economic growth in emerging economies like Malaysia. This is because financial development promotes economic activity such as foreign direct investment (FDI) and efficiency in banking services and promotes stock market participation through sale and purchase of quoted shares of companies. Thus, financial development can promote growth and increase energy consumption (Sadorsky 2011; Shahbaz et al. 2013a, b) and then leads to CO2 emission (Zhang and Chen 2011). On the other hand, financial development also improves environmental quality (Tamazian et al., 2009; Jalil and Feridun 2011), through efficient use of energy. Thus, the development stage of an economy determines the role of financial development on its environmental quality. In addition, Frankel and Rose (2002), Sadorsky (2011), Shahbaz et al. (2013a, b),and Aslan et al. (2014) also argue that financial sectors direct firms to use environmentally friendly technology in production process in order to reduce environmental degradation. The financial sector achieved this by ensuring that business entrepreneurs comply with requirement of environmental friendly practices (Frankel and Rose 2002) before accessing credit facilities to stimulate their business. Hence, the financial system can penalize firms that do not comply with such requirement by preventing them from accessing credit facilities. Through this practice, financial systems can reduce energy emissions and enhance technological invention in energy-related industry (Sadorsky 2010) and as a result promote environmental quality. Moreover, the theoretical link between financial development and energy use has three effects (Çoban and Topcu 2013): (i) the direct effect which suggests that development in financial system leads businesses to borrow at cheaper rates in order to buy durable goods which production requires greater energy demands; (ii) the business effect which points out that certain improvements in the financial system of the economy helps businesses to access financial capital at lesser costs, leading to increase in energy demand; and (iii) the wealth effect that expands stock market operations, consumer confidence, energy consumption, economic growth, and CO2 emissions. Before the work of Çoban and Topcu (2013), Grossman and Krueger (1995) had earlier related the level of economic growth with environmental performance. In agreement with Kuznets (1955), the findings of Grossman and Krueger (1995)
show that the relationship between economic growth and CO2 emissions is inverted U-shape, implying that the early stage of economic growth is linked with greater environmental hazard up to a certain threshold beyond which additional expansion of output level reduces CO2 emissions and improves environmental quality. Later on, the nexus between financial development and environmental quality was introduced in the literature (see for instance, Bello and Abimbola 2010; Ozturk and Acaravci 2013; Shahbaz et al. 2013a; Boutabba 2014). Bello and Abimbola (2010) and Ozturk and Acaravci (2013) show that financial development has insignificant impact on CO2 emissions. While Bello and Abimbola (2010) used capital market indicator, Ozturk and Acaravci (2013) used money market indicator for financial development. Furthermore, Boutabba (2014) provides evidence of positive impact of financial development on environmental quality. On the other hand, Shahbaz et al. (2013a) reveal a negative sign between financial development and environmental quality. These divergence empirical results may be owing to differences in variables, frequency of data employed, and the state of economic development of the country under consideration. Despite these inconclusive theoretical and empirical perspectives on the relationships between financial development, energy demand, and carbon dioxide emissions, investigating financial development and sectoral CO2 emissions has eluded literature and thus constitute a literature gap that provide the medium to contribute to scientific knowledge.
Methodology The paper uses the extended Cobb-Douglas production function as its theoretical framework and Autoregressive Distributed Lag (ARDL) approach as the empirical model. As such, besides labor and capital, the production function was extended to include energy consumption as production input which is directly related to CO2 emissions, an indicator for environmental quality (see for example, Ang 2008; Shahbaz et al. 2013a, b; Omri et al. 2015). On the other hand, the technological component of the model provides a safe ground to incorporate financial development. As such, the augmented Cobb-Douglas production function following Shahbaz et al. (2013b) and Omri et al. (2015) is given as: Y t ¼ AK at EC λt Lβt eμt
ð1Þ
where Yt is real income; Kt, ECt, and Lt stand for capital, energy consumption, and labor inputs, respectively. The term A represents technological progress, and et is the error term assumed to be normally distributed with zero mean and constant variance. On the other hand α , λ, and β refer to the
Environ Sci Pollut Res
output elasticity for domestic capital, energy demand, and labor, respectively. Given the technological level at a particular point of time, there exists a direct relationship between energy demand and CO2 emissions (Omri et al. 2015). As such, ECt = aCOt, where COt refers to CO2 emissions, so that the extended model can be rewritten as follows: Y t ¼ AK αt aCOλt Lβt eμt
ð2Þ
Within an extended Cobb-Douglas production function, we allow technological progress to be endogenously determined by financial development and trade openness. This is because financial development facilitates the inflow of foreign direct investment and transfer of better technology. Financial development also facilitates economic growth through efficient capital formation while trade openness improves technological progress and its diffusions. If financial development and trade openness represent technological progress, they can be captured in the model as follows: At ¼ γFDt θ TOt ϕ
ð3Þ
where γ indicates the time-invariant constant; FDt represents financial development; TOt represents trade openness; θ and φ are the elasticity of financial development and trade openness, respectively. By substituting Eq. (3) into Eq. (2), we obtained the following model (4): yt ¼ γ:fd t θ tot ϕ k αt coλt l βt eμt
ð4Þ
Following Shahbaz and Lean (2012) and Omri et al. (2015), we leave the impact of labor to be constant. If we take the logarithm, then the linearized extended Cobb-Douglas production function is given as: lnyt ¼ γ þ λlncot þθlnfd t þϕlntot þαlnk t þξt
ð5Þ
development, trade openness, and capital, respectively, while lncojt denotes per-capita CO2 emissions and CO2 emissions from other sectors. As such, j denotes lncopt for per-capita CO2 emissions, lncomt for CO2 emissions from manufacturing sector, lncoat for CO2 emissions from agricultural sector, lncott for CO2 emissions from transportation sector, and lncoogt for CO2 emissions from oil and gas sector. The symbol εt refers to the stochastic error term expected to be normally distributed with zero mean and constant variance. The expected sign of the coefficient of long-run impact with respect to the explanatory variables are projected as follows: we expect the coefficients π2, π3, π4, π5 and π6 >0. However, higher level of income as well as higher level of financial development could mitigate CO2 emissions, and then we would expect that π2, π4 <0. Furthermore, efficient trade relation with the rest of the world by adhering to principles of carbon footprint can assist in mitigating CO2 emissions; as such, we also expect that π5 < 0. Therefore, Eq. (6) represents the long-run model. In order to estimate Eq. (6), we employed the autoregressive distributed lag (ARDL) framework of Pesaran et al. (2001) to develop a model for the relationship between financial developments with each sectoral CO2 emissions. Some of the advantages of this model include efficient property for small sample size, combination of I(0) and I(1) variables in the same framework, and simultaneous estimation of long-run and short-run model without loss of degree of freedom. Hence, to ascertain the validity of the long-run Eq. (6), we need to determine whether the variables are cointegrated. Cointegration of the long-run model can be tested by employing the following unrestricted error-correction model (UECM). For instance, see Wang et al. (2011), Jalil and Feridun (2011), Çoban and Topcu (2013), Shahbaz et al. (2013a, b) and Rafindadi and Ozturk (2016). However, expanding CO2 emissions into sectoral level is one of the major differences between this work and their studies: n−1
n
n
i¼1 n
i¼0 n
i¼0 n
Δlncojt ¼ ϕ1 þ ∑ ϕ2i Δlncojt−i þ ∑ ϕ3i Δlnyt−i þ ∑ ϕ4i Δlnect−i þ ∑ ϕ5i Δlnfd t−i þ ∑ ϕ6i Δlntot−i þ ∑ ϕ7i Δlnk t−i i¼0
where γ is the intercept; λ , θ , φ , α are coefficients of explanatory variables; and ξt is the residual term. Thus, from (5), we can estimate the empirical model of our interest. Following Lean and Smyth (2010), Shahbaz and Lean (2012), Shahbaz et al. (2013b), and Omri et al. (2015), we divide non per-capita variables by population so that all variable will be in per-capita terms as follows: lncojt ¼ π1 þ π2 lnyt þπ3 lnect þπ4 lnfd t þπ5 lntot þπ6 lnk t þεt
ð6Þ
where lnyt , ln ect , ln fdt , ln tot , and ktare the per-capita variables of real income, energy demand, domestic credit to private sector by banks as an indicator for financial
i¼0
i¼0
ð7Þ
þβ 1 lncojt−1 þ β 2 lnyt−1 þ β 3 lnect−1 þ β 4 lnfd t−1 þβ5 lntot−1 þ β 6 lnk t−1 þ et
where lncojt , ln yt , ln ect , ln fdt , ln tot , and ln kt are as defined earlier, et is the residual, and Δ is the first difference operator. From Eq. (7), we can derive the long-run coefficients as follows: π1 ¼ 1=β1 ; π2 ¼ −β 2 =β1 ; π3 ¼ −β3 =β 1 ; π4 ¼ −β4 =β1 ; π5 ¼ −β5 =β1 ; π6 ¼ −β 6 =β1
Environ Sci Pollut Res
On the other hand, to estimate the short-run or the errorcorrection model, we have the following: n−1
n
n
i¼1 n
i¼0 n
i¼0 n
Δlncojt ¼ ϕ1 þ ∑ ϕ2i Δlncojt−i þ ∑ ϕ3i Δlnyt−i þ ∑ ϕ4i Δlnect−i þ ∑ ϕ5i Δlnfdt−i þ ∑ ϕ6i Δlntot−i þ ∑ ϕ7i Δlnk t−i ð8Þ i¼0
i¼0
i¼0
þηECT t−1 þ μt
Results and discussions
where h ECT t−1 ¼ εt−1 ¼ lncojt−1 − π1 þ π2 lnyt−1 þ π3 lnect−1 i þπ4 lnfd t−1 þπ5 lntot−1 þπ6 lnk t−1
measurement for carbon emissions from transportation sector while CO2 emissions from liquid fuel consumption (% of total) is used as the measurement for CO2 emissions from oil and gas sector. Total population variable was then used to divide non per capita variable to ensure that all the variables are in per-capita terms to provide suitable bases for the adoption of extended Cobb-Douglas production function as the theoretical framework of this paper.
ð9Þ
ECTt − 1 refers to the error-correction term while η is the coefficient of error-correction term that measures the speed of adjustment to long-run equilibrium. If η is negative and significant, then cointegration exists. In order to test for cointegration, we first estimate the Fstatistics and compare with the upper-bound critical values in the Narayan (2005) table. Cointegration exists when Fstatistics is greater than the upper bounds, and it does not exist when the F-statistics is lower than the lower-bound critical values of Narayan. On the other hand, the outcome is inconclusive when the F-statistics falls within the upper and lower bounds. Thus, testing for cointegration requires setting of a null hypothesis. Hence, the null hypothesis is given by Ho: β1 = β2 = β3 = β4 = β5 = β6 = 0 while the alternative hypothesis is given as H1: β1 ≠ β2 ≠ β3 ≠ β4 ≠ β5 ≠ β6 ≠ 0. Furthermore, the datasets for our estimation were collected from Word Bank’s World development Indicators (WDI) and Food and Agricultural Organization (FAO) databases for the period of 1980–2014. The variables include financial development, real income, energy consumption, trade openness, capital stock, CO2 emissions per capita, CO2 emissions from manufacturing and construction, CO2 emissions from agriculture, CO2 emissions from transportation, and CO2 emissions from oil and gas sector. We used domestic credit by banks to private sectors as the measurement for financial development, GDP per capita (constant 2005 US$) as the measurement for real income, energy use (kg of oil equivalent per capita as the measurement for energy consumption, trade (% of GDP) as the measurement for trade openness, gross fixed capital formation (% of GDP) as the measurement for capital stock, CO2 emissions (metric tons per capita) as an indicator for per-capita CO2 emissions, CO2 emissions from manufacturing industries and construction (% of total fuel combustion) as the measurement for carbon emissions from manufacturing and construction sector, CO2 emission from agriculture % of total as an indicator for CO2 emissions from the agricultural sector, CO2 emissions from transport (% of total fuel combustion) as the
The result discussion begins with a pretesting for the stationarity of the variables using the augmented Dickey and Fuller (1981) and Phillips and Perron (1988) unit root test. We also report the ADF-GLS test of Elliot et al. (1996) and Zivot and Andrews (1992) that handle some of the biasness of ADF and PP, particularly the power problem. Thus, a variable is said to be stationary when its mean, variance, and covariance are time invariant. This is necessary in order to verify the level of integration of the series and to also avoid spurious result. Furthermore, our empirical framework of ARDL does not allow the use of I (2) series. Only I (0) or I (1) or the integration of the two is allowed (Pesaran et al. 2001). Tables 1 and 2 show the results of the unit root tests with all the variables stationary. After ensuring the stationarity of the variables, a descriptive statistics and correlation matrix were carried out on the variables and the results are presented in Table 3. The statistics of Jarque-Bera test shows that that all the variables are normally distributed. On the other hand, the correlation analysis reveals not only the strength of relationship but also the direction of the relationship among the variables. The result of the correlation was also presented in the lower part of Table 3. For our main variable of interest, the results show that a positive correlation exist between financial development and CO2 emissions from manufacturing and construction, CO2 emissions from transportation, and CO2 emissions from oil and gas sector. Moreover, a negative correlation exists between financial development and CO2 emissions from agriculture and per-capita CO2 emissions. Furthermore, a mixed correlation exists among the control variables. For instance, a negative correlation exists between gross capital formation and income and energy while a positive correlation exists between trade openness and financial development. Next, we present the results of the cointegration tests in Table 4. The lag selection of the models was estimated based on the Akaike information criterion (AIC) and Schwarz Bayesian criterion (SBC), and the results were chosen base on the best model. Financial development and CO2 emissions from manufacturing and construction, CO2 emissions from agriculture, CO2 emissions from oil and gas, and CO2 emissions per capita were estimated based on AIC while financial development and CO2 emissions from the transportation sector was
Environ Sci Pollut Res Table 1 Results of unit root tests Variables
ln copt ln comt ln coat ln cott ln coogt ln yt ln ect ln fdt ln tot ln kt
ADF
PP
Level q
1st difference q
Level q
1st difference q
−1.5057(0.808)0 −2.9698(0.155)0 −1.9091(0.628)0 −2.8512(0.191)2 −2.6963(0.244)0 −1.7418(0.710)0
−6.4091(0.000)***0 −5.4565(0.001)***0 −5.4203(0.001)***0 −7.0422(0.000)***0 −8.2079(0.000)***0 −4.7565(0.003)***0
−1.5655(0.786)2 −3.0656(0.131)3 −2.0228(0.568)1 −2.4969(0.328)1 −2.6677(0.255)2 −1.9372(0.613)2
−6.4013(0.000)***1 −5.7420(0.000)***6 −5.413(0.0001)***3 −7.0422(0.000)***0 −8.8458(0.000)***3 −4.7565(0.003)***0
−1.2124(0.892)0 −2.4854(0.333)0 0.0550(0.995)0 −2.3973(0.374)1
−5.4969(0.001)***1 −5.4096(0.001)***0 −3.5079(0.056)*2 −4.2719(0.009)***0
−1.2124(0.892)0 −2.5154(0.319)3 −0.2134(0.990)1 −1.9648(0.599)1
−6.6384(0.000)***6 −5.4019(0.001)***3 −3.797(0.029)**14 −4.2250(0.011)**3
Note: Probability values are in parenthesis; unit root is with constant and trend while q is the lag length ***, **, and * represent significance at 1, 5, and 10% levels, respectively
estimated based on SBC, because they provide the best result for these models. The AIC performs relatively well when the sample size is small compared to a large sample size (Acquah 2010; Sbia et al. 2014). On the other hand, the SBC imposes more restriction on the number of parameters to be estimated compared to AIC; thus, the SBC is more parsimonious in lag length selection and reduces loss of degree of freedom (Ito 2009). The results of the bound test for cointegration in Table 4 include the five models estimated, their optimal lags, and Fstatistics. The results suggest that cointegration exists between financial development and sectoral indicators of CO2 emissions and per-capita CO2 emissions; as such, we accept the alternative hypothesis and reject the null hypothesis of no cointegration. The existence of cointegration is justified by the higher values of the F-statistics over their upper-bound Table 2 ADF-GLS and ZivotAndrews unit root test
Variables
ln copt ln comt ln coat ln cott ln coogt ln yt ln ect ln fdt ln tot ln kt
critical values based on Narayan (2005) and at conventional 5% level of significance. For example, the F-statistic value for per-capita CO2 emission model is 4.577, which is greater than the upper bound 4.443 at 5% level of significance. Similarly, all the sectoral CO2 emission models possess F-statistics that are greater than their upper bound at either 1 or 5% level of significance. For instance, the model for CO2 emissions from manufacturing and construction, CO2 emissions from transportation, and CO2 emissions from oil and gas sector all possess F-statistics greater that their upper bounds at both 1 and 5% level of significance. Thus, the null hypothesis set earlier as Ho: β1 = β2 = β3 = β4 = β5 = β6 = 0 can be rejected. Endogeneity problem is a situation where the explanatory variable of a model is correlated with the error term is addressed in these models with the help of lagged dependent
ADF-GLS
Zivot-Andrews
Level
1st difference
Level
Break point
1st difference
Break point
15.7783(0)*** 7.4514(0)*** 12.5257(0)*** 2.7636(2)*** 8.3084(0)*** 15.1846(0)*** 16.1428(0)*** 24.0187(0)** 58.4885(0)*** 6.7196(1)**
5.3485(0)*** 6.0555(0)*** 5.4289(0)*** 5.6454(0)*** 6.8474(0)*** 5.3484(0)*** 3.1073(1)*** 5.5103(0)** 3.5327(2)*** 6.8195(0)**
−3.7386(0)* −6.2236(2)*** −7.9377(0)*** −4.2353(2)*** −6.4896(0)*** −3.2777(1) −3.8426(0)** −4.1325(3) −3.8753(0)* −4.6892(1)
1991 2006 1990 1999 2004 1991 1991 1996 1998 1998
−7.7937(0)** −4.4397(3) −6.8352(0)*** −7.8580(0)* −9.9659(0)*** −6.0873(0)*** −3.2383(2) −6.6901(1)* 4.9512(0)* −5.2799(0)**
1997 2004 1992 1994 2007 1998 1999 1991 1987 1998
Values in parenthesis are the lag lengths of variable. The ADF-GLS unit root includes constant and linear trend and the Zivot-Andrews break also includes the intercept and trend ***, **, and * indicate 1, 5, and 10% level of significance
Environ Sci Pollut Res Table 3
Descriptive statistics and correlation matrix ln copt
ln comt
ln coat
ln cott
ln coogt
ln yt
ln ect
ln fdt
ln tot
ln kt
Mean
1.5118
−13.5784
−7.5880
−13.6426
−12.9479
Median Maximum
1.6882 2.0778
−13.5539 −12.8524
−7.5371 −7.3907
−13.6856 −13.1842
−12.9118 −12.0267
8.3389
7.4582
−12.2559
−11.8193
−13.5356
8.4268 8.9045
7.5479 7.9441
−12.2543 −11.855
−11.7611 −11.564
−13.5364 −12.9037
Minimum Std. Dev.
0.7031 0.4788
−14.3705 0.4453
−7.8824 0.1671
−14.1089 0.3098
−13.7326 0.57897
7.7485 0.3636
6.7717 0.4040
−12.5519 0.1970
−12.2686 0.1907
−14.0986 0.41767
Skewness
−0.430
−0.19511
−0.6694
0.0620
0.0616
−0.1882
−0.3314
0.3094
−0.8904
0.0469
Kurtosis Jarque-Bera
1.6632 3.6862
2.0451 1.5518
1.9208 4.3124
1.7613 2.2599
1.6478 2.6884
1.6736 2.7721
1.6477 3.3073
1.9610 2.13267
2.9893 4.6247
1.3549 3.9592
Probability ln copt ln comt ln coat ln cott ln coogt ln yt ln ect ln fdt
0.1583 1 −0.9311 0.7538 −0.9733 −0.9596 0.9857 0.9864 −0.1499
0.4603
0.1157
0.3230
0.2607
0.2501
0.1913
0.3443
0.0990
0.1381
1 −0.6124 0.9556 0.9697 −0.9631 −0.9504 0.2811
1 −0.7103 −0.6369 0.6915 0.7121 −0.11745
1 0.9724 −0.9853 −0.9746 0.2695
1 −0.9795 −0.9815 0.2906
1 0.9894 −0.2144
1 −0.1939
1
ln tot
−0.2668 −0.7712
0.5032 0.8744
0.1069 −0.3825
0.4059 0.8048
0.4028 0.8994
−0.3615 −0.8201
−0.3015 −0.8449
0.4915 0.3416
1 0.4253
1
ln kt
variables of each model. The presence of lagged dependent variable ensures that independent variables are mutually exclusive from the error term. As such, the next step involves the estimation and analysis of the long-run and short-run results following the same lag length selection criteria with the cointegration tests. In Table 5, panel A presents the estimated long-run results. The result shows that there exists a positive and significant relationship between financial development and CO2 emissions per capita in the long run. As a result, financial development increases CO2 emission and reduces environmental quality. Concisely, a 10% increase in financial Table 4 Results of the cointegration tests
development triggers per-capital CO2 emissions by 1.6% in the long run. Similarly, energy demand and gross capital formation were significant and positively related to per-capita CO2 emissions in the long run. However, income and trade openness do not possess statistical power to draw any inference. On the other hand, the impact of financial development on sectoral CO2 emissions from the manufacturing and construction sector reveals a negative and significant relationship, suggesting that financial development do not take place at the expense of CO 2 emissions and environmental quality.
Bound testing for cointegration Models (in logarithm) copt = f(yt, ect, fdt, tot, kt) comt = f(yt, ect, fdt, tot, kt) coat = f(yt, ect, fdt, tot, kt) cott = f(yt, ect, fdt, tot, kt) coogt = f(yt, ect, fdt, tot, kt)
Diagnostic tests Optimal lags (2, 2, 2, 1, 0, 2) (1, 2, 0, 2, 2, 0) (2, 1, 1, 1, 2, 2) (1, 0, 0, 0, 0, 2) (1, 1, 0, 0, 2, 0)
Significance level 1% level 5% level 10% level
F-statistics 4.5770 [0.009]*** 7.2876 [0.001]*** 4.7008 [0.008]*** 6.5627 [0.002]*** 6.1543 [0.006]*** Critical values Lower bounds (0) 4.257 3.037 2.578
R2 0.9952 0.9306 0.9192 0.9737 0.9674
Upper bounds (1) 6.040 4.443 3.858
Note: The critical values for the upper and lower bounds were obtained from Narayan (2005) ***Indicates significance at 1% level
DW tests 2.5622 2.0627 2.2623 2.0158 2.2145
Environ Sci Pollut Res
Hence, 10% increase in financial development reduces CO2 emissions from this sector by 3.5% and improves environmental quality. Similarly, income and trade openness were both significant with evidence of negative sign for their coefficient, suggesting that they reduce CO2 emissions in the manufacturing and construction sector. However, energy and gross capital formation were not significantly related to coefficient of CO2 emissions from the sector in question. In addition, the long-run result for financial development and sectoral CO2 emissions from the agricultural sector, although positive, did not reveal a statistical relationship, suggesting that financial development does not possess power to either increasing nor mitigating CO 2 emissions in the Table 5
agricultural sector of the Malaysian economy. However, energy consumption and gross capital formation were both significant and positively related to carbon emissions from the agricultural sector, but income and trade openness possess lesser power of explaining carbon emission in this sector. Moreover, the elasticity of financial development is positive and significantly related to carbon emissions of the transportation sector in the long run. To be precise, a 10% increase in financial development will increase carbon emissions from the transportation sector by 3.4%, suggesting that financial development does not mitigate carbon emissions in the transportation sector. The long-run result also reveals evidence of negative and significant relationship between energy demand,
Results of the long-run and short-run models of sectoral CO2 emissions
Regressors
Dependent variables
Long-run model (panel A) ln yt ln ect ln fdt ln tot
ln copt (2,2,2,1,0,2)
ln comt (1, 2, 0, 2, 2, 0)
ln coat (2, 1, 1, 1, 2, 2)
ln cott (1, 0, 0, 0, 0, 2)
ln coogt (1, 1, 0, 0, 2, 0)
−0.0098 (−0.0249) 1.2575 (3.6376)*** 0.1688 (2.4181)** 0.0378 (.5516)
−1.3545 (−3.1378)*** 0.4065 (1.0374) −0.3596 (−2.8040)** −0.3518 (−2.1478)**
−0.5771 (−1.2144) 1.0303 (2.3077)** 0.1207 (0.6638) 0.1041 (0.6592)
0.3519 (0.8044) −1.4599 (−3.4703)*** 0.3429 (1.9520)** −0.5140 (−3.6921)***
0.9721 (1.1175) −2.2962 (−2.8694)*** 0.5007 (2.1609)** 0.4116 (1.3808)
ln kt
0.1042 (1.8261)* −3.8813 (4.0558)*** Δln copt
0.0465 (0.4530) −13.2216 (−8.4124)*** Δln comt
0.3443 (2.657)** −3.1049 (1.8251)* Δln coat0.5031
−0.6225 (−4.4255)*** −15.9055 (−10.9958)*** Δln cott
0.1367 (0.9041) 8.7746 (3.2226)*** Δln coogt
0.3131 (1.5799) 0.7183 (2.0937)** 1.2697 (3.8068)*** 0.6663 (3.4138)*** −1.3298 (−4.6619)*** 0.2674 (2.8892)*** –
− −0.4414 (−0.5243) 1.1881 (1.4613) 0.5472 (1.0659) – 0.6882 (3.2052)*** 0.6882 (3.2052)***
(2.5055)** 0.5912 (1.1807) – −0.3598 (−1.0045) – −0.1716 (−1.2105) –
– 0.2467 (0.8419) – −1.0233 (−3.7640)*** – 0.2404 (2.3278)** –
– 1.7549 (1.9730)* – −1.6607 (−2.9747)*** – 0.3622 (2.1318)** –
0.0371 (0.5188) – −0.0568 (−0.9447) 0.0671 (1.1451) −3.8082 (−2.1455)** −0.9812 (−3.5399)***
0.7783 (2.1149)** 0.7783 (2.1149)** 0.0625 (0.4503) – −17.7999 (−6.3996)*** −1.3463 (−7.7868)***
0.3414 (1.7947)* 0.3414 (1.7947)* 0.1812 (1.7328)* 0.1177 (1.1400) −2.1696 (−1.4192) −0.6988 (−3.1283)***
−0.3603 (−3.4370)*** – −0.1293 (−1.7234)* 0.3152 (3.6898)*** −11.1487 (−5.0928)*** −0.7009 (−5.4719)***
−0.2529 (−0.7311) −0.7050 (−1.9213)* 0.0988 (0.8698) – 6.3462 (2.6310)** −0.7233 (−4.4304)***
5.7557 [0.016] 1.4854 [0.223] 0.5944 [0.743] 1.2570 [0.262]
0.1689 [0.681] 0.0522 [0.819] 0.9003 [0.638] 0.4340 [0.510]
2.2414 [0.134] 0.2227 [0.637] 1.0963 [0.578] 0.0258 [0.872]
0.0254 [0.873] 23.2633 [0.000] 0.8059 [0.668] 2.0772 [0.150]
1.7863 [0.181] 1.9810 [0.159] 18.7031 [0.000] 0.0028 [0.958]
Constant Short-run model (panel B) Δln cojt Δln yt Δln yt−1 Δln ect Δln ect − 1 Δln fdt Δln fdt − 1 Δln tot Δln tot − 1 Δln kt Δln kt−1 Constant ECTt − 1 Diagnostic tests (panel C) χ2sc χ2ff χ2nom χ2het
Note: Figures in parenthesis () indicate t statistics. χ2 sc, χ2 ff, χ2 nom, and χ2 het are tests for serial correlation, functional form, normality, and heteroscedasticity, respectively. For the diagnostic test, values in [] are p values ***, **, * denotes significance at 1, 5, and 10%, respectively
Environ Sci Pollut Res
trade openness, and gross capital formation and carbon emissions from the transportation sector. Furthermore, financial development shows evidence of positive and significant sign with the indicator of carbon emissions from the oil and gas sector, suggesting that financial development increases carbon emissions from the oil and gas sector of the Malaysian economy. A 10% increase in financial development increase carbon emissions from the oil and gas sector by 5%, with other factors influencing carbon emission held constant. On the other hand, energy demand provides evidence of negative and significant relationship with carbon emission from oil and gas sector, suggesting that it has not taken place at the expense of environmental quality. However, income, trade openness, and gross capital formation show evidence of non-statistical relationship with carbon emission from oil and gas sector in the long run. The implication of these findings is that financial development at present has not fully advanced to an extent that will bring about sustainable economic growth and development; as such, further financial deepening is needed to enhance
financial development to the level where it can mitigate CO2 emissions, reduce greenhouse gasses and global warming, and improve environmental quality. The level of capital is positively related to CO2 emissions from the agricultural sector; this may signal the possibility of higher CO2 footprint embedded in imported capital goods from trade with other countries such as machines and equipment used in the agricultural sector of the economy. As a result, policy makers must be fully aware on the implication of imported CO2 emissions from capital inflows in order to implement the required policy measure. In general, our long-run findings of the relationships between financial development and CO2 emissions from transportation, oil and gas, and emissions per capita are consistent with the findings of Bello and Abimbola (2010), Çoban and Topcu (2013), Boutabba (2014) and Shahbaz et al. (2015). On the other hand, the long-run model between financial development and CO2 emissions from manufacturing and construction sector is consistent with the work of Jalil and Feridun (2011) and Shahbaz et al. (2013a) while our result
16 12 8 4 0 -4 -8 -12 -16 86
88
90
92
94
96
98
00
CUSUM
02
04
06
08
10
12
14
5% Significance
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 86
88
90
92
94
96
CUSUM of Squares
Fig. 2 Model of CO2 emissions per capita
98
00
02
04
06
5% Significance
08
10
12
14
Environ Sci Pollut Res 16 12 8 4 0 -4 -8 -12 -16 86
88
90
92
94
96
98
00
CUSUM
02
04
06
08
10
12
14
5% Significance
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 86
88
90
92
94
96
98
00
CUSUM of Squares
02
04
06
08
10
12
14
5% Significance
Fig. 3 Model of CO2 emissions from the manufacturing and construction sector
for financial development and CO2 emissions from agricultural sector is in agreement with the work of Ozturk and Acaravci (2013). On the other hand, in panel B of Table 5, we present the results of the short-run model. Financial development, income, and energy demand were found to be significant and positively related to per-capita carbon emissions in the short run while gross capital formation has lesser power of drawing inference in the short run. Even more so, financial development and trade openness were significant and positively related to carbon emissions from the manufacturing and construction in the short run while the impact of income and energy are not statistically different from zero. The result of financial development and carbon emissions from agriculture in the short run also corroborates the long-run result; however, trade openness and gross capital formation were positive and significantly related to carbon emissions in the agricultural sector in the short run. Again, carbon emissions from the transportation sector and the oil and gas sector are positive and significantly different
from zero, indicating that financial development leads to carbon emissions from the transportation sector and oil and gas sector. Income is positively related to carbon emissions in the two sectors but only significant in the oil and gas sector while energy demand is also significant but negatively related to carbon emissions in both transportation and oil and gas sector. Moreover, the results of the error correction model for each carbon emission indicator is negative, less than one in absolute terms and significant. In addition, diagnostic tests were carried out for all the models and presented in panel C of Table 5. This includes tests for serial correlation, functional form, normality test, and heteroscedasticity. The results reveal that the models passed the major diagnostics tests. The diagnostic tests were further strengthened by a stability test of cumulative sum of recursive residuals (CUSUMs) and cumulative sum of squares of recursive residuals (CUSUMsq) forwarded by Brown et al. (1975). The results of the stability tests as presented in Figs. 2, 3, 4, 5, and 6 suggest that our models are relatively stable, consistent, and reliable. This is because the graphical
Environ Sci Pollut Res 16 12 8 4 0 -4 -8 -12 -16 86
88
90
92
94
96
98
00
CUSUM
02
04
06
08
06
08
10
12
14
5% Significance
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 86
88
90
92
94
96
98
CUSUM of Squares
00
02
04
10
12
14
5% Significance
Fig. 4 Model of CO2 emissions from the agricultural sector
illustrations show that the models are within the critical bounds at 5% level of significance. Additionally, the results of the Granger Causality tests for the per-capita and sectoral CO2 emissions are presented in Table 6. A long-run causality runs from financial development and other explanatory variables such as income, energy consumption, trade openness, and capital to CO2 emissions from manufacturing and construction, CO2 emissions from transportation sector, and CO2 emissions from the oil and gas sector. A long-run causal relationship also runs from the per-capital CO2 emissions, financial development, and other control variables to energy consumption and trade openness. Moreover, a longrun causality runs from CO2 emissions from the manufacturing and construction and financial development to the level of capital and trade openness. Long-run causality also runs from CO2 emissions from transportation and financial development to energy consumption and trade openness while a causal link runs from CO2 emissions from the oil and gas sector, financial
development, and others explain variable to energy consumption, trade openness, and the level of capital. Furthermore, a short-run causality runs from income and energy consumption to CO2 emissions per capita and a shortrun causality also runs from per-capita CO2 emissions to trade openness. A short-run causality run from the level of capital to CO2 emissions from the transportation sector. A bidirectional short-run causal relationship runs from income to energy consumption for per-capita CO2 emissions model. A short-run causality also runs from CO2 emissions from the manufacturing and construction and financial development to trade openness and a short-run causality runs from energy consumption to trade openness in the same model. A short-run causality runs from income to energy in the model of CO2 emissions from agricultural sector. Furthermore, a short-run causality runs from CO2 emissions of transportation sector and financial development to trade openness and from capital to CO2 emissions from transportation sector. Again, a shot-run causal
Environ Sci Pollut Res 16 12 8 4 0 -4 -8 -12 -16 86
88
90
92
94
96
98
00
CUSUM
02
04
06
08
10
12
14
5% Significance
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 86
88
90
92
94
96
98
CUSUM of Squares
00
02
04
06
08
10
12
14
5% Significance
Fig. 5 Model of CO2 emissions from the transportation sector
relationship runs from CO2 emissions from oil and gas sector and financial development to trade openness and from financial development to the level of capital.
Conclusion This paper estimates the impact of financial development on sectoral CO2 emissions in Malaysia. The theoretical framework for the paper is Cobb-Douglas production theory. The ARDL model was used as the empirical framework. The dataset employed ranges from 1980 to 2014 with income, capital, and trade openness serving as the control variable. Recent literature had only related financial development with CO2 emissions without including disaggregated CO2 emissions. Thus, expanding the impacts of financial development to sectoral CO2 emissions is one of the contributions of this research work. The knowledge of sectoral carbon emissions
assist in identifying which sector’s carbon emissions is due to financial development. This will then provide guide and direct policy makers in decision making with respect to sectoral CO2 reduction measures. Therefore, disaggregated CO2 emissions can direct the efficient allocation of scarce resource to handling specific sector’s environmental problem. We first verified the integration level of the series and then test the existence of long-run relationships before estimating the long- and short-run models. Our result suggest that financial development increase CO2 emissions in the transportation sector and oil and gas sector and also increases CO2 emissions per capita in the long run. On the other hand, financial development mitigates CO2 emissions from the manufacturing and construction sector and fails to influence CO2 emissions from agricultural sector in the long run. However, the short-run result reveals that financial development provoked CO2 emissions in the entire sector in question except agricultural sector whose coefficient is negative but do not provide evidence of
Environ Sci Pollut Res 16 12 8 4 0 -4 -8 -12 -16 86
88
90
92
94
96
98
00
CUSUM
02
04
06
08
06
08
10
12
14
5% Significance
1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 -0.2 -0.4 86
88
90
92
94
96
98
CUSUM of Squares
00
02
04
10
12
14
5% Significance
Fig. 6 Model of CO2 emissions from the oil and gas sector
statistical significance. On the average, we concluded that financial development increased CO2 emissions and reduced environmental quality in Malaysia. However, due to data limitations, we have considered only disaggregated CO2 emissions from manufacturing and construction, CO2 emissions from agricultural sector, CO2 emissions from transportation sector, and CO2 emissions from oil and gas sector while per-capita CO2 emissions are used to capture the overall economic impact of financial development on CO 2 emissions. In addition, CO 2 emissions from manufacturing and construction are captured as a single sector. We could not separately capture the impact of financial development on each sector due largely to data limitation. This may probably be the reason why our result for the impact of financial development on the manufacturing and construction sector is a bit surprising despite its scientific contribution to knowledge. Thus, we recommend further investigation of this phenomenon by future studies.
Policy implication The policy implications of our findings is that financial development has not fully developed to a level that will bring about sustainable economic growth and development; thus, additional financial deepening is required to boost financial development to a stage where it can mitigate CO2 emissions and improve environmental quality. Furthermore, technological transfer and its diffusion can make financial development to be more volatile to be controlled by the policy makers. The implication of this is that financial development will continue to increase sectoral CO2 emissions. However, since trade openness which is less volatile compared to financial development has proved evidence of mitigating sectoral CO2 emissions, we recommend that policy makers should pay attention to quality of international trade rather than the quantity, particularly the trade of major export products that generate revenue to the government.
Δln fdt
Δln tot
Δln kt
– 0.2666 (1.0247) 0.2654 (1.3081) 0.0362 (0.1692) 0.5998*** (3.4184) 0.0551 (0.2597) Δln comt – 0.2031 (0.7592) −0.1065 (−0.4522) −0.1257 (−0.6131) 0.3431** (2.1051) 0.0814 (0.4498) Δln coat – 0.1922 (0.7063) 0.0361 (0.1670) −0.2490 (−0.8722) 0.4256** (2.3465) 0.2291 (1.2174) Δln cott – 0.1744 (0.6419) 0.1158 (0.4503) 0.0215 (0.0981) 0.6073*** (4.2487) 0.1564 (0.8408) Δln coogt – 0.1058 (0.3999) 0.2654 (1.3081) 0.0362 (0.1692) 0.5998*** (3.4184) 0.0814 (0.4498) Δln ect 0.2309 (1.0110) −0.1140 (−0.6200) – 0.9130 (1.0302) 0.1729 (0.5724) 1.6670** (2.4239) Δln ect 0.0420 (0.1338) −0.0654 (−0.3817) – 0.6875 (0.7453) 0.3628 (0.9970) 1.8607*** (2.8476)
Δln yt −0.1816 (−0.5383) – 0.0949 (0.4927) −0.6539 (−1.4552) 0.2614 (1.7066) 0.0243 (0.0698) Δln yt 0.2177 (0.4483) – 0.4119*** (3.0530) 0.0062 (0.0158) −0.0714 (−0.4562) 0.1491 (0.6472)
***, **, and * indicate significance at 1, 5, and 10%, respectively
−1.0119** (−2.2965) −0.4317** (−2.237) – 0.6875 (0.7453) 0.3628 (0.9970) 1.9679** (2.5422) Δln ect 0.3206 (0.9961) −0.1599 (−0.9932) – 0.2192 (0.2885) 0.1031 (0.2956) 1.8607** (2.8476) Δln ect 0.3973 (0.9425) −0.1663 (−0.9753) – 0.7147 (0.8336) 0.2189 (0.5513) 1.7177** (2.5192)
1.1389* (1.9169) – 0.4119*** (3.0530) 0.0062 (0.0158) −0.0714 (−0.4562) 0.3867 (1.2534) Δln yt 0.6736 (1.2598) – −0.1211 (−0.8778) −0.4679 (−1.6999) −0.1375 (−1.1186) 0.1491 (0.6472) Δln yt −0.3313 (−0.4925) – 0.2595** (2.2809) −0.0730 (−0.2585) 0.1219 (0.9313) −0.0528 (−0.2350) Δln fdt 0.1310 (1.5662) 0.0560 (0.8315) 0.0469 (0.4980) – −0.5547** (−2.7138) −0.6610 (−1.4202) Δln fdt 0.1326 (1.1764) 0.0480 (0.7814) 0.0392 (0.5388) – −0.4798* (−2.0370) −0.8403** (−2.1362)
0.0228 (0.1743) 0.0351 (0.6145) 0.0392 (0.5388) – −0.4798* (−2.0370) −1.6238** (−2.8284) Δln fdt 0.2522 (1.6145) 0.0698 (0.8943) 0.0483 (0.4233) – −0.3843* (−1.8311) −0.8403** (−2.1362) Δln fdt 0.2588 (1.1551) 0.0590 (0.6514) 0.2301* (2.0018) – −0.3199 (−1.2855) −0.7903* (−1.8496) Δln tot −0.2483 (−1.5546) 0.1424 (1.1072) 0.1042 (0.5798) −0.2971 (−0.7084) – 0.0273 (0.1603) Δln tot −0.3458 (−1.4768) 0.1290 (1.0110) 0.0534 (0.3539) −0.1116 (−0.2510) – 0.1621 (0.8494)
0.4213 (1.6556) 0.1369 (1.2291) 0.0534 (0.3539) −0.1116 (−0.2510) – −0.1356 (−0.7961) Δln tot 0.2083 (0.8335) 0.1911 (1.5284) 0.2164 (1.1832) 0.0468 (0.2173) – 0.1621 (0.8494) Δln tot 0.3628 (1.1809) 0.1651 (1.3286) 0.1051 (0.6669) −0.0907 (−0.2316) – 0.0774 (0.3410) Δln kt −0.1830* (−2.0043) 0.0406 (0.5523) 0.0287 (0.2797) −0.0639 (−0.2666) −0.1119 (−1.4954) – Δln kt −0.0246 (−0.1722) −0.0105 (−0.1357) −0.1547 (−1.6740) 0.0187 (0.0688) 0.0867 (0.8087) –
0.1757 (0.0778) −0.0307 (−0.4303) −0.1547 (−1.6740) 0.0187 (0.0688) 0.0867 (0.8087) – Δln kt −0.1275 (−0.8610) 0.0325 (0.4397) 0.0174 (0.1607) 0.0172 (0.1603) 0.0319 (0.3305) – Δln kt 0.2804 (1.5103) 0.0492 (0.6556) 0.0200 (0.2101) −0.0467 (−0.1973) −0.0745 (−0.6796) –
ECTt − 1 −0.1916* (−1.7309) 0.1178 (.0968) −0.6388** (−2.3636) 0.0036 (0.5183) −0.2774*** (−4.4168) −0.2671** (−2.2791) ECTt−1 −0.4218* (−2.0039) 0.0886 (1.6575) −0.9855*** (−4.4208) 0.0026 (0.1853) −0.2196*** (−2.9857) −0.2165** (−2.6756)
−0.1509 (−0.9757) 0.0352 (0.2917) −0.9855*** (−4.4208) 0.0026 (0.1853) −0.2196*** (−2.9857) 0.0003 (0.0048) ECTt − 1 −0.1069* (−1.8615) 0.0779 (0.9839) −0.2062 (−1.2037) −0.0200** (−2.1659) −0.2122*** (−3.5151) −0.2165** (−2.6756) ECTt − 1 0.0809 (1.1988) 0.1028 (0.7023) −0.6508*** (−2.8273) −0.0555 (−1.7773) −0.0949 (−1.6848) −0.3245** (−2.1428)
ECTt − 1
Δln ect
Δln copt Δln yt
Long run
Short run
VECM Granger causality tests for per capita and sectoral CO2 emissions
Values in parenthesis () are t statistics
Δln coogt Δln yt Δln ect Δln fdt Δln tot Δln kt
Δln cott Δln yt Δln ect Δln fdt Δln tot Δln kt
Δln coat Δln yt Δln ect Δln fdt Δln tot Δln kt
Δln comt Δln yt Δln ect Δln fdt Δln tot Δln kt
Δln copt Δln yt Δln ect Δln fdt Δln tot Δln kt
Variables
Table 6
Environ Sci Pollut Res
Environ Sci Pollut Res
The results had also shown that financial development can mitigate CO2 emissions in the manufacturing and construction sector. Thus, paying more attention to the abatement technology of this sector will be a useful policy to further reduce emissions from the sector for sustainable economic growth and development. However, CO2 emissions from transportation and the oil and gas sector exceed CO2 mitigations from the manufacturing and construction sector. This further implies that financial development will continue to invoke CO2 emissions, until proper mitigation strategies are put in place in the transportation and oil and gas sectors. Moreover, the positive link between the level of capital and CO2 emissions from the agricultural sector may suggest the possibility of higher CO2 embodiment on imported capital used in the agricultural sector. As such, the government most takes proactive measure to reduce CO2 embodied on imported capital, particularly for agricultural use. Finally, since energy consumption was consistently significant in both the long-run and short-run models, it implies that it plays an important role in CO2 emission as economic growth requires energy as its catalyst. However, the control of energy consumption can be less volatile compared to financial development, because financial development includes technical transfer and diffusion. Thus, we further recommend energy conservation measures by policy makers to enhance efficiency in the utilization of energy. This will reduce the contribution of energy to sectoral CO2 emissions and help in achieving sustainable economic growth and development.
Acknowledgements We would like to thank anonymous referees for their useful comments and suggestions. The usual disclaimer applies. Funding for this project comes from the Putra Grant (Grant No. GPIPB/2014/9440901) provided by Universiti Putra Malaysia, Malaysian (UPM).
References Alam A, Azam M, Abdullah AB, Malik IA (2015) Environmental quality indicators and financial development in Malaysia. unity in diversity 22:8392–8404 Alam MM, Murad MW, Noman AHM, Ozturk I (2016) Relationships among carbon emissions, economic growth, energy consumption and population growth: testing environmental Kuznets curve hypothesis for Brazil, China, India and Indonesia. Ecol Indic 70: 466–479 Acquah HD (2010) Comparison of Akaike information criterion (AIC) and Bayesian information criterion (BIC) in selection of an asymmetric price relationship. Journal of Development and Agricultural Economics 2:1–6 Ang J (2008) The long-run relationship between economic development, pollutant emissions, and energy consumption: evidence from Malaysia. J Policy Model 30:271–278 Aslan A, Apergis N, Topcu M (2014) Banking development and energy consumption: evidence from a panel of middle eastern countries. Energy 72:427–433
BNM (2016) Financial sector blueprint, bank Negara Malaysia: available at http://www.bnm.gov.my Bello AK, Abimbola OM (2010) Does the level of economic growth influence environmental quality in Nigeria? A test of environmental Kuznets curve (EKC) hypothesis. Pakistan Journal of Social Sciences 7:325–329 Brown RL, Durbin J, Evans JM (1975) Techniques for testing the constancy of regression relations over time. J R Stat Soc 37:149–163 Boutabba MA (2014) The impact of financial development, income, energy and trade on carbon emissions: evidence from the Indian economy. Econ Model 40:33–41 Çoban S, Topcu M (2013) The nexus between financial development and energy consumption in the EU: a dynamic panel data analysis. Energy Econ 39:81–88 Dickey DA, Fuller WA (1981) Likelihood ratio statistics for autoregressive time series with a unit root. Econometrica 49:1057–1072 EIA (2015) Energy Information Administriation of United State: available at http://www.eia.gov/ Elliott G, Rothenberg TJ, Stock JH (1996) Efficient tests for an autoregressive unit root. Econometrica 4:813–836 Frankel J, Rose A (2002) An estimate of the effect of common currencies on trade and income. Q J Econ 117:437–466 Goldsmith RW (1969) Financial structure and development. Yale University Press, New Haven Grossman G, Krueger A (1995) Economic growth and the environment. Q J Econ 110:352–377 Ito K (2009) The Russian economy and the oil price: a co-integrated VAR approach. Transit Stud Rev 16(1):220–227 Jalil A, Feridun M (2011) The impact of growth, energy and financial development on the environment in China: a cointegration analysis. Energy Econ 33:284–291 King RG, Levine R (1993) Finance, entrepreneurship, and growth: theory and evidence. J Monet Econ 32:513–542 Kuznets S (1955) Economic growth and income inequality. Am Econ Rev 45:1–28 Lean HH, Smyth R (2010) On the dynamics of aggregate output, electricity consumption and exports in Malaysia: evidence from multivariate Granger causality tests. Appl Energy 87(6):1963–1971 MCkinnon RI (1973) Money and capital in economic development. The Brookings institutions, Washington, D.C. Omri A, Daly S, Rault C, Chaibi A (2015) Financial development, environmental quality, trade and economic growth: what causes what in MENA countries. Energy Econ 48:242–252 Ozturk I, Acaravci A (2013) The long-run and causal analysis of energy, growth, openness and financial development on carbon emissions in Turkey. Energy Econ 36:262–267 Narayan PK (2005) The saving and investment nexus for China: evidence from cointegration tests. Appl Econ 37:1979–1990 Pesaran MH, Shin Y, Smith RJ (2001) Bounds testing approaches to the analysis of level relationships. J Appl Econ 16:289–326 Phillips PCB, Perron P (1988) Testing for a unit root in time series regression. Biometrika 75:335–346 Rafindadi AA, Ozturk I (2016) Effects of financial development, economic growth and trade on electricity consumption : evidence from post-Fukushima Japan. Renew Sust Energ Rev 54:1073–1084 Sbia R, Shahbaz M, Hamdi H (2014) A contribution of foreign direct investment, clean energy, trade openness, carbon emissions and economic growth to energy demand in UAE. Econ Model 36:191–197 Sadorsky P (2010) The impact of financial development on energy consumption in emerging economies. Energy Policy 38:2528–2535 Sadorsky P (2011) Financial development and energy consumption in central and eastern European frontier economies. Energy Policy 39:999–1006 Shahbaz M (2013) Does financial instability increase environmental degradation? Fresh evidence from Pakistan. Ecol Model 33:537–544
Environ Sci Pollut Res Shahbaz M, Lean HH (2012) Does financial development increase energy consumption? The role of industrialization and urbanization in Tunisia. Energy Policy 40:473–479 Shahbaz M, Adebola S, Mahmood H, Arouri M (2013a) Does financial development reduce CO2 emissions in Malaysian economy? A time series analysis. Econ Model 35:145–152 Shahbaz M, Khan S, Tahir MI (2013b) The dynamic links between energy consumption, economic growth, financial development and trade in China: fresh evidence from multivariate framework analysis. Energy Econ 40:8–21 Shahbaz M, Mallick H, Mahalik MK, Loganathan N (2015) Does globalization impede environmental quality in India? Ecol Indic 52:379– 393 Schumpeter JA (1911) The theory of economic development. Harvard University Press, Cambridge, MA Tamazian A, Chousa JP, Vadlamannati KC (2009) Does higher economic and financial development lead to environmental degradation: evidence from BRIC countries. Energy Policy 37:246–253
UNFCCC (2015) United Nations Framework Convention on Climate Change: available at www.unfccc.int/ Wang SS, Zhou DQ, Zhou P, Wang QW (2011) CO2 emissions, energy consumption and economic growth in China: a panel data analysis. Energy Policy 39:4870–4875 Wang S, Zhou C, Li G, Feng K (2016) CO2, economic growth, and energy consumption in China’s provinces: investigating the spatiotemporal and econometric characteristics of China's CO2 emissions. Ecol Indic 69:184–195 WDI (2015). World Bank’s World Development Indicators: available at http://www.worldbank.org Zhang C, Chen X (2011) The impact of global oil price shocks on China’s stock returns: evidence from the ARJI(-ht)-EGARCH model. Energy 36:6627–6633 Zivot E, Andrews DWK (1992) Further evidence on the great crash, the oil-price shock, and the unit-Root hypothesis. J Bus Econ Stat 10: 251–270