RESEARCH ARTICLE

Impact of financial development and energy consumption

on environmental degradation in 184 countries using a dynamic

panel model

Sher Khan

& Muhammad Kamran Khan

& Bashir Muhammad

Received: 4 September 2020 /Accepted: 12 October 2020

Springer-Verlag GmbH Germany, part of Springer Nature 2020

Abstract

This study aimed to examine the impact of financial development and energy consumption on CO

emissions by employing

balanced panel data from the period 1990–2017 for 184 countries worldwide. This study applied seemingly unrelated regression

(SUR), two-step difference, and the system GMM model for data analysis. The examined results of SUR, two-step difference,

and system GMM show that energy consumption positively impacts the CO

emissions worldwide; on the other hand, the

examined results of two-step difference and the system GMM model indicate that financial development helps to reduce the CO

emissions and the results of SUR indicate that financial development positively impacts the CO

emissions. The examined results

regarding economic growth indicate a positive effect on the CO

emission and the square of economic growth verifies the validly

of the environmental Kuznets curve in 184 countries. This study has significant implications for policy makers with regard to

environment improvement, clean energy conservation, and an efficient financial system. Further directions are suggested based

on the examined results.

Keywords Financial development

Energy consumption

emissions

Environmental Kuznets curve

Two-step GMM and

two-step system GMM

Introduction

Financial development plays a major role in energy use and

economic development in developing and developed coun-

tries. A large numbers of countries

in the world have

abundant energy resources and have more energy production

than their required level, while others countries are facing a

shortage of energy, and most of these countries depended on

primary energy sources. Similarly, some countries in the

world have strong financial institutions; however, their effects

on the environment are detrimental.

Financial development is one of the most important drivers

of energy consumption and carbon emissions, and they are

believed to have profound impacts on one another.

However, the findings regarding the nexus between financial

development and energy use remain in an early sta ge

(Shahbaz and Lean 2012; Ziaei 2015), while findings regard-

ing the nexus between carbon emissions and financial devel-

opment have provided widely different conclusions.

There are two nexus between energy consumption and fi-

nancial development. The first one is related to the negative

relationship (inhibitory effects) between financial develop-

ment and energy consumption. Through this inhibitory effect,

financial development dis courages energy consumptio n

which in return leads to lower emission of carbon. For exam-

ple, a few studies have validated the inhibitory strand between

For example, most of the developed and developing countries such as the

United States, Canada, Japan, Germany, Australia, France, UK, Italy, China,

Russia, India, South Africa and some countries in the Middle East/North

Africa, Latin America, and central Asia regions. Several countries are also

facing an energy crisis or shortage of energy. Most of those countries are from

Africa and Asia regions, some are from other regions (see World Bank 2018

access to electricity percentage of population)

Responsible Editor: Nicholas Apergis

* Muhammad Kamran Khan

kamrankhanaup@gmail.com

School of Economics, Henan University, Jinming Road,

Kaifeng Postal Code 475004, Henan, China

School of Economics and Management, Northeast Normal

University, Changchun, Jilin, China

Environmental Science and Pollution Research

https://doi.org/10.1007/s11356-020-11239-4

finance and energy consumption for individuals and economic

group countries (Top cu and Pay ne 2017; Ouyang and Li

2018; Farhani and Solarin 2017; Gómez and Rodríguez

2019).

The second strand is related to the positive relationship

called promoting effects that assume financial development

encourages energy consumption by enabling consumers and

enterprises to obtain easier access to finance resources to pur-

chase big-ticket products that lead to an increase in energy

consumption both in production and daily life usage. For in-

stance, the direct relationship was identified in (Perry

Sadorsky 2010;Wangetal.2011;Arourietal.2012;

Shahbaz et al. 2013 a, b; K asman and Duman 2015;

Zaman and Abd-el Moemen 2017;Khanetal.2019;

Muhammad 2019). They detected a positive effect of fi-

nancial developm e nt (different indicat ors ) on ener g y con-

sumption. Similarly, in another study, involving nine

Central and Eastern European countries, Sadorsky

(2011 ) reconfirmed the positive relationship link between

finance and e nergy consumption in sele cted region.

Several other papers confirmed the positive nexus or promot-

ing effects between variables (e.g., Al-mulali and Lee 2013)

Omri and Kahouli 2014;Ziaei2015; Mukhtarov et al. 2018,

2020). This proposes that finance is one o f the important

drivers of energy consumption in developed countries.

Nevertheless, the contrary effect may also hold, which sug-

gests that better finance promotes renewable energy sectors.

This may be true for developed or some developing econo-

mies because their residents value the ambient environment

more and are willing to switch to low energy use or less emit-

ting but financially more costly energy sources. Several stud-

ies have validated the positive hypothesis view (e.g, Dogan

and Seker 2016; Nasreen et al. 2017; Saidi and Mbarek 2017;

Shahbaz et al. 2019).

However, th ere are some studie s that found a non-

significant relationship between financial development and

carbon emissions. For instance, Ozturk and Acaravci (2010)

found a non-significant impact of finance on gas emissions in

Turkey. Furthermore, Dogan and Turkekul (2016) found non-

significant effects of financial development on carbon emis-

sions in the USA, Lu (2018) for 12 Asians countries, Shahbaz

et al. ( 2018) for Kuwait by using financial development

indicators, and Charfeddine and Kahia (2019) for 24 MENA

countries.

This study had three main objectives: (I) to investigate the

impact of financial development on energy consumption and

emissions; (II) to investigate the impact of energy con-

sumption on CO

emissions and financial development; (III)

to investigate the impact of CO

emissions on energy

consumption and financial development in 184 countries

worldwide. Financial development plays a very important

role in environmental deg radation worldwide both in

positive and negative ways. Finance development promotes

and stimulates emissions of carbon by financing different

energy projects. This may be true for developing countries

or less developing countries because their residents value the

ambient environment less and are not willing to shift to low

emission energy resou rces due to high interest rates and

complicated procedures for ob taining a l oan from th e

banking sector. Several studies have validated the positive

hypothe sis view, for example, Boutabba (2014) for India,

Haseeb et al. (2018) for BRICS countries, Tsaurai (2019)for

12 West African countries, Charfeddine and Kahia (2019)for

24 MENA countries, Ganda (2019) for OECED countries.

Despite the importance of the topic only a few papers ex-

amined individual countries and regions (e.g., Sadorsky 2010,

2011; Shahbaz and Lean 2012; Al-mulali and Sab 2012a, b;

Ziaei 2015;Furuoka2015; Destek 2018;Khanetal.2019a;

Khan et al. 2019b; Alvarado et al. 2018;Khanetal.2020a,

Khan et al. 2020b). This study added two types of contribu-

tions to the energy–finance and finance–carbon emissions lit-

erature. First, this study investigate the impact of financial

develop ment on energy consumption and CO 2 emissions;

the impact of energy consumption on CO

emissions and fi-

nancial development, the impact of CO

emissions on energy

consumption and financial development for a panel of 184

countries over the period 1990–2017. Second, it uses bank

based financial development variables such as domestic credit

to the private sector provided by financial institutions as a

proxy for financial development by employing country paired

fixed effect, difference GMM, system GMM and dynamic

seemingly unrelated regression (SUR) techniques to provide

highly robust, comprehensive and reliable results.

Literature review

Financial development, energy consumption, and

economic growth

Sadorsky (2010) and Levine (1999) examined the relationship

between financial development (different indicator) and output

per capita with energy consumption. The examined result indi-

cates that financial development positively impacts energy con-

sumption. The association between output growth and financial

development is multifaceted in both the theoretical and empir-

ical literature (Shahbaz and Lean 2012; Claessens and Laeven

2003; Kaminsky and Schmukler 2003). Financial liberalization

is considered a risk sharing element that might lower the equity

cost and upsurge investment, eventually being the cause of

boosted economic gro wth (Sadorsky 2010 ;Beakertand

Harvey 2000). Shahbaz and Lean (2012) argue that without

studying the existing economic condition, steps taken for finan-

cial liberalization and development could be detrimental to the

economy. The rivalry between national and international banks

within a country makes the financial market elastic and

Environ Sci Pollut Res

produces additional and novel opportunities for investment.

This elasticity increases the association between financial de-

velopment and output growth (Shahbaz and Lean 2012;

Sadorsky 2010, 2011; Mankiw and Scarth 2008).

According to Khan et al. (2019), the relationship between

output growth and energy use is not defensible according to a

sample bivariate method alone. They recommended adding of

the financial indicators, such as credit provided by banks and

financial corporations to private sectors. They further argued

that populations, urban population, trade opens, and labor

force can effects the consumption of energy. They put

forward that financial institutions variables and economic

growth positivel y effects energy consumption in the

aggregate global panel. Conversely, Karanfil (2009)provides

evidence that exchange rate and interest rate affects the energy

use through energy prices. In this regard, Stern (2000)desig-

nated the omission of significant indicators from the equation.

More recently, in two extensive studies, using mean group

(MG) and common correlated effect (CCE) estimator, Teng

et al. (2020) and Destek (2018) investigated the effects of

financial indicators (different type) and economic growth on

energy consumption by including three indexes—stock mar-

ket index, bank indicators index, and bond markets indicators

index. They found a negative and significant relationship be-

tween bond market financial indictors with energy consump-

tion, and a negative but insignificant relationship between

aggregate financial index, banking market index, and stock

market development index with energy consumption. Using

different indicators of financial development (e.g., credit to

private sectors by financial institutions and broad money

(M2)) and by employing the Johansen coin tegration and

vector Error correction model, Kakar (2016)investigatedthe

effects of financial development and economic output growth

on energy consumption in Pakistan and Malaysia. The empir-

ical results indicate a long running relationship between finan-

cial development and energy consumption. They put forward

that a unidirectional relationship exists between energy use

and financial development, while no short run equilibrium

were found between financial development and energy use

in the case of Pakistan. In the case of Malaysia, bidirectional

causality was found to exist between broad money (supply)

and energy use as well as unidirectional causality between

financial development and economic growth.

Furtherm ore , using data from a mixed group of countr ies ,

including 12 East Asia, 13 European and Oceania countries,

Ziaei (2015) investigated the effects of financial indicators

(different types) on energy consumption by employing a

panel VAR model and including the domestic credit to pri-

vate sector and stock traded turnover ratio of financial de-

velopment. The empirical results revealed that the influence

of energy use on the stock market in the case of European

countries is well pronounced but to a higher extent than in

Asian countries; furthermore, capital market shock

positively and significantly influenced the deviation of en-

ergy use in Oceania and Asian countries, while in European

countries, the results revealed that energy use shocks on the

stock market are more pronounced relative to the credit

market. They put forward that countries with a developing

assets market tend to have greater effects on energy con-

sumption with regard to stock vari ables. Similarly, the em-

pirical results in an extensive study conducted by Furuoka

(2015), who used panel data approaches such as panel fully

modified regression and various tests such as IPS test, panel

unit root, cointegration, and causality test, revealed a long

running equilibrium relationship between energy use and

financial development. They suggested a unidirectional re-

lationship exists running from energy use to finance, but the

opposite was not found in the region. In another extensive

paper on the nexus between financial development and en-

ergy, carried out by Khan et al. (2014), found a bidirectional

relationship between the two variables. Furthermore, by

employing the autoregressive distributed lag bounds testing

method, Shahbaz and Lean (2012) found a long running

relationship among economic growth, energy use, and fi-

nancial development by including the financial credit pro-

vided by financial institutions, industrialization a nd urban-

ization in the model for Tunisia. They argued that urbaniza-

tion and industrialization play important roles in increasing

energy consump tion. Furthermore, Al-Mula li and Sab

(2012a) investigated the effects of energy utilization on fi-

nance and output growth for 30 Sub-Saharan African coun-

tries using panel data approaches, a nd they originate that

energy consumption rises with financial development and

output per capita growth; however, with the consequence of

high greenhouse gases emissions (pollutants). Another

study conducted by Al-Mulali and Sab (2012b), involving

19 countries, revealed the same findings for selected coun-

tries. Conversely, Sadorsky (2011) used the dynamic panel

dataestimationin11CentralandEuropeancountriesforthe

period of 1990–2009. He detected the positive effect of

financi al de vel o pm en t (dif fer e nt indi c at ors ) on ener gy con-

sumption in the selected countries. In a similar study, re-

garding 22 emerging countries, Sadorsky (2010)usedthe

dynamic model system GMM method and different

variables of financial development such as stock market

capitalization, the r atio of deposit money bank assets,

stock market turnover ratio, and stoc k market total value

traded. The empirica l results indicate that increases in

financial development increase the energy demand.

Likewise, in the case of Malaysia, Islam et al. (2013)put

forward that an increase in financial development increases

energy consumption. Similarly, for Pakistan, Shahbaz et al.

(2013a) found a significant and direct relationship between

financi al devel o pm e nt and e ner g y consum pti on . They also

found a bidirectional causal relationship between energy

use and financial development.

Environ Sci Pollut Res

Despite the prevailing literature on the given topic, there

remains a lack of empirical studies on the association between

financial development and energy consumption as well as

financial development and CO

emissions. Research on the

relationship between financial development and energy con-

sumption remains in an early stage (Furuoka 2015; Ziaei

2015; Sadorsky 2010). This study will extend the pioneering

studies (Sadorsky 2010, 2011;Al-mulaliandSab2012a, b;

Ziaei 2015) to investigate the aforementioned relationship.

This study is different from aforementioned studies in many

aspects. First, the aforementioned studies were carried out on

30 SAA countries, 9 Central and Eastern Europe countries, 22

emerging countries, 12 East Asia, and 13 European and

Oceania countries, respectively, using different methodolo-

gies, while this study was conducted on 184 countries.

Second, this study incorporated different explanatory and con-

trol variables into the model to investigate the relationship.

Third, compared wi th the study of (Al-Mulali and Sab

2012a, b) on the effects of energy consumption on financial

development, our study used different methodologies such as

the two-step GMM and system GMM, dynamic seeming un-

related regression (DSUR), and paired country fixed effects to

examine the relationship among the variables.

Financial development and carbon emissions

In literature, the relati onship between carbon emissions

and financial development has been reported for different

countries and regions. From the positive view, the nexus

between financial development–carbon emissions is inves-

tigated by many researchers regarding different countries

and panels, by employing different methodologies and

different types of financial development and carbon

emissions indicators to test the relationship. For instance,

Jalil and Feridun (2011) provided evidence that financial

development reduced environmenta l pollution in China.

Similarly, another study on China conducted by Zhang

(2011 ) provides e vidence that an increase in financial de-

velopment increases carbon emissions in China. In addi-

tion, for Sub-Saharan African countries, Al-Mulali and Sab

(2 012a) investiga ted the effects of c arbon emissions o n

financial development and found that energy

consumption boosted by financial development is the

cause of high CO

emissions. Some other studies report a

positive association, such as Amine Boutabba (2014)for

India, Sadorsky (2011) for central and Eastern Europe, A l-

Mulali and Sab (2012b) for 19 countries, Komal and Abbas

(2015 ) for Pakistan, Katircioğlu and Taşpinar (2017)for

Turkey, Haseeb et al. (2018) for BRICS countries,

Salahuddin et a l. (2018) for Kuwait while using FDI,

Cetin et al. (2018) for Turkey, Shahbaz et al. (2018)for

France while using FDI, Tsaurai (2019) for 12 West

African countries, Charfeddine and Kahia (2019) for 24

MENA countries (CO

and FD). More recently, an exten-

sive study conducted by Ganda (2019) found positive and

significant relationships between finance and carbon emis-

sion OECD countries by employing the FDI variable.

From the negative view, in the case of BRICS countries,

Tamazian et al. (2009) found negative effects of financial

development on carbon emissions. Similar results were also

detected by Tamazian et al. (2009) for 21 transitional coun-

tries, Jalil and Feriduin (2011) in the case of China, Shahbaz

et al. (2013a) in the case of Indonesia, Saahuddin et al. (2015)

for GCC countries. Some other studies reported negative as-

sociations, such as Dogan and Seker (2016) for 23 countries,

Abbasi and Riaz (2016) in the case of the emerging economy

of Pakistan, Shahbaz et al. (2016) for Pakistan, Nasreen et al.

(2017) for South Asian countries, Saidi and Mbarek (2017)in

the case of 19 emerging countries, and Katircioglu and

Taspinar (2017) for Turkey. More recently, an extensive study

conducted by Shahbaz et al. (2018) found negative and

significant relationships between fina nce and carbon

emission for France by employing financial development

variables. Furthermore, in another extensive study, Zaidi

et al. (2019) found a negative relationship between financial

development and carbon emissions in 17 APEC countries. In

the same vein, using the same methodology, Zafar et al.

(2019) also confirmed the negative relationship between fi-

nance and energy use in 27 OECD countries.

Regarding the non-significant view, Ozturk and Acaravci

(2010) found a non-significant impact of finance on green-

house gas emissions in Turkey. Furthermore, Dogan and

Turkekul (2016) also found non-significant effects of financial

development on CO

emissions in the USA, Lu (2018)for12

Asians countries, Salahhuddin et al. (2018) for Kuwait using

financial development indicators, and Charfeddine and Kahia

(2019) for 24 MENA countries.

In the aforementioned prevailing literature, the relation-

ship between carbon emissions and financial development

has been reported for different countries. However, these

papers reported a range of conflicting results. For example,

their results differ greatly in terms of signs, magnitudes,

and significance of the estimates of the finance–emissions

nexus. Several researchers found different results, for ex-

ample, some are positive and negative, while others are a

non-significant relationship between financial develop-

ment with carbon emissions (Ozturk and Acaravci 2010;

Dogan and Turkekul 2016;Lu20 18 ). Further mor e, differ-

ent researchers use different types of financial develop-

ment and carbon emissions to test the relationship. For this

reason, they present inconclusive evidence for the different

countries and regions. Therefore, this research revisits the

aforementioned relationship using GMM estimators to in-

vestigate the effects of financial development on CO

emissions in developed, developing, and emerging

countries.

Environ Sci Pollut Res

Methodology

The study investigated the association between consumption

of energy, financial development, and CO

emissions for 184

countries of a global panel by utilizing balanced panel data

from the period 1990–2017. To investi gate the aforemen-

tioned objective, we utilized the fixed effect model, two step

difference and system GMM, and seemingly unrelated regres-

sion (SUR).

Dynamic econometric models

This study used the dynamic panel approach for the examina-

tion of the relationship between financial development, carbon

dioxide emissions and consumption of energy. The Arellano

and Bond (1991) proposed model GMM has been applied

because the simple OLS or fixed effects model may not give

paramount conclusions and may lead to different econometric

quandaries. Difference GMM utilizes the first differences of

regressor and the dependent variables to transform the regres-

sion for abstracting the country concrete effects and make

regressor time invariant. First, the differenced lagged depen-

dent variable was instrumented with precedent levels so that

the autocorrelation quandary can be eliminated. However, in

some cases, the lagged levels of regressor impecunious instru-

ments in the first difference regressor lowered the efficiency.

To improve assessment efficiency, the Sys-GMM estimators

were used for simultaneity partialness and country concrete

effects abstraction. To improve efficiency results, this study

applied the system GMM estimator of Arellano and Bover

(1995)andBlundellandBond(1998). In our study model,

we inclined to transform the model into first difference for

dispensing country categorical effect and employed the lagged

levels of independent variables as instruments for eschewing

simultaneity biasness (Arellano and Bond 1991). However, in

this modeling there was arguing debate that these models

would probably give results with confusing conclusions when

the independent variables are sedulous in nature (Arellano and

Bover 1995). However, Arellano and Bover (1995)suggested

system GMM estimator caliber as well as there being merged

different equipollence. Sys-GMM is suitable for this study for

two reasons: First, GMM is utilized for controlling country

specific effects and endogeneity, as well as omitted variables

partialness on consumption of energy, emission of carbon, and

financial development. Second, it is proposed for those situa-

tions where the duration is short in a study that has quite a

consequential number of individuals (Roodman 2009), but

this reason in the current study is not true. Consequently, we

are utilizing two step GMM and system-GMM. The three-way

association between consumption of energy, financial devel-

opment, and CO

emissions are empirically observed by uti-

lizing three equations as follows.

¼ a

i;t−1

þ a

CO2

þ a

FDPV T

þ a

ln GDPPCðÞ

þ a

þ ε

ð1Þ

CO2

¼ a

CO2

i;t−1

þ a

FDPV T

þ a

FDB

þ a

GDP

þ a

þ ε

ð2Þ

FDPVT

¼ a

FDPVT

i;t−1

þ a

CO2

þ a

CAP

þ a

þ ε

ð3Þ

FDB

¼ a

FDB

i;t−1

þ a

CO2

þ a

CAP

þ a

þ ε

ð4Þ

In the above Eqs. (1), (2), (3), (4), the EC, CO

, FDPVT,

and FDB denotes, consumption of energy per capita, carbon

dioxide per capita, and both types of financial development

(FDPVT & FDB) to the private sector. EC

it-1,

CO2

it-1

FDPVT

it-1

, and FDB

it-1

are the lag terms of all dependent

variables specified in Eqs. (1)to(4)andareutilizedasinde-

pendent variables to quantify the effect of the prior year on the

current year. GDP

is economic growth squares, and CAP is

fixed capital. Xit shows the control variables that may affect

all dependent variables in Eqs. (1)to(4); it includes national

expenditure, trade openness, population, infrastructures, sav-

ing; ⍺

, ⍺

⍺

,and⍺

denote coefficients of the afore-

mentioned explanatory variables, whereas subscripts i (i = 1.. .

N) and t (t = 1990. .. . 2017) index cou ntry and time,

respectively.

Furthermore, this study used the GMM and system GMM

as a base line model to carry out the analysis because it per-

formed better than the static models such as fixed effects and

seemingly unrelated regressions and solved many problem

such as endogeneity and omitted variables bias. Likewise,

the GMM estimator is divided into one- and two-step differ-

ence GMM and system GMM. This study used the two-step

GMM and two-step system GMM because two-steps perform

better in treatmen t autocorrelation and heteroscedasticity

problem compared to one-step GMM. Furthermore, the two-

step estimators employ optimal weighting matrices as well.

Data source and variable description

In this study, financial development (FDPVT & FDB), energy

consumption (EC), and CO

emissions (CO

) have been taken

as dependent and independent variables in Eqs. (1)to(4)to

determine their effect on each other, while (GDP), square of

GDP, and fixed capital stock are independent variables. The

control variables in this study were labor force, merchandise

trade, trade openness, urbanization, gross national expendi-

ture, infrastructure, total population, and saving.

Financial institutions variables such as domestic credit

to the private sector by bank and financial corporations

(FDPVT and FDB) as a percentage of GDP (Khan et al.

Environ Sci Pollut Res

2019;Muhammad2019) has been used as a proxy to mea-

sure financial development. Sadorsky (2011) indicates that

bank based financial development, a variable that segre-

gates credit issue d by private financ ial institutio ns, is a

significant financial development variable. Furthermore,

many variables have been reported in the literature but

the domestic credit share of GDP has been deemed a suit-

able indicator t o measure financial development and has

been extensively used in the literature (Ziaei 2015;

Baloch et al. 2019;ChiuandLee2019;Ganda2019).

Energy consumption has been measured in kg of oil equiv-

alent per capita, real GDP per capita w as measured in con-

stant 2010 international dollars (GDPPC), and CO

emis-

sions in metric tons per capita. Urbanization was measured

by proportion of urban population in the total population.

Trade openness was measured as the percentage of exports

and imports of GDP and capital stock was measured as

fixed capital formation as a share of GDP. Labor force

was measured by total labor force of populations.

Merchandise trade was measured by the sum of merchan-

dise exports and imports divided by t he value of GDP as a

portion of GDP, infrastructure was mobile cellular sub-

scription per 100 people, and gross saving was percentage

of GDP. All of these data are available fro m the World

Bank World Development Indicators.

Results and discussion

This section discusses the results in detail. First, an estimation

of the dynamic techniques for a sample of 184 countries were

analyzed using the Arellano–Bond two-step GMM (Arellano

and Bond 1991) and two-step system GMM proposed by

(Blundell and Bond 1998). We postulated that two step

GMM and two-system GMM are robust methods and stan-

dard error are consistent and equitable. Therefore, the investi-

gation can be conducted based upon the GMM estimations

results. Furthermore, this study estimated the seemingly unre-

lated regressions (SUR) and fixed effects models for compar-

ison purposes to check the results validity. This study only

presented the GMM estimations results because empirically

and theoretically GMM is considered an efficient and consis-

tent estimator (see the previous section for details).

Furthermore, the parameter estimates of base line and SUR

approaches persisted homogeneous in magnitude and sign in

most results. Furthermore, the estimated statistics for AR (1)

were significant at the 1% level in Tables 1, 2, 3 and 4,re-

spectively, whereas those for AR (2) were insignificant at 1%

level, which designated that the results were not affected by

second-order autocorrelation. The estimated results of the

Sargan test were insignificant in all tables from (1) to (4),

suggesting that the alternative hypothesis is rejected while

accepting the null hypothesis of instrumental variables as

being mutually valid, validating that the selection of instru-

mental indicators in the equations were appropriate.

Table 2 exhibits that for Eqs. (1–4), the coefficients of

energy consumption is positive and significant at the 1% level,

indicating that 1% unit increase in energy use results in a

0.001% increase in carbon emissions. The empirical result

implies that energy use has a positive effect on carbon emis-

sions; the fixed effects and seemingly unrelated regression

confirmed the reliability and validity of these findings. The

overall results revealed that a rise in primary energy use

deteriorated environmental quality by consequently

stimulating environmental degradation, perhaps as t hese

countries may depend on primary energy or most of them

use obsolete technologies that consume high amounts of

energy, causing an increase carbon emissions. Previous

studies such as Muhammad ( 2019) found similar findings

for developed countries and the MENA region, Khan et al.

2019 from the global prospective, Saidi and Hammami

(2015a, b) for different regions and global panel, and Zaman

and Abd-el Moemen (2017) in low, middle, and high income

countries. This study suggests that the government of every

country, particularly some developed countries and under-

developed countries, should adopt energy conservation poli-

cies to ensure efficient use of energy resources in their states,

shorten energy consumption, and boost economic growth with

minimal damage to the environment. These goa ls can b e

achieved through increasing the use of clean technology or

increases the renewable energy sources.

The estimated coefficient of financial development

(FDPVT) is −0.006 and −0.019 in models (2) and (3) indicat-

ing that financial development has a negative effect on CO

emissions. The results suggest that financial development dis-

courages CO

emissions in 184 countries; the fixed effects and

seemingly unrelated regression did not confirm the reliability

and validity of these findings. The results of model (3), which

designate that financial development encourages CO

emis-

sions in the aggregate global panel of 184 countries, are con-

trary to those of models (2) and (4). Conversely, regarding the

variable financial development by banks (FDB), model (3)

results denote a positive association with CO

emissions,

while the coefficients of the other three models have negative

signs and are statistically significant, which can be explained

the same as for financial development (FDPVT) above. The

overall results concluded that financial development plays a

crucial role in reducing CO

emissions globally because most

of the countries in our sample may use updated technologies

due to globalization, which is supported by the strong finan-

cial institutions that may help lead to less energy consumption,

resulting in decreased carbon emissions and improved envi-

ronmental quality. Dogan and Seker (2016), Nasreen et al.

(2017), Saidi and Mbarek (2017), and Shahbaz et al. (2019)

reported similar results. Furthermore, the results of this study

suggest that underdeveloped countries in our sample should

Environ Sci Pollut Res

improve their financial systems in order to reduce the emis-

sions to minimum levels.

The results of “economic growth” of model (1), (2), and (3)

demonstrate that the economic growth positively and signifi-

cantly effects carbon emissions. The results of all three models

are further verified by model (4) SUR, which designates that

mass productions of output is responsible for the carbon emis-

sions in the aggregate global panel countries, and these results

are consistent with the results of (Muhammad 2019: Khan

et al. 2019; Chaabouni and Saidi 2017; Shahbaz et al.

2013a, b; Al-Mulali 2011; Zaman and Abd-el Moemen

2017), who put forward that high GDP per capita is associated

with high CO

emissions.

Moreover, the control variable “trade openness” signifi-

cantly effects CO

emissions and the association is positive

in model (2) and (3). These results accentuate that openness of

trade liberalization in countries would increase CO

emissions

due to technology transfer, the SUR model (4) attested the

results validity, and the model (1) results indicated a negative

association with CO

emissions. The economic growth per

capita squares is negative and statistically consequential in

models (1), (2), and (3), proving the inverted U-shaped asso-

ciation of the environmental Kuznets curve in aggregate glob-

al panel countries. This implicatively insinuates that incre-

ments in GDP per capita would first increment carbon emis-

sions and then decrease as economic growth increases, and

model (4) results also verified the results validity of all three

models. Furthermore, the coefficients values of modeld (2)

and (3) indicate that the urban population has a negative and

paramount effect on CO

emissions, while the results of fixed

effects model (1) is opposite.

Table 2 displays that for all regressions (1, 2 and 3, 4), the

coefficients of carbon emissions are positive and significant at

1%. The empirical result implies that CO

emissions have a

positive effect on energy utilization; the fixed effects and

seemingly unrelated regression confirmed the reliability and

Table 1 The influence of consumption of energy and financial development on CO

emissions in an aggregated global panel of 184 countries

Dependent variable: (1) (2) (3) (4)

emission FE Two-step GMM Two-step sys GMM SUR

Consumption of energy 0.001*** 0.001*** 0.001*** 0.002***

(0.001) (0.001) (0.001) (0.001)

Financial develop-Pvt 0.007 −0.006*** −0.019*** 0.019***

(0.005) (0.001) (0.001) (0.004)

Financial develop-B −0.010** −0.002*** 0.008*** −0.041***

(0.001) (0.001) (0.001) (0.005)

Log (GDP per capita) 1.311*** 1.111*** 0.567*** 0.791***

(0.159) (0.002) (0.001) (0.066)

Trade openness −0.005*** 0.001*** 0.000*** 0.006***

(0.001) (0.001) (0.001) (0.001)

Square of GDP −0.001*** −0.001*** −0.001*** −0.001**

(0.001) (0.000) (0.000) (0.001)

Urban pop 0.001** −0.001*** −0.001*** −0.001

(0.001) (0.001) (0.001) (0.001)

CO2

it-1

0.529*** 0.594***

(0.001) (0.001)

Constant −8.292*** −8.568*** −4.428*** −5.668***

(1.310) (0.046) (0.008) (0.481)

Time dummy Yes

AR1 −2.743 (0.006)

AR2 −1.403 (0.161)

Sargan test 139.849 (1.000)

Observations 2694 2457 2607 2694

R-squared 0.334 0.836

Standard errors value is under the coefficients value in parenthesis. ***, **, and * indicate significance at the 1%, 5%, 10% level, respectively. FE, SUR,

2 steps GMM, and 2 steps S.GMM are fixed effects, seemingly unrelated regression, two-step difference GMM, and two-step system GMM, respec-

tively. CO2

it-1

,EC2

it-1

,andFDPVT

it-1

and FDB

it-1

are the first-order lag terms of all the dependent variables in Tables 1, 2, 3,and4, respectively. Sargan

test refers to the over-identification test for the restrictions in GMM estimation

Environ Sci Pollut Res

validity of these findings. Previous studies such as

Muhammad (2019), Saidi and Hammami (2015a, b), Esso

and Keho (2016), Khan et al. (2019), and Hwang and Yoo

(2014) are consistent with our findings, which contend that

emissions increase with the consumption of energy. The

overall results concluded that higher CO

emissions are

associated with higher energy use, implying little

decompiling. The results suggest that our sample countries

should reduce emissions by using the advanced technology

as well as renewable energy consumption.

The coefficients of financial development (FDPVT) are

2.362, 1.006, and 2.769 and financial development (FDB)

are 0. 009 , 0.006, a nd 0.279 for models (1), (2), and (3)s

respectively, and they revealed that financial development

both positively and significantly affects consumption of

energy and the results validity are verified by model (4),

which shows financial development encourages consump-

tion of energy in the aggregated global panel. The estimat-

ed results revealed that a 1% increase in the financial de-

velopment causes an increase in the consumption of energy

in 184 countries. The theoretical analysis indicates that

finance has two kinds of effects on energy use, the

positive—called promoting effects—and the negative—

called inhibitory or discouragi ng effects. Our f indings sup-

port the promoting effects, su ggesting that finance pro-

motes fossil fuel consumption or outdated technology,

and hence more finance leads to more primary energy con-

sumption resulting in deteriorated environmental quality.

The results of this study are consistent wi th prior s tudies,

for example, (Sadorsky 2010, 2011; Shahbaz et al. 2013a,

b;OmriandKahouli2014;Zaiwei2015; Mukhtarov et al.

2018, 2020), while the results contradict the studies of

(Mielnik and Goldemberg 2002;Destek2018).

The estimated coefficient of economic growth displays that

for all regressions (2 and 3, 4), the coefficients of economics

are positive and significant at 1%. The empirical result implies

that economic growth has a positive effect on energy utiliza-

tion; the seemingly unrelated regression confirmed the reli-

ability and validity of these findings. The previous studies

such as (Muhammad 2019:Khanetal.2019; Saidi and

Hammami 2015a , b; Chaabouni and Saidi 2017;Apergis

and Payne 2010; Shahbaz et al. 2013a, b; Zaman and Abd-el

Moemen 2017) reported the same results for different coun-

tries and regions.

Table 2 The effects of CO

emission and financial development on consumption of energy in an aggregated global panel of 184 countries

Dependent variable: (1) (2) (3) (4)

Energy consumption FE Two-step GMM Two-step sys GMM SUR

Carbon dioxide 191.5*** 65.32*** 38.25*** 326.9***

(6.472) (0.023) (0.023) (4.480)

Financial development- Pvt 2.362*** 1.006*** 2.769*** 4.178***

(0.587) (0.004) (0.007) (0.673)

Financial development-B 0.009*** 0.006*** 0.279*** 0.339***

(0.001) (0.001) (0.005) (0.007)

GDP per capita 1.514 107.9*** 128.5*** 223.8***

(64.31) (0.468) (0.485) (27.23)

Urban pop 0.001 −0.001*** −0.001*** −0.001***

(0.001) (0.001) (0.001) (0.001)

Trade openness 0.748 0.648*** −4.695*** −1.770**

(0.853) (0.009) (0.009) (0.894)

Government expenditure 7.802*** −1.012*** 0.300*** −5.862***

(1.151) (0.008) (0.009) (1.762)

it-1

0.741*** 0.855***

(0.001) (0.001)

Constant 1755*** −695.0*** −836.2*** −814.5**

(546.9) (6.076) (4.189) (327.0)

Time dummy Yes

AR1 −2.962(0.003)

AR2 .0174 (0.986)

Sarggan test 124.464 (1.00)

Observations 2643 2395 2543 2643

R-squared 0.337 0.839

Environ Sci Pollut Res

The coefficients of control variables urban population and

the trade openness with respect to the energy consumption are

statistically significant positive in models (2, 3, and 4) and (3

and 4), respectively. With regard to urban population, the

model (1) coefficients are statistically insignificant, while with

regard to trade openness, the model (3) and (4) coefficients

have a statistically positive relationship with energy consump-

tion. The estimated coefficients of models (2) and (4) revealed

that gross national expenditure has negative effects on con-

sumption of energy, while the results of models (1) and (3)

indicated a positive relationship.

Table 3 shows that the estimated coefficients are statistical-

ly significant. As for the results of models (3) and (4), the

relationship between consumption of energy with respect to

financial development (FDPVT) is statistically significant and

negative, indicating that one unit percent increase in consump-

tion of energy can significantly reduce financial development

(FDPVT) by approximatelay −0.003 to −0.005%. Expressed

differently, a decrease in energy consumption stimulates

financial development, perhaps since our most sampled coun-

tries may have adopted energy conservation policies in which

they may have ensured efficient use of energy resources in

their countries. The results of models (1) and (4) did not verify

the results validity and reliability.

We concluded that lower energy consumption spurs finan-

cial development in the aggregated global panel of 184 coun-

tries, which are new contributions to the current literature. Our

results are contrary with the studies of Khan et al. (2019)for

global panel, Pao and Tsai (2010) for BRIC countries, Al-

mulali and Sab (2012) for Sub-Saharan Africa (SSA)

countries, and Halicioglu (2009) in the case of Turkey, who

finds that energy consumption has a crucial role to help in-

crease economic growth and financial development. These

results suggest that financial institutions of every country

should focus on discouraging primary energy consumption

and promote the renewable energy sectors in order to further

lower primary energy consumption and improve the environ-

mental quality.

Table 3 The influence of consumption of energy and CO

emission on financial development by private sectors (FDPVT) credit in an aggregated

global panel of 184 countries

Dependent variable: (1) (2) (3) (4)

Financial development FE Two-step GMM Two-step sys GMM SUR

Consumption of energy 0.003*** −0.005*** −0.003*** 0.003***

(0.001) (0.001) (0.001) (0.001)

Carbon dioxide −0.983*** 0.971*** 0.963*** −1.442***

(0.322) (0.024) (0.034) (0.258)

GDP Per capita 24.40*** 8.992*** 10.56*** 22.85***

(2.523) (0.357) (0.134) (0.787)

Fixed capital 0.152 0.281*** 0.411*** 0.224*

(0.094) (0.011) (0.008) (0.133)

Population 0.001 −0.001*** −0.001*** −0.001***

(0.001) (0.001) (0.001) (0.001)

Government expenditure 0.136** 0.237*** 0.421*** 0.603***

(0.065) (0.005) (0.006) (0.071)

Saving −0.346*** −0.096*** −0.023*** −0.168**

(0.067) (0.004) (0.005) (0.086)

Infrastructures 0.063*** 0.014*** −0.001 0.111***

(0.019) (0.001) (0.001) (0.015)

it-1

0.739*** 0.865***

(0.001) (0.002)

Constant −174.4*** −90.80*** −130.8*** −214.1***

(22.25) (2.785) (1.246) (10.93)

Time dummy Yes

AR1 −3.263 (0.001)

AR2 −1.919 (0.055)

Sargan test 131.944(1.00)

Observations 2394 2150 2295 2394

R-squared 0.325 0.527

Environ Sci Pollut Res

The results of CO

emissions show that the estimated co-

efficients are statistically significant. As for the results of

models (3) and (4), the relationship between CO

emissions

with respect to financial development (FDPVT) is statistically

significant and positive, indicating that one unit percent in-

crease in carbon emission increases financial development,

perhaps since our sampled countries may use primary energy,

which emits a high amount of pollutants, resulting in deterio-

rated environmental quality and is further spurred by financial

development. The results of models (1) and (4) did not verify

the results validity and reliability.

The estimated coefficients of economic growth in (1), (2),

and (3) demonstrate that the economic growth positively and

significantly effects financial development. The reliability and

validity of all models from (1) to (3) is further verified by the

SUR model. These results are consistent with the findings of

(Chaabouni and Saidi 2017; Khan et al. 2019; Zaman and

Abd-el Moemen 2017) who found that high output per capita

is associated with high financial development.

Furthermore, the coefficient values of models (2) and (3)

designate that the fixed capital has a positive and paramount

impact on financial development. The coefficient value of

model (4) additionally verified the validity of models (2 and

3), whereas fixed effects model results are statistically non-

significant. Conversely, the estimated coefficient of the con-

trol variable “population” is negative and highly significant in

models (2), (3), and (4), respectively. This result is further

validated by model (4). Moreover, the coefficients of control

variables government national expenditure and the saving

with respect to the financial development are statistically sig-

nificant positive and negative in all models from (1) to (4)

while the coefficients of infrastructure variable are positive

and statistically significant i n model (1), (2), and (4)

respectively.

Table 4 demonstrates the impact of carbon dioxide

emissions and energy consumption effects on financial de-

velopment in 184 countries by employing the financial

development provided by banks to the private sector

Table 4 The influence of consumption of energy and CO

emissions on financial development by banks (FDB) in an aggregated global panel of 184

countries

Dependent variable: (1) (2) (3) (4)

Financial development FE Two-step GMM Two-step sys GMM SUR

Consumption of energy 0.002*** −0.005*** −0.003*** 0.003***

(0.001) (0.001)) (0.001) (0.001)

Carbon dioxide −0.157 1.100*** 0.680*** −2.404***

(0.313) (0.032) (0.025) (0.238)

Log (GDP per capita) 0.002*** 0.001*** −0.001*** 21.63***

(0.000) (0.001) (0.001) (0.726)

Fixed capital 0.350*** 0.339*** 0.546*** 0.450***

(0.091) (0.007) (0.009) (0.122)

Population 0.001 −0.001*** 0.001*** −0.001***

(0.001) (0.001) (0.001) (0.001)

Government expenditure 0.099 0.202*** 0.340*** 0.399***

(0.063) (0.003) (0.006) (0.0643)

Saving 0.0163 −0.123*** −0.036** −0.105

(0.0180) (0.005) (0.004) (0.076)

Infrastructures −0.272*** 0.014*** 0.002** 0.125***

(0.0652) (0.001) (0.001) (0.014)

it-1

0.752*** 0.889***

(0.001) (0.002)

Constant 2.323 −21.49*** −43.95*** −188.3***

(7.137) (0.443) (0.480) (9.927)

Time dummy Yes

AR1 −2.213((0.001)

AR2 −1.818(0.04)

Sargan test 136.112 138.224

Observations 2400 2150 2300 2400

R-squared 0.299 0.522

Environ Sci Pollut Res