Environmental Science and Pollution Research (2022) 29:33106–33116

https://doi.org/10.1007/s11356-021-17895-4

RESEARCH ARTICLE

What causes environmental degradation in Pakistan? Embossing

the role of fossil fuel energy consumption in the view of ecological
footprint
Usman Saleem Yousaf1 · Farhan Ali2 · Babar Aziz3 · Saima Sarwar3

Received: 26 July 2021 / Accepted: 28 November 2021 / Published online: 13 January 2022
© The Author(s), under exclusive licence to Springer-Verlag GmbH Germany, part of Springer Nature 2021

Abstract
For decades, environmental degradation has become a universal challenge, and for sustainable environment quality, a true and
broader proxy is vital. Pakistan is an ecological deficient country in the world, being the sixth-largest economy (population-
wise). This study investigates the prime sources of environmental degradation through ecological footprint in Pakistan. The
yearly time-series data spanning 1972 to 2020 is utilized for a set of regressors as fossil fuel energy consumption, trade open-
ness, arable land, industrial share to GDP, economic growth, and population growth. We use various econometric techniques,
the bounds test, ARDL (short and long run) model, FMOLS, and Granger causality test. Bounds test confirms the existence
of cointegration among variables included in our model. The ARDL estimates suggest that fossil fuel energy consumption,
trade openness, and population growth are the leading factors affecting the environment. Fossil fuel consumption and popula-
tion growth significantly damage the environment in the short and long run. Contrasting to that, trade openness is substantial
to the environment quality. The FMOLS approves the robustness of the cointegrating findings. Moreover, a unidirectional
causal relationship from economic growth to the ecological footprint (GDP → EFP). And also, the ecological footprint of
arable land (EFP → AL) is witnessed. At the same time, bidirectional causality is found between growth rate and fossil energy
consumption (GDP ↔ FEC). Lastly, we recommend some policy options to improve environmental quality in Pakistan.

Keywords Ecological footprint · Environmental degradation · Bounds test · ARDL · FMOLS · Granger causality · Pakistan

Introduction

As per statistics of world poverty and rapid population

growth, reduction in arable land creates intense pressure on
Responsible Editor: Roula Inglesi-Lotz environmental degradation and establishes a cost of almost
365 billion per year, lack of water supply, sanitation, and
* Farhan Ali hygiene condition. Massive migration from rural to urban
[email protected]
creating pressure on big cities and smog issue has increased
Usman Saleem Yousaf too much (that is why air pollution, soil degradation vari-
[email protected]
able, real state sector converting agriculture land into hous-
Babar Aziz ing societies is used) as per statistics of air quality index.
[email protected]
Oxide of carbon is continuously increasing due to excessive
Saima Sarwar use of CNG and diesel (that is why the share of carbon oxide
[email protected]
is used). The population of Pakistan may reach 234 million
1
Department of Economics, Government Associate College, by 2025, and it will increase the use of vehicles from 11
Sharaqpur, Punjab 39460, Pakistan to 35 million, which will become the source to add 25%
2
The Center for Economic Research, Shandong University, more pollution. About 59,000 tons of municipal waste is
Jinan, Shandong 250100, People’s Republic of China generated daily, and 50% of this waste goes into freshwater
3
Department of Economics, Government College University, bodies daily, and only 8% of wastewater is treated, which is
Lahore, Punjab 54000, Pakistan

13
Environmental Science and Pollution Research (2022) 29:33106–33116 33107

negligible. By summing up, we can say that environmental included (Tillaguango et al. 2021; Alvarado et al. 2021).
degradation has become a challenge for the economy. Secondly, the most important variable is arable land (% of
There are many environmental issues in India like air pol- land area), which was not used earlier. The importance is
lution, water pollution, garbage, and pollution of the natural ignored because as more and more land is driven under cul-
environment are all challenges for India. The situation was tivation or the artificial forest is sown in the land, as in the
worse between 1947 and 1995. Pollution is a significant case of Pakistan, the concept of ten billion tree tsunami with
challenge for India as Delhi, Bombay, and many other cities the help of a united nation environment program (UNEP) is
are the most polluted world cities. Environmental issues are introduced. It will bring a contribution to controlling envi-
the leading causes of disease, health issues, and long-term ronmental degradation. The industry contributed 24% in the
impact for India. 1990s, and in 2020, it is approximately 22% of GDP; Paki-
Sustainable environment quality has become a vital stan fossil fuel energy consumption was at the level of 61%
need, especially for developing countries like Pakistan. It in 2015, and it comprises coal, oil, petroleum, and natural
has always played a crucial role in human life. A sustain- gas products, so it was the dire need to include these vari-
able environment can be achieved by controlling ­CO2 emis- ables simultaneously along with population growth which
sion and other nitrogenous waste from vehicles, automo- was ignored earlier (see Table 1).
biles, greenhouse gases, etc. All these play an essential role In this regard, to cater for the variable of carbon emission,
in affecting the ecological footprints and turn to increase which is not a comprehensive indicator of the environment,
environmental degradation. Hence, the decrease in carbon we have introduced the ecological footprints for Pakistan
emissions and other nitrogenous waste is required because to explain environment degradation, as it has already been
industrialization has drastically affected the environment advocated by many researchers (Suki et al. 2020; Gormus
over time in big cities and around the cities. A cause of the and Aydin 2020; Usman and Hammar 2020; Usman et al.
recent wave of smog has badly affected the health of people. 2020a, b, c, 2021; Khalid et al. 2020). The ecological foot-
A rapid increase in population growth is a cause to decrease print is a more recent and advanced indicator for environ-
in area under cultivation, and an uncontrolled and unprec- mental degradation. It consists of five types of land foot-
edented increase in the housing sector put huge pressure on prints: forest land footprint, grazes land footprint, cropland
energy demand. footprint, fishing footprint, built-up footprint, and carbon
Secondly, industries have tried to reduce without con- footprint (Tillaguango et al. 2021; Alvarado et al. 2021).
sidering the environmental cost. World Energy Council This study has investigated the role of fossil fuel, reduc-
has estimated that the need for energy by the world will tion in arable land, high population growth rate, trade open-
increase from 50 to 80% by 2020 and will be more than 1/3 ness, and industrial growth on the ecological footprint in
by 2035. The demand for energy from all-natural resources Pakistan. Based on the unit root test results as suggested by
will increase by 13% from oil, 17% from coal, 48% from gas, Shrestha and Bhatta (2018), the ARDL model is utilized for
66% from nuclear, and 77% from fossil fuels. Fossil fuels are econometric analysis. It provides a definitive solution to the
exhausting day by day, and we cannot reproduce them; they cointegrating relation in the form of a single equation. How-
are non-renewable. That is why it is the source of concern, ever, the ARDL model could also consider a dependent vari-
especially for underdeveloped countries like Pakistan, as it able and a set of regressors adopting no unit root, one unit
is a critical challenge for increasing fuel demand, climate root, or both together. Furthermore, for robustness, FMOLS
change, and economic growth (Ali et al. 2019, 2020). is also applied. The finding estimates suggest that fossil fuel
In the recent literature, primarily, the researchers have energy consumption, trade openness, and population growth
used ­CO2 emission as an indicator of environmental deg- are the leading factors affecting the environment. Fossil fuel
radation for India, China, Iran, Malaysia, and other Asian consumption and population growth significantly damage
countries (Kwakwa et al. 2020; Sharif et al. 2020; Yang et al. the environment in the short and long run. Contrasting to
2020; Zhang et al. 2021; Adedoyin et al. 2020; Adedoyin that, trade openness is substantial to the environment qual-
and Zakari 2020; Ahmad et al. 2016a; Ahmed and Du 2017; ity. The novelty of this paper is that it has taken ecological
Ahmed and Wang 2016; etc.). In this regard, many stud- footprint (combined indicators) as a proxy for environmental
ies have found that it is a minor part of the total environ- degradation of Pakistan (instead the using just C ­ O2 emis-
ment rather than a comprehensive indicator; it also does not sion as an indicator of environmental degradation, which
capture the complete picture of environment degradation has already been criticized by many as mentioned above)
(Ahmed and Wang 2019; Balsalobre-Lorente et al. 2019; along with economic growth, fossil fuel energy consump-
Wang and Dong 2019). It is observed that ­CO2 emission tion, reduction in arable land trade openness, and high popu-
has been extensively used in literature before; basically, lation growth rate.
it does not capture the critical aspect of cropping, water, Pakistan is among the ecologically deficit countries,
and soil, which are the primary indicators that should be which means we are consuming more than we have (Khan

13
33108 Environmental Science and Pollution Research (2022) 29:33106–33116

Table 1  Summary of existing literature reviews on environmental degradation

Authors Duration Region/country Variables Technique Results

Abbas et al. (2020) 198–42017 Pakistan FDI, GDP, POP, IND ARDL Positive
Zhang et al. (2021) 1985–2018 Pakistan HC, NR, ­CO2, GDP, EFP DARDL Positive
Adedoyin et al. (2020) 1985–2017 UK CO2, GDP, EU, uncertainty ARDL Mixed
Adedoyin and Zakari (2020) 1990–2014 BRICS Coal consumption, coal rent, PMG Sig. impact
­CO2
Ahmad et al. (2020) 1984–2016 Selected EFP, NR, TECH, GDP Second-generation panel Positive
cointegration
Ahmed and Wang (2019) 1971–2014 India HC, EFP, GDP Granger causality. EKC found
Ahmad et al. (2020) 1971–2014 China NR, HC, URB, EFP Bayer and Hack cointegration Mixed
Ahmed et al. (2020) 1971–2014 G7 Urb, HC, EFP, GDP Causality Yes
Ahmad et al. (2016) 1971–2014 India CO2, EU, GDP ARDL, cointegration EKC found
Ahmed and Du (2017) 1971–2011 Iran CO2, FDI, GDP, EU Cointegration Positive
Ahmed and Wang (2016) 1992–2011 Croatia CO2, GDP, ARDL, VECM EKC found
Ahmad and Shahbaz et al. 1980–2013 Pakistan DFR, GDP, Urb, EU, TOP, ARDL EKC found
(2014) POP
Ajmi et al. (2013) 1960–2004 G7 EU, ­CO2, GDP Causality Mix
Mamun et al. (2014) 1980–2009 World CO2, GDP, Causality Mix
Begum et al. (2015) 1980–2009 Malaysia GDP, EU, POP, ­CO2 DOLS, causality EKC not found
Baek et al. (2009) 1960–2000 Selected TRD, Y, ­CO2 Cointegration Mixed
Farhani et al. (2014) 1971–2008 Tunisia CO2, GDP, TRD Cointegration, ARDL Yes

FDI, foreign direct investment; GDP, gross domestic product; TRD, trade, CO2, carbon dioxide emission; POP, population; EU, energy used;
URB, urbanization; HC, human capital; EFP, ecological footprint; DFR, deforestation; TOP, trade openness; Y, income; NR, natural resources;
IND, industrial share

and Hussain 2017). To the best of our knowledge, as the degradation in Pakistan for the period of 1984 to 2017
rapidly growing economy of Asia with the sixth-largest observed a negative and significant impact on environmental
population with a high population growth rate, this country degradation. They found that FDI, per capita GDP, indus-
has massive potential in fossil fuel energy. The present study trialization, and the total population positively and signifi-
fastens fossil fuel energy, industrial growth trade openness, cantly affect environmental degradation. Still, rural popula-
population growth rate, and ecological footprint help facili- tion and agricultural land have an insignificant association
tate the existing literature regarding Pakistan, suggesting a with environmental degradation. The role of economic pol-
solution to control the increasingly degrading environment. icy uncertainty in the energy consumption–emission nexus
The division of this study is planned as follows: “A brief states that growth of C
­ O2 emissions, while prolonged use of
review of the literature” explains the review of recent stud- EPU exhibits controversial influence, where ­CO2 emissions
ies in this regard; “Data, model, and methodology” explores continue to rise. Zhang et al. (2021), Adedoyin et al. (2020),
the source of data as well as a description of variables and Adedoyin et al. (2020), and Ahmed et al. (2020a) studied the
model construction; empirical analysis and result in discus- interaction between economic growth, pollutant emissions,
sion present the empirical analysis, results, and discussion; coal rent, and balanced panel data from 1990 to 2014 using
after that, conclusion and highlights of policy implications pooled mean group with dynamic autoregressive distributed
are discussed in “Conclusion and policy implications.” lag. The study found that coal rents have a significant but
negative impact on ­CO2 emissions, unlike coal consump-
tion. This study suggests that stricter environmental-energy-
A brief review of the literature related regulations must reduce ­CO2 emissions for green
growth and sustainable development. The same has been
Several studies have examined the impact of energy and found by Ahmad et al. (2020), Ahmed and Wang (2019),
growth on environmental degradation, but their results are and Ahmad et al. (2020) linkages between natural resources,
inconclusive. A pairwise review of different studies has technological innovations, economic growth, and the result-
been mentioned below to explore the published work of the ing ecological footprint in emerging economies using data
researcher is. Abbas et al. (2020) investigated the impact from 1984 to 2016, employing the second-generation panel
of socioeconomic factors on dimensions of environmental cointegration test. They found a stable, long-run relationship

13
Environmental Science and Pollution Research (2022) 29:33106–33116 33109

between the ecological footprint, natural resources, techno- not find the valid EKC hypothesis for Malaysia. Begum
logical innovations, and economic growth. Furthermore, et al. (2015), Baek et al. (2009), Farhani et al. (2014), and
the quadric term showed a negative impact on the ecologi- Javid and Sharif (2016) investigated the dynamic relation-
cal footprint in the presence of the environment Kuznets ship between carbon dioxide emissions, output (GDP),
curve (EKC) hypothesis, whereas human capital mitigates energy consumption, and trade for Tunisia over the period
environmental deterioration. Interestingly, the link between 1971–2008 using cointegration and the ARDL methodology.
urbanization and human capital helps in alleviating environ- They found a causal long-run relationship between the vari-
mental degradation. Ahmad et al. (2016b), Ahmed and Du ables. There are unidirectional Granger causality relation-
(2017), and Ahmed et al. (2020b) all explored the relation- ships, from GDP, squared GDP, and energy consumption
ships between carbon emissions, energy consumption, and to ­CO2 emissions in the short run. It is a simple study to
economic growth in India using data from 1971 to 2014. investigate the causality among the variables. For groups of
The autoregressive distributed lag model is employed. They countries, Khan et al. (2017) and Nasreen et al. (2017) tried
found that a long-run cointegration relationship exists and to investigate the link between greenhouse gas, financial
that the environmental Kuznets curve is valid. Energy (total development, energy consumption, trade, and urbanization
energy, gas, oil, electricity, and coal) consumption positively in 34 upper–middle-income countries from Asia, Europe,
relates to carbon emissions and the relation between eco- Africa, and America (South and North) by using panel data
nomic growth and carbon emissions. Ahmad et al. (2014) from 2001 to 2014. VECM is applied to find the causality
explored the effectiveness of the environmental Kuznets among the variables and found a causality among all these
curve (EKC) hypothesis for Pakistan using time-series data variables. The GMM is applied to explore a direct associa-
from 1980 to 2013 with deforestation as a dependent vari- tion of per capita greenhouse gas emission with financial
able for environmental degradation, and economic growth, development in Europe and found the relationship of vari-
energy consumption, trade openness, and population as ables changes from region to region and country to country
independent variables. The autoregressive distributed lag (Ozturk and Acaravci 2013; Shahbaz 2013). In this study,
(ARDL) bounds testing approach is applied. They found the we examined the causal relationship between financial devel-
existence of cointegration among the variables. However, in opment, trade, economic growth, energy consumption, and
the long run, the decreasing effect of economic growth on carbon emissions in Turkey for the 1960–2007 period. The
deforestation confirms the EKC hypothesis for deforestation bounds F-test for cointegration is applied to find the coin-
in Pakistan and economic growth and energy consumption tegration among the study variables and a long-run rela-
Granger cause deforestation. Carbon emission and ecologi- tionship between variables. But an increase in foreign trade
cal footprint are more suitable for environmental degradation to GDP ratio results from an increase in per capita carbon
rather than deforestation. Ajmi et al. (2013) investigated the emissions. The financial development variable has no sig-
relationships between energy consumption, carbon dioxide nificant effect on per capita carbon emissions. Shahbaz et al.
emissions, economic growth, and gross domestic product (2014) explored the relationship between trade openness and
in the G7 countries. They used a novel approach to detect energy consumption for 91 low-income countries from 1980
causalities in regulatory policy and technology. They found to 2010 panel; cointegration technique is applied to check
bidirectional causality between GDP and energy consump- the causal relationships among the variables; cointegration
tion for Japan, unidirectional causality among GDP, and is found among trade openness, and energy consumption
energy consumption for Italy and the United Kingdom. This and inverted U shape for high-income countries confirm the
study is a simple judgment of causality among the variables. EKC hypothesis but U shape for middle-income and low-
Mamun et al. (2014) investigated the relationship between income countries. Katircioglu (2013) conducted a simple
­CO2 emissions per capita and economic growth across the study to investigate the causality among imports and energy
world for 1980–2009 period; a very comprehensive study consumption in Singapore using time-series analysis from
found that the different economies rustic, industrial, and 1960 to 2011 and found a long-run relationship between
highly developed countries in few decades reducing ­CO2 import growth and energy consumption and a unidirectional
emission have a significant effect on growth and also found causality the energy growth and imports.
that EKC is a general phenomenon of the world especially Up to the best of our knowledge, existing literature indi-
the transformation from regular to service sector economy cates no such study that has investigated the relationship
that produced more pollution. Dynamic impacts of GDP between fossil fuel, reduction in arable land, industrial
growth, energy consumption, and population growth on growth, and trade openness on the ecological footprint of
­CO2 emission for Malaysia using time-series data analy- Pakistan. The novelty of this paper is that it has taken eco-
sis from 1980 to 2009 found that per capita ­CO2 emission logical footprint (combined indicators) as a proxy for envi-
decreased with economic growth by using dynamic ordi- ronmental degradation of Pakistan (instead the using just
nary least square. Still, the important thing is that they do ­CO2 emission as an indicator of environmental degradation,

13
33110 Environmental Science and Pollution Research (2022) 29:33106–33116

which has already been criticized by many as mentioned form, as Hossain (2011) followed, where β0 represents inter-
above) along with economic growth, fossil fuel energy con- cept term, β1, β2, β3, β4, β5, and β6 indicate the elasticities of
sumption, reduction in arable land, trade openness, and high explanatory variables, and et indicates the stochastic error
population growth rate. This study is a humble attempt to fill term.
the gap in existing literature regarding the case of Pakistan.
EFPt = 𝛽0 + 𝛽1 lnGDPt + 𝛽2 ln FECt + 𝛽3 lnALt
+ 𝛽4 ln TOPt + 𝛽5 ln INDt + 𝛽6 ln POPt + et
Data, model, and methodology (2)
Equation (3) is formulated for the short-run estimates
This study aims to find what caused environmental deg- where the error correction term (ECM) required to be nega-
radation in Pakistan by using time-series data from 1972 tive and significant refers to the speed of adjustment or con-
to 2019. The variable of ecological footprint is defined as vergence of the indicators to the equilibrium.
the sum of aggregate cropland, building up land, grazing ∑n
ΔEFPt = 𝛼 + 𝜆iΔ(EFP)t−1
land, fishing grounds, and forest products (Tillaguango ∑n
i=1
∑n ∑n
et al. 2021; Alvarado et al. 2021). This data is taken from +
i=1
𝛹 iΔln(GDP)t−i +
i=1
𝛿iΔ ln (FEC)t−i +
i=1
𝜂i ln (AL)t−i

Global Footprint Network (GFPN 2020). Data of GDP per

∑n ∑n
+ 𝜈iΔln(TOP)t−i + 𝜔iΔln(IND)t−i
(3)
i=1 i=1

capita is estimated in GDP per capita (constant 2010 US$) +

∑n
𝜋iΔln(POP)t−i + +φ1 (EFP)t−1
from (WDI 2020); fossil fuel consumption is an aggregate
i=1

+ φ2 ln (GDP)t−1 + φ3 ln (FEC)t−1 + φ4 ln (AL)t−1 + φ5 ln (TOP)t−1

of three leading essential indicators like coal per capita
(kWh), oil per capita (kWh), and natural gas per capita + φ6 ln (IND)t−1 + φ7 ln (POP)t−1 + βECMt−1 + et

(kWh) as a complete form; data is taken from Energy The parameters λ, ψ, δ, η, υ, ω, and π symbolize the
Transitions: Global and National Perspectives and BP short–run, and φi indicates the long-term relationship among
Statistical Review of World Energy; arable land is esti- the variables in Eq. (3).
mated as arable land (% of land area). Data is extracted
from (WDI 2020); trade openness is calculated as a ratio Augmented Dicky Fuller test
of exports and imports to GDP (%) and obtained from
American Economic Review (105(10), 3150-3182); the We initiate the unit root testing for multi-purposes. For
industrial share is estimated as the industry (including instance, firstly, for time-series data to ensure all the
construction), value added (% of GDP) extracted from series are stationary (but not non-stationary), secondly,
(WDI 2020), and population growth rate data is also none of the series reflects I(2) integration order, to evade
extracted from WDI (2020), as shown in Table 2. spurious estimates, and thirdly, to choose an appropriate
Our basic econometric model is as follows: econometric technique for analysis. For this resolve, the
EFPt = 𝛽0 + 𝛽1 GDPt + 𝛽2 FECt + 𝛽3 ALt augmented Dicky Fuller (ADF) test is exploited. It is
(1) the most commonly used test. If all the variables follow
+ 𝛽4 TOPt + 𝛽5 INDt + 𝛽6 POPt + et
the I(0) order of integration, then the usual least-squares
Equation (1) provides the long-run estimates through the (OLS) method is better to choose. If show mixed, I(0)
ARDL model of the involved exogenous variables on envi- and I(1, stationary at 1st difference) or only I(1), then
ronmental degradation via ecological footprints of Pakistan. cointegration for long-run relationship is better to apply
To make the results more comprehensive and easier to inter- (Shrestha & Bhatta 2018).
pret, we transformed our model into a natural logarithmic

Table 2  Variables, description, and data sources

Variables Description Data sources

EFP Ecological footprint as a dependent variable Global footprint network (GFPN)

lnGDP Natural log of GDP per capita (constant 2010 US$) WDI
lnFEC Natural log of fossil fuel energy consumption per capita BP Statistical Review of World Energy
lnAL Natural log of arable land (% of land area) WDI
lnTOP Natural log of trade openness ratio (exports/imports) to GDP American Economic Review
lnIND Natural log of industrial share (including construction) to GDP WDI
lnPOP Natural log of the total population WDI

13
Environmental Science and Pollution Research (2022) 29:33106–33116 33111

Table 3  Descriptive statistics both together, unlike other bounds tests to have one unit
Indicators Mean Std. dev. Minimum Maximum root, I(1).

EFP 0.006 0.015 −0.031 0.036 Autoregressive distributed lag model

lnGDP 0.020 0.018 −0.019 0.065
lnFEC 0.028 0.028 −0.031 0.090 Pesaran and Shin (1996) devised a model to find a life-
lnTOP 3.301 0.178 2.488 3.575 long relationship among comprised indicators. As pre-
lnIND −0.002 0.052 −0.231 0.086 viously, unit root and cointegration testing were the
lnAL 3.683 0.023 3.641 3.761 only ways to circumvent counterfeit results. In contrast,
lnPOP 18.63 0.389 17.93 19.23 the ARDL provides an alternative way to avoid forged
outcomes caused by missing variables by adding lag

Table 4  ADF (augmented Indicators ADF test statistic at level (with ADF statistic at 1st diff. (with Conclusion
Dicky Fuller) test estimates for trend and intercept) trend and intercept)
unit root check
H0: The variable has a unit root (i.e., non-stationary)
EFP −2.43 −4.59* Stationary at 1st diff.
lnGDP −1.67 −4.28* Stationary at 1st diff.
lnFEC −1.46 −5.03* Stationary at 1st diff.
lnAL −4.78* Stationary at level
lnTOP −4.88* Stationary at level
lnIND −3.02 −7.68* Stationary at 1st diff.
lnPOP −3.05 −4.27* Stationary at 1st diff.

*Significant at 1% level

Bounds test of dependant variable as one of the regressors and lags

of other regressors. Consequently, it is considered as
Pesaran et al. (2001) advised a test, an extension of
the remedy to spurious estimates (Ghouse et al. 2018).
ARDL modeling testifying to the presence of a lasting
However, ARDL also benefits the dependent variable
connection between two sides of the model: F-statis-
not requiring integration order I(1). Finally, the ARDL
tic to test the null hypothesis ­( H 0 : φ 1 = φ 2 = φ 3 = φ 4 =
model (1, 2, 1, 0, 0, 1) is chosen by opting for Akaike
φ 5 = φ 6 = 0, i.e., no enduring relationship; see Eq. (3))
automated lag choice criteria.
and alternative hypothesis ­( H 1: φ 1 = φ 2 = φ 3 = φ 4 = φ 5 =
φ 6 ≠ 0, i.e., there is enduring relationship). F-statistic
Granger causatives test
beyond tabulated critical upper bound statistically sig-
nifies the cointegration (rejection/acceptance of H ­ 0/H 1
When cointegration is found in the model, there are three
accordingly). When cointegration is verified, the ECM
possibilities: One variable affects the other, another variable
equals nil and the long-term estimates are generated by
affects the first variable in a unifacial manner, or both vari-
standardizing coefficients φ 2 – φ 6 on φ 1. The short-run
ables mutually influence each other. For this purpose, testing
evaluations are arbitrated based on I(1). It tests the coin-
the pairwise Granger causality, a common approach in time-
tegration regardless of integrating order, I(0), I(1), or
series data, referred to predict the values of an economic
variable (i.e., economic growth) by using lagged values of
another variable (i.e., energy consumption). Two variables
are said to be Granger causes each other if the past value
Table 5  Autoregressive distributed lag (ARDL) bounds test estimates of one variable can forecast the present value of another
Test statistic Value Sign. level I(0) I(1) variable. However, a contemporaneous correlation between
the two variables cannot mean the Granger cause can also
F-statistic 16.69 10% 2.75 3.79 be witnessed.
K 6 5% 3.12 4.25 Granger (1969) has exposed a technique named Granger
1% 3.93 5.23 causality test to explore such causatives analytically.

13
33112 Environmental Science and Pollution Research (2022) 29:33106–33116

Table 6  ARDL short and long Indicators ARDL short-run estimates ARDL long-run estimates FMOLS estimates
run and FMOLS estimates
Coefficients t-statistic Coefficients t-statistic Coefficients t-statistic

EFP (−1) −0.300** −2.179

lnGDP −0.105 −0.873 0.087 0.713 −0.086 −1.105
lnGDP (−1) 0.218* 1.809
lnFEC 0.251*** 2.800 0.193*** 2.966 0.283*** 5.413
lnAL −0.076 −0.758 −0.058 −0.767 −0.14* −1.76
lnTOP −0.049** −2.048 −0.038** −2.000 −0.050*** −1.982
lnIND −0.011 0.778 −0.022 1.070 0.292 −1.391
lnIND (−1) 0.078* 1.927
lnPOP 0.15** 2.27 0.11** 2.24 0.07 1.41
ECT (−1) −1.30*** −10.66
Constant 0.432 1.109 0.431* 1.716
Trend 0.01** 2.153 0.01 1.537
Adj. R2 0.322 0.130
F-statistic 3.432***
LM test 2.044
Prob. F(2, 40) 0.146
Breusch Pagan Godfrey test
F-statistic 1.005
Prob. F(5, 42) 0.427

*Significant 10% level

**Significant at 5% level
***Significant at 1% level

Results and discussion

This study adopts the following steps of the econometric

procedure: starting from descriptive statistics (Table 3), (sec-
ond) unit root test by augmented Dicky Fuller test (Table 4),
(third) ARDL bound test (Table 5), (fourth) ARDL short and
long run, error correction term (ECT), and fully modified
ordinary least square (FMOLS) estimation for robustness
(Table 6).
We start with the general tabular display of our dataset’s
essential characteristics (descriptive statistics) (1972–2020).
Unit root testing is considered a prerequisite for time-series
data analysis to rule out spurious outcomes. We employ an
augmented Dicky Fuller test for order of integration of the
included variables with intercept and trend. Supposing for a
maximum of 2 lags, we use the Akaike lags selection crite-
rion. The results in Table 4 reveal that none of the variables
is of I(2), as it is the essential requirement of econometric
analysis, but a mix of I(0) and I(1). In this case, the use of
the conventional regression may provide biased and inef-
ficient results. Therefore, we follow the ARDL approach,
performing better even for a small sample size producing
consistent outcomes. Ecological footprint (EFP), economic
growth (GDP), fossil fuel energy consumption (FEC), indus-
Fig. 1  (a) Cusum graph representation of ARDL model and (b) trial share to GDP (IND), and population growth do not have
cusum squares graph representation of ARDL model

13
Environmental Science and Pollution Research (2022) 29:33106–33116 33113

Table 7  Pairwise Granger causality estimates

Sr. no. Null hypothesis ­(H0) F-statistic (prob.) Decision Concluding remarks

1 lnFEC does not Granger cause EFP 0.38 (0.69) H0 accepted No Granger cause
EFP does not Granger cause lnFEC 1.26 (0.30)
2 lnGDP does not Granger cause EFP 2.48 (0.09)* H0 rejected lnGDP Granger causes EFP
EFP does not Granger Cause lnGDP 1.32 (0.28) H0 accepted No Granger cause
3 lnIND does not Granger cause EFP 1.04 (0.36)
EFP does not Granger cause lnIND 1.78 (0.18)
4 lnTOP does not Granger cause EFP 0.86 (0.43)
EFP does not Granger Cause lnTOP 1.69 (0.20)
5 lnAL does not Granger cause EFP 0.44 (0.65)
EFP does not Granger cause lnAL 2.93 (0.07)* H0 rejected EFP Granger causes lnAL
6 lnGDP does not Granger cause lnFEC 3.18 (0.05)* lnGDP and lnFEC both
lnFEC does not Granger cause lnGDP 5.9 (0.01)** Granger cause each other
7 lnIND does not Granger cause lnFEC 0.17 (0.85) H0 accepted No Granger cause
lnFEC does not Granger cause lnIND 0.12 (0.89)
8 lnTOP does not Granger cause lnFEC 1.68 (0.20)
lnFEC does not Granger cause lnTOP 0.54 (0.59)
9 lnAL does not Granger cause lnFEC 2.14 (0.13)
lnFEC does not Granger cause lnAL 0.53 (0.59)
10 lnIND does not Granger cause lnGDP 1.03 (0.37)
lnGDP does not Granger cause lnIND 0.57 (0.57)
11 lnTOP does not Granger cause lnGDP 1.47 (0.24)
lnGDP does not Granger cause lnTOP 0.64 (0.53)
12 lnAL does not Granger cause lnGDP 1.99 (0.15)
lnGDP does not Granger cause lnAL 1.27 (0.29)
13 lnTOP does not Granger cause lnIND 0.54 (0.59)
lnIND does not Granger cause lnTOP 0.05 (0.96)
14 lnAL does not Granger cause lnIND 0.20 (0.82)
lnIND does not Granger cause lnAL 1.66 (0.20)
15 lnAL does not Granger cause lnTOP 0.34 (0.71)
lnTOP does not Granger cause lnAL 0.73 (0.49)

*Significant 10% level

**Significant at 1% level

I(0) but I(1), whereas arable land (AL) and trade openness findings are partially consistent with the findings of Abbas
have I(0). et al. (2020). The error correction term (ECT) that needs
In Table 5, the F-statistic (16.69) is significant at 1%, to be negative and significant infers that our model con-
5%, and 10% significance level. It validates the cointegration verges to the equilibrium with the speed of 130% in one
among our variables showing higher value than upper bound year. Adjusted R2 refers to the goodness of fit, as it shows
(Pesaran’s tabulated value) against all significance levels. 32% variation in the environment as ecological footprints.
Table 6 depicts all the main findings of our study as F-statistic confirms the significance of the model. The lower
ARDL short and long run and FMOLS estimates. All the part of Table 6 suggests no serial correlation and heterosce-
models reveal similar results. The estimated coefficient of dasticity through the Breusch Godfrey serial correlation LM
fossil fuels energy consumption (FEC) is statistically sig- test and Breusch Pagan Godfrey tests as diagnostics.
nificant and positive at a 1% level in all models, meaning Fig. 1a, b (cumulative sum of recursive residuals (cusum)
that FEC increases EFP in both the short and long run. The and cumulative squares sum of recursive residuals (cusum
statistical significance of trade openness suggests that trade square)) graphs depict and confirm the stability of the model
openness decreases ecological footprint. FMOLS as robust at a 5% level of significance, respectively. The blue line
estimates verify similar results. The population growth also (which needs to be lying between the two red lines in both
significantly increases its environmental footprint. Our the cusum and cusum square graphs) surrounded by the red

13
33114 Environmental Science and Pollution Research (2022) 29:33106–33116

lines verifies that the model is reliable and stable. Table 7 Export-oriented industries should be provided opportuni-
provides Granger causality findings. A unidirectional causal- ties and incentives to enhance the export–import ratio and
ity is found from economic growth to ecological footprints share to GDP.
(lnGDP → EFP) and EFP to arable land (EFP → lnAL). It
describes that lnGDP causes EFP, and EFP causes lnAL.
Two directional causalities are observed between economic Author contribution Usman Saleem Yousaf proposed this paper and
wrote the introduction, literature review, and variables description.
growth and fossil fuels energy consumption (lnGDP ↔ Farhan Ali analyzed the data, wrote, and explained the rest part of the
lnFEC), as both lnGDP and lnFEC cause each other. paper. Babar Aziz and Saima Sarwar provided guidance and approved
the final version.
Cusum and Cusum squares graphical representation
for model stability Data availability The datasets used and analyzed during the current
study are available from the corresponding author on reasonable
request.

Declarations
Conclusion and policy implications
Ethics approval Not applicable
Environmental degradation has been a universal challenge
for decades. For sustainable environment quality, a broader Consent to participate I am free to contact any of the people involved
in the research to seek further clarification and information
and true representative is vital. Pakistan is an ecological
deficient country in the world, being the sixth-largest econ- Consent for publication Not applicable
omy (population-wise). We primarily examine the causes
of environmental degradation in Pakistan using ecological Conflict of interest The authors declare no competing interests.
footprints, a broader indicator for environmental degrada-
tion. Based on the fundamental properties (i.e., stationar-
ity test) of our time-series data spanning (1972–2019), we
chose the autoregressive distributed lag (ARDL) model, a References
better performing method than traditional regression and
cointegration tests. A fully modified least square (FMOLS) Abbas A, Kausar S, Yaseen M, Mayo Z, Zainab M, Mahmood M,
is also employed. Robust analysis indicates that fossil fuel Raza H (2020) Impact assessment of socioeconomic factors on
dimensions of environmental degradation in Pakistan. SN Appl
consumption, trade openness, and population growth rate are Sci. https://​doi.​org/​10.​1007/​s42452-​020-​2231-4
the leading factors affecting environmental quality in Paki- Adedoyin F, Zakari A (2020) Energy consumption, economic expan-
stan. And the findings suggest that fossil fuel consumption sion, and ­CO2 emission in the UK: the role of economic policy
and population growth significantly increase the ecologi- uncertainty. Sci Total Environ. https://d​ oi.o​ rg/1​ 0.1​ 016/j.s​ citot​ env.​
2020.​140014
cal footprint and degrade the environment. However, trade Adedoyin F, Gumede M, Bekun F, Etokakpan M, Balsalobre LD (2020)
openness contributes to reducing environmental footprints Modelling coal rent, economic growth, and C ­ O2 emissions: does
and improving the quality of the environment. Our findings regulatory quality matter in BRICS economies? Sci Total Environ.
are partially consistent with that of Abbas et al. (2020). https://​doi.​org/​10.​1016/j.​scito​tenv.​2019.​136284
Ahmad N, Du L (2017) Effects of energy production and ­CO2 emis-
Furthermore, we also go for the Granger causality test. sions on economic growth in Iran: ARDL approach. Energy
Granger causality test infers that economic growth helps to 123:521e537
cause ecological footprints and arable land in uni-direction. Ahmad A, Zhao Y, Shahbaz M, Bano S, Zhang Z, Wang S, Liu Y
At the same time, economic growth and fossil fuels con- (2016a) Carbon emissions, energy consumption, and economic
growth: an aggregate and disaggregate analysis of the Indian
sumption cause each other in two directions. economy. Energy Policy 96(2016):131–143
As far as policy options are concerned, reducing fossil Ahmad N, Du L, Lu J, Wang J, Li H, Hashmi M (2016b) Modelling
fuels consumption, imported oil, gas, and low-quality coal the ­CO2 emissions and economic growth in Croatia: is there any
consumption for economic growth and development should environmental Kuznets curve? Energy. https://​doi.​org/​10.​1016/j.​
energy.​2016.​12.​106
be a more significant matter of concern for policymakers. Ahmad M, Jiang P, Majeed A, Umar M, Khan Z (2020) Muhammad
The focus should be on indigenous, abundantly available, S (2020) The dynamic impact of natural resources, technologi-
environment-friendly (solar, wind) energy sources. Popula- cal innovations and economic growth on ecological footprint: an
tion growth is required to control on an emergency basis, advanced panel data estimation. Res Policy 69:101817
Ahmed Z, Wang Z (2019) Investigating the impact of human capital on
as it is already top in the region. The government should the ecological footprint in India: an empirical analysis. Environ
take the necessary steps, i.e., awareness, proper guidance, Sci Pollut Res. https://​doi.​org/​10.​1007/​s11356-​019-​05911-7
and primary health education. Trade restrictions should be
relaxed, and free trade agreements should be appreciated.

13
Environmental Science and Pollution Research (2022) 29:33106–33116 33115

Ahmed K, Shahbaz M, Qasim A, Long W (2014) The linkages between Khan M, Yaseen M, Ali Q (2017) The dynamic relationship between
deforestation, energy, and growth for environmental degradation financial development, energy consumption, trade, and green-
in Pakistan. Ecol Indic 49(2014):95–103 house gas: comparison of upper-middle-income countries from
Ahmed Z, Wang Z, Ali S (2019) Investigating the non-linear rela- Asia, Europe, Africa, and America. J Clean Prod. https://​doi.​
tionship between urbanization and ­CO2 emissions: an empirical org/​10.​1016/j.​jclep​ro.​2017.​05.​129
analysis. Air Qual Atmos Health 12:945–953. https://​doi.​org/​10.​ Kwakwa PA, Alhassan H, Adu G (2020) Effect of natural resources
1007/​s11869-​019-​00711-x extraction on energy consumption and carbon dioxide emission
Ahmed Z, Asghar M, Malik M (2020a) Nawaz K (2020) Moving in Ghana. Int J Energy Sect Manag 14:20–39. https://​doi.​org/​
towards a sustainable environment: the dynamic linkage between 10.​1108/​IJESM-​09-​2018-​0003
natural resources, human capital, urbanization, economic growth, Mamun M, Sohag K, Mia M, Uddin G, Ozturk I (2014) Regional
and ecological footprint in China. Res Policy 67:101677 differences in the dynamic linkage between C ­ O 2 emissions,
Ahmed Z, Zafar M, Ali S (2020b) Danish (2020) Linking urbanization, sectoral output, and economic growth. Renew Sust Energ Rev
human capital, and the ecological footprint in G7 countries: an 38(2014):1–11
empirical analysis. Sustain Cities Soc 55:102064 Nasreen S, Anear S, Oxturk I (2017) Financial stability, energy
Ajmi A, Hammoudeh S, Nguyen D, Sato J (2013) On the relation- consumption, and environmental quality: evidence from South
ships between C ­ O2 emissions, energy consumption and income: Asian economies. Renew Sust Energ Rev 67:1105–1122. https://​
the importance of time variation. Energy Econ. https://​doi.​org/​ doi.​org/​10.​1016/j.​rser.​2016.​09.​02113​64.​0321
10.​1016/j.​eneco.​2015.​02.​007 Ozturk I, Acaravci A (2013) The long-run and causal analysis of
Ali F, Khan KA, Raza A (2019) Determining the relationship energy, growth, openness and financial development on carbon
between energy consumption and economic growth in Paki- emissions in Turkey. Energy Econ 36(2013):262–267
stan. Ukrainian. J Ecol 9(3):322–328. https://​doi.​org/​10.​15421/​ Pesaran MH, Shin Y, Smith RJ (2001) Bounds testing approaches to
2019_​98 the analysis of level relationships. J Appl Econ 16(3):289–326
Ali F, Khan KA, Jahan S (2020) Evaluating energy security perfor- Shahbaz M (2013) Does financial instability increase environmen-
mance in Pakistan and India through aggregated energy security tal degradation? Fresh evidence from Pakistan. Econ Model
performance indicators (AESPI ). Eur Online J Nat Soc Sci 33(2013):537–544
9(2):425–442 Shahbaz M, Nasreen S, Ling C, Sbia R (2014) Causality between
Alvarado R, Tillaguango B, Dagar V, Ahmad M, Işık C, Méndez P, trade openness and energy consumption: what causes high, mid-
Toledo E (2021) Ecological footprint, economic complexity, dle and low-income countries? Energy Policy. https://​doi.​org/​
and natural resources rents in Latin America: empirical evi- 10.​1016/j.​enpol.​2014.​03.​029
dence using quantile regressions. J Clean Prod 318:128585. Sharif A, Baris-Tuzemen O, Uzuner G, Ozturk I, Sinha A (2020)
https://​doi.​org/​10.​1016/j.​jclep​ro.​2021.​128585 Revisiting the role of renewable and non-renewable energy
Baek J, Cho Y, Koo W (2009) The environmental consequences of consumption on Turkey’s ecological footprint: evidence from
globalization: a country-specific time-series analysis. Ecol Econ Quantile ARDL approach Sustain Cities. Sustain Cities Soc.
68(2009):2255–2264 https://​doi.​org/​10.​1016/j.​scs.​2020.​102138
Balsalobre-Lorente D, Gokmenoglu KK, Taspinar N, Cantos-Cantos ShresthaMB, Bhatta GR (2018) Selecting appropriate methodologi-
JM (2019) An approach to the pollution haven and pollution cal framework fortime series data analysis. J Financ Data Sci
halo hypotheses in MINT countries. Environ Sci Pollut Res 4(2):71–89. https://​doi.​org/​10.​1016/j.​jfds.​2017.​11.​001
26:23010–23026. https://​doi.​org/​10.​1007/​s11356-​019-​05446-x Suki NM, Sharif A, Afshan S, Mohd SN (2020) Revisiting the envi-
Begum R, Sohag K, Abdullah S, Jaafar M (2015) ­CO2 emissions, ronmental Kuznets curve in Malaysia: the role of globalization
energy consumption, economic and population growth in in a sustainable environment. J Clean Prod. https://​doi.​org/​10.​
Malaysia. Renew Sust Energ Rev 41(2015):594–601 1016/j.​jclep​ro.​2020.​121669
Farhani S, Chaibi A, Rault C (2014) ­C O 2 emissions, output, Tillaguango B, Alvarado R, Dagar V, Murshed M, Pinzón Y, Mén-
energy consumption, and trade in Tunisia. Econ Model dez P (2021) Convergence of the ecological footprint in Latin
38(2014):426–434 America: the role of the productive structure. Environ Sci Pollut
Gormus S, Aydin M (2020) Revisiting the environmental Kuznets Res. https://​doi.​org/​10.​1007/​s11356-​021-​14745-1
curve hypothesis using innovation: new evidence from the top Usman M, Hammar N (2020) Dynamic relationship between techno-
10 innovative economies. Environ Sci Pollut Res 27:27904– logical innovations, financial development, renewable energy,
27913. https://​doi.​org/​10.​1007/​s11356-​020-​09110-7 and ecological footprint: fresh insights based on the STIRPAT
Hossain MS (2011) Panel estimation for C ­ O2 emissions, energy con- model for Asia Pacific Economic Cooperation countries. Envi-
sumption, economic growth, trade openness, and urbanization ron Sci Pollut Res. https://​doi.​org/​10.​1007/​s11356-​020-​11640-z
of newly industrialized countries. Energy Policy 39:6991–6999. Usman M, Kousar R, Makhdum MSA (2020a) The role of financial
https://​doi.​org/​10.​1016/j.​enpol.​2011.​07.​042 development, tourism, and energy utilization in environmental
Javid M, Sharif F (2016) Environmental Kuznets curve and deficit: evidence from 20 highest emitting economies. Envi-
financial development in Pakistan. Renew Sust Energ Rev ron Sci Pollut Res 27:42980–42995. https://​doi.​org/​10.​1007/​
54(2016):406–414 s11356-​020-​10197-1
Katircioglu S (2013) Interactions between energy and imports in Usman M, Kousar R, Yaseen MR, Makhdum MSA (2020b) An
Singapore: empirical evidence from conditional error correction empirical nexus between economic growth, energy utilization,
models. Energy Policy 63:514–520 trade policy, and ecological footprint: a continent-wise compari-
Khalid K, Usman M, Mehdi MA (2020) The determinants of envi- son in upper-middle-income countries. Environ Sci Pollut Res
ronmental quality in the SAARC region: a spatial heterogeneous 27:38995–39018. https://​doi.​org/​10.​1007/​s11356-​020-​09772-3
panel data approach. Environ Sci Pollut Res. https://​doi.​org/​10.​ Usman M, Makhdum MSA, Kousar R (2020c) Does financial inclu-
1007/​s11356-​020-​10896-9 sion, renewable, and non-renewable energy utilization acceler-
Khan UA & Hussain A (2017) Analysis of ecological footprint of ate ecological footprints and economic growth? Fresh evidence
rural-urban households in Islamabad Department of Environ- from15 highest emitting countries. Sustain Cities Soc. https://​
mental Economics. Working Paper No 13, 13 doi.​org/​10.​1016/j.​scs.​2020.​102590

13
33116 Environmental Science and Pollution Research (2022) 29:33106–33116

Usman M, Khalid K, Mehdi MA (2021) What determines the envi- era of globalization? A global perspective. Air Qual Atmos
ronmental deficit in Asia? Embossing the role of renewable and Health. https://​doi.​org/​10.​1007/​s11869-​020-​00885-9
non-renewable energy utilization. Renew Energy. https://​doi.​ Zhang L, Godil D, Bibi M, Khan M, Sarwat S, Anser M (2021) Car-
org/​10.​1016/j.​renene.​2021.​01.​012 ing for the environment: how human capital, natural resources,
Wang J, Dong K (2019) What drives environmental degradation? and economic growth interact with environmental degradation in
Evidence from 14 Sub-Saharan African countries. Sci Total Pakistan? A dynamic ARDL approach. Sci Total Environ. https://​
Environ. https://​doi.​org/​10.​1016/j.​scito​tenv.​2018.​11.​354 doi.​org/​10.​1016/j.​scito​tenv.​2021.​145553
World Bank countries classification (2019) World bank countries
classification explored at, https://​datah​elpde​sk.​world​bank.​org/​ Publisher’s note Springer Nature remains neutral with regard to
knowl​edgeb​ase/​artic​les/​906519-​world-​bank-​count​r y-​and-​lendi​ jurisdictional claims in published maps and institutional affiliations.
nggro​ups. Accessed 12 Nov 2020
World Energy Council, Survey of World Energy Resources, (2013).
London; 2013
Yang B, Jahanger A, Khan MA (2020) Does the inflow of remit-
tances and energy consumption increase ­CO2 emissions in the

13