




































American Interdisciplinary Journal of Business 

and Economics 
ISSN: 2837-1909| Impact Factor : 6.71 

Volume. 11, Number 3; July-September, 2024; 

Published By: Scientific and Academic Development Institute (SADI) 

8933 Willis Ave Los Angeles, California 

https://sadijournals.org/index.php/AIJBE| editorial@sadijournals.org 

 

 

1 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

IMPACT OF ENERGY COMMODITIES AND ECONOMIC GROWTH ON 

ENVIRONMENTAL OUTCOMES: INSIGHTS FROM NIGERIA 

 
John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams 

Department of Economics, University of Cape Town, South Africa. 

DOI: https://doi.org/10.5281/zenodo.12799659 

Abstract: This study explored the relationship between energy commodities, economic growth, and Nigeria's 
carbon dioxide (CO ) emissions spanning from 1981 to 2021. Employing the Vector Error Correction 

Mechanism (VECM) process, the study revealed a negative correlation between fossil fuel consumption, 
economic growth, and carbon dioxide emissions. Long-term elasticities indicate that carbon dioxide emissions 

would rise by 24 and 211% if both fossil fuel consumption and economic growth decreased by 1%, 
contradicting the Environmental Kuznets Curve (EKC) theory in Nigeria. Nevertheless, a positive correlation 

was observed between carbon dioxide emissions and the total annual population. As per the error correction 
model (ECM = -2.64441), two years are required for carbon dioxide emissions to return to long-term 

equilibrium, with 26.4% of a shock in the variable resolved within a year. Upon closer examination of the 

impulse response function, it is evident that GDPPC and FFC will exert a negative short- and long-term impact 
on CO  emissions. The study proposes that Nigeria's government should implement a comprehensive strategy 

to bolster investments in renewable energy. This encompasses creating a stable policy environment, 
establishing ambitious targets for renewable energy capacity, providing financial incentives, and introducing 

feed-in tariffs, given that the country's consumption of fossil fuels has not yet reached a point where emissions 

are increasing. 

Keywords: CO  emission, fossil fuel consumption, GDP per capita, VECM. 
 
 

INTRODUCTION   

Energy commodities, encompassing nuclear, chemical, mechanical, thermal, radiation, and electrical energy, 

contribute to economic growth by enhancing productivity and employment. They exist in various forms—liquids, 

solids, and gases—yet their environmental impact is intricate, particularly with the combustion of fossil fuels 

contributing to global warming. The production,  invariably  result  in  significant  environmental 

consequences.  

The consumption of fossil fuels can lead to localized air pollution and climate change (Han et al., 2019). Recent 

research by the WEF (2022) indicates that certain pollutants related to fossil fuels actually have a cooling effect. 

According to Bölük and Mert (2015), natural gas is transportation,    and     consumption     of     energy    almost 

less aggressive than oil, accounting for only half of the CO  emissions compared to coal. On average, the 

combustion of oil (in the form of petroleum) releases approximately 33% less carbon dioxide (CO ) per unit of 

energy produced compared to the combustion of coal.  



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

2 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

In many emerging nations like Nigeria, fossil fuels remain the primary source of energy (Sugiawan and Managi, 
2019). Despite their numerous benefits, such as providing thermal power plants with more precise operational 

control and monitoring (Vincent and Ezaal, 2022), these systems face various challenges that have been 
extensively studied. Robinson et al. (2007) note that Nigeria is not an exception to the escalating environmental 
concerns. Okafor and Joe-Uzoegbu (2010) emphasize the environmental impact of urbanization in Nigeria, where 
rural communities rely on traditional biomass for energy, resulting in greenhouse gas emissions. This imbalance 
contributes to global warming and environmental degradation, with Nigeria experiencing some of the highest CO
 emissions worldwide. 

Figure 1. Carbon dioxide emissions, energy consumption and economic growth. 
 
Elevated carbon dioxide emissions are primarily associated with economic growth, as proposed by the 
Environmental Kuznets (1955) Curve (EKC) hypothesis and supported by studies such as those conducted by 
Han et al. (2018), Acheampong (2018), Abbas et al. (2019), and Esmaeili et al. (2023). The EKC theory posits 
that income contributes to environmental degradation in the early stages of development but diminishes once 
certain income levels are reached. However, the EKC exhibits diverse shapes, suggesting different policy 
implications. The validity of this hypothesis is debated due to variations in methodology, independent variables, 
examined sectors, and modeling. Few empirical studies focus on Nigeria, and no recent dataset has been employed 
to analyze Nigerian CO  emissions, unlike the studies by Chuku (2011), Ogundipe (2013), Alege and Ogundipe 
(2013), and Okon (2021). To our knowledge, as of the time of writing, no paper concentrating on Nigerian CO  
emissions has utilized a more recent dataset.  
Nigeria's economy remains susceptible to the risks associated with climate change due to the country's escalating 
energy consumption and the ensuing CO  pollution. Figure 1 illustrates the shift in energy consumption from 
negative to positive after 1995, resulting in a 7.4% increase in CO  emissions in 1996 compared to the negative 
rates of 24.1% and 6.7% in 1994 and 1995, respectively. Concurrently, due to the rising energy consumption, 
GDP per capita rose from 18.9% in 1995 to 27.03% in 1996, and the overall economy expanded from -0.1% in 
1995 to 4.2% in 1996. The year 2010 marked the most significant change in CO  emissions (46%) and the highest 
GDP per capita (19.3%) during the research period, with an 8% expansion in the economy. Conversely, the lowest 
changes occurred in 1989, with a -0.35% change in energy consumption and a -41.7% change in CO  emissions. 
Although both CO  emissions and GDP deviated from their 1981 values, the growth rate of CO  emissions 
exceeded that of GDP. This suggests limited evidence of absolute decarbonization, indicating that the nation's 
CO  emissions were not proportional to economic growth. Figure 1 demonstrates that from 1981 to 2021, 
Nigeria's economy did not follow a low-carbon trajectory. Policymakers need to comprehend the directional and 
causal relationship between energy commodities, economic growth, and the environment.  



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

3 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

The increase in CO  emissions over the past 70 years is also attributed to the expansion of the human population. 
A growing population results in increased demands for commodities, energy, and food, leading to higher 
emissions from transportation, industry, and agriculture. However, to minimize emissions per person, additional 
measures such as enhancing energy efficiency, transitioning     to      renewable      sources,     and     altering 
consumption habits must be implemented alongside population policies (The Conversation, 2023). Globally, CO
 emissions are distributed unevenly, with high- and upper-middle-income countries, housing slightly less than 

half of the world's population, responsible for over 80% of global CO  emissions. According to Our World in 
Data (2023), the average person in high-income countries emits over ten times as much CO  as the average person 
in lowincome countries.  
This study addresses several gaps in existing literature. Firstly, it focuses on Nigeria from 1981 to 2021, as data 
before the 1980s are incomplete. Secondly, it employs the Vector Error Correction Model (VECM) to explore 
longitudinal cointegration and causal links between the environment, energy commodities, and economic growth, 
providing fresh empirical data for the ongoing discussion about their relationship.  

LITERATURE REVIEW    
The links between energy consumption, CO  emissions, and economic growth are the subject of three broad 
genres of literature. The first discusses whether the relationship between economic growth and CO  emissions is 
consistent with the environmental Kuznets curve (EKC) theory. According to this theory, some pollutants and 
percapita income have an inverted U-shaped connection (Grossman and Krueger, 1995). The Environmental 
Kuznets Curve (EKC) is a relationship between income change and environmental quality, based on Kuznets' 
work. It suggests that rapid industrialization leads to increased pollution and resource use, putting pressure on the 
environment. As income increases, people value the environment more, leading to a decline in pollution levels. 
The EKC hypothesis reveals how environmental quality changes as a country's fortunes change, with an inverted 
U-shaped curve when pollution indicators are plotted against income per capita (Dinda, 2004). Richer consumers 
put more pressure on lawmakers to enact environmental laws and regulations, in addition to being prepared to 
spend more money on eco-friendly goods. The majority of the examples where emissions have decreased while 
income has increased can be attributed to institutional reforms at the local and national levels, including 
environmental laws and market-based incentives aimed at halting environmental degradation.  
In their study, Özokcu and Özdemir (2017) verified the "inverted U shape theory." However, Friedl and Getzner 

(2003) hypothesize a long-term link that takes the form of an N or another shape rather than an inverted U between 

CO  emissions and per-capita income. Although He and Richard (2010) and Agras and Chapman (1999) maintain 

that there is no correlation between CO  emissions and economic growth in their non-existence theory, the 

primary issue with these early investigations on the EKC hypothesis   is   that   they   may    be    biased    by   

missing variables. This happens when one or more independent variables that correlate with one or more of the 

included independent variables and have an impact on the dependent variable are excluded from a statistical 

model (Tong et al., 2020).  
Recently, the Granger causality test, an econometric technique particularly well-suited for time series and panel 
data analyses, was proposed to examine the connection between economic growth and carbon emissions. For 
example, Hossain (2012) discovered that in newly industrialized nations, there was unidirectional short-run 
causality between economic growth and carbon dioxide emissions, as well as between urbanization and economic 
growth. Wang et al. (2016) discovered that economic growth was a Granger cause of CO  emissions in China 
between 1995 and 2012, and Hamit-Haggar (2012) found a unidirectional causality relationship between the 
economy and greenhouse gas emissions in both the short and long runs in their investigation of the Canadian 
industrial sector. In contrast to Omri (2013), who discovered only a one-way Granger causation linking CO  
emissions to economic growth in some European, Central Asian, Latin American, and Caribbean countries, 
Salahuddin and Gow (2014) showed a two-way Granger causal association between the two components.  
Abubakar and Cudjoe (2021) estimate the short-run and long-run impacts of energy consumption on Nigeria's 
environment through total CO  emissions using error correction models and normalized estimations. Results show 
that GDP has a significant long-run tendency to reduce total CO  emissions in Nigeria, confirming the Kuznets 



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

4 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

curve hypothesis for climate. The research also supports the suggestion that environmental destruction increases 
with per capita income during early economic development stages and decreases with an increase after reaching 
a plateau. Rafindadi (2016) modeled the relationships between economic development, energy use, and emissions. 
The model had collinearity issues because the study takes into account CO  emissions as a function of income, 
income squared, and income cubed in addition to other explanatory variables like energy consumption.  
In his paper, Okon (2021) used the auto-regressive distributed lag approach to investigate the applicability of the 
Environmental Kuznets Curve in Nigeria from 1970 to 2018. According to the bounds test, there exists an 
equilibrium relationship over a long period of time between the gross domestic product per capita, the square of 
the GDP per capita, waste, combustible renewable energy, alternative and nuclear energy, adjusted savings, or 
net forest depletion. However, neither short-run nor long-run results are consistent with the Environmental 
Kuznets Curve hypothesis, nor there is no evidence of an inverse U-shaped link between growth and fluorinated 
greenhouse gas emissions in Nigeria. Omisakin (2009) tested the Environmental Kuznets Curve (EKC) 
hypothesis in Nigeria, finding no long-term causal relationship between carbon emissions and income. The 
regression line shows an "U-shaped" pattern, suggesting that income increases carbon emissions before rising 
again. Other studies have shown a long-term relationship between environmental pollution indicators, per capita 
income, institutional variables, and trade. Alege and Ogundipe (2013) found no EKC in Nigeria due to its early 
development stages. Egbetokun et al.'s (2020) study found that SPM and CO have an EKC, while other 
environmental contamination measures did not significantly affect economic development. Table 1. Variables 
measurement and sources of data.  
Source: Researcher’s compilation, 2023.  

METHODOLOGY   
Granger causality, a linear regression model, and cointegration tests will all be used to examine the collected data. 
Using the unit root test, the initial step will be to determine the stationarity and order of integration among the 
study variables. Using cointegration tests, the second phase will look at the long-term relationship between the 
research variables. In the third, regression analysis will be used to examine how the independent study variables 
affect CO  emissions. The Granger causality test will be used in the fourth step to determine the causal relationship 
between the research variables. 

Data and sources    
The Central Bank of Nigeria's (CBN) statistical bulletins and the World Bank 2023 Development Indicators 
(WDI) provided the data used in the study's empirical analysis. Table 1 lists the variables' names, meanings, and 
measurements.  

Model specification 
Based on the empirical literature in energy economics, it makes sense to write the long-term relationship between 
CO  emissions, energy commodities, and economic growth in the form of a linear logarithmic quadratic. This will 
allow us to test the EKC hypothesis in the following way (Equation 1): 

CO2 = 𝑓(FFC, GDPPC, GFCF, POP)                                                              (1)   

In this case, POP stands for population, GFCF for gross fixed capital formation, GDPPC for gross domestic 
product per capita, FFC for fossil fuel energy consumption, and C02 for carbon dioxide emissions per capita. 
Equation 2 can be expressed as follows in the natural log form for C02, GDPPC, GFCF, and POP as well as in 
the econometric model:   

LCO2𝑡 = 𝛽0 + 𝛽1FFC𝑡 + 𝛽2LGDPPC𝑡 + 𝛽3LGFCF𝑡 + 𝛽4LPOP𝑡 + 𝜀𝑡     (2)   

Stationary test   
Since it can affect a series' behavior, stationarity is a significant phenomenon. Regressing X on Y in Equation (3) 
will result in spurious or gibberish regression if X and Y are two non-stationary series (Yule, 1926).  

  

𝑌𝑡 = 𝛽0 + 𝛽1𝑋𝑡 + 𝜀𝑡                                                                                            (3)   



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

5 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

The series is considered non-stationary if it has a unit root. The series is stationary if it doesn't have a unit root. 
The purpose of the stationarity test is to determine if an autoregressive model has a unit root or not. To ascertain 
the sequence of the variables’ integration, the unit root test is helpful. To verify if the provided series is stationary, 
the augmented Dickey-Fuller test (ADF) and PhillipsPerron test (PP) have been employed.  

Co-integration test    
To determine whether there is a co-integration relationship between the two variables' non-stationary series, the 
Johansen-Juselius test is used. We can determine whether there is co-integration between two non-stationary 
series using the Johansen-Juselius co-integration procedure. This indicates that 0<rank (π) = r <n, which is the 
maximum rank that the matrix π can have. In terms of the vector or matrix of adjustment parameters  and the 

vector or matrix of cointegrating vectors ′, π can be expressed as =𝛼𝛽′, where (r) is the number of co-integration 

vectors and (n) is the number of variables. Based on a likelihood ratio test (LR), this procedure uses the Trace test 

and the Maximum Eigenvalues test (λ𝑚𝑎𝑥) to calculate the number of co-integration vectors between variables. 
The definition of a trace test is:   

n   ˆ ) 

trace(r) =−T i=r+1log(1 i   

In contrast to the alternative hypothesis, which states that there are r co-integration vectors, the null hypothesis 
states that there are ≤r co-integration vectors.   

  

The Maximum Eigenvalues test ( max )is defined as:  

max (r,r+1) =−T log(1 r+1) The null hypothesis that the number of co integration vectors = r against the 
alternative those they r+1.  

Granger-causality   
The results of the stationarity and co-integration tests will determine the Granger-Causality test's application in 
the following ways: The following vector auto-regression (VAR) should be estimated in order to perform the 
conventional Granger-Causality test to determine whether the series (FFC), (LGDPPC), (LGFCF), (LPOP), and 
(LCO ) are stationary.  

  

   

    

  

𝑖= = 

  

  

  

  

  

  

  



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

6 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

  

  
  

  

  

In models (Equations 4 and 11), the subscripts denote time periods and  is a white noise error. The constant 

parameter  represents the constant growth rate of FFCt in Equation 4;  LCO2t in Equation 5; LGDPPCt in 

Equation 6; LGFCFt in Equation 8 and POPt in Equation 10. We can obtain eight tests from this analysis: the first 

examines the null hypothesis that the FFCt does not Grangercause LCO2 t and the second test examine the null 

hypothesis that the LCO2t does not Granger-cause FFCt . The third examines the null hypothesis that the 

LGDPPCt does not Granger-cause LCO2t and the fourth test examines the null hypothesis that the LCO2t does 

not Granger-cause LGDPPCt 

. The fifth examines the null hypothesis that the LGFCFt does not Granger-cause LCO2t  and the sixth test 

examines the null hypothesis that the LCO2t does not Granger-cause LGFCFt . The seventh examines the null 

hypothesis that the LPOPt  does not Granger-cause LCO2t and the eight test examines the null hypothesis that the 

LCO2t does not Granger-cause POPt .   

Vector error correction model (VECM)  
The conventional VECM is written compactly as (Equation 12):  

  

𝑘−1 𝑘−1 𝑘−1 

𝛥𝑌 = 𝛼 + ∑ 𝛾𝛥𝑌 𝑡−1 + ∑ 𝜂𝛥𝑋 + ∑ 𝜑𝛥𝑅 + 𝜆𝐸𝐶𝑇 + 𝜇                    (12)  

𝑖=1 𝑖=1 1=𝑖 
where ECTt−1= OLS residual with a lag derived from the long-run cointegrating formula (Equation 13):  

 𝑌 𝑡 = 𝜎 + 𝜂𝑗𝑋𝑡 + 𝜉𝑚𝑅𝑡 + 𝜇𝑡                                                                          (13)  
And express as (Equation 14):  

  

𝐸𝐶𝑇 𝑡−1 = [𝑌𝑡−1 − 𝜂𝑗𝑋𝑡−1 − 𝜉𝑚𝑅𝑡−1]                                                         (14)   

= coefficient of the ECT and the speed at which changes to X and R cause Y to stabilize. The specific VECM 

for this study is as follows (Equations 15 to 19):  

= -2.94); P-value= Probability value, * signifies stationarity.  

Source: Researcher’s calculations from Eviews 10, 2023.   

This model was selected because empirical studies show that when economic variables show individual 
cointegration, or a strong longterm relationship, the VECM performs well for model estimation. Another 
advantage is its capacity to integrate the short-run dynamic and long-run equilibrium models into a single, 
efficient system. It also guarantees accuracy, conceptual rigor, and data integrity (Abubakar and Cudjoe, 2021).  

RESULTS AND FINDING 
The empirical estimation result and a suitable justification are presented in this section to support the study's 
argument. 



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

7 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

  

 

 

 

 

 
Stationarity test   
The study initiated by examining the stationarity of relevant variables using tests detailed in the methodology 
section. The Augmented Dickey-Fuller (ADF) and Phillips and Perron (PP) unit root tests were employed to 
determine if the variables are stationary.  Table  2  presents  compelling evidence that all our variables are 
integrated at order one (that is, I(1)). The data reveals that for each variable, at least one of the tests does not reject 
the null hypothesis of the unit root at levels, indicating non-stationarity. In contrast, it is found that every variable 
in the first difference is stationary. The subsequent step involves confirming whether our variables of interest 
exhibit a long-run relationship since all the variables in our model are integrated of order one, according to at 
least one of the tests employed.   

Endogeneity analysis   
 In order to ascertain whether variables are exogenous or endogenous, Endogeneity analysis is necessary. To 

verify it, apply the paired Granger causality test. Table 3 displays the outcomes of the pairwise Granger causality 

tests. The null hypothesis is rejected at F-statistic critical values of 1, 5, and 10%. First, the study indicates that 

fossil fuel consumption does not Granger-cause CO  emissions in Nigeria based on the pairwise Granger causality 

test.   

*indicates lag order selected by the criterion. Source: Researcher’s calculations from Eviews 10, 2023.  

 

Table 3. Pairwise granger causality test (Lags: 3).  

  

 

Source: Researcher’s calculations from Eviews 9, 2023.  

  

  

  

Table 4. VAR Lag order selection criteria.  

  



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

8 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

  
Secondly, CO  emissions in Nigeria, a proxy for the environment, do not have any feedback from gross domestic 
product per capita, a proxy for economic growth, and instead Granger-cause it. Third, gross fixed capital 
formation is a proxy for investment without feedback and population growth, Granger-cause CO  emissions in  
Nigeria. Fourth, gross fixed capital formation, a stand-in for investment that lacks feedback, Granger-causes gross 
domestic product per capita (a proxy for economic growth).  

 Lag selection  

  
The Vector Error Correction Model (VECM), the Phillips and Perron (PP), the Augmented Dickey-Fuller (ADF), 
and the co-integration tests are sensitive to the number of lags when they are run. Thus, the Schwarz Information 
Criterion (SIC) and Akaike Information Criterion (AIC) were used to determine the actual amount of lags used. 
Table 4 displays the proper lag length for each variable. Table 4 shows that the AIC value at lag 3 is the lowest 
and is likewise lower than the SIC value at lag 1. To estimate Equation (1), the model (Lag 3) is selected as a 
result. Below is the cointegration result.  

Hypothesized   Trace   0.05     Hypothesized   Max-

Eigen   

0.05   Prob.**   

No. of CE(s)   Statistic   Critical Value   Prob.**   No. of CE(s)   Statistic   Critical Value     

None *   115.2717   69.81889   0.0000   None *   46.24334   33.87687   0.0011   

At most 1 *   69.02839   47.85613   0.0002   At most 1 *   31.44267   27.58434   0.0151   

At most 2 *   37.58572   29.79707   0.0052   At most 2 *   22.25237   21.13162   0.0347   

At most 3   15.33335   15.49471   0.0529   At most 3   12.49434   14.26460   0.0934   

At most 4   2.839018   3.841466   0.0920   At most 4   2.839018   3.841466   0.0920   

  

Cointegration test  
The relevant hypothesis is that there is no long-run relationship in order to ascertain whether the variables are 
cointegrated over the long term, such as:   
Hypothesis: λ1 = λ2 = λ3 = λ4 = 0 (no long-term association exists).  



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

9 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

Hypothesis 1: λ1 ≠ λ2 ≠ λ3 ≠ λ4 ≠ 0 (a long-term relationship exists)  
The next step is to run a cointegration test after confirming that all variables are integrated to order one and that 
I(1) cannot be refused. Johansen (1988) and Johansen and Juselius (1990) suggested the multivariate cointegration 
technique, which is used with multivariate time series to find stable long-term links between carbon dioxide 
emissions, GDP per capita, gross fixed capital formation, energy use from fossil fuels, and population. Because 
the cointegration vectors will be used for the subsequent vector error correction model (VECM), it should be 
emphasized that the cointegration test is conducted before the VECM.   
The Johansen test, for instance, rejects the existence of one or fewer cointegrating relations but fails to reject the 
existence of at most two in model 1. This suggests that there are two cointegrating equations in model 1.  
Using trace test statistics, the null hypothesis is rejected because the probability value is less than 5% (P-value = 
0.00) and the trace statistic value is greater than the critical value (115.2717>69.81889). This suggests the 
existence of at least one cointegrating vector. A second evaluation states that since the trace statistic value is 
higher than the essential values, we reject the null hypothesis for asterisks ranked one through two. Every 
associated probability value is less than five percent. In summary, the results indicate that two equations are 
cointegrated to order one (1) at the 0.05 critical level, and there is at least one cointegrating vector.  
Based on the Max-Eigen results, the null hypothesis that there are no cointegrating equations is likewise rejected. 
This is because the probability value is less than 5% (Pvalue = 0.00) and the Max-Eigen Statistic is bigger than 
the important value (46.24334>33.87687). This suggests the existence of at least one cointegrating vector. Based 
on the trace statistical test and the Max-Eigen test, the series are cointegrated to the same order (1), as seen in 
Table 5. This study also takes advantage of a series that has a long-standing relationship. This study will estimate 
the VECM using trace value statistics since it offers a more accurate alternative hypothesis that specifies the 
number of cointegrating vectors. Consequently, one may contend that there is a long-term relationship between 
the variables and that both their short- and long-term dynamics can be found using the VECM model.   

Vector Error Correction Model (VECM) Estimation 
Using the same variables, two distinct vector autoregression models (VAR and VEC) were made to determine 
which one more accurately captured the relationship between Nigeria's economic growth, energy commodities, 
and environmental factors in the real world. Despite not being as structural as the VAR, the VEC model functioned 
well as a limited substitute. Meanwhile, as Table 5 illustrates, the cointegration relationship between the variables 
made the VAR ineffective. The optimum model to apply in this situation is the Vector Error Correction Model 
(VECM). Table 6 displays the outcomes of the vector error correction model (VECM) for the cointegrated series' 
first, second, and third differences. It also includes the error-correction terms from Equation 20. The results are 
displayed in two sections: the first section displays the cointegrating equations, and the second section displays 
the outcomes of the vector error correction models. Table 7 displays the regression's result. The target equations 
D(LCO ), D(FFC), and D(LGFCF) have error correction terms that are negative (-0.26), 0.62), and (-0.29), 
respectively, according to Table 6 above, but D(LGDPPC) and D(LPOP) have positive (0.01) and (0.01) error 
correction terms, respectively. You can see that the VEC model can explain about 69% of the changes in the 
variables that you can depend on, 43% of the changes in the target variable D(LCO ), and 48% of the changes in 
the D(FFC), D(LGDPPC), D(LGFCF), and D(LPOP) equations. This suggests that all five models fit the data.  

must be evaluated. This is due to the fact that within-group designs and two samples are the ideal settings for the 

tstatistic's application. This makes the simultaneous model of t-statistic-based result interpretation insufficient. 

Second, t-statistics are inappropriate for sample sizes greater  than   or   equal   to   30  (n ≥ 30).  The  independent 

variables are not homogeneous; they have variations for both groups (Engle and Granger, 1987). The impact of 

the explanatory factors on Nigeria's carbon dioxide emission is estimated using the simultaneous equation in the 

study using ordinary least squares (OLS).  

The error correction term (ECT) in Table 7 indicates the   

-statistic   0.205142  Prob. F (3,17)  0.8914   
Obs*R-squared   1.292661  Prob. Chi-Square (3)  0.7309   

Source: Researcher’s calculations from Eviews 9, 2023.  



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

10 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

  

  

  

   
Figure 2. Normality test.  

  
 rate of correction of the disequilibrium between the longrun 
and short-run estimations. The value indicates that only 
about 26% of errors generated in the previous period are 
corrected in the current period for the equation. With a p-
value of 0.06 at a 5% confidence level and a standard error 
of 0.133009, this value is significant.  
  

  

Model checking  
  
The null hypothesis (H0) is accepted if the probability is 
more than 5%, indicating no serial correlation in the longrun 
model in Table 8. The normality test shows a kurtosis  

   of 2.81 and skewness of 0.36, indicating normal  
Figure 3. Plot of CUSUM. Distribution. The heteroscedasticity test shows continuous   variance, indicating 
continuous variance. The stability test   shows the Cusum of squares plots do not pass the 5%  critical line, 
indicating the model is stable and suitable for economic study.  

Autocorrelation Residual LM Test   

Test for normality  
A normal model is indicated by residual skewness and kurtosis, and confirmed by JB test (Figure 2).   

 

 

 

 

 



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

11 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

₂ 

Test for stability      
Figure 4. Plot of CUSUMSQ.  The  Figures  3  and  4 show the results of the stability tests Ihugba et al.           

Table 9. Breusch-Pagan-Godfrey tests for Heteroscedasticity.  

  

  

Source: Researcher’s calculations from Eviews 9, 2023.  

Table 10. ARCH tests for Heteroscedasticity.   

F-statistic   0.318784  Prob. F (3,31)  0.8117   
Obs*R-squared   1.047440  Prob. Chi-Square (3)  0.7898   

 Source: Researcher’s calculations from Eviews 9, 2023.  

 Table 11. Wald tests and short-run test.  

  

Dependent variable: DLCO   

Variables  Chi-square test  Prob.  Relationship   

Source: Researcher’s calculations from Eviews 9, 2023.  
(CUSUM and CUSUMSQ). They show that the estimates, variance, residuals, and square residual are stable 
because they are all within the 5% critical boundaries for both the CUSUM and the CUSUMSQ.  Both the 
CUSUM and CUSUMSQ tests accept parameter stability as one of their null assumptions (Tables 9 and 10).  

Simultaneous equation short-run simulation and analysis   
The results of the short-run test are presented in Table 11. The Chi-square joint statistics probability values show 
that, aside from LGFCF, there is a short-run relationship between the explanatory variables and the independent 
variable according to our findings in Table 11. If the p-value of the chi-square test for (FFC) fossil fuel energy 
consumption, (LGDPPC) gross domestic product per capita, and (LPOP) annual total population is less than 0.05, 
the null hypotheses ( 0): β5=0 will be rejected, therefore they cause LCO  in the short run, while (LGFCF) gross 
fixed capital formation as a proxy for investment does not cause LCO  in the short run. The VECM systems 
Granger causality tests results dos not conform to the Pairwise Granger causality tests except for LGDPPC and  
LPOP.  The   next   step   is to conduct exante forecasting involving impulse response and variance decomposition 
tests.   

Impulse response function    
According to Table 12, Nigeria's carbon dioxide emissions forecast are on the positive side, with sporadic 
variations brought on by innovations and shocks. The findings demonstrate that (FFC) fossil fuel energy 
consumption, (LGDPPC) gross domestic product per capita, and  
(LGFCF) gross fixed capital formation will all contribute to explaining the countries increased carbon dioxide 
emissions. A one-standard deviation positive own shock will result in a short-term change from 0.14 to 0.07 and 
a long-term increase at a decreasing rate to 0.037. Second, projections indicate that the energy consumption of 
fossil fuels (FFC) has a short-term negative impact on carbon dioxide emissions (-0.01) and a long-term positive 
impact (0.01). This indicates that FFC has a long-term beneficial effect on carbon dioxide emissions.  

F-statistic  

Obs*R-squared  

0.892884  

19.51504  
Prob. F (20,16)   0.6001   

Prob. Chi-Square (20)   0.4886   

Scaled explained SS  4.024741  Prob. Chi-Square (20)   1.0000   

D(FFC)  11.58  0.00  Short-run causality  

D(LGDPPC)  5.84  0.05  Short-run causality  

D(LGFCF)  5.20  0.07  No Short-run causality  

D(LPOP)  6.77  0.03  Short-run causality  

ALL   25.41   0.00   Short - run causality   



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

12 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

fixed capital formation result in higher carbon dioxide emissions over a five-year period. Simulations show that 
carbon dioxide emissions rise by 0.034 in the short term and 0.032 in the long term for every standard deviation 
increase in LGFCF. Accordingly, the amount invested has a significant impact on carbon dioxide emissions. Fifth, 
projections indicate that, despite both short- and long-term declines, Nigeria's annual population will not be a 
cause for concern when it comes to carbon dioxide emissions.  

Variance decomposition   
To predict the error variance effects for each endogenous variable in a system, variance decomposition is used. 
Any change in time in a simple linear equation corresponds to a change in the dependent variable 
(Wickremasinghe 2011). This study's forecast consists of three time periods: short-term (two years), medium-
term (five years), and long-term (ten years), all based on the Monte Carlo method and Cholesky's ordering. LCO
, FFC, LGDPPC, LGFCF, and LPOP are the outcomes of the variance decomposition forecast for endogenous 

variables.  
In the short run, impulses, innovations, or shocks to carbon dioxide emissions account for 85.1% of fluctuations 
in carbon dioxide emissions. However, carbon dioxide emissions own shock fluctuations continuously increase 
to 140.2% in the long run. Meanwhile, shocks to fossil fuel energy consumption account for 0.1% of fluctuations 
in carbon dioxide emissions in the short run. The fluctuations in carbon dioxide emissions due to fossil fuel energy 
consumption increase in the long run to 1.5%. In the short run, shocks to gross domestic product per  capita  
account for 0.3%, gross fixed capital formation accounts for 1.6%, and annual population accounts for 12.9%. In 
the long run, shocks to gross domestic product per capita increase to 5.9%, gross fixed capital formation increases 
to 9.6%, and the annual population accounts for 92.7%. Shocks to carbon dioxide emissions will account for the 
highest fluctuations in Nigeria’s carbon dioxide emissions, followed by its own shock (Table 13).   

DISCUSSION  
The primary objective of this paper was to investigate the relationship between CO  emissions, energy 
commodities, and economic growth between 1981 and 2021 in Nigeria. The vector error correction model 
(VECM) method was utilized to estimate Equation 2 using annual data. Both the Phillips and Perron (1988) and 
Dickey and Fuller (1981) Augmented Dickey-Fuller (ADF) unit root-testing methods are used to check the time 
series properties of the variables in Equation 1. In Equation 1, all the series seem to have a unit root in their levels, 
but their first differences show that they are stationary.  
The analysis of the data collected has revealed some notable findings that make this study a significant 
contribution to knowledge in the area of carbon dioxide emissions in Nigeria. Firstly, The findings does not 
support the Kuznets Curve (EKC) hypothesis for climate change by demonstrating that economic growth as 
measured by GDP per capita has a significant negative long-run tendency to reduce total CO  emissions in 
Nigeria. The adjustment term (-2.11356) in the second year from the estimated result in Table 7 is statistically 
significant. The findings indicate that when income rises, emissions fall, and vice versa. The World Bank (2018), 
divides Nigeria's income distribution into five quintiles. In 2018, the second 20% of the population held an 
income share of 11.60%, while the third 20% of the population held an income share of 16.20 percent. 22.70 
percent of the population, or the fourth 20%, had an income share. The richest 20% of the population owned 
42.40% of the total income. Nigeria's 2022 Gini coefficient for nations with high levels of wealth inequality was 
35.1%. Nigeria is ranked 100th out of 163 countries worldwide and 11th in West Africa with this score (Harmon, 
2023).  
This can be as a result of changes in consumption patterns, energy efficiency, technology, or income inequality. 
The findings can also be attributed to institutional reforms at the local and national levels, including 
environmental laws and market-based incentives aimed at halting environmental degradation not necessarily 
increase in income. The finding of the negative effect of income on CO  emissions agrees with the findings of 
Friedl and Getzner (2003), He and Richard (2010), Agras and Chapman (1999) and Alege and Ogundipe (2013) 
but disagrees with the findings of Dinda (2004), Özokcu and Özdemir (2017), Abubakar and Cudjoe (2021) and 
Okon (2021). The pairwise Granger causality test also indicates that CO  emissions in Nigeria, a proxy for the 
environment, do not have any feedback from gross domestic product per capita, a proxy for economic growth, 



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

13 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

and instead Granger-cause it. This finding is line with the study of Omisakin (2009), Omri (2013) and Wang et 
al. (2016). Salahuddin and Gow (2014) findings do not agree with our pairwise Granger causality test result.  
Secondly, this study found that, in the long run, fossil fuel consumption had a negative effect on CO  emissions. 
The results show that fossil fuel use and carbon dioxide emissions are inversely correlated; a decrease in fossil 
fuel use corresponds to a decrease in atmospheric carbon dioxide emissions. Even when it is the biggest cause of 
air pollution in developed countries, It impacts positively in the current year but negatively after a year of fossil 
fuel consumption. The findings suggest that the quantity of fossil fuel used for energy production, transport, or 
industrial processes is still low in the country. In 2014, the nation's share of global energy consumption from 
fossil fuels was 18.9%, lower than the 79.4% global average for the same year. The pairwise Granger causality 
test indicates that fossil fuel consumption does not cause CO  emissions in Nigeria. The findings of the negative 
effect of fossil fuel consumption on CO  emissions do not agree with the findings of Abubakar and Cudjoe (2021).  
Thirdly, investments proxied by gross fixed capital formation have the long-term possibility of reducing total CO
 emissions if increased, and finally, the total annual population has a significant positive effect on total CO  

emissions in the long run. This signifies that an increase in total annual population has a possibility of rising 
Nigerian total CO  emissions in the future.  Also, the implication of   the finding is that increasing total annual 
population threatening Nigeria’s effort to meet the global goal for O2 emission reduction as outlined in the 2015 
Paris Climate Change Conference. This finding agrees that the increase in CO  emissions during the past 70 years 
has also been attributed to the expansion of the human population.  

Conclusion 
This paper empirically analyzes the dynamic relationships between CO  emissions, energy commodities, and 
economic growth in Nigeria, using CO  emissions as a proxy for the environment. The long-run relationship, with 
CO  emissions as the dependent variable, is examined to test the short-run and long-run elasticities of CO  
emissions with respect to explanatory variables.  
Contrary to the typical positive correlation between income and emissions, indicating higher emissions per capita 
in wealthier nations, our findings reveal the opposite trend. However, this relationship is not constant, suggesting 
that emissions rise at varying rates based on income levels. In high-income countries, consumptionbased 
emissions tend to exceed production-based emissions, while the reverse is observed in low-income countries. 
This implies that high-income countries are net importers of emissions, while low-income countries are net 
exporters.  
The significant impact of fossil fuel usage on the environment is acknowledged in our study. Surprisingly, our 
findings indicate a significantly negative impact on the environment of Nigeria, as fossil fuel usage influences 
the amount of CO  emissions. This relationship mirrors the social and economic development of the nation. Given 
the high prices and limited supply of fossil energy in Nigeria, insufficient to meet the demands of its over 200 
million inhabitants and expanding economy, there is a pressing need for a substantial increase in energy 
efficiency. Additionally, the creation of new energy consumption structures, particularly those based on 
affordable renewable sources like solar energy, is essential for sustainable growth over time.  
Furthermore, our empirical results challenge the Environmental Kuznets Curve (EKC) theory of climate change, 
which posits that higher income can lower a nation's environmental pollution once a certain threshold is reached. 

RECOMMENDATIONS AND POLICY CONSEQUENCES   
The study's conclusions lead to the recommendation that, to mitigate environmental degradation in Nigeria, 
governments should support initiatives educating and training rural residents to use fewer non-renewable energy 
sources. Despite nonrenewable resources being widely utilized   for     fuel,   industrial   production,   and   
residential energy consumption in Nigeria without currently causing substantial environmental harm, the 
suggestion is for the nation to prioritize energy sources causing minimal environmental damage. Policymakers, 
serious about preventing long-term environmental damage, should enact policies promoting the use of 
environmentally friendly machinery, vehicles, utilities, and equipment.  



John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

14 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

Considering fossil fuel consumption has not yet reached a point where CO  emissions are increasing, Nigerian 
policymakers should focus more on adopting renewable energy sources to reduce emissions from other sources. 
All energy-related investments and developments in the country should prioritize renewable energy and include 
it as a key performance indicator in investment appraisal considerations.  
While concerns about economic growth are valid, this study suggests a bi-directional causal relationship between 
GDPPC and CO  in Nigeria. Policymakers should consider both factors when making decisions, emphasizing the 
need for a comprehensive strategy to boost renewable energy investments. This includes creating a stable policy 
environment, setting ambitious targets for renewable energy capacity, providing financial incentives, and 
implementing feed-in tariffs. The government should invest in research and development, workforce 
development, and public-private partnerships, encouraging private sector participation. Risk mitigation 
instruments should be introduced to reduce perceived risks associated with renewable energy projects. 
Infrastructure development should incorporate grid integration and energy storage. Sustainable finance initiatives, 
such as green bonds, public investment, community engagement, and awareness campaigns, should be 
established.  
International support can be leveraged through climate finance, technology transfer, and streamlined permitting 
processes. Performance monitoring should be instituted to ensure projects meet their objectives, attracting 
investors and accelerating project development.  
The study's general conclusions propose that, to reduce poverty and lower CO  emissions in Nigeria, the 
government should directly deliver goods and services, including free medical services, subsidized housing, and 
education. Implementing negative income taxes to supplement the earnings of the poor and providing a guaranteed 
income are additional measures suggested.  

CONFLICT OF INTERESTS   
The authors have not declared any conflict of interests.  

REFERENCES  

Abbas M, Dalia S, Fausto C, Nanthakumar L, Masoumeh K (2019). Carbon dioxide (CO2) emissions and 

economic growth: A systematic review of two decades of research from 1995 to 2017. The Science of the 

Total Environment. 649(1):31-49.  

Abubakar   SI,  Cudjoe  D   (2021).  The Environmental Impact of energy consumption in Nigeria: Evidence from 

CO  Emissions. Retrieved from https://www.researchsquare.com/article/rs-420727/v1  

Acheampong AO (2018). Economic growth, CO2 emissions and energy consumption: What causes what and 

where? Energy Economics 74:677-692.  

Agras J, Chapman D (1999). A dynamic approach to the environmental Kuznets curve hypothesis. Ecological 

Economics 28(2):267-277.  

Alege PO, Ogundipe AA (2013). Environmental quality and economic growth in Nigeria: A fractional co-

integration analysis. International Journal of Development and Sustainability 2(2):20-29.  

Bölük G, Mert M (2015). The renewable energy, growth and environmental Kuznets curve in Turkey: An ARDL 

approach. Renewable and Sustainable Energy Reviews 52:587-595.  Chuku AC (2011). Economic 

development and environmental quality in Nigeria: Is there an environmental Kuznets curve? MPRA 

Paper No. 30195, posted 18 Apr 2011 11:38 UTC pp. 83-98. 

https://mpra.ub.unimuenchen.de/30195/Cooperation   

https://www.researchsquare.com/article/rs-420727/v1


John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

15 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

Dickey D, Fuller W (1981). Likelihood ratio statistics for autoregressive time series with a unit root. Econometrica 

49:1057-1072.  

Dinda S (2004). Environmental Kuznets curve hypothesis: A survey. Ecological Economics 49(4):431-455.  

Egbetokun S, Osabuohien E, Akinbobola T, Onanuga OT, Gershon O, Okafor V (2020). Environmental pollution, 

economic growth and institutional quality: Exploring the nexus in Nigeria. Management of Environmental 

Quality 31(1)18-31.  

Engle R, Granger W (1987). Co-Integration and error correction: representation, estimation, and testing. 

Econometrica 55:251-276.   

Friedl B, Getzner M (2003). Determinants of CO  emissions in a small open economy. Ecological Economics 

45(1):133-148.  

Grossman GM, Krueger AB (1995). Economic growth and the environment. Quarterly Journal of Economics 

60(2):353-375. http://www.jstor.org/stable/2118443   

Hamit-Haggar M (2012). Greenhouse gas emissions, energy consumption and economic growth: A panel 

cointegration analysis from Canadian industrial sector perspective. Energy Economics 34(1):358-364.  

Han J, Tianyi D, Chao Z, Xuepeng Q (2018). Correlation analysis of CO  emissions, material stocks and economic 

growth nexus: evidence from Chinese provinces. Journal of Cleaner Production 180:395-406.  

Han S, Lin C, Zhang B, Farnoosh A (2019). Projections and recommendations for energy structure and industrial 

structure development in China through 2030: A system dynamics model. Sustainability 11(18):1-20.   

Harmon C (2023). Gini coefficient shows progress in Nigeria’s wealth distribution under democracy. Accessed 

from https://nairametrics.com/2023/03/21/gini-coefficient-shows-progressin-nigerias-wealth-

distribution-under-democracy.  

He J, Richard P (2010). Environmental Kuznets curve for CO  in Canada, Ecological Economics 69(5):1083-

1093.  

Hossain S (2012). An econometric analysis for CO  emissions, energy consumption, economic growth, foreign 

trade and urbanization of Japan. Low Carbon Economy 3(3A):92-105.  

Johansen S, Juselius K (1990). Maximum likelihood estimation and inference on cointegration — with 

applications to the demand for money. Oxford Bulletin of Economics and Statistics 52(2):169-210.  

Johansen S (1988). Statistical analysis of cointegrating vectors. Journal of Economic Dynamics and Control 

12(2/3):231-54. Kuznets S (1955). Economic growth and income inequality. American Economic Review 

45(1):1-28.  

Okafor ECN, Joe-Uzoegbu CKA (2010). Challenges to development of renewable energy for electric power 

sector in Nigeria. International Journal of Academic Research 2(2):211-216.  

http://www.jstor.org/stable/2118443


John D. Thompson, Emma L. Harrison, Michael J. Adams, and Sarah N. Williams (2024) 
 

16 
American Interdisciplinary Journal of Business and Economics | 

https://sadijournals.org/index.php/AIJBE 

 

Okon EO (2021). Nigeria: Is there an environmental Kuznets Curve for fluorinated gases?, Open Economics, 

ISSN 2451-3458, De Gruyter, Warsaw 4(1):57-71.  

Omisakin OA (2009). Economic growth and environmental quality in Nigeria: Does environmental Kuznets curve 

hypothesis hold? Environmental Research Journal 3(1):14-18.  

Omri A (2013). CO  emissions, energy consumption and economic growth nexus in MENA countries: Evidence 

from simultaneous equations models. Energy Economics 40:657-664.  

Özokcu   S,   Özdemir   Ö (2017).    Economic     growth,     energy,    and environmental Kuznets curve. Renewable 

and Sustainable Energy Reviews 72:639-647.  

Phillips PC, Perron P (1988). Testing for a unit root in time series regression. biometrika, 75(2):335-346.  

Rafindadi AA (2016). Revisiting the concept of environmental Kuznets curve in period of energy disaster and 

deteriorating income: Empirical evidence from Japan. Energy Policy 94:274-284.  

Robinson P, Shaheen EI, Shaheen EI (2007). Environmental pollution control. In ‘Practical Advances in 

Petroleum Processing’ pp. 395-447.  

Salahuddin M, Gow J (2014). Economic growth, energy consumption and CO  emissions in Gulf Cooperation 

Council countries. Energy 73:4458.  

Sugiawan Y, Managi S (2019). New evidence of energy-growth nexus from inclusive wealth. Renewable and 

Sustainable Energy Reviews 103:40-48.  

Tong T, Ortiz J, Xu C, Li F (2020). Economic growth, energy consumption, and carbon dioxide emissions in the 

E7 countries: a bootstrap ARDL bound test. Energy, Sustainability and Society 10(1):1-17.  

Vincent M, Ezaal O (2022). Fossil fuel consumption, economic growth, and environmental degradation: is the 

'energy consumption- growth' nexus sustainable in Nigeria? IOSR Journal of Economics and Finance 

13(2):51-68.  

Wang S, Zhou C, Li G, Feng K (2016). CO , economic growth, and energy consumption in China's provinces: 

Investigating the spatiotemporal and econometric characteristics of China's CO  emissions. Ecological 

Indicators 69:184-195.  

Wickremasinghe G (2011). The Sri Lankan stock market and the macro  economy: an empirical investigation. 

Studies in Economics and Finance 28(3):179-195.  

Yule U (1926). Why do we sometimes get non-sense correlations between time-series? A Study in Sampling and 

the Nature of TimeSeries. Journal of the royal statistical society 89(1):1-63. 


