The Impacts of Energy Consumption by Sector and Foreign Direct Investment on CO₂ Emissions in Malaysia

Abstract

The global push for sustainable development has brought environmental issues to the fore. As minimizing environmental deterioration implies reducing energy consumption, this has come under intense discussion among economists and policymakers. This is because it may affect productivity, and thus slow economic growth ensues. Most earlier studies focused on overall energy consumption rather than energy use by sector to see how it can affect CO₂ emissions. However, little research has been conducted on the connection between energy use in particular sectors, such as agriculture and transportation, and CO₂ emissions. Therefore, this study aims to investigate the impacts of energy use in Malaysia’s transportation, industrial, and agricultural sectors and foreign direct investment (FDI) on CO₂ emissions between 1989 and 2019. The ARDL technique is employed, and the results demonstrate that energy consumption in the transportation sector has a greater impact on CO₂ emissions than in the industrial sector. On the other hand, energy use in the agricultural sector reduces CO₂ emissions. These findings may help policymakers formulate the right policies in specific sectors to benefit communities. They will be at low risk of suffering from diseases attributed to environmental degradation.

1. Introduction

Climate change has come to the fore and merits serious global attention as it dramatically affects human health and global food security. It stems from CO₂ emissions, which account for the majority of greenhouse gas emissions [1,2]. China is the world’s greatest CO₂ emitter, accounting for 32.8% of total CO₂ emissions, followed by the US (12.7%), the European Union (EU) countries (7.4%), India (6.9%), and Japan (0.3%) [3]. CO₂ emissions related to energy rose by 60% from 20.5 Gt in 1990 to 33.0 Gt in 2021 [3].

Climate change brought on by burning fossil fuels and forest fires can result in environmental degradation, which serves as a barrier to sustainable development. Climate change and consequent global warming occur when CO₂ is released into the atmosphere. Global warming can negatively affect natural resources, leaving governments and policymakers baffled about how to address the problem. Some suggest that we formulate policies to lower CO₂ emissions, suggesting that production may be cut, which brings up another issue: sluggish economic growth. This is due to the fact that every drop in CO₂ emissions also results in a decrease in energy use. Energy is indispensable for producing output and thus providing job opportunities. This is why some government initiatives fail to protect the environment, which presents a considerable obstacle to sustainable development.

The effect of energy use on CO₂ emissions has been studied in a wide range of earlier research, such as [4,5], etc. Only a few studies looked specifically at energy use by sector. For instance, Shaari et al. [1] concentrated on agriculture, while Solaymani [6] investigated the connection between Malaysia’s transportation industry and CO₂ emissions. This current research will fill the gap by examining the effects of energy consumption in the industrial, transportation, and agriculture sectors on CO₂ emissions. The industrial sector makes up a bigger share of Malaysia’s total GDP than the agricultural and transportation sectors. It uses various non-renewable sources, such as coal and oil, which pose an increasing threat to the environment.

FDI inflows, which can also serve as a determinant of CO₂ emissions, refer to the value of cross-border transactions involving direct investment during a specific period, establishing a close relationship between different economies. FDI inflows play an important role in boosting economic growth in most developing countries, especially Malaysia. This is due to the fact that technology transfer stemming from FDI inflows can help increase productivity. However, this is a topic that sparks debate as to whether technology transfer can reduce or aggravate environmental degradation. Hence, previous studies obtained mixed results on the relationship between FDI inflows and CO₂ emissions. Some ascertained that FDI could be harmful to the environment as higher production ensues, which can release more CO₂ into the air [7,8]. More non-renewable energy sources, such as oil and coal, are required to produce output, culminating in greater environmental degradation. However, some found that FDI can conserve the environment as more FDI inflows bring green technology that can reduce CO₂ emissions [9,10]. The mixed findings pave the way for this study to re-examine the matter in a bid to reach a conclusion on whether FDI inflows can mitigate or aggravate environmental degradation. Therefore, this study aims to investigate the impact of energy consumption by sector and FDI on CO₂ emissions in Malaysia.

Malaysia was chosen for this study as it is still developing and is anticipated to transition from an upper middle-income economy to a high-income economy. Certainly, more energy is needed during the transition to support the local economy. As a result, policies to reduce energy consumption can be detrimental. Figure 1 shows an upward trend of energy consumption in Malaysia for 35 years, from 1984 to 2019. During this period, the highest energy consumption was recorded in 2019 at 66,484 Ktoe, and it decreased by 2.67% in 2009 due to an economic recession, suggesting that higher economic growth reflects higher energy consumption. As economic growth decreases, less energy is needed. Thus, it is expected that CO₂ emissions will also decline; hence, Figure 2 shows a drop in CO₂ emissions in the same year. Malaysia uses coal to produce power and energy, emitting a significant amount of CO₂ into the atmosphere. Transportation accounted for the largest share of total energy consumption in 2018 (36.4%), followed by industry (29.5%), non-energy usage (20.5%), residential and commercial (12%), and agriculture (1.6%).

Net FDI inflows are shown in Figure 3 from 1984 to 2018 as a share of Malaysia’s GDP. Although FDI plays a crucial role in fostering economic growth, creating job opportunities, and bringing new technology, the trend in FDI inflows remained uncertain over the period. Malaysia experienced the biggest FDI inflows, at 8.76%, in 1992, and the lowest, at 0.06%, in 2009, during the American financial crisis. Malaysia’s economy saw a marked decrease in economic growth and was plunged into a recession. If we compare the trend in FDI inflows and CO₂ emissions, it is difficult to infer whether FDI might reduce or decrease environmental degradation. However, during Malaysia’s economic recession in 2009 and 2020, FDI inflows dropped markedly. At the same time, we can observe a drop in CO₂ emissions, implying a positive relationship between FDI inflows and CO₂ emissions. Technology brought into the country due to FDI inflows, which emitted CO₂, was affected during the two years.

Figure 1. Total Energy Consumption from 1984 to 2018. Source: Malaysia Energy Commission. Malaysia Energy Statistics Handbook [11].

Figure 1. Total Energy Consumption from 1984 to 2018. Source: Malaysia Energy Commission. Malaysia Energy Statistics Handbook [11].

Figure 2. Energy Consumption by sector in 2018. Source: Malaysia Energy Commission. Malaysia Energy Statistics Handbook [11].

Figure 2. Energy Consumption by sector in 2018. Source: Malaysia Energy Commission. Malaysia Energy Statistics Handbook [11].

Malaysia increased total CO₂ emissions by nine times, from 28 Mt in 1980 to 262.2 Mt in 2020. The nation released 264.685 Mt of CO₂ in 2019, the highest level ever. Due to the COVID-19 epidemic, the economy entered a recession, which led to a drop in CO₂ emissions in 2020. GDP decreased by 5.6%. Figure 4 illustrates the alarming trend of increasing CO₂ emissions in Malaysia from 1980 to 2020. The trend will continue if no action is taken to preserve the environment. As it may have a detrimental effect on human health and sustainable development, it is bad news for Malaysia’s next generation.

Figure 3. FDI inflows as a percentage of GDP. Source: World Bank Development Indicator [12].

Figure 3. FDI inflows as a percentage of GDP. Source: World Bank Development Indicator [12].

This paper is organized as follows: (1) The first section provides an overview of Malaysia’s energy use and CO₂ emissions. (2) Previous research on CO₂ emissions as a dependent variable is reviewed in the second section. (3) The method of analysis is described in the section that follows. (4) The next section presents the study’s findings. (5) The conclusion concerning whether energy consumption by sector affects CO₂ emissions is presented in the last section, which also includes policy recommendations.

2. Materials and Methods

The effects of energy use and economic expansion on CO₂ emissions are extensively documented in the previous literature, including [14,15,16]. According to Khan et al. [14], economic expansion and energy use have a negative impact on the environment. Radmehr et al. [17] and Karaaslan and Camkaya [18] examined the causal relationships between energy consumption, economic development and CO₂ emissions. Investigating possible relationships between economic development, energy use, and CO₂ emissions in the European Union (EU) region from 1995 to 2014 using the Generalized Method of Moments (GMM) technique, Muhammad [19] discovered bidirectional connections between economic growth and CO₂ emissions as well as between the use of renewable energy and economic growth.

Muhammad [19] also discovered that energy consumption and economic growth could have a negative impact on the environment in developed nations and the Middle East and North Africa (MENA) region using the same methodology and data from 2001 to 2017. Environmental damage was discovered to be caused by energy consumption, and economic development can mitigate environmental degradation in developing nations. Ardakani and Seyedaliakbar [20] also looked into the associations between energy use, economic growth, and CO₂ emissions in the MENA region between 1995 and 2014. They discovered the existence of the environmental Kuznets curve (EKC) in Oman, Qatar, and Saudi Arabia, suggesting that greater GDP prevents environmental damage in the final stage, supported by Ozgur et al. [21] and Li and Haneklaus [22]. They also discovered a U-shaped curve between gross domestic product (GDP) per capita and CO₂ emissions in Algeria and Bahrain, indicating that raising GDP in the final stage aggravates environmental degradation.

Ozgur et al. [21], Karaaslan and Camkaya [18] as well as Cai et al. [23] employed the Autoregressive Distributed Lag (ARDL) method in their studies. Cai et al. [23] looked at how the G7 countries’ economic development and the use of clean energy affected CO₂ emissions. The study produced mixed results, showing a one-way relationship between energy consumption and economic growth in Canada, Germany, and the US; a one-way relationship between CO₂ emissions and energy consumption in Germany; and a one-way relationship between clean energy consumption and CO₂ emissions in the US. From 2008 to 2018, Pejovic et al. [24] also found mixed results when exploring the relationship between economic growth, CO₂ emissions, and energy consumption in the 27 EU nations and the Western Balkans. The results of the panel vector autoregression (PVAR) and GMM methods showed various linkages: (1) a bidirectional linkage between economic growth and CO₂ emissions, (2) a negative bidirectional linkage between CO₂ emissions and energy consumption, (3) a positive linkage running from economic growth to CO₂ emissions, (4) and a positive linkage running from CO₂ emissions to economic growth.

The EKC’s existence in India between 1970 and 2016 was investigated by Ozgur et al. [21]. According to the study’s findings, which used the ARDL technique, nuclear energy can lower CO₂ emissions. The findings in India confirmed the EKC since they suggested that, after the turning point, increasing GDP might lead to less environmental damage. In 11 nations, Wang et al. [25] investigated whether financial development could reduce the effects of using renewable energy on CO₂ emissions. The study used data from 1990 to 2015, employing the Driscoll–Kraay and Dumitrescu–Hurlin methods. The findings revealed that financial development is important for determining how much renewable energy consumption affects CO₂ emissions. Moreover, there are mutual linkages between financial development and the use of renewable energy as well as between the use of renewable energy and CO₂ emissions.

In the G7 nations, Li and Haneklaus [22] looked into the relationship between the use of clean energy, economic expansion, trade openness, urbanization, and CO₂ emissions. The ARDL technique was employed to analyze data spanning 40 years from 1979 to 2019. The findings supported the EKC. In addition, using more renewable energy can lower CO₂ emissions both in the long run and short run. The results also showed that environmental degradation could result from other factors, such as trade openness and urbanization.

In order to determine whether economic growth, health expenditure, and renewable and non-renewable energy use may have an impact on environmental deterioration in Turkey, Karaaslan and Camkaya [18] conducted their research. Besides the ARDL technique, the Toda–Yamamoto approach was also used to determine the variables’ causal link from 1980 to 2016. As a result of economic expansion and the use of non-renewable energy sources, CO₂ emissions increase, while spending on health and using renewable energy sources cause a decrease in CO₂ emissions.

From 2008 to 2018, Pejovic et al. [24] explored the relationship between economic growth, CO₂ emissions, and energy consumption in 27 EU nations and the Western Balkans. The results of the panel VAR and GMM methods showed various linkages: (1) a bidirectional linkage between economic growth and CO₂ emissions, (2) a negative bidirectional linkage between CO₂ emissions and energy consumption, (3) a positive linkage running from economic growth to CO₂ emissions, and (4) a positive linkage running from CO₂ emissions to economic growth.

Chandran and Tang [26] discovered that income and energy for the transportation sector resulted in a long-term, considerable rise in CO₂ emissions for Malaysia, Indonesia, and Thailand. With data from 1971 to 2008, the authors conducted a multivariate co-integration analysis on a subset of ASEAN members. The findings revealed that transportation leads to greater CO₂ emissions. Shaari et al. [1] employed the ARDL approach to focus on energy consumption in agriculture, and the results revealed that energy in the sector reduces environmental degradation in Malaysia.

Several studies, such as Shaari et al. [27], Wang and Huan [7], Huang et al. [9] as well as Demena and Afesorgbor [10], investigated the impact of FDI and CO₂ emissions. However, the results are inconsistent as Shaari et al. [27] examined the impacts of FDI and economic growth on CO₂ emissions in 15 developing countries. Employing a panel ARDL, the results showed that FDI does not influence CO₂ emissions, but economic growth can increase environmental degradation in developing countries. However, Wang and Huan [7] and Huang et al. [9] argued that FDI harms the environment. Wang and Huan [7] investigated the impact in the East Asian region using a panel data analysis. Their results indicated that FDI, economic growth, and trade openness could contribute to greater CO₂ emissions between 2011 and 2020. The results are slightly different from the findings of Huang et al. [9]. According to the study, the diminished impact of FDI on CO₂ emissions in the G20 countries from 1996 to 2018 could have been due to environmental regulations. With evidence from meta-data analysis, Demena and Afesorgbor [10] found that FDI can reduce CO₂ emissions.

In a nutshell, most literature concentrated on energy consumption in general without specifying sectors. The impact of energy consumption is different from one sector to another. Hence, the failure to address specific sectors may lead to the wrong policies, thus affecting productivity.

3. Research Methodology

3.1. Data and Method

This study investigates the impacts of energy consumption in various sectors, particularly industry, transportation and agriculture, and FDI, on environmental degradation in Malaysia. There is an urgent need to address this issue since CO₂ emissions in the region exhibit a steady increase in tandem with higher demand for energy in those sectors. Hence, larger energy consumption is expected to contribute to increasing environmental degradation. Data ranging from 1989 to 2019 on energy consumed in the three sectors, trade openness, FDI, economic growth, exports, imports, and population growth, were collected to achieve the objective of this study. The autoregressive distributed lag (ARDL) approach is employed in this study to estimate the long-run and short-run impacts of energy consumption on environmental degradation. The ARDL model is an ordinary least square (OLS)-based model that includes lags, and it can be used even in the presence of mixed order of integration. The secondary data were extracted from countryeconomy.com, the Malaysia Energy Information Hub, and the World Bank.

3.2. Description of Variables

This current study includes nine variables based on the IPAT model and previous studies to examine the impacts of energy consumption in three different sectors on environmental degradation. The independent variables used in this study consist of energy demand in the agricultural sector (A), energy demand in the transportation sector (T), energy demand in the industrial sector (I), imports (M), exports (X), economic growth (Y), foreign direct investment (FDI), and population growth (P). At the same time, CO₂ emissions are treated as a dependent variable. The proxies for all variables, units of measurement, and the sources of data collection are presented in

Table 1. Description of variables.

Variable	Proxy	Measurement	Source
Energy consumption in the agricultural sector (A)	Total final energy demand in the agricultural sector	Tonnes of oil equivalent (toe)	Malaysia Energy Information Hub
Energy consumption in the transportation sector (T)	Total final energy demand in the transportation sector	Tonnes of oil equivalent (toe)	Malaysia Energy Information Hub
Energy consumption in the industrial sector (I)	Total final energy demand in the industrial sector	Tonnes of oil equivalent (toe)	Malaysia Energy Information Hub
Exports (X)	Exports of goods and services as a share of GDP	Percent (%)	World Bank
Imports (M)	Imports of goods and services as a share of GDP	Percent (%)	World Bank
Foreign direct investment (FDI)	FDI inflows as a percentage of GDP	Percent (%)	World Bank
Economic growth (Y)	GDP at constant prices	Local currency units (LCU)	World Bank
Population growth (P)	Total population growth	Percent (%)	World Bank
CO2 emissions (CO2)	Total CO2 emissions	Metric tonnes (Mt)	countryeconomy.com

. CO₂ emissions are proxied for environmental degradation owing to its having the largest contribution to the total greenhouse gas emissions. The transportation and industrial sectors have been given attention in this study since they consume the largest and second-largest share of the total energy in Malaysia. In contrast, the energy used in the agricultural sector accounts for the smallest share.

3.3. Econometric Model

The IPAT model describes the idea of how environmental degradation (I) is dependent on three main factors: population (P), affluence (A), and technology (T). According to the model, as affluence increases, environmental degradation intensifies. In addition, population growth can also contribute to harming the environment. Using more technologies, especially non-green technologies, may also result in a lower quality environment. Therefore, the model is as follows:

I = P · A · T

Previous studies, such as Mahmood et al. [28], Al-mulali and Sheau-Ting [29], and many more, used CO₂ emissions as a proxy for environmental degradation, population growth as a proxy for population, GDP as a proxy for affluence, and energy use as a proxy for technology.

$${\mathrm{CO}}_2=\mathsf{\alpha }+{\mathsf{\beta }}_1\mathrm{A}+{\mathsf{\beta }}_2\mathrm{T}+{\mathsf{\beta }}_3\mathrm{I}+{\mathsf{\beta }}_4\mathrm{X}+{\mathsf{\beta }}_5\mathrm{M}+{\mathsf{\beta }}_6\mathrm{FDI}+{\mathsf{\beta }}_7\mathrm{Y}+{\mathsf{\beta }}_8\mathrm{P}+{\mathsf{\mu }}_{\mathrm{t}}$$

(1)

In time-series data analysis, Equation (1) should be transformed into logarithms to examine the percentage change in environmental degradation due to a percentage change in energy consumption. In addition, natural logarithmic transformation is carried out to deal with the problem of potential non-linearities in the relationship between the dependent and independent variables. Therefore, a new equation is as follows:

$${\mathrm{LNCO}}_2=\mathsf{\alpha }+\mathrm{LN}{\mathsf{\beta }}_1\mathrm{A}+\mathrm{LN}{\mathsf{\beta }}_2\mathrm{T}+\mathrm{LN}{\mathsf{\beta }}_3\mathrm{I}+\mathrm{LN}{\mathsf{\beta }}_4\mathrm{X}+\mathrm{LN}{\mathsf{\beta }}_5\mathrm{M}+\mathrm{LN}{\mathsf{\beta }}_6\mathrm{FDI}+\mathrm{LN}{\mathsf{\beta }}_7\mathrm{Y}+\mathrm{LN}{\mathsf{\beta }}_8\mathrm{P}+{\mathsf{\mu }}_{\mathrm{t}}$$

(2)

whereby

• LNCO₂ represents the natural logarithm of CO₂ emissions;
• LNA represents the natural logarithm of energy consumption in the agricultural sector;
• LNT represents the natural logarithm of energy consumption in the transportation sector;
• LNI represents the natural logarithm of energy consumption in the industrial sector;
• LNX represents the natural logarithm of exports;
• LNM represents the natural logarithm of imports;
• LNFDI represents the natural logarithm of FDI;
• LNY represents the natural logarithm of GDP;
• LNP represents the natural logarithm of population growth.

The expected signs for β₁, β₂, and β₃ depend on the development of these three sectors. β₄ and β₅ are expected to have a negative relationship with CO₂ emissions. Meanwhile, the expected sign for β₆ will exhibit the presence of either the Pollution Haven Hypothesis for a negative sign or the Halo Effect Hypothesis for a positive sign. Lastly, both β₇ and β₈ are expected to have positive signs.

3.4. Unit Root Test

This study uses secondary data on macroeconomic variables for its analysis. Thus, conducting a unit root test is vital to examine the stationarity of data. The augmented Dickey–Fuller approach is used to complete the analysis because it has been successfully applied in much earlier research. Econometric techniques can often be used to examine whether there are periodic changes in the mean and variance for time-series data. The unit root test null hypothesis indicates that data are not stationary, while the alternative hypothesis indicates that data are stationary. To ensure that the ARDL approach can be employed, we have to make sure that the variables used in this study are integrated of mixed order I(0) and I(I), suggesting that all the variables must be stationary at the first difference. The equation for the unit root test is as follows:

$$\Delta {\mathrm{X}}_1={\mathsf{\alpha }}_1+{\mathsf{\beta }}_1\mathsf{\Gamma }+{\mathsf{\beta }}_2{\mathrm{X}}_{\mathrm{t}-1}+{{\displaystyle \sum }}_{\mathrm{i}-1}^{\mathrm{m}}{\mathsf{\alpha }}_2\Delta {\mathrm{X}}_{\mathrm{t}-1}+{\mathsf{\mu }}_{\mathrm{t}}$$

(3)

• X represents the variable used in this study to test for stationarity;
• $\mathsf{\Gamma }$ represents the linear trend;
• $\Delta {\mathrm{X}}_{\mathrm{t}-1}$ represents the lag difference;
• ${\mathsf{\alpha }}_1$ represents the intercept;
• t represents the time trend.

The null and alternative hypotheses for stationarity are as follows:

H0. 

α = 0 integrated of mixed order I(0) and I(I), suggesting.

H1. 

α ≠ 0 (there is no unit root or stationarity).

3.5. Autoregressive Distributed Lag (ARDL) Technique

A co-integration test is crucial for determining whether there is a long-term link between the following variables: CO₂ emissions, energy consumption in the agricultural, transportation, and industrial sectors, exports, imports, FDI, economic growth, and population growth. Due to the importance of this analysis, the ARDL bounds testing approach can be employed. This approach is selected due to its flexibility in conducting a co-integration test, as stated by Pesaran and Shin [30]. If unit root test results reveal that the variables are integrated of mixed order, then the ARDL bound test can be performed to check whether there is a co-integration. With the results obtained, we can compare the F-statistic with the upper and lower bounds, as suggested by Pesaran et al. [31], to confirm the existence of co-integration. If the F-statistic exceeds the critical upper bound, it can be inferred that there is a co-integration. However, if the F-statistic does not, the variables are not co-integrated. The results remain inconclusive if it falls between the upper and lower bounds. We can estimate the long-run and short-run relationship by following the bound test results indicating a co-integration. Furthermore, the error correction model (ECM) determines the speed of the adjustment of the long-run equilibrium. The long-run and short-run models can be written as follows:

$$\begin{array}{l}{\Delta \mathrm{LNCO}}_{2\mathrm{t}}={\mathsf{\beta }}_1+{\mathsf{\theta }}_0{\mathrm{LNCO}}_{2\mathrm{t}-1}+{\mathsf{\theta }}_1{\mathrm{LNA}}_{2\mathrm{t}-1}+{\mathsf{\theta }}_2{\mathrm{LNT}}_{\mathrm{t}-1}+{\mathsf{\theta }}_3{\mathrm{LNI}}_{\mathrm{t}-1}+{\mathsf{\theta }}_4{\mathrm{LNX}}_{\mathrm{t}-1}+{\mathsf{\theta }}_5{\mathrm{LNM}}_{\mathrm{t}-1}+{\mathsf{\theta }}_6{\mathrm{LNFDI}}_{\mathrm{t}-1}+{\mathsf{\theta }}_7{\mathrm{LNY}}_{\mathrm{t}-1}+{\mathsf{\theta }}_8{\mathrm{LNP}}_{\mathrm{t}-1}\\ +{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{a}}{\mathsf{\beta }}_{\mathrm{i}}{\Delta \mathrm{LNCO}}_{2\mathrm{t}-\mathrm{i}}}+{\displaystyle \sum _{\mathrm{i}=0}^{\mathrm{b}}{\mathsf{\gamma }}_{\mathrm{i}}}{\Delta \mathrm{LNA}}_{\mathrm{t}-\mathrm{i}}+{\displaystyle \sum _{\mathrm{i}=0}^{\mathrm{c}}{\mathsf{\delta }}_{\mathrm{i}}\Delta }{\mathrm{LNT}}_{\mathrm{t}-\mathrm{i}}+{\displaystyle \sum _{\mathrm{i}=0}^{\mathrm{d}}{\mathsf{\lambda }}_{\mathrm{i}}}{\Delta \mathrm{LNI}}_{\mathrm{t}-\mathrm{i}}+{\displaystyle \sum _{\mathrm{i}=0}^{\mathrm{e}}{\mathrm{J}}_{\mathrm{i}}}{\Delta \mathrm{LNX}}_{\mathrm{t}-\mathrm{i}}+{\displaystyle \sum _{\mathrm{i}=0}^{\mathrm{f}}{\mathsf{\psi }}_{\mathrm{i}}{\Delta \mathrm{LNM}}_{\mathrm{t}-\mathrm{i}}}+{\displaystyle \sum _{\mathrm{i}=0}^{\mathrm{g}}{\mathsf{\xi }}_{\mathrm{i}}{\Delta \mathrm{LNFDI}}_{\mathrm{t}-\mathrm{i}}}+{\displaystyle \sum _{\mathrm{i}=0}^{\mathrm{h}}{\mathrm{v}}_{\mathrm{i}}{\Delta \mathrm{LNY}}_{\mathrm{t}-\mathrm{i}}}\\ +{\displaystyle \sum _{\mathrm{i}=0}^{\mathrm{i}}{\mathsf{\Xi }}_{\mathrm{i}}{\Delta \mathrm{LNP}}_{\mathrm{t}-\mathrm{i}}}{+\mathsf{\upsilon }}_{\mathrm{t}}\end{array}$$

(4)

where ∆ is the first difference operator, and u_t is the white-noise disturbance term. The residuals for the UECM should be serially uncorrelated, and the model should be stable. This validation can be addressed with diagnostic tests. The final version of the model represented from Equation (4) can also be viewed as an ARDL of order (a b c d e f g h i). The short-run effects are captured by the coefficients of the first differenced variables.

The null of no co-integration in the long-run relationship is defined as follows:

• H0: θ₀ = θ₁ = θ₂ = θ₃ = θ₄ = θ₅ = θ₆ = θ₇ = θ₈ = 0 (there is no long-run relationship).
• H1: θ₀ ≠ θ₁ ≠ θ₂ ≠ θ₃ ≠ θ₄ ≠ θ₅ ≠ θ₆ ≠ θ₇ ≠ θ₈ ≠ 0 (there is a long-run relationship).

Consider the case where the calculated F-statistic is lower than the lower bound critical value. In that situation, we do not rule out the possibility of co-integration. However, suppose that the calculated F-statistic exceeds the upper bound critical value of at least the 10% significance level. We, therefore, reject the null hypothesis that co-integration does not exist.

4. Findings

The results of descriptive statistics for a sample of 29 observations are shown in

Table 2. Descriptive Statistics.

	Mean	Median	Maximum	Minimum	Std. Dev.	Observations
LNCO2	5.0713	5.2100	5.5785	4.2337	0.4281	29
LNX	4.5008	4.5152	4.7983	4.1790	0.1978	29
LNM	4.3699	4.4006	4.6112	4.0553	0.1739	29
LNY	8.7398	8.6444	9.3264	7.9673	0.4430	29
LNFDI	−3.4030	−3.3524	−2.4304	−6.9078	0.8346	29
LNP	0.6889	0.6931	1.7918	0.2624	0.3299	29
LNA	5.7692	5.9687	6.9791	4.1271	0.9460	29
LNT	9.5416	9.6352	10.1268	8.6666	0.1141	29
LNI	9.4026	9.4851	9.8546	8.6672	0.3206	29

. The mean for LNT is 9.5416, whereas the mean for LNFDI is −3.4030. With a difference of 4.4774, LNFDI has the biggest difference between its highest and minimum values, while LNM has the least. All variables have standard deviations that are almost zero, which shows that the data points are quite close to the means.

A unit root test based on ADF is conducted to examine the stationarity of all variables, and the results are presented in Table 3. The intercept and no trend results show that LNA, LNX, LNY, LNM, and LNP have unit roots and are not stationary at the level. However, the other variables (LNCO2, LNFDI, LNA, and LNT) have no unit root and are stationary at the level. All of the variables become stationary at the first difference. The results with an intercept and trend show that only LNFDI, LNY, and LNP have unit roots and are not stationary at the level, and the rest (LNCO2, lNA, LNX, LNM, and LNT) have no unit roots and are stationary at the level. Nevertheless, the results indicate that all variables have no unit root and are stationary at the first difference. In a nutshell, the variables used in this study are integrated of mixed order—I(I) and I(0)—indicating that the ARDL technique can be applied.

Prior to examining the short-run and long-run impacts of energy use on environmental degradation, a bound test must be performed to check whether there is a co-integration among the variables used in this study. The ARDL bounds testing technique is employed, and the results reported in

Table 4. Bound Test.

Lag Model: 1, 1, 1, 1, 1, 0, 1, 0, 1
F-Statistic	8.0225 ***
Critical Value	Lower Bound	Upper Bound
10%	1.85	2.85
5%	2.11	3.15
1%	2.62	3.77

Note: *** denotes significance at 1%.

show that the F-statistic of 8.0225 is higher than the upper bound of 3.77, implying the null hypothesis that there is no co-integration is rejected. This indicates that we can proceed to long-run and short-run estimations. In addition, the optimal lag of 1, 1, 1, 1, 1, 0, 1, 0, 1 is automatically selected based on AIC.

Table 3. Unit Root Test Results.

Variable	Intercept		Intercept + Trend
	Level	1st Difference	Level	1st Difference
LNCO2	−4.3170 ***	−4.6001 ***	−2.0792	−6.0661 ***
LNA	−2.0774	−7.9965 ***	−2.6111	−7.9298 ***
LNX	−0.5393	−3.4845 **	−1.3776	−4.1872 **
LNFDI	−4.7318 ***	−6.5092 ***	−5.2192 ***	−6.3848 ***
LNY	−1.3664	−2.0881 ***	−4.6330 ***	−0.5123 ***
LNM	−0.5123	−4.5845 ***	−2.7753	−5.2327 ***
LNA	−2.6917 *	−3.5198 **	−2.0735	−3.6777 **
LNT	−2.7064 *	−4.9034 ***	−1.8423	−5.3775 ***
LNP	−1.3292	−9.0692 ***	−4.9297 ***	−8.9065 ***

Note: ***, **, and * denote significance at 1%, 5%, and 10%, respectively.

Table 3. Unit Root Test Results.

Variable	Intercept		Intercept + Trend
	Level	1st Difference	Level	1st Difference
LNCO2	−4.3170 ***	−4.6001 ***	−2.0792	−6.0661 ***
LNA	−2.0774	−7.9965 ***	−2.6111	−7.9298 ***
LNX	−0.5393	−3.4845 **	−1.3776	−4.1872 **
LNFDI	−4.7318 ***	−6.5092 ***	−5.2192 ***	−6.3848 ***
LNY	−1.3664	−2.0881 ***	−4.6330 ***	−0.5123 ***
LNM	−0.5123	−4.5845 ***	−2.7753	−5.2327 ***
LNA	−2.6917 *	−3.5198 **	−2.0735	−3.6777 **
LNT	−2.7064 *	−4.9034 ***	−1.8423	−5.3775 ***
LNP	−1.3292	−9.0692 ***	−4.9297 ***	−8.9065 ***

Note: ***, **, and * denote significance at 1%, 5%, and 10%, respectively.

Table 5 provides the results of long-run relationships between population growth, energy consumption by sector, economic growth, FDI, exports, imports, and CO₂ emissions. The results reveal a significant and negative relationship between energy consumption in the agricultural sector and CO₂ emissions in the long run, with a coefficient value of 0.0417. A 1% increase in energy consumption in the sector can contribute to a 0.04% lower level of CO₂ emissions. There is also a significant and positive relationship between energy consumption in the industrial sector and CO₂ emissions in the long run. The coefficient value of 0.1500 indicates that a 1% rise in energy consumption in the sector may lead to a 0.15% rise in CO₂ emissions. A significant and positive relationship between energy consumption in the transportation sector has been observed with a coefficient value of 0.4582, suggesting that as energy consumption in the sector goes up by 1%, CO₂ emissions may increase by 0.46%. Exports can also contribute to larger CO₂ emissions in the long run. A 1% rise in exports results in CO₂ emissions rising by 0.44%. Economic growth also has a significant and positive relationship with CO₂ emissions. Economic growth escalates by 1%, prompting a rise of 0.43% in CO₂ emissions. Population growth, FDI, and imports do not have a long-term effect on CO₂ emissions. Therefore, a 1% increase in population growth, FDI, and imports does not cause CO₂ emissions to increase in the long run.

Table 6. Short-run estimation results.

Variable	Coefficient	Std. Error	T-Statistic	Probability
LNA	−0.0181 *	0.0096 *	−1.8890 *	0.0833
LNI	0.0200	0.0826	0.2422	0.8127
LNP	−0.0208	0.0218	−0.9535	0.3592
LNT	0.1490 *	0.0833 *	1.7874 *	0.0991
LNX	0.0598	0.1567	0.3817	0.7094
LNFDI	0.0286 ***	0.0076 ***	3.7899 ***	0.0026
LNY	0.2199 ***	0.0620 ***	3.5467 ***	0.0040
LNM	−0.1910	0.1498	−1.2747	0.2265
C	−3.8120 ***	0.6383 ***	−5.9718 ***	0.0001
ECT	−0.7610 ***	0.14027 ***	−5.4254 ***	0.0002

Note: *** and * denote significance at 1% and 10%, respectively.

provides the results of short-run relationships between the variables. The findings reveal a significant negative relationship between energy consumption in the agricultural sector and CO₂ emissions in the short run. There is also a significant and positive relationship between energy consumption in the transportation sector and CO₂ emissions. FDI and economic growth have significant impacts on CO₂ emissions. Energy consumption in the industrial sector, population growth, and imports, on the other hand, do not significantly impact CO₂ emissions. The coefficient of the ECT is −0.7610 and significant, confirming that there are long-run relationships between energy consumption by sector, population growth, exports, imports, economic growth, and CO₂ emissions in the long run.

Table 5. Long-run estimation results.

Variable	Coefficient	Std. Error	T-Statistic	Prob
LNA	−0.0417 **	0.0174 **	−2.3955 **	0.0338
LNI	0.1500 **	0.0610 **	2.4613 **	0.0300
LNP	−0.0274	0.0295	−0.9294	0.3710
LNT	0.4582 ***	0.4796 ***	8.4378 ***	0.0000
LNX	0.4406 ***	0.1227 ***	3.5915 ***	0.0037
LNFDI	0.0244	0.0139	1.7565	0.1045
LNY	0.4337 ***	0.0655 ***	6.6229 ***	0.0000
LNM	−0.2510	0.1682	−1.4918	0.1616
C	−5.0092 ***	0.4796 ***	−10.4448 ***	0.0000

Note: *** and ** denote significance at 1% and 5%, respectively.

Table 5. Long-run estimation results.

Variable	Coefficient	Std. Error	T-Statistic	Prob
LNA	−0.0417 **	0.0174 **	−2.3955 **	0.0338
LNI	0.1500 **	0.0610 **	2.4613 **	0.0300
LNP	−0.0274	0.0295	−0.9294	0.3710
LNT	0.4582 ***	0.4796 ***	8.4378 ***	0.0000
LNX	0.4406 ***	0.1227 ***	3.5915 ***	0.0037
LNFDI	0.0244	0.0139	1.7565	0.1045
LNY	0.4337 ***	0.0655 ***	6.6229 ***	0.0000
LNM	−0.2510	0.1682	−1.4918	0.1616
C	−5.0092 ***	0.4796 ***	−10.4448 ***	0.0000

Note: *** and ** denote significance at 1% and 5%, respectively.

Diagnostic tests, such as Breusch–Godfrey serial correlation, Ramsey RESET stability, heteroscedasticity, and Jarque–Bera, are indispensable for checking whether the model used in this study suffers any problem and whether the model is stable. The results reported in

Table 7. Diagnostic tests results.

Test Statistic	F-Statistic	Probability
Breusch-Godfrey Serial Correlation LM	2.3376	0.1469
Ramsey RESET stability	1.9956	0.1854
Heteroscedasticity	0.4696	0.9159
Jarque–Bera	0.8107	0.6668

show that the p-values for all of them are insignificant at 5%, indicating that the alternative hypotheses are rejected. This means that the model used in this study does not suffer from any diagnostic problems. In addition, CUSUM and CUSUMQ tests are also performed to check the stability of our model. The results presented in Figure 5 show that the model is stable due to the plots falling within the two critical lines. Therefore, our inference concerning the impacts of energy consumption in the agricultural, transportation, and industrial sectors on environmental degradation is reliable.

5. Discussions

The findings of this study reveal that higher use of energy in the agricultural sector can harm the environment. They are similar to the findings of Shaari et al. [1]. Energy use in the region’s sector is still low, and some farmers, especially in rural areas, still use traditional methods of producing outputs, which may reduce CO₂ emissions. In comparison to the other sectors, agriculture consumes the smallest share of the total energy, especially non-renewable energy such as oil. Therefore, developing the sector by introducing technologies that consume non-renewable energy may harm the environment. In addition, this study also finds that higher energy consumption in the industrial sector can also put the country at risk of suffering greater environmental degradation. The sector is responsible for the second-largest share of total energy consumption. The use of non-renewable energy sources, such as oil and coal, in the industrial sector exhibits an upward trend, contributing to greater CO₂ emissions in Malaysia. Therefore, it is imperative to pay more attention to this sector in order to reduce CO₂ emissions. Energy consumption in the transportation sector is also found to positively impact CO₂ emissions, supported by Chandran and Tang [26] and Nasreen et al. [32]. The sector consumes the largest share of the total energy in Malaysia, followed by the industrial sector. Thus, its impact on CO₂ emissions is greater than the industrial sector. As has been explained by Yuaningsih et al. [33], this is because there is higher demand for passengers using transport services due to increasing income and rapid economic development; this conclusion is supported by Shaari et al. [34], Ridzuan et al. [35] and Ridzuan et al. [36], who also found the same outcomes for Malaysia.

Economic growth is also responsible for greater environmental degradation as it entails using more non-renewable energy that can release more CO₂ into the air. Due to Malaysia’s reliance on oil, coal, and gas to generate its economic activity, it is no wonder that CO₂ emissions exhibit a steady increase. Malaysia is not a developed country, so economic growth still precedes environmental conservation. This does not correspond with the results of Rahman et al. [2] that higher economic growth should reduce CO₂ emissions in newly industrialized countries. Greater FDI inflows can also harm the environment in Malaysia. Zubair et al. [37] argue that FDI inflows may reduce CO₂ emissions in Nigeria as FDI inflows will result in using more green technologies to reduce CO₂ emissions. However, FDI inflows in Malaysia do not contribute to an increase in the use of green technology that can cushion the impact on environmental degradation. Exports can also contribute to greater environmental degradation, supported by Wang et al. [38]. Malaysia’s exports depend highly on electrical and electronic goods, which can release CO₂ during production. Furthermore, the sector consumes various non-renewable energy sources, such as coal and oil, that can also emit CO₂.

6. Conclusions

This study examines the effects of energy consumption by sector on CO₂ emissions in Malaysia from 1989 to 2019. It employs the ARDL approach, and the results reveal that energy consumption in the transportation and industrial sectors, exports, FDI, and economic expansion have positive impacts on CO₂ emissions, whereas energy consumption in the agriculture sector has a significant negative impact. This indicates that an increase in energy use in agriculture does not harm the environment. Population growth and imports do not significantly impact CO₂ emissions.

Policymakers must develop appropriate policies in light of the findings that energy use in the transportation and industrial sectors can worsen environmental deterioration. Malaysia continues to rely on non-renewable energy, such as coal and oil. Since agriculture does not impact the environment, it is not necessary to implement any environmental policies in this sector. However, we need to focus more on the industrial and transportation sectors. In comparison to the industrial sector, the transportation sector has a greater impact on CO₂ emissions. To lessen environmental degradation, improvements must be made to five fuel diversifications that the government has already implemented, particularly in the transportation and industrial sectors. Despite the policies, non-renewable energy is still widely used in those sectors in Malaysia. Therefore, we must switch to utilizing more renewable energy sources, like solar, biodiesel, and hydro. In order for consumers to purchase electric vehicles, their prices must be reasonable. Additionally, firms in the two sectors that produce a lot of CO₂ emissions should be subject to carbon taxes and pricing. Most economists also supported the idea that the policies may lower CO₂ emissions while providing profits to the government.

Due to higher economic growth and exports increasing CO₂ emissions, it is impossible to reduce them in a bid to reduce environmental degradation. However, the country can consider using carbon capture and storage technology. Carbon will be stored deep underground in geological formations, leading to a reduction in environmental degradation. FDI inflows might be detrimental to the environment, suggesting that technology brought into the country is not environmentally friendly. Thus, transferring green technology as a result of FDI inflows is of utmost importance. The government may also consider imposing additional taxes on firms that import non-green technology.

This study is still not perfect, as it has several limitations. There are several potential variables that can be included in this study, such as corruption, financial development, and so forth. Therefore, including these variables in the future may help improve the findings of this study and shed more light on formulating policies. In addition, this study only focuses on Malaysia using time-series data analysis. In future research, it is better to explore the impact of energy consumption on CO₂ emissions in many countries employing panel data analysis.