Exploring the Relationship between Energy and Food Security in Africa with Instrumental Variables Analysis

Abstract

The well-being of human populations and their sustainable development are strongly predicated on energy and food security. This is even more true of Africa due to often suboptimal food production, undernourishment, and extreme poverty. This article researches the relationship between energy and food security using Cobb–Douglas production functions based on the World Development Indicators data for 28 African countries. The methodological approach includes cross-sectional dependence and unit root tests, instrumental variables two-stage least-squares and generalized method of moments, and panel Driscoll–Kraay standard errors. Results suggest that the promotion of energy security promotes food security. This is possible because food production and distribution are energy-intensive. Therefore, energy is fundamental to achieving food security and zero hunger. The availability, affordability, accessibility, and acceptability of energy can thus help to fix the growing agricultural production shortage in Africa. An important policy focus should be on achieving energy security.

1. Introduction

Energy and food security are essential preconditions for adequate well-being for human populations and to ensure sustainable development. Energy security, while highly context-dependent, consists primarily of the availability, affordability, accessibility, and acceptability of energy in a particular society [1,2]. In developing countries, it has been found to pertain to different parts of the energy system, namely the supply, conversion, distribution, and demand subsystems [3,4].

The notion of food security, on the other hand, is more straightforward and has generally been defined as a ‘situation that exists when all people, always, have physical, social and economic access to sufficient, safe and nutritious food that meets their dietary needs and food preferences for an active and healthy life [5].’ It is of high importance, as the estimated human population growth will increase the global need for food by an estimated 70% by 2050 [6]. As such, food security is an essential precondition for sustainable development. It is also known to be predicated on improving food quality and quantity, food safety, and socio-cultural and environmental acceptance [7].

While energy and food security are each highly relevant as policy goals, their interrelation and mutual influence are often not clear enough in different contexts. A better understanding of their relationship might prove to be especially important for the global South, in which African countries dominate, due to often unsustainable food production practices, the prevalence of extreme poverty, and undernourishment [8]. Food security in terms of sustainability is necessary to reach several of the United Nations Sustainable Development Goals, which are aimed mainly at the global South’s challenges [9]. The connection between energy security and sustainable development in economic, societal, and environmental terms is also widely accepted [10].

This article aims to contribute to understanding the energy–food security nexus by exploring the relationship between energy and food security on a select dataset of 28 African countries to arrive at conclusions and policy implications that are relevant for Africa and could also be applied to the global South more broadly. Thus, the research question is how potential energy security improvements influence food security in Africa. The study is limited to the 28 African countries due to the relevance and availability of quality data. Our study makes incremental contributions to the literature. First, we used a computed food security index comprising all various classes of foods compared to previous studies, which mainly recognize an aspect of food. We argued that relying mainly on one aspect of food will not give a comprehensive picture of food status. Second, we applied two-stage least-squares (2SLS) and generalized method of moments (GMM), as well as panel Driscoll–Kraay standard errors, which consider cross-sectional dependency, endogeneity, and heteroskedasticity [11]. We argue that the presence of heteroskedasticity shows that the scattering of the model is dependent on at least one independent variable. This adds to the problem of the model and creates a scenario of deviation between effective and actual results.

Regarding the study’s limitations, this article focuses on the interrelation between food and energy security and does not explore the related relationship between water security and energy security, where important prior research has been conducted [12,13,14,15]. This article also does not directly explore the relationship between armed conflicts, food, and energy security [16,17], which has become additionally relevant with the Ukraine crisis [18,19]. The relatively small sample size does not represent a significant limitation since green energy adoption is still at an infant stage in the global South.

This article is structured so that after the introduction, in the second part, the authors will review existing literature and prior research on the topics of energy and food security and their interrelation. The data, modelling, and methodology will be described in the third part. The methodology will include cross-sectional dependence and unit root tests, instrumental variables 2SLS and GMM, and panel Driscoll–Kraay standard errors [11]. The fourth part will present results, including pre-regression and main and alternative analyses. Additionally included will be a discussion of heteroskedasticity, autocorrelation, and spatial correlation problem. Finally, conclusions will be stated in the fifth part, and main policy implications will be proposed.

2. Literature Review

Prior relevant research can be grouped into three main spheres, namely research on energy security, research delving into food security, and research at the intersection of both. Due to the vast literature, this review will mainly focus on state-of-the-art studies produced in the last few years, which build on and complement previous studies. At the same time, it will be shown that the present article is novel. The existing research has not delved into exploring the relationship between energy and food security in Africa, using instrumental variables analysis in a way this article does. Moreover, the research specifically quantitatively exploring the energy–food binomial in the global south has been relatively scarce.

Regarding energy security, Axon and Darton have explored its relationship with sustainability and risk [20], while Yu and others have critically analyzed the energy security metrics themselves [21]. Studies have been performed that delve into the issue of energy security in different parts of the world and, more specifically, the global South, using a diverse methodology. Brunet and others focused on a select dataset on African countries to explore the possibility of solar energy reducing poverty [22], while Lee and others explored the impact of energy security on income inequality for 68 countries [23].

Alemzero and others have assessed energy security in Africa using a principal composite analysis and a multi-dimensional approach [24]. Santon performed research regarding energy integration and regional energy security in South America, arguing that integrated scenarios reduce the need to increase installed capacity and lower socio-environmental impacts [25]. Filho and others focused on the use of renewables as a tool for ensuring energy security in small island developing states [26]. Okpanachi and others have researched energy security in light of energy regime reconfiguration in Morocco [27]. Some researchers have focused on energy security in countries such as Ukraine [28] and China [29].

Research regarding food security has primarily focused on the global South. Akbari and others published an extensive literature review earlier this year [7]. A state-of-the-art of science in food security has been discussed by Cole and co-authors [6]. In terms of the latest research on specific topics in food security, Bezner Kerr and others thus aimed to address social and environmental aspects of food production in light of food security, emphasizing low-income states [30]. Several studies, such as one performed by O’Hara and Toussaint, were produced regarding the impact of the pandemic on food security [31]. Agyei and others have noted the influence of the COVID-19 pandemic on the food prices in sub-Saharan Africa [32]. Nicholson and others critically evaluated and conceptualized the notion of food security and the outcomes in agricultural systems models [33]. Lukwa and co-authors researched the potential influence of informal savings groups on the promotion of food security [34]. Fujimori and others explored how climate change mitigation measures can impact food security [35]. Horn and others performed a case study of Sweden, focusing on the links between food trade, security, and climate change [36].

Research has also been conducted on the interplay of climate change, the environment, and food security. Thus, Molotoks and others researched the impact of climate change, population, and use of land on food security at the global level [37]. Wang and others critically analyzed the interplay between food security and environmental sustainability in waste management [38].

The most relevant for the present article is research that has been performed, analyzing the interplay of energy and food security. Studies have been conducted regarding specific countries, such as Poland, on how food security and agriculture might influence energy security [39], as well as studies aimed at reconciling energy and food security from a regulatory perspective, using qualitative methodology [40]. Researchers have often focused on the food–energy–water nexus as part of the research on environmental policies [41,42,43,44] or concerning climate change more broadly [45].

Specifically, regarding the food–energy binomial, Candelise and others have explored the effect of electricity prices on food security, finding both immediate and income-mediated impacts [46]. Alsaleh and others have researched bioenergy and shown that increasing it in the energy mix could improve food security in the European Union [47]. Falchetta has identified a lack of electricity access as one of the main obstacles to food security in rural sub-Saharan Africa [48]. Chandio and others have found that after reaching a certain threshold, financial development improves agricultural production and that energy security can play a role in aiding the production of crops, thus increasing food security [49]. Guo and Tanaka have explored the fuel versus food issue in the context of biofuels [50]. Hesari and others have explored the oil price and food price interrelationship, showing that agricultural food prices respond positively to oil shocks [51]. Gafa and Ebgendewe have focused on energy security in rural West Africa and have noted its potential influence on food security [52]. Mironova and others explored fiscal controls’ influence on energy and food security [53]. Xia and Yan explored the energy–food nexus in light of the climate policy [54].

The study is novel in that it uses available data to establish and argue the relationship between energy security and food security using a state-of-the-art quantitative methodology and does so concerning Africa. While studies on energy and food security in the global South and Africa have focused mainly on the water–energy–food nexus, this study will be aimed specifically at the potential of energy security positively influencing food security, thus focusing on the energy–food binomial. Consistent estimates will be produced, and the reliability of estimations will be ensured. Endogeneity, heteroskedasticity, autocorrelation, and spatial correlation will be accounted for.

3. Data, Modeling, and Methodology

3.1. Data

We sourced our data from world bank databases (World Development Indicators—WDI) with a panel dataset of 28 African countries (Burkina Faso, Burundi, Central African Republic, Chad, Ethiopia, Guinea, Guinea-Bissau, Liberia, Madagascar, Malawi, Mali, Mozambique, Niger, Rwanda, Sierra Leone, Somalia, Togo, and Uganda.) from 1991 to 2019. First, we constructed the food security, which comprises agricultural products such as cassava, maize, potatoes, rice, wheat, fruit, pulse, vegetables, cereal, and roots and tubers (See Appendix A). At the same time, the energy security index comprises energy availability, applicability, acceptability, and affordability. Second, we considered variables such as gross domestic product, labor force, and CO₂ emissions (see details in

Table 1. Variables measurement and source.

Variables	Meaning	Unit and Measurement	Sources
Fsec	Food security	PCA estimation	WDI, 2022
Esec	Energy security	PCA estimation	WDI, 2022
Gdp	GDP per capita	GDP per capita (constant 2015 USD)	WDI, 2022
Labor	Labor force	Labor force participation rate, total (percentage of total population ages 15+)	WDI, 2022
CO2	CO2 emissions	CO2 emissions (metric tons per capita)	WDI, 2022

).

3.2. Modeling

The Cobb–Douglas production functions can establish the relationship between energy security and food security [55]. The production function is based on a single good with two factors of production:

$$Y=A{K}^{\beta } L^{\alpha }$$

(1)

where $Y$ is the production outputs, $A$ is the total productivity, $K$ is the capital input, and $L$ is the labor inputs, while $\beta$ and $\alpha$ are the coefficients of the output of capital ($K$) and labor ($L$), respectively. However, Zhang and others [56] argued that energy has come to stay as a factor of production, given that it drives economic development. Therefore, we introduced energy use (E) into the Cobb–Douglas production:

$$Y=A{K}^{\beta } L^{\alpha } E^{\sigma }$$

(2)

Following the prior studies (Ogbolumani and Nwulu [41]; Fetanat et al. [57]; Lu et al. [58]), we estimate the role of energy security on food security in Africa: The baseline equation is expressed as:

$$Fse{c}_{it}={\beta }_0+{\beta }_{1}Ese{c}_{it}+{\beta }_{2}Gd{p}_{it}+{\beta }_{3}Labo{r}_{it}+{\beta }_4{{\mathrm{C}\mathrm{O}}_2}_{it}+{\epsilon }_{it}$$

(3)

where $Fsec$ is the food security, $Esec$ is the energy security, $Gdp$ is the gross domestic product, $Labor$ is the labor force, ${\mathrm{C}\mathrm{O}}_2$ is the CO₂ emissions, and $\epsilon$ is the error term. The coefficients of ${\beta }_0-{\beta }_4$ are estimation parameters for $Fsec$, $Esec$, $Gdp$, $Labor$, and ${\mathrm{C}\mathrm{O}}_2$, while $i$ and $t$ are the country’s cross-section and time, respectively.

3.3. Methodology

3.3.1. Cross-Sectional Dependence and Unit Root Tests

Panel datasets often face the problem of cross-sectional dependence, which hinders the reliability of estimations. Hence, we applied the Pesaran [59] cross-sectional dependence test based on the average of pairwise correlation coefficients of ordinary least-squares (OLS) residual generated from standard augmented Dickey–Fuller [60] for each individual.

$$CD=\sqrt{\frac{2T}{N\left(N-1\right)}}\left({\displaystyle \sum }_{i=1}^{N-1}{\displaystyle \sum }_{j=i+1}^N{\rho }_{ij}\right) \to N\left(0,1\right)$$

(4)

Unlike the first-generation unit root test, Pesaran’s Cross-sectionally augmented Im–Pesaran–Shin (CIPS) unit root accounts for cross-sectional dependence [61]. It rejects the null hypothesis of a unit root in cross-section dependence [62]. This can be written as: Given this, we constructed the unit root hypothesis below:

$$\begin{gathered} \Delta y_{it}={\alpha }_i+{\beta }_i{y}_{it-1}+{\gamma }_i{\mathrm{ƒ}}_t+{\epsilon }_{it} i=1,2\dots N and t=1,2\dots T \\ {H}_0={\beta }_i=0 for all i \left(series not stationary\right) \\ {H}_0={\beta }_i<0, i=1,2\dots N_1, {\beta }_i=0, i=N_1+1, N_2+2\dots N. \left(series stationary\right) \end{gathered}$$

(5)

CIPS statistics are constructed below:

$$CIPS=N^{-1}{{\displaystyle \sum }}_{i=1}^{N}CAD{F}_i$$

(6)

3.3.2. Instrumental Variables (2SLS)

IV-2SLS provides consistent estimates, assuming the explanatory variables are endogenous. So, the implementation will require selecting a set of instruments to ensure identification. Hence, the instrumental variables are exogenous and partially correlated with the endogenous explanatory variables [63]. The IV-2SLS can be expressed as:

$$\mathsf{\psi }\left(\mathsf{\beta }\right)={(\mathrm{y}-\mathrm{X}\mathsf{\beta })}^{\prime }\mathrm{Z}{(\mathrm{Z}\prime \mathrm{Z})}^{-1}{ \mathrm{Z}}^{\prime }\left(\mathrm{y}-\mathrm{X}\mathsf{\beta }\right)$$

(7)

where $\mathrm{Z}$ is the matrix of instruments, and $\mathrm{y}$ and $\mathrm{X}$ are dependent and explanatory variables, respectively.

Then the coefficients computed in two-stage least squares are given by:

$${\mathrm{b}}_{\mathrm{T}\mathrm{S}\mathrm{L}\mathrm{S}}={(\mathrm{X}\prime \mathrm{Z} {(\mathrm{Z}\prime \mathrm{Z})}^{-1}\mathrm{Z}\prime \mathrm{X})}^{-1} \mathrm{X}\prime \mathrm{Z}{(\mathrm{Z}\prime \mathrm{Z})}^{-1} \mathrm{Z}\prime \mathrm{y}$$

(8)

And the standard estimated covariance matric of these coefficients may be given as:

$${{\displaystyle \sum }}_{\mathrm{T}\mathrm{S}\mathrm{L}\mathrm{S}}^{{\displaystyle ^}}={\mathsf{\delta }}^2{(\mathrm{X}\prime \mathrm{Z} {(\mathrm{Z}\prime \mathrm{Z})}^{-1} \mathrm{Z}\prime \mathrm{X})}^{-1}$$

(9)

where ${\mathsf{\delta }}^2$ is the estimated residual variance (square of the standard error of regression).

3.3.3. Instrumental Variables (GMM)

According to Adams and Acheampong [64], the two-stage generalized method of moment (IV-GMM) model corrects the endogeneity problem and produces consistent results. The IV-GMM is a robust alternative model to autocorrelation and produces more reliable and efficient results in unaccounted heteroscedasticity. Therefore, we applied IV-GMM to correct endogeneity and unaccounted heteroscedasticity in models (4) and (5). The IV-GMM can be expressed given the dynamic endogenous growth specification of the form:

$$y_{it}={\alpha }_0{y}_{i,t}+{\alpha }_t\left(y_{i,t-1}-{y^T}_{i,t-1}\right)+{\beta }^{\prime }{X}_{it}+{\mathsf{\eta }}_i+{\xi }_t+{\epsilon }_{it}$$

(10)

Therefore, we use the lag components of regressors as instruments, and differenced transformation is given by:

$$\left(y_{it}-y_{i,t-1}\right)-\left(y_{it}-y_{i,t-2}\right)={\alpha }_0\left(y_{i,t-1}-y_{i,t-2}\right)+\beta \left(X_{it}-X_{i,t-1}\right)+\left({\epsilon }_{it}-{\epsilon }_{i,t-1}\right)$$

(11)

so that:

$$\Delta y_{it}=\alpha \Delta y_{i,t-1}+\Delta X_{it}{\beta }^{\prime }+\Delta {\mu }_{it}$$

(12)

3.3.4. Panel Driscoll–Kraay Standard Errors

We employ the Driscoll–Kraay standard errors [11] estimation technique to capture the autocorrelation, heteroskedasticity, and CD in the midst of missing data or unbalanced panel settings. Therefore, we construct OLS-Driscoll–Kraay standard errors estimation following a linear model expression below:

$$y_{it}=x_{it}\beta +{\epsilon }_{it}, i=1,2\dots N, t=1,2\dots .T$$

(13)

where $y_{it}$ is the dependent variable, while $x_{it}$ is a scalar and represents the independent variables with a (K + 1) × 1 vector. $\beta$ defines the coefficients with the (K + 1) × 1 vector, and $i$ is the cross-sectional units at time t.

Snowballing all other observations, the expression is as follows:

$$y={\left[y_{1t1,1}\dots y_1{T}_{1,} y_2{t}_{2 }\dots y_N{T}_{N,}\right]}^{\prime } and X={\left[x_{1t1,1}\dots x{T}_{1,} x_2{t}_2\dots x_N{T}_{N,}\right]}^{\prime }$$

(14)

Here, the assumption is that $x_{1t1}$ is not correlated or linked to the scalar error term ${\epsilon }_{it}$ for all s, t (strong exogeneity). However, ${\epsilon }_{it}$ may show heteroscedasticity, autocorrelation, and cross-sectional dependence. This postulation may only hold assuming β are always estimated using OLS regression [65].

Therefore:

$$\beta ={(X^{\prime }X)}^{-1} X^{\prime }y$$

(15)

For details, the coefficient estimates of the Driscoll–Kraay standard errors (DKSE) are calculated by taking the square roots (S^T) of the asymptotic covariance matrix elements following [11]:

$$V\left(\beta \right)={(X^{\prime }X)}^{-1}{S}_T {(X^{\prime }X)}^{-1}$$

(16)

4. Results

4.1. Pre-Regression Analysis

Table 2. Descriptive statistics.

Variable	Obs	Mean	Std. Dev.	Min	Max
Fsec	522	0	1.902	−18.128	3.341
Esec	522	−0.323	0.813	−4.647	0
Gdp	522	1.39	5.951	−47.503	37.535
Labor	522	6,070,325.8	7,974,551.1	385,654	53,950,175
CO2	522	1488.651	1908.175	150	16,280

, a description of the statistics, and the results show that the coefficient of food security is zero, suggesting that the selected countries have not achieved food security. The negative mean value of energy security (−0.323) shows that there is energy insecurity, while there is an average mean of 1.39% increase in gross domestic product (GDP) per capita per annum. Similarly, the labor force (6,070,326) participation is growing exponentially. However, the environmental quality is threatened by annual increases in CO₂ emissions (1488.651, metric tons).

Table 3. Pairwise correlations.

Variables	(1)	(2)	(3)	(4)	(5)
(1) Fsec	1.000
(2) Esec	−0.044	1.000
	(0.317)
(3) Gdp	0.225 ***	−0.078 **	1.000
	(0.000)	(0.073)
(4) Labor	0.380 ***	−0.351 ***	0.170 ***	1.000
	(0.000)	(0.000)	(0.000)
(5) CO2	0.434 ***	−0.247 ***	0.172 ***	0.861 ***	1.000
	(0.000)	(0.000)	(0.000)	(0.000)

*** p < 0.01, ** p < 0.05.

shows the pair correlation analysis, and the results show a positive relationship between energy security and food security. Similarly, GDP per capita and labor force are positively related to food security. However, CO₂ emissions are negatively related to food security, suggesting that CO₂ emissions reduce food security.

Table 4. Cross-sectional dependence and unit root tests.

	Pesaran CSD	CIPS
Variables	Statistical Value	Level	1st Different
Fsec	12.988 *** (0.000)	−3.058 ***	−5.595 ***
Esec	6.490 *** (0.000)	1.609	2.713 ***
Gdp	1.002 (0.3165)	−4.327 ***	−5.434 ***
Labor	−1.253 (0.2103)	−3.006 ***	−3.021 ***
CO2	0.356 *** (0.002)	−2.016	−4.817 ***

*** p < 0.01.

shows the cross-sectional dependence and unit root tests, and the results confirm the rejection of the null hypothesis of no cross-sectional dependence among the variables Fsec, Esec, Gdp, Labor, and CO₂. Given the cross-sectional dependence among the series, we tested the CIPS unit root test, and the results reject the alternate hypothesis of stationarity among the variables; Esec and CO₂ at a level value, except for the Fsec, Gdp; and labor at a 1% significant level. However, at the first-different value, we confirm the acceptance of the null hypothesis of stationarity for Fsec, Esec, Gdp, Labor, and CO₂ at a 1% significance level.

4.2. Main Results

Table 5. Instrumental variables (2SLS) regression.

Variables	Coef.	St.Err.	t-Value	p-Value	95% Conf	Interval
Esec	0.232 **	0.105	2.21	0.027	0.026	0.437
Gdp	0.05 **	0.013	3.93	0.001	0.025	0.075
Labor	0.001	0.001	0.48	0.632	0.001	0.001
CO2	−0.001 ***	0.001	−5.06	0.001	−0.001	−0.001
Constant	−0.666 ***	0.097	−6.84	0.001	−0.857	−0.475
Mean dependent var	−0.027		SD dependent var		1.905
R-squared	0.215		Number of obs		504
Chi-square	140.776		Prob > chi2		0.000

*** p < 0.01, ** p < 0.05.

shows the main results, and the results show that the coefficient of energy security (0.232 **) is positive and significant at a 5% level.

Similarly, the coefficient of GDP per capita (0.05 **) is positive and significant at a 5% level, an indication that GDP per capita increases food security in these states. This means that a one-point increase in GDP per capita will increase food security by 0.30% in the selected states. This is consistent with the assumption that income increases and increases household demand for goods, including food [66]. This is consistent with Candelise et al. [46], who found that income improves access to food on a dataset of 54 developing economies.

However, the coefficient of CO₂ emissions (0.001 ***) is negative and significant at a 1% level, suggesting that CO₂ emissions reduce food security in the selected states. On the other hand, an increase in CO₂ emissions causes food security reduction by over 1.91%. Land tenure richness is reduced due to the CO₂ emissions and ultimately affects food security. This is partly consistent with Koondhar et al. [67], who found that an increase in agricultural carbon emissions reduces the agricultural proceed of cereal.

4.3. Alternative Analysis

Although the two-stage least squares model corrects the endogeneity problem [68,69], to validate the findings in Table 5, we applied alternative instrumental variables GMM, and the results are presented in

Table 6. Instrumental variables (GMM) regression.

Variables	Coef.	St.Err.	t-Value	p-Value	95% Conf	Interval
Esec	0.385 ***	0.112	3.43	0.001	0.165	0.604
Gdp	0.066 ***	0.022	3.02	0.003	0.023	0.11
Labor	0.001	0.001	−0.55	0.582	0.001	0.001
CO2	−0.001 ***	0.001	−4.14	0.001	−0.001	−0.001
Constant	−0.776 ***	0.084	−9.29	0.001	−0.939	−0.612
Mean dependent var	−0.027		SD dependent var		1.905
R-squared	0.195		Number of obs		504
Chi-square	85.223		Prob > chi2		0.000

*** p < 0.01.

. The results show that the coefficient of energy security and GDP per capita is positive and significant, suggesting that energy security and GDP per capita are associated with increased food security. This is consistent with the main findings in Table 5. However, the coefficient of CO₂ emissions is negative and significant, indicating that CO₂ emissions reduce food security. This is not quantitatively different from the main findings in Table 5. Therefore, we conclude that our model is not suffering from an endogeneity problem, and the findings are robust compared to the alternative model.

4.4. Heteroskedasticity, Autocorrelation, and Spatial Correlation

Due to the peculiarity of the panel data, there are likely heteroskedasticity, autocorrelation, and spatial correlation problem in modelling [70,71,72]. Given these facts, we applied panel Driscoll–Kraay standard errors to correct the heteroskedasticity, autocorrelation, and spatial correlation problem, and the findings are presented in

Table 7. Regression with Driscoll–Kraay standard errors.

Variables	Coef.	Std.Err.	T	P > t	95%Conf.	Interval
Esec	0.163 **	0.094	1.720	0.096	−0.031	0.356
Gdp	0.049 **	0.024	2.030	0.052	−0.001	0.099
Labor	0.001	0.001	0.100	0.920	−0.001	0.000
CO2	−0.001 **	0.001	−2.070	0.048	−0.001	0.001
Constant	−0.675 ***	0.180	−3.750	0.001	−0.044	−0.306

*** p < 0.01, ** p < 0.05.

. The results show that the coefficient of energy security and GPD per capita is positive and significant, suggesting that energy security and GDP per capita are associated with increased food security. This is consistent with the main findings in Table 5. However, the coefficient of CO₂ emission is negative and significant, indicating that CO₂ emissions are reducing food security. This is not quantitatively different from the main findings in Table 5. Therefore, we conclude that there is no heteroskedasticity, autocorrelation, or spatial correlation problem in the model, and the model is fit.

5. Discussion

The study investigated the role of energy security on food security in the region of Africa. Interestingly, we found that energy security matters in eradicating hunger in Africa. This suggests that making energy available and affordable could affect agricultural production and distribution. This is partly consistent with Alsaleh et al. [47], given that sub-Saharan Africa is the global hotspot for poverty, with 80% of agricultural production coming from smallholder farmers. However, the major obstacle to agricultural productivity by the smallholder farmers is the lack of rural electrification [48].

Growing the food demand in the region, which was further worsened by the outbreak of COVID-19 and increasing climate change, has had dire consequences. Over 650 million Africans—50% of the population—lack economic or physical access to sufficient food to meet minimum needs daily [72]. Africa indeed is in dire need of strategies to boost its agricultural production. Chandio et al. [49] argued that energy for productive uses increases efficiency and crop yields through the mechanization of land clearing, preparation, and harvesting. Therefore, energy can be used to boost agricultural outputs, which is consistent with our findings.

To make energy accessible for agricultural production, there is a need for a constant energy supply known as energy availability. Africa as a region lacks sufficient electricity supply in most of the rural areas because it is capital-intensive. Therefore, the energy supply in Africa should not be left in the hand of the governments, but a public–private partnership should be created to feed all financial obligations for the expansion of energy infrastructure [48]. Furthermore, the smallholder farmers are a significant source of food in Africa and are considered the most disadvantaged population because of their income level. According to Harris et al. [73], an average smallholder farmer in sub-Saharan Africa lives on less than USD 1.90 per person per day, which is unlikely to them lift out of poverty. Invariably, smallholders are unlikely to afford energy that will help them improve their production capacity. Consistent with our findings, making energy affordable will help smallholder farmers to access energy easily.

Similarly, we found a positive relationship between GDP per capita and food security, suggesting that GDP per capita increases food availability in Africa. This is consistent with Cabdelise et al. [46], who found a positive relationship between income and access to food in 54 developing countries. This is possible because the higher the household income, the higher demand for food and food production. However, we found that there is a negative relationship between carbon emissions and access to food. This finding is consistent with Koondhar et al. [67], who found that agricultural carbon emission reduces the agricultural proceed of cereal. This is possible given the riches of natural resources that have been bestowed upon the African region. This prompted a significant increase in the exploration of natural resources and hampering the environment with carbon emissions rise. Therefore, it is expected that agricultural resources will be affected, and consequently, agricultural outputs might be reduced.

6. Conclusions and Policy Implications

In recent times, agricultural production in Africa has been lower than in the rest of the world, threatening the food security in the region. That mean that African countries have not produced enough food that can cater for the population. This is owing to many factors such as energy, income, labor, climate, soil quality, and disease [74,75,76]. Given these facts, this study examined the role of energy security in eradicating hunger in the region of Africa.

We applied two-stage least squares, instrumental variable GMM, and panel Driscoll–Kraay standard errors. This accounts for endogeneity, heteroskedasticity, autocorrelation, and spatial correlation. The two-stage least squares regression confirmed that energy security positively affects food security. Put differently, this means that energy security helps to eradicate hunger in African economies. No doubt, energy is a prime factor for agricultural production; however, energy is relatively expensive for the smallholder farmers who are major stakeholders in agricultural sectors in Africa. Therefore, energy provision and affordability matter for the farmers in this region. The cheaper the energy becomes, the more likely the smallholder farmers will purchase and produce food. Similarly, GDP per capita also helps to promote food security. On the other hand, this means that the share of household income influences food production and distribution. This is partly consistent with the notion that the higher the household income, the higher the demand for food, which directly causes the food production and supply to increase. However, the CO₂ emissions are affecting food security as well. This means that the results from instrumental variables GMM and panel Driscoll–Kraay standard errors regression confirm the robustness of the alternative model and suggest that our model is not suffering from endogeneity, heteroskedasticity, autocorrelation, and spatial correlation.

Our findings show that availability, affordability, accessibility, and acceptability of energy can help fix the growing agricultural production shortage in Africa. An important policy focus should thus be on achieving energy security. While further research might be necessary to prove the effectiveness of specific policy solutions to heighten energy security, its importance for food security is quite clear. Different policy options could be understood as starting points for further research aimed at specific countries or broader geographical areas.

Energy security could be achieved through competitive markets. Governments of the states of the global South could initiate monetary and fiscal policies that tend to lower interest rates to allow easy and cheap access to renewable energy loans and grants, which has a significant impact on the energy business. The governments could boost energy access and affordability by providing bailouts to stakeholders in energy sectors, creating subsidies, taxing the public, and giving the money to the companies or tariffs, reducing energy importation. Soft paternalist approaches such as nudging could be used to supplement other policy efforts to increase food security by increasing energy security.

With the effectiveness of policies aimed at energy security in mind, the governments in Africa and the global South more broadly can reasonably expect smart investments in energy security to have a positive effect and contribute to heightened food security, thus helping to ensure sustainable development and aid humanity in reaching the relevant UN Sustainable Development Goals.

Regarding the study’s limitations, we acknowledge that our sample size may be small. However, it does not represent a significant limitation, given that green energy adoption in Africa is still at an infant stage. We, therefore, consider our current sample size appropriate for providing insights into the benefits of energy security. The article did not explore the interrelationship of water security and energy security or armed conflicts concerning energy security. We believe future studies can use large data over the years to examine how energy security affects food security beyond African countries. Further research might also study the impact of energy security on water security using a similar dataset.