COVID-19 and Microeconomic Resilience in Sub-Saharan Africa: A Study on Ethiopian and Nigerian Households

Abstract

The severity of the COVID-19 pandemic on overall welfare depends on the resilience of microeconomic units, particularly households, to cope and recover from the shocks created by the pandemic. In Sub-Saharan Africa, where the pandemic has been less pervasive, the pandemic is expected to increase food insecurity, vulnerability, and ultimately poverty. To accurately measure the welfare impact of the pandemic on the macroeconomy, it is important to account for the distributional impact on households and the ability of households to cope with it, which reflects their microeconomic resilience. In this paper, we seek to determine the differential impacts of the COVID-19 pandemic on household microeconomic resilience in Sub-Saharan Africa. We use direct measurements of economic indicators to measure the impact of the pandemic on 6249 households across Ethiopia and Nigeria. Given that resilience is a latent variable, the FAO’s Resilience Index Measure Analysis (RIMA) framework is utilized to construct the resilience index. We hypothesize that the pandemic created differential economic impacts among households and ultimately household microeconomic resilience. Study findings show that government containment measures improved household microeconomic resilience, while self-containment measures lowered microeconomic resilience. Additionally, households that relied on wage employment and non-farm businesses as their main source of livelihood were found to be more microeconomic resilient.

1. Introduction

Most economic crises are caused by a demand shock, a supply shock, or a financial shock [1]. However, when a health crisis affects the economy—from individual households to businesses—and causes far-reaching negative consequences, this type of crisis can also be labeled as a shock to the system. The COVID-19 pandemic was a major health crisis that transitioned into an economic crisis due to the disruptions it caused to the functioning of the economy and the considerable adverse impact it had on human welfare [2,3,4]. In response to the COVID-19 pandemic, many countries scrambled to contain the spread of the virus and lessen the adverse health impacts of the disease by implementing policy measures that included strict lockdowns and travel restrictions [5,6,7]. Although the pandemic spread more slowly in low-income countries and there have been fewer reported cases and deaths in Sub-Saharan Africa than in other parts of the world [8,9,10], the region has not been spared from the negative impact of the pandemic [11,12]. The pandemic was expected to increase the pervasiveness of food insecurity, vulnerability, and ultimately poverty in low-income countries [13,14]. While high-income countries have sustained the majority of the COVID-19 related health burden [10,15], the pandemic posed a serious threat to millions of people in countries where the health systems are weakened, and the social protection systems are inadequate to protect its most vulnerable population from the harsh realities of economic shocks [16,17,18]. Coupled with the ongoing pandemic, countries in Sub-Saharan Africa were already impacted by other shocks. For example, Nigeria’s oil-reliant economy was dealing with a global decrease in oil prices, while Eastern Africa was experiencing its worst locust invasion in many decades.

Resilience is the ability of individuals, households, communities, or economies to lessen welfare losses following the occurrence of a shock or disaster. Generally, resilience is assessed in one of three ways: (i) the ability to cope; (ii) the ability to quickly recover; and (iii) the capacity to develop new adaptive abilities [19,20,21,22,23,24]. With two-thirds of the world’s poor residing in Sub-Saharan Africa, the COVID-19 pandemic is threatening the lives and livelihoods of millions of people. The severity of the pandemic on overall welfare depends on the ability of households to cope and recover from the shocks created by the pandemic. To accurately measure the welfare impact of the pandemic on the macroeconomy, it is crucial to measure its distributional impact on households and their ability to cope with it. This is important, because when the welfare impact of an economic shock on microeconomic units, such as households, is being accounted for, appropriate policy responses need to be devised and directed effectively to reduce the adverse effect on welfare [25]. Additionally, while poor people generally lose fewer assets than non-poor people during an economic crisis, the impact on their welfare is relatively worse and often lost in an aggregated economic analysis [26,27]. For example, a $100 loss means different things to a rich person and a poor person. Poor people are disproportionately impacted because they depend on fewer assets, they are more vulnerable, and they cannot fully rely on their adaptive capacity to smooth consumption and income. Thus, if the objective of economic analysis is to assess the welfare impact of shocks, such as a pandemic, on the microeconomy, then it would be a misrepresentation to focus mainly on economic aggregates who have enough wealth to lose instead of all economic agents.

There is a growing literature on the effect of the COVID-19 pandemic on African households using the lens of resilience. Much of the literature focuses on the resilience of local food systems and food security in Africa [28,29,30]. The focus of this study is on the economic resilience of households in Sub-Saharan Africa, an area that has received relatively little attention in previous research. Thus, our research is significant in the following ways. Firstly, we explore the impact of the pandemic on household microeconomic resilience among the world’s most vulnerable populations using real-time data. This is important because the severity of the pandemic on the economy depends on who is impacted the most. Secondly, by using the baseline data for our empirical analysis, our study captures a vulnerable time during the COVID-19 pandemic when most countries were scrambling to contain the virus and many individuals and households lost their source of income and livelihood due to the lockdowns, closure of businesses, and restricted travel. Lastly, we specifically identified the impact of the pandemic on households of varying income levels, employment status, and livelihood sources. It should be noted that we do not account for the macroeconomic impacts of the pandemic or the resilience of the broader economy. By isolating microeconomic effects, we focus on the inherent ability of households to cope with, and recover from, the impacts of the pandemic. Thus, this research seeks to answer the following question: what are the differential impacts of the COVID-19 pandemic on household microeconomic resilience in Sub-Saharan Africa? To answer our research question, we rely on direct measurements of economic indicators to measure the impact of the pandemic on 6249 households across Ethiopia and Nigeria, as well as the households’ ability to cope with the economic shock created by the pandemic. We propose an empirical framework to test the microeconomic resilience of households, which relies on the distribution of losses across households, their access to basic services, and their ability to smooth the shock through social protection, savings, borrowing, and other coping mechanisms.

The economic impact created by the COVID-19 pandemic can keep people trapped in a vulnerable state and move them back into poverty cycles. This is one of the reasons why eradicating poverty in developing countries is difficult. Hence, the need to understand the impact of the pandemic on households and their ability to cope with it is, and will remain important, especially in developing countries, as these countries are likely to have limited access to the COVID-19 vaccine compared to developed countries. This deficiency will have a long-term impact on socioeconomic recovery in developing countries by worsening global inequality and reversing decades of hard-won progress made in economic development.

2. The Resilience Construct

The concept of resilience has gone beyond its initial use in ecology [31] and has been developed further by scholars in different disciplines and applied in different fields, including economics, engineering, and agriculture. The concept has also been incorporated into many policy conversations surrounding sustainable development, economic development, disaster, and risk management [32,33]. It provides a practical framework in which models that can improve society’s ability to cope with shocks can be formed [20]. Although the concept of resilience is widely used among scholars, there are variations in its interpretations due to its abstractedness and the difficulty of operationalizing it [20,32,34,35,36].

A few studies have used the term “economic resilience” in explaining the resilience of economic systems. Economic resilience is the ability of an economic system to maintain a current economic state or recover quickly to its pre-existing level of growth or output after being impacted by some type of external shock [34,37,38,39,40]. Economic literature highlights two types of resilience constructs: ‘equilibrium’ and ‘non-equilibrium’, which are also referred to as ‘engineering’ and ‘ecological’ resilience [32,41]. The notion of equilibrium or engineering resilience describes the ability of a system to absorb shocks that can steer it away from its growth path or the ability to return to its pre-existing state or growth path it would have been in had the shock never happened, so that the shock has no long-term impact on the system or its institutions [22,32,33]. Non-equilibrium or ecological resilience, also known as complex or adaptive resilience, represents the ability of a system not to return to its pre-existing non-equilibrium state following a shock but to transform to a better growth path. Thus, resilience is not only about the ability to absorb shocks (resistance), but also about the ability to harness the new opportunities that arise from the shock, to transform (recovery), reorganize changed system structures and institutional arrangements (reorientation), and the development of new trajectories (renewal) [42].

2.1. Economic Resilience

Economic resilience can be a combination of macro and micro-economic resilience. Microeconomic resilience focuses on individuals, households, and firms. Macroeconomic resilience focuses on a combination of all economic units.

Generally, the impact of an economic shock depends on the number of shock occurrences and the length of exposure to the shock. Economists have long understood the importance of disaggregating economic impact into microeconomic and macroeconomic analysis, for several reasons. Firstly, the macroeconomy is comprised of microeconomic units making economic decisions, all of which interact in the broader economy. Secondly, the impact of these economic decisions is best tackled at the most fundamental level. Macroeconomic resilience is determined by the ability of microeconomic units—households and firms—to cope with shocks, the availability of resources, and the ability to mobilize these resources for rebuilding quickly. Macroeconomic resilience is the ability of a country to endure shocks and recover quickly from them [43]. Prior studies have used a variety of macro indicators to measure resilience, such as fiscal deficit to GDP ratio, unemployment and inflation rates, external debt, government size, trade openness, and property rights [21,34,37,42]. Microeconomic resilience focuses on the ability of economic units within the system to cope with and recover from economic shocks. Microeconomic resilience is determined by pre-capita income, susceptibility to shocks, the distributional impact of losses across households, the ability to smooth income, access to basic services, and the social protection system available in the economy. Economic shocks have considerable adverse impact on welfare depending on the magnitude of the shock and susceptibility to the shock; and for households, consumption matters [26,44]. Thus, the impact on household welfare can be minimized by reducing the susceptibility of households to economic shocks.

2.2. Measuring Economic Resilience

There are several challenges to measuring economic resilience [45]. Firstly, resilience is a combination of adaptive measures and real economic impacts, both of which depend on the severity of the shock and the pre-existing conditions of the economy [46]. Secondly, resilience is dynamic because it captures attributes and actions that pre-exist and ensue before, during, and after a shock [47]. Lastly, resilience is determined by individual socioeconomic attributes and behavior, which are often unknown until faced with a shock. Thus, any attempt to frame a resilience construct should consider its multifaceted nature and heterogeneity in individual socioeconomic factors [45]. Several studies have attempted to measure economic resilience empirically. The authors in [48] used the results of a survey questionnaire to examine resilience to water service disruptions. In doing so, they examined two types of resilience: inherent, which is individual firms’ ability to substitute other inputs for the scarce input (water), and adaptive, which is the ability to conserve the scarce input (water). A ‘refined’ computable general equilibrium (CGE) model was employed to measure the sectoral and macroeconomic impacts of a potentially destructive earthquake in Portland, Oregon. It was found in [48] that the micro impacts of resilience to water service disruptions feeds into the macro impacts through price and quantity effects. The resilience index developed by [49] described four elements that make up regional economic resilience—resistance, recovery, reorientation, and renewal. Other scholars have used the work of [49] in evaluating regional resilience. The authors in [35] used a modified version of the resilience/sensitivity index developed by [49,50] to study regional economic resilience in the wake of major earthquakes in Japan. The modified version [35], measured resilience as the change in the employment level of the prefecture in the year the earthquake occurred and the preceding year, relative to the change in employment for all selected prefectures in the same period. The authors in [35] found that prefectures that had the most resilience were also likely to have the strongest recovery, while prefectures with the least resilience were more likely to have weak recovery. While these approaches present a practical way of measuring economic resilience at the business and regional levels, they do not take into account economic resilience at the household or individual level, and as such do not fit the scope of our research.

Another branch of literature considers an economic framework for selecting proper indicators of an effective microeconomic resilience index. An example is [26], who used economic models and processes to uncover the channels through which natural disasters lead to welfare losses. While innovative, the Hallegatte framework has several weaknesses. Firstly, the framework focused only on the welfare impact of the areas affected by a natural disaster, ignoring unaffected areas that may incur spillover economic impacts from the affected area. Secondly, the framework starts with an “ex-ante welfare impact” equation that incorporates risk and vulnerability and then extends to an “ex post welfare impact” with microeconomic resilience as the last component. This intricate top-down process of measuring economic resilience is difficult to conceptualize empirically. Lastly, the framework includes indicators for measuring economic resilience that can be difficult to obtain for developing countries. A lack of data would require the use of multiple proxies to construct a resilience index which can lead to misrepresentations.

A more applicable approach to measuring household microeconomic resilience is the Resilience Index Measure Analysis (RIMA) developed by [51]. The FAO’s approach uses latent variable models to measure household resilience as a function of several observable household characteristics. For example, [52] utilized the FAO’s RIMA model in estimating the impact of household resilience on future food security in Tanzania and Uganda. The authors in [52] also tested the four dimensions of the RIMA to see which dimension was a major contributor to household resilience. The study found adaptive capacity to be the primary contributor to household resilience. Moreover, resilience decreased the likelihood of being food insecure if a shock/disaster occurred and also sped up household recovery in the post-disaster period.

The main takeaway from the studies examined in this section is that measuring economic resilience should be linked to the specific shock and context being studied. Additionally, the methodologies and frameworks employed should be adjusted to reflect the variables that will be utilized to define the resilience index [53]. Thus, this paper utilized the RIMA framework to construct the resilience index used in this study. To construct the resilience index, we represented each of the four resilience indicators with variables from our dataset. Each of these resilience indicators contributes to households’ capacity to absorb and recover from shocks. We provide a more detailed description of the RIMA framework in Section 4.

3. Conceptual Framework for Microeconomic Resilience

We present a conceptual framework based on the FAO’s RIMA model. The RIMA methodology consists of the following four “pillars” or indicators that influence household resilience: Access to Basic Services (ABS), Assets (AST), Social Safety Nets (SSN), and Adaptive Capacity (AC).

Because resilience is a multidimensional construct that cannot be easily observed or directly measured, it must be measured via proxies [52]. Furthermore, because resilience refers to an individual’s or household’s response when faced with a shock, the concept is an outcome construct—that is, being the outcome of a measure of welfare [52]. This conceptualization lies at the center of the fundamental economic problem, such that, when faced with scarcity, it is crucial to utilize the available resources efficiently at any given point in time during the crisis and post-crisis period. Thus, ref. [52] lists three principles that must be considered when attempting to measure resilience:

• Resilience is an outcome variable that measures the ability of an “agent”, such as a given household, to maintain function when faced with a shock;
• Resilience is not a static construct, that is, it varies over time; thus, any empirical assessment must be time-related, and a proper time period must be established;
• The empirical framework should capture all the agent’s possible channels to resilience, even though these channels may differ across agents. Thus, the empirical framework must account for heterogeneity in agents’ ability to cope with the shock.

We hypothesize that the pandemic created differential economic impacts among households, which in turn affected household microeconomic resilience. It is assumed that households have no control over exogenous shocks and respond to shocks using available adaptive mechanisms. Due to the heterogeneity in direct household losses, we assume that certain attributes make any given household more resilient than other households even though they are all faced with the same shock. Thus, it is important to identify how these attributes contribute to resilience.

4. Data Collection and Methods

4.1. Data Description

The study utilized longitudinal household survey data from the COVID-19 High-Frequency Phone Survey of Households (HFPS-HH) conducted by the World Bank during the pandemic. The phone interviews were tailored to fit the changing nature of the pandemic. The goal of the phone interviews was to observe the socioeconomic effects of the pandemic in real time and use the data to help design policies to mitigate the negative impacts of the pandemic. Survey responses were recorded as discrete (binary and categorical data), and respondents were asked to respond to a set of ‘Yes or No’ options or different categories depending on which question was asked. Survey questions included household respondents’ location, knowledge about the pandemic, access to basic needs, employment status, type of household income sources, food insecurity, and coping mechanisms. The phone interviews started in April 2020 and followed monthly subsequently, ending in April 2021 for Nigeria and in June 2021 for Ethiopia. The sample of households was drawn from the World Bank Living Standards Measurement Study–Integrated Surveys on Agriculture (LSMS-ISA) initiative. These include the Ethiopia Socioeconomic Survey (ESS) and the Nigeria General Household Survey (GHS). The survey responses are anonymized and available through the World Bank Microdata Library.

4.2. Country Selection

Two countries—Nigeria and Ethiopia—were selected because (i) they are the first and second most populous countries in Africa and (ii) Ethiopia has the most comprehensive social protection program in Sub-Saharan Africa. Thus, including both countries in the analysis provides an effective sample of the socioeconomic implications of the pandemic on households in the Sub-Saharan region. Additionally, the selection of both countries was based on data availability at the time of the initiation of this study. Since then, additional survey rounds have been conducted and made publicly available for both countries and other Sub-Saharan African countries including Burkina Faso, Chad, Kenya, Malawi, Mali, and Uganda. Given that most countries in the Sub-Saharan region face similar economic issues, the results derived from this study may be applicable to other Sub-Saharan African countries. By accounting for the differential impacts of the pandemic on Sub-Saharan households, governments and policymakers are better able to understand the welfare conditions of their citizenry and thus, make well-informed, data-driven policies.

4.3. Data Collection

The surveys were intended to be representative of the population at all levels—national, regional, and local (urban and rural). Phone survey interviews were conducted monthly for 12 months. Each month, households were asked a series of essential questions on the pathways through which the pandemic is expected to impact households.

In Ethiopia, the sample of HFPS-HH was drawn from the 2018/2019 Ethiopia Socioeconomic Survey (ESS). The ESS is a representative sample of all Ethiopian households at the national and regional levels. In the ESS, survey interviews were conducted for 6770 households in urban and rural areas. Households were asked to provide their phone numbers or a reference phone number household they could be contacted on if they were unreachable. At least, one phone number was obtained for 5374 households—4626 own phone numbers and 995 reference phone numbers. These households formed the sampling frame for the HFPS-HH.

In the baseline round (round 1), all 5374 households were called to account for attrition. There were 2184 households in the rural areas, and 3437 households in the urban areas. A total of 3249 households (978 in rural areas, 2271 in urban areas) were completely interviewed in round one (Appendix A).

In Nigeria, the HFPS-HH is referred to as the COVID-19 National Longitudinal Phone Survey 2020 (NLPS). The sample of NLPS was drawn from Wave 4 of the 2018/2019 Nigeria General Household Survey-Panel (GHS-Panel). The GHS is representative of the six geopolitical zones that make up the country. A total of 4976 households were interviewed in Wave 4 of the 2018/2019 GHS-Panel. About 90 percent of these households provided at least one phone number while others provided reference phone numbers for individuals who maintain close contact with the household to help in locating households who may have moved in subsequent survey rounds. To establish the sampling frame of the NLPS, a total of 3000 households were selected from the 4934 households that made up the GHS-Panel.

In the baseline round (round 1), all 3000 households were contacted (2033 in rural areas, 967 in urban areas). Overall, 1950 households were completely interviewed in round one (Appendix A).

4.4. Variables and Estimation Model

The RIMA framework above (Figure 1) is useful in identifying a structural equation approach (four-step procedure) for constructing the resilience index. Resilience is itself a latent variable. Thus, to measure resilience, we separately measured the resilience indicators. In the first step, household observable variables were selected for each resilience indicator or “pillar”. In the second step, factor analysis was used to transform the data so that an array of observed variables could be summarized in one of the four “pillars”. We ran the factor analysis using the principal component factor method to derive the factor loadings for the original variables. Varimax rotation was applied to rotate the factors so that they were more interpretable. This was done by maximizing the distance between factors orthogonally so that they were uncorrelated with each other. Factor scores (weights) for the variables were derived from the factor loading. We then multiplied the weights for each variable with the value of the original variable and summed these up to create each resilience indicator, such that, each indicator was a standardized weighted linear combination of its weighted averages and its components. In the third step, the indicators were summed up to create the resilience index. The process of selecting variables for making up each indicator in the resilience framework can be somewhat complex due to resilience being a multidimensional concept, such that an individual variable can apply to more than one indicator. This issue, however, was simplified in our framework where each resilience indicator is a measurement of specific household microeconomic characteristics. In the fourth and final step, a regression model was used to measure the impact of the independent variables on household microeconomic resilience.

4.4.1. Access to Basic Services (ABS)

The generation of the variables in the ABS indicator was based on the responses to these questions on access to basic needs: “In the last week, has your household been able to buy enough of [ITEM]?” to which survey respondents had to answer ‘Yes or No. Some of the selected variables include the following: (i) able to buy enough medicine; (ii) able to buy enough food; (iii) children engaged in any education or learning activities during the outbreak; (iv) access to financial services.

ABS_i = _ωabs_1_i + _ωabs_2_i + _ωabs_3_i + …………. + _ωabs_n_i

(1a)

4.4.2. Assets (AST)

There was no information available on assets in the COVID-19 High-Frequency Phone Survey of Households (HFPS-HH) dataset. Therefore, we decided not to use any proxies for it, as doing so may have led to biased estimates of the true relationship between household resilience and the other independent variables included in the study.

4.4.3. Social Safety Nets (SSN)

The generation of the variables in the SSN indicator was based on the response to this question on social safety nets: “Since [OUTBREAK_MONTH] has any member of your household received any assistance from any institution such as the government, international organizations, religious bodies in the form of?” to which survey respondents answered ‘Yes or No’. The SSN indicator includes the following variables: household received any assistance in the form of (i) free food; (ii) cash or food for work; (iii) direct cash transfers (iv) none.

SSN_i = _ωssn_1_i + _ωssn_2_i + _ωssn_3_i + _ωssn_4_i

(1b)

4.4.4. Adaptive Capacity (AC)

The generation of the variables in the AC indicator was based on the response to this question: “How did your household cope with the LOSS?” to which survey respondents answered ‘Yes or No’. The variables in the AC indicator included the following: (i) sale of assets (ag and non-ag); (ii) engaged in additional income-generating activities; (iii) received assistance from friends and family; (iv) borrowed from friends and family; (v) took a loan from a financial institution; (vi) credited purchases; (vii) delayed payment obligations; (viii) sold harvest in advance; (ix) reduced food consumption; (x) reduced non-food consumption; (xi) relied on savings; (xii) received assistance from NGOs; (xiii) took advanced payment from employer; (xiv) received assistance from government; (xv) was covered by insurance policy; (xvi) did nothing.

AC_i = _ωac_1_i + _ωac_2_i + _ωac_3_i + …………. + _ωac_13_i

(1c)

In the third step, microeconomic resilience was constructed as a linear combination of the resilience indicators [51,52,54] listed above such that:

$$R_i^{\mathit{micro}}{=A\hat{B}S}_i{+S\hat{S}N}_i{+A\hat{C}}_i$$

(2)

where;

R_i^micro = Microeconomic resilience; ${A\hat{B}S}_i$ = Access to Basic Service; ${S\hat{S}N}_i$ = Social Safety Nets; ${A\hat{C}}_i$ = Adaptive Capacity.

Lastly, in the fourth step, we estimated the following empirical model:

R_i^micro = β₀ + β₁MeC_i + β₂X_i + ε_i

(3)

where:

R_i^micro represents household microeconomic resilience. MeC is a vector of household microeconomic characteristics such as employment status, type of employment, current employment activities, source of livelihood, etc. X is a vector of household control variables.

4.5. COVID-19 and Microeconomic Resilience

In a bid to contain the spread of the COVID-19 virus, many countries in Sub-Saharan Africa implemented strict lockdown measures and curfews and imposed the closures of schools and businesses. About half of the working population in Sub-Saharan Africa work in sectors that were affected by the lockdown [44]. Many of these workers operate in smaller firms, earn lower wages, work under informal conditions, and thus do not have the resources to cope with extended periods of lockdown [15,55]. To determine the prevalence of income changes, households were asked whether their income had changed since the outbreak in 2020.

We note that if someone has experienced a loss of income from a particular source, it implies that they had been earning income from that source prior to the survey. Additionally, not all income loss was caused by the pandemic. Survey results show in Nigeria that about 75% of the sampled population were employed before the outbreak in mid-March (Figure 2a); 73% of the population live in households that have lost income since the outbreak (Figure 2b), and 85% report losing their jobs due to the pandemic (Figure 2c). Compared to Ethiopia where 37% of the sampled population were employed before the outbreak (Figure 2a); 50% of Ethiopian households had lost income (Figure 2b), while 61% reported losing their jobs due to the pandemic (Figure 2c).

Employment and income changes can have a dire impact on the household economy. Of particular concern is food security. Estimates show that in 2020 about 2 billion of the global population did not have access to adequate food—an increase of 320 million people from 2019 [56]. The severity of food insecurity is worse for the world’s poor and those who are vulnerable and is likely to worsen further due to the socioeconomic impacts of the pandemic. Figure 3 shows the incidence of food insecurity as measured by the Food Insecurity Experience Scale (FIES) at the outbreak of the pandemic.

In Ethiopia, male-headed households tend to be more food insecure than female households, and food insecurity is more prevalent in urban areas than rural areas (Figure 3a). In Nigeria, compared to urban households, more rural households tend to go a full day without eating, ran out of food or skipped a meal due to a lack of resources or money (Figure 3b). Income shocks do not automatically translate into significant spending shocks, as many individuals will rely on saving for consumption smoothing [44]. Figure 3c shows that across countries, a majority of the households did nothing to smooth their consumption to cope with the income shock resulting from the pandemic. However, households were more likely to rely on their savings or reduce nonfood and food consumption than other income loss coping strategies.

Figure 4 shows the contribution of the RIMA pillars to household microeconomic resilience. In Ethiopia, Access to Basic Services (ABS) contributed substantially to the ability of a household to absorb and recover from the shock created by the pandemic. Adaptive Capacity (AC) played a less important role, while the role of Social Safety Nets (SSN) in household microeconomic resilience is insignificant. In Nigeria, Adaptive Capacity (AC) contributed significantly to microeconomic resilience. Social Safety Nets (SSN) appeared to be the least important to microeconomic resilience.

5. Empirical Analysis and Results

The results for our estimation of Equation (3) are presented in

Table 1. Benchmark regression results.

	(1)	(2)
	Ethiopia	Nigeria
Currently employed	2.135 (1.551)	0.999 (0.883)
Able to work	2.162 * (1.253)	-
Currently employed * Able to work	−1.792 (1.684)	-
Food insecurity	−0.361 *** (0.113)	0.372 (0.255)
COVID-19 threat perception to household finance 1	-	−1.679 (1.228)
Possibility of falling sick 2	-	3.145 *** (0.639)
Controls
Self-containment measures	−0.174 (0.280)	−0.133 *** (0.037)
Government containment measures	−0.033 (0.071)	0.740 *** (0.100)
Knowledge of COVID-19	2.345 (1.573)	-
Knowledge * Satisfaction with government response 3	-	−5.243 *** (1.031)
Area	3.875 *** (0.665)	−0.398 (0.795)
Head of Household Gender 4	−0.562 (0.640)	-
_const	−5.456 *** (1.922)	4.190 *** (1.563)
R2	0.06	0.07
Adjusted R2	0.05	0.07
Obs.	857	1902

Note: OLS results with robust standard errors in parentheses. Dependent variable is household microeconomic resilience. *** p < 0.01, * p < 0.10. Footnotes ¹, ² and ³: Source data only available for Nigeria. Footnote ⁴: Source data only available for Ethiopia.

,

Table 2. Regression results—Type of household employment.

	(1)		(2)
Type of Employment	Ethiopia	Type of Employment	Nigeria
Government	−1.140 ** (0.457)	Own non-farm business	4.866 ** (2.088)
Public	2.201 (1.575)	Other non-farm business	0.314 (1.318)
Private	−0.449 (0.540)	Family farm	−1.930 *** (0.645)
NGO	4.244 ** (2.152)	Private employee	0.751 (1.372)
Domestic	−0.876 (1.215)	Government employee	0.817 (2.327)
Self-employed	−0.280 (0.424)	Apprentice	0.904 (2.278)
Unpaid worker	0.079 (1.232)	-	-
Employer	2.328 (2.237)	-	-
Cooperative	0.293 (2.594)	-	-
Casual worker	−1.012 (1.177)	-	-
Knowledge of COVID-19	3.508 *** (1.083)	Knowledge of COVID-19 * satisfaction with government response	−4.936 *** (0.998)
Government-containment measures	0.044 (0.028)	Government-containment measures	0.734 *** (0.099)
Self-containment measures	−0.509 *** (0.163)	Self-containment measures	−0.120 *** (0.035)
Area	2.782 *** (0.390)	Area	−1.048 (0.788)
Gender	−0.003 (0.351)	-	-
_const	−3.186 *** (1.079)	_const	5.602 *** (0.962)
R2	0.03	R2	0.08
Adjusted R2	0.02	Adjusted R2	0.07
Obs.	3222	Obs.	1915

Note: OLS results with robust standard errors in parentheses. Dependent variable is household microeconomic resilience. *** p < 0.01, ** p < 0.05, * p < 0.10.

,

Table 3. Regression results—The main activity of household employment.

	(1)	(2)
Employment Activity	Ethiopia	Nigeria
Agriculture	−0.756 (0.500)	−1.283 * (0.716)
Manufacturing	0.305 (0.791)	−1.690 (4.510)
Electrical 5	-	1.586 (2.967)
Trade	−0.181 (0.699)	2.757 * (1.580)
Transportation	−0.360 (0.915)	−0.004 (1.205)
Hospitality Service	−0.938 (1.037)	-
Government	−1.186 ** (0.484)	2.818 (4.685)
Personal services	0.490 (0.737)	3.531 (3.448)
Construction	0.423 (0.875)	3.513 (2.927)
Education and Health	1.385 (1.235)	-
Professional activities 6	-	5.553 (3.806)
Controls	Yes	Yes
_const	−2.924 *** (1.062)	5.646 *** (0.974)
R2	0.03	0.07
Adjusted R2	0.02	0.07
Obs.	857	1915

Note: OLS results with robust standard errors in parentheses. Dependent variable is household microeconomic resilience. *** p < 0.01, ** p < 0.05, * p < 0.10. Source data for Nigeria groups trade and hospitality services into one variable. Footnotes ⁵ and ⁶: Source data only available for Ethiopia.

and

Table 4. Regression results—Household source of livelihood.

	(1)	(2)
Household Source of Livelihood	Ethiopia	Nigeria
Farm business	−2.176 *** (0.390)	2.135 *** (0.627)
Non-farm business	0.811 ** (0.408)	3.997 *** (0.698)
Wage employment	0.814 ** (0.361)	1.604 * (0.941)
Domestic remittances	−0.564 (0.565)	−1.413 (1.204)
Foreign remittances	−1.547 ** (0.616)	−2.754 ** (1.261)
Investment and Savings	−0.889 * (0.516)	−1.851 * (1.076)
Pension	−1.397 ** (0.596)	0.685 (1.665)
Government	0.979 (0.650)	5.980 (5.914)
NGO/Charity	−0.050 (1.176)	10.402 (7.511)
Other 7	-	8.116 (7.929)
Non-family 8	-	2.666 ** (1.132)
Controls	No	No
_const	2.172 *** (0.337)	−1.388 (0.970)
R2	0.03	0.04
Adjusted R2	0.02	0.04
Obs.	3246	1960

Note: OLS results with robust standard errors in parentheses. Dependent variable is household microeconomic resilience. *** p < 0.01, ** p < 0.05, * p < 0.10. Footnotes ⁷ and ⁸: Source data only available for Nigeria.

. To identify the impact of the COVID-19 pandemic on household microeconomic resilience, we regressed several independent variables on microeconomic resilience using the Ordinary Least Square (OLS) method. The independent variables included in the regression models are employment status, type of employment, current employment activities, and source of livelihood. The independent variables were selected following examples from previous literature on economic resilience [35,49] and in relation to economic principles. For example, it is very unlikely that an economy facing some kind of shock will create more jobs and new income-earning opportunities. A shock to the economy results in a decline in employment and economic activities, leading to a reduction in income and potential income-earning opportunities. These knock-off effects are likely to impact household microeconomic resilience. Control variables included containment measures taken by households and the government, knowledge of the COVID-19 virus, and demographic information such as head of household gender and area (urban/rural). Head of household gender was coded 1 if the respondent was male and 0 otherwise. Area was coded 1 if the household lives in an urban area. We constructed the food security variable and the COVID-19 containment measure variables using responses to the survey questions. For the food insecurity variable, survey respondents had to answer ‘Yes or No’ to these questions: ‘Have you or any other adult in your household had to skip a meal?’; ‘Have you or any other adult in your household ran out of food?’; ‘Have you or any other adult in your household gone without eating for a whole day?’. For the government-containment measure variable, respondents were asked ‘What steps has the government/local authorities taken to curb the spread of the coronavirus in your area?’. The respondents were then asked to indicate ‘Yes or No’ on a series of responses which included “advised citizens to stay home, advised to avoid gatherings, restricted travel within the country, restricted international travel, closure of schools, curfews/lockdown, closure of non-essential businesses, provide food to needy”. For the self-containment measure variable, respondents were asked what measures they had taken to reduce the risk of the pandemic. The respondents were then asked to indicate ‘Yes or No’ on a series of responses which included avoid travel, staying at home, and avoid gatherings. The components of each of these variables—food insecurity, government-containment measure, and self-containment measures—were combined, respectively, using factor analysis and extracting the principal components of all the responses to these variables.

Table 1 presents two sets of results that show the effect of current employment status, food insecurity, and household control variables on household microeconomic resilience. Due to the lockdown restrictions and closure of most businesses in the wake of the COVID-19 pandemic, many workers were unable to work normally from their workplaces or homes. Thus, being currently employed does not guarantee that a person can work as usual. To identify the effect of being currently employed and able to work as usual on household microeconomic resilience, an interaction term was included in the benchmark model. Data on the threat perception of COVID-19 to household finances, the possibility of falling sick with COVID-19, and satisfaction with government response to the pandemic was only available for Nigeria. The results suggest that in Ethiopia, being able to work as normal from either the workplace or from home significantly improved household microeconomic resilience in the wake of the pandemic. However, the interaction of being currently employed and able to work as normal had no impact on household microeconomic resilience. Ethiopian households that faced food insecurity were less resilient to the shocks created by the pandemic compared to households that were food secure. While urban households were more resilient than rural households, the head of household gender had no significant impact on microeconomic resilience. In Nigeria, data on whether the respondent/head of household could work normally from either the workplace or home had too many missing observations, therefore, the “able to work” variable was omitted from the regression analysis. In Nigeria, there is evidence to suggest that concerns over the potential illness of a family member due to COVID-19 enhanced microeconomic resilience. This implies that many households in Nigeria prepared for ‘rainy days’ by engaging in alternative economic activities to minimize the impact of lost income resulting from a family member falling sick.

Although self-containment measures lessened microeconomic resilience, government containment measures seemed to increase it. These effects are consistent with government response to the COVID-19 pandemic in many Sub-Saharan African countries. Strict lockdown measures, curfews, and the closure of businesses affected about half of the working population in these countries; the majority of which work in the informal sector under informal conditions and earn low wages. Thus, self-containment measures by these workers lessened their microeconomic resilience because they do not have the resources or ability to handle long periods of lockdown. However, government containment measures including government fiscal and monetary policies utilized during the pandemic were instrumental in easing the impact of containment measures on microeconomic resilience, thereby increasing household microeconomic resilience. This finding is consistent with [55,57,58,59] studies on the economic effects of COVID-19 containment measures. Both studies showed larger short-term economic losses in countries that utilized fewer fiscal and monetary policies during the pandemic.

Table 2 disaggregates the employment variable by type of employment and allows a more specific assessment of the true impact of employment on household microeconomic resilience. Disaggregating the employment variable by types of employment seems to add considerable explanatory power to the model without very large differences in point estimates. In Ethiopia, government-employed households were significantly less resilient. The opposite effect can be observed with non-governmental employed households, which showed significantly higher microeconomic resilience. The coefficient signs on the control variables remained the same, and the statistical signs were as expected. In Ethiopia, having knowledge of COVID-19 was positively associated with household microeconomic resilience. As expected, engaging in self-containment measures to reduce the risk of contracting the COVID-19 virus was negatively associated with household microeconomic resilience. Consistent with the sign and statistical significance in the benchmark regression model, urban households were more resilient than rural households. In Nigeria, households that owned non-farm businesses displayed greater resilience, whereas households that relied on family farming were less resilient. Engaging in self-containment measures, being aware of COVID-19, and satisfaction with the government’s response to the COVID-19 crisis came at an opportunity cost that lowered microeconomic resilience. This finding suggests that Nigerian households, cognizant of the risks posed by COVID-19 and the government’s efforts to control its spread, were willing to adopt preventive measures to mitigate their exposure to the virus. Despite the resulting loss of income and work opportunities, this precautionary behavior ultimately reduced their microeconomic resilience.

Estimating the effects of current employment status and types in Table 1 and Table 2, however, obscures how different employment activities may be influencing results. The results in Table 3 indicate that Ethiopian households where the respondent/head of household was employed in an organization involved in governmental activities, exhibited lower resilience. Nigerian households where the head of household worked in a business involved in agricultural activities demonstrated lower microeconomic resilience. Conversely, households that engaged in trade (buying and selling of goods, repair of goods) and hospitality services showed the opposite effect, as their involvement in these sectors increased their microeconomic resilience. The reason behind these findings could be that when the government enforces stringent containment measures such as shutting down businesses, households employed by the government or those working in organizations associated with government activities are more likely to be adversely affected by such measures. Additionally, in most developing countries, the majority of the rural poor engage in agriculture. In these countries, agriculture is mainly subsistence and consists of smallholder farmers who are extremely vulnerable to the different shocks that directly or indirectly impact production or income generation.

The results in Table 4 show the impact of household source of livelihood on household microeconomic resilience. It is interesting to observe the similarities and heterogeneity in the effects of households’ livelihood sources on their microeconomic resilience. In both countries, households that rely on nonfarm business as the main source of livelihood were the most resilient to the shocks created by the pandemic. The same effect is present, only weaker (in Nigeria), with wage employment. Households that rely on foreign remittances and investment and savings were less resilient, an indication of the effect of restrictions on international travel. In Ethiopia, households that depend on farm businesses and pensions as their primary sources of livelihood also exhibited lower resilience in the aftermath of the pandemic. In contrast, farm businesses contributed significantly to improving household microeconomic resilience in Nigeria. A seemingly similar effect is also found in households that receive assistance from nonfamily members. The impact of farm businesses on microeconomic resilience among Nigerian households can be attributed to the youth ‘movement into agriculture’. To find a lasting solution to its unemployment problem, youth unemployment has become a vital component of Nigeria’s agricultural policy agenda. In Nigeria, where about 69 percent of the youth dwell in rural areas and depend on agriculture as their primary source of livelihood; agriculture accounts for over 70 percent of rural employment and over 85 percent of total rural income revenue [60]. Aside from perceiving agriculture to be profitable, many young men and women in Nigeria view agriculture as a means to earn an income and provide for their family expenses including children’s education, health expenses, etc. [61].

6. Discussion

Although a significant amount of work has been done on the economic impact of the COVID-19 pandemic, most of it has been studied from a global perspective. Few studies have examined the microeconomic impact of the pandemic at the household level. This study provides an empirical analysis of the impact of the pandemic on household microeconomic resilience in Sub-Saharan Africa. We note that these countries were already facing complex development challenges that existed before the pandemic and would probably persist after the pandemic. Our analysis focused on Ethiopia and Nigeria, and our findings showed that government containment measures improved household microeconomic resilience. Households that rely on wage employment and non-farm businesses as their main source of livelihood were found to have more microeconomic resilience. The impact of agriculture on household microeconomic resilience seemed to differ across both countries and two trajectories: (i) agriculture as a type of employment for the household, and (ii) agriculture as the primary source of household livelihood. As a type of household employment, agriculture lessened household microeconomic resilience, but as the primary source of household livelihood, agriculture increased microeconomic resilience in Nigeria and lessened microeconomic resilience in Ethiopia.

The impact of agriculture on household microeconomic resilience highlights the trade-off between subsistence farm work and agricultural wage work for income and consumption smoothing. The primary determining factor of this trade-off is the demographic composition of the household—the ratio of men to women and/or employed to non-employed household members—and the existence of an agricultural labor market in which the wage rate differs for different categories of labor, particularly between men and women [62]. Firstly, many Sub-Saharan African countries struggle with generating enough employment to absorb their excess labor supply. Hence, there is the creation of an agricultural labor market to absorb the excess labor supply and provide income and employment for the unemployed rural and urban poor [63]. Additionally, in some developing countries, land ownership can be particularly difficult for the rural and urban poor. Thus, it is not unusual to see “farmers or households with access to land employ non-working members of other households to work on their farms” [63] (p. 242). Although, even in developing regions where land ownership is common, many rural and urban poor households hire out household farm labor to work on other farms to earn more income. Secondly, different household members have different wage-earning potentials. Male household members are more likely to have a greater comparative advantage in agricultural wage employment than female household members. As a result, subsistence farm work is mostly carried out by women, children, and the elderly in the household. Nonetheless, in numerous developing countries, the agricultural labor market is underdeveloped, and the wages earned from agricultural employment are inadequate to bolster household microeconomic resilience. Such that it is more advantageous for households to engage in subsistence farm work to improve microeconomic resilience.

There are several implications for these results, all of which can be studied further. Firstly, there is the challenge of whether a single indicator can adequately reflect resilience at either the macroeconomic or microeconomic level. At the microeconomic level, resilience is largely determined by several practical measures. Consequently, microeconomic resilience is not merely a function of a single resilience measure, but several resilience measures implemented by both government, individuals, and households. However, given that resilience indices are somewhat novel and economic shocks sporadic, there are not many studies that empirically test the potential contribution this new construct might provide [64]. Secondly, some of the indicators included in the resilience index indicate pre-existing economic conditions that can be barely improved even in the aftermath of the shock. For example, the assets and adaptive capacity of households in low-income countries would probably remain the same or most likely decrease after recovery from the economic shock created by the pandemic. Further research is needed to assess the distribution of asset losses among households during the pandemic.

7. Conclusions

Our findings draw attention to some critical existing issues in many Sub-Saharan countries. For example, the labor market is often characterized by an “excess supply- limited demand” problem, that is, there are fewer jobs to go around than those who can work. In addition, there is a duality in the labor market in these countries. The formal sector determines the amount of skilled labor to hire for work; then any unhired labor by the formal labor market is absorbed by the informal sector, which consists of both smallholder farms and non-farm businesses. Our results present a perspective on microeconomic resilience that can be used by governments and policymakers to target resources and policies in the event of an economic shock. It also provides evidence for further research in the aftermath of the pandemic. The ability of these countries to be resilient will depend on the policies and strategies that prioritize the creation of jobs and entrepreneurial prospects, and long-term strategies that foster sustainable economic development. Governments and policymakers should focus on expanding cash transfer programs, providing grants for small businesses, income support, and improving the social protection systems. Additionally, policies that encourage smallholder farming in rural areas should be applicable and inclusive to urban and peri-urban areas to accommodate the working urban poor. As long as there exists a lag between labor supply and demand in these countries, both farm and non-farm businesses can boost microeconomic resilience for the unemployed and underemployed. Research has also shown that urban and peri-urban agriculture is an important component in reducing urban poverty and food insecurity.

Our study is subject to several limitations. Firstly, our research does not take into consideration vulnerability, which is a pre-existing condition that predisposes individuals and households to be adversely impacted by a shock [65]. Several of the indicators included in resilience indices are also included in vulnerability indices. Literature suggests that resilience is a by-product of vulnerability and one of the ways to reduce vulnerability [45,66,67]. Secondly, our empirical analysis is only limited to two countries. Thus, while the results from the analysis can be indicative of outcomes in similar countries, it is not entirely representative of the entire Sub-Saharan region. Thirdly, due to the lack of information on household assets in the HFPS-HH dataset, our resilience construct does not include assets. Households may use their assets in various capacities as part of a coping strategy to mitigate the impact of an economic shock. Households may decrease their consumption to conserve their assets—assets smoothing—or they may sell their assets to sustain consumption—consumption smoothing. Thus, the exclusion of assets in the resilience construct may underestimate household microeconomic resilience. Lastly, the R-square from the regression results are significantly low for each empirical model. While there is no universal criterion to determine a “good” or “bad” R-square, its interpretation can vary depending on the context of the analysis and the field of study. Although the R-square is a measure of the explanatory power of the model, it is not a measure of fit. Typically, a higher R-squared value implies a better fit of the model to the data, but it is important to consider other metrics of model quality to gain a more comprehensive understanding of its performance. In social sciences, it is implausible to include all the relevant predictors into a model to explain an outcome variable. This is because socioeconomic events are complex and multidimensional, and therefore, it is harder to come up with complete, well-defined models, making them more difficult to predict than physical processes.

Notwithstanding the limitations of this study, we have drawn important conclusions about the impact of the pandemic on household microeconomic resilience. We present results on the microeconomic burden of the COVID-19 pandemic in Sub-Saharan Africa, where the economic burden to protect its most vulnerable population from the harsh realities of economic shocks can be felt the most.