COVID-19 and food insecurity in Africa: A review of the emerging empirical evidence

Abstract

The coronavirus disease (COVID-19) risks rolling back many of the efforts and global successes recorded in reducing poverty and food insecurity. We undertake a systematic search and review of the growing microeconomic literature on the association between COVID-19 and food insecurity in Africa, discussing its implications for food policy and research. Furthermore, we review the various coping strategies households employ to build resilience to COVID-19. The evidence indicates that COVID-19 is associated with an increase in food insecurity both ex-ante and ex-durante. Given the covariate nature of COVID-19 and associated control mechanisms, current evidence is short of providing clear causal learning. We provide some potential interesting areas where future efforts can be geared to improve learning on the relationship between COVID-19, food insecurity, and building resilience to shocks.

1. Introduction

The Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), hereafter referred to as the coronavirus disease (COVID-19), continues to spread globally as there has been little success in effectively containing the virus. Originally described as a pandemic by the World Health Organization, there are emerging ideas of COVID-19 transiting from a pandemic to an endemic (Steere-Williams, 2022). This is not to say that its effects are getting mild or COVID-19 will come to a natural end as this assumption and endemic fatalism have been associated with misplaced complacency (Katzourakis, 2022). Different variants have been reported even with increasing vaccine rollout in many countries. There have been many talks on preventing the emergence of not only more dangerous but also more transmissible variants by ensuring vaccine equity for many developing countries through large-scale vaccination and other non-clinical preventive strategies. However, COVID-19 has so far been associated with economic downturns and poverty in many developing countries, especially in Africa (Laborde et al., 2020; Zeufack et al., 2020). Most African governments responded to COVID-19 by recommending and enacting lockdowns, travel restrictions, shelter-in-place, physical distancing measures, and some hygienic procedures to control the spread of the virus and save their health infrastructures (Durizzo et al., 2021). These containment measures have been associated with an increased risk of food insecurity, in a way that shifts focus to viewing COVID-19 as a hunger pandemic.

In this paper, we synthesise the growing empirical literature on COVID-19 and food insecurity in Africa, discussing its implications for research and food policy. In doing so, we review the various mechanisms in the way of this relationship. Particularly, we document the association between COVID-19, income and employment losses as well as price increases for various staple commodities consumed by poor households. We also explore the various ways households are coping with COVID-19-induced food insecurity such as reducing their consumption (food rationing) and relying on lower-quality diets (negative coping mechanisms). Relatedly, we highlight the role of informal support through remittances, emergency, and existing social protection interventions, and food stockpiling in sustaining household resilience during the pandemic.

This review underscores a strong association between COVID-19 and food insecurity in Africa. Various containment measures and policies have been associated with reduced food consumption and food insecurity (Bloem and Farris, 2022). In the early days of the pandemic, there were indications of resilient food systems as little or no changes in food consumption and household dietary diversity were observed in some countries like Ethiopia, Liberia, and Malawi (Aggarwal et al., 2022; Hirvonen et al., 2021). Over time, these associations have been outstandingly negative with insights from different countries and contexts. In some countries, there was also a locust outbreak that coincided with COVID-19 (Kassegn and Endris, 2021; Salih et al., 2020). We report significant income losses, which could explain these reductions in food security. Various containment policies have been associated with reduced income streams of many rural households that depend on numerous sources of income for their livelihoods. Similarly, work-related losses have also been reported in many countries (Bargain and Aminjonov, 2021). Many households in developing countries depend largely on hands-on labour income given the huge informal sector prevalent in these countries. Many individuals in many African countries have lost jobs due to lockdown restrictions and stay-at-home policies.

Besides income and employment-related losses, significant price effects have also been reported due to panic buying, hoarding, and stockpiling. We review and provide evidence that prices for many staple foods have heightened during the pandemic (Aggarwal et al., 2022; Agyei et al., 2021; Dietrich et al., 2022; Hirvonen et al., 2021). We then explore the various mitigating strategies households are using to cope with COVID-19-induced food insecurity. Households are using adverse food coping strategies such as reducing their food intake and relying on less nutritious foods (Tabe-Ojong et al., 2022). They are also relying on support from friends, family, and the government (Dasgupta and Robinson, 2021, 2022; Maredia et al., 2022). The support from family and friends takes the form of remittances (Akim, Ayivodji and Kouton, 2021). Remittances and social protection relief have been shown to reduce food insecurity arising from COVID-19 (Abay et al., 2021a; Akim, Ayivodji and Kouton, 2021). Furthermore, livestock and the adoption of improved storage technologies have been highlighted to be effective buffers in reducing food insecurity (Huss et al., 2021). We discuss these results and insights in light of food policy and research. Most studies employ state-of-the-art empirical methods to get at their findings, but data limitations prevent them from implying causality in the strictest sense about these relationships. That notwithstanding, pursuing the causal question is hard and may not be very important given that COVID-19 affected everyone somehow, making it almost impossible to obtain a valid counterfactual group. Documenting associations through a descriptive understanding of the changes in measures of food insecurity associated with the onset and/or continuation of the COVID-19 pandemic is sufficient given that the shock is large such that simple logic may be enough to connect the dots. Geographically, there seems to be multiple empirical evidence from some countries like Ethiopia, Kenya, Uganda, and Nigeria, while learning is limited in many countries in other regions of Africa such as Central Africa, North Africa, and some parts of Southern Africa.

This review offers the following contributions to the growing literature on the relationship between COVID-19 and food insecurity. First, it synthesises the microeconomic literature on COVID-19 and food insecurity, making it easy to learn from this relationship at a broader scale. While this review is not the first to synthesise the relationship between COVID-19 and food insecurity in the context of low-income countries, it is far more comprehensive in coverage, scope, and time than papers by Picchioni, Goulao and Roberfroid (2021) and Bloem and Farris (2022). Moreover, given the quick growth of evidence on this relationship, this review adds more empirical insights, exploring more thematic issues with a special focus on Africa where the infection rates are low as compared to other regions in the world. Second, we take a step ahead after documenting evidence of food insecurity by exploring and highlighting various channels in the way of this relationship such as income, employment, and prices. Third, we also review the various coping mechanisms households are using to reduce COVID-19-associated food losses and insecurity. We also consider the role of remittances and social protection in relieving households and building resilience. Finally, we identify critical gaps in the literature and discuss their implications for policy and research agenda-setting.

The rest of the article is structured as follows. Section 2 provides an overview of the pandemic in Africa and the various containment policies rolled out and enacted by various governments. Section 3 looks at the methodology of selecting various articles for the review and the road map. The findings of the review are presented in section 4, while section 5 delves into some aspects of mitigating COVID-19-induced food insecurity using instruments such as social protection, coping strategies and remittances. The article discusses these findings and offers some deep thoughts and ideas for both research and policy.

2. Coronavirus pandemic in Africa and containment measures

COVID-19 originated in China in late December 2019 and has since then spread around the world, causing a pandemic. While more cases per million have been observed in higher-income countries, Africa is equally affected, recording just under 12 million cases and over 250,000 deaths as of April 2022 (Hasell et al., 2020). The first infection in Africa was reported in Egypt on 14 February 2020, while the first death was reported on 27 March 2020 in Burkina Faso (Lone and Ahmad, 2020). Even as new evidence emerges that the majority of COVID-19 cases have not been reported due to low screenings (WHO, 2021), it is still arguably realistic that Africa has been relatively less affected by the pandemic in comparison to other regions. Confirmed cases account for only about 0.3 per cent of the global total and confirmed deaths account for 4.2 per cent of the global total as of August 2022 (Hale et al., 2021).

Nonetheless, at the onset of COVID-19, the response by some African countries was as strong as other countries even when they (African countries) had lower exposure levels given their lower international connections to China and therefore lower risk of importing early transmissions (Wu, Leung and Leung, 2020). African countries implemented some control measures, including lockdowns, closure of education facilities, cancellation of public events, curfews, and restrictions on domestic and international travel among others. Considerable variation across countries exists as to how much such measures were mandatory and how they were applied to the general population (Haider et al., 2020). Uganda reopened its schools in January 2022 almost 2 years after they were closed (Blanshe and Dahir, 2022), a COVID-19 prevention policy said to have been the longest of its kind, leaving extensive learning losses on schoolchildren (Sandefur, 2022).

The Oxford COVID-19 Government Response Tracker (OxCGRT) monitors the level of government responses across a variety of indicators and aggregates them into an index referred to as the stringency index. This index indicates the degree of stringency of different policy actions by various governments to contain the spread of COVID-19. The index value ranges from 0 to 100, with a higher value indicating greater limitations and vice versa. Additionally, the OxCGRT gives a summary of the overall government response, aggregating indicators across four dimensions (containment and closure policies, economic policies, health system policies, and vaccination policies). Figure 1 illustrates heterogeneity in the stringency index among African countries. In 18 of the 50 countries shown in Figure 1, the mean stringency index was above 50, implying some countries implemented strict measures. The majority of the countries implemented less strict measures, possibly due to the lower number of officially reported cases.

Despite the lower number of officially reported cases, the COVID-19 pandemic will most likely leave Africa in a much worse-off situation than many other regions. This is partly because of the existing fragile social support systems in most African countries which would have made scale-up not as comprehensive as higher-income countries. In a review of social protection coverage during the pandemic, (Gentilini, 2022) notes that while in high-income countries, emergency transfers reached up to 50 per cent of the population, in African countries, only about 10 per cent of the population received some form of emergency support. Growth projections indicate that between 2020 and 2022, economic growth in Africa would reduce by over 5 per cent with recessions expected in the mineral and oil–dependent countries (Zeufack et al., 2020). Furthermore, initial evidence indicates that foreign direct investments would also be negatively affected (Hayakawa et al., 2022; Moosa and Merza, 2022). Similarly, remittances to Africa have declined significantly (Ratha et al., 2020). All these compounding effects of COVID-19 are likely to leave Africa in more precarious conditions, worsening poverty and food security among other social development declines.

3. Methodology

We conduct a systematic search and review study (Grant and Booth, 2009). Given the global nature of COVID-19 and the (at the moment) almost impossible attempt to introduce clear variation in exposure to COVID-19, we do not attempt to conduct a systematic review with explicit motivations for evaluating causal claims. We aim to only provide the state of the art of what is known by utilising the available literature on the relationship between COVID-19, prevention mechanisms, and food security without imposing strict measures on quality and biases in the literature. A systematic search and review methodology enables us to achieve this. This review methodology enables us to combine the strength of a careful and robust search process to undertake a critical evaluation of the literature and provide the best synthesis of the literature. As such, we show the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram that portrays our overall search and inclusion strategy (Supplementary Material). However, as opposed to systematic reviews, we do not assess bias (such as sampling biases and internal validity) in the studies included.

3.1. Search strategy

3.1.1. Data sources

Our review relies on multiple sources. First, we searched for published papers indexed in two academic databases, namely Web of Science and Scopus. The two databases are the most comprehensive and precise social science databases for literature reviews (Gusenbauer and Haddaway, 2020). Next, we searched for grey literature in the form of published and unpublished working papers and reports. For these, we searched the websites of research organisations and institutions that have substantial coverage of agricultural issues in the African region. We thus scanned the websites of the World Bank (Policy Research Working Paper Series), the National Bureau of Economic Research (NBER Working Paper Series), the Institute for the Study of Labour (IZA Discussion Papers Series), the International Food Policy Research Institute (IFPRI Discussion Papers) and papers deposited on the Social Science Research Network (SSRN Papers). Our search covered the period of January 2020 to 20 November 2022.

3.1.2. Search terms

Our topical search strategy was based on a combination of various words and phrases that were used to describe the coronavirus disease including COVID-19, SARS-cov-19 or coronavirus. The outcome of interest was food insecurity in its variants, including food security, food insecurity, food prices, food disruption, food consumption, and food access. The search results were combined using the appropriate Boolean operators AND/OR. We selected publications in English only and filtered publications by all African countries.¹ Our initial search provided 4,304 documents.

3.2. Inclusion and exclusion

Studies on COVID-19 and food insecurity in low-income countries are many with varying quality and depth. While this review sought to include studies that could account for pre-COVID-19 food insecurity conditions, it also included cross-sectional and descriptive studies that are carefully conducted. We included randomised controlled trials (RCTs), studies implementing difference in differences, instrumental variables, and panel fixed effects, and repeated cross-sectional studies. Randomised controlled studies do not imply that some individuals were exposed to COVID-19 and others not but rather that these studies follow RCTs of other interventions (Aggarwal et al., 2022; Hirvonen et al., 2021; Huss et al., 2021; Stein et al., 2022) which were not linked with COVID-19. Simulation studies can often be of good quality and, in some instances, reveal an underlying relationship that can be missed by quantitative studies. Nonetheless, we excluded them from this review. We do not include purely qualitative studies. Through screening, we eliminated records that did not match these search intentions, duplicates, and those that did not clearly and precisely measure food insecurity. Altogether, we retained 44 studies for full-text review. Key summary insights from the reviews are provided in Supplementary Material. The PRISMA diagram in Supplementary Material shows the search, exclusion, and inclusion strategy.

4. Overview of the literature

4.1. Continental coverage

The review includes at least one study from 44 out of the 54 African countries, with Nigeria (16 studies), Kenya (15), Uganda (11), and Ethiopia (10) having the highest number of studies reviewed (Figure 2). Thirteen of the 44 studies are multi-country studies with one study by Agyei et al. (2021) using administrative data on food prices to cover 35 countries.

Comparing the country-level average stringency in 2020 and 2021, we observe a correlation between the number of studies per country and average stringency. In Figure 3, we plot a scatterplot of the correlation between the number of studies and stringency and observe that the four countries with the highest number of studies also experienced higher stringency overall. It is therefore most likely that as countries instituted more stringency, their population was made more economically vulnerable and most likely associated with food insecurity (Egger et al., 2021). At the end of 2020, official reports suggested that close to 350 million people were severely food insecure, with increases more likely related to COVID-19-related food access disruptions (FAO, UNECA, & AUC, 2021).

4.2. Measuring food security

Throughout the literature, different measures of food insecurity have been used with the most common being the Food Insecurity Experience Scale (FIES). In Figure 4, we show the number of studies per the different food security indicators used. The most commonly used FIES score is constructed based on eight questions about the experience of food insecurity (worry, healthy, few foods, skipping, eating less, running out of food, going hungry and not eating for whole days). Given the interest to keep surveys short since most surveys were through mobile phones, some studies only used one or more of these eight questions. Other proxies of food insecurity from the reviewed studies include household dietary diversity score (Hirvonen et al., 2021), hunger scale, food gap (Abay et al., 2021a), food expenditures (Mahmud and Riley, 2021), production and yields, consumption pattern index (Maredia et al., 2022; Mueller et al., 2022), food insecurity access scale and food disruption (Tabe-Ojong et al., 2022), food prices (Dietrich et al., 2022; Hirvonen et al., 2021), hunger scale and food consumption score (Aggarwal et al., 2022) and the coping strategy index in various lengths (Huss et al., 2021; Tefera, Tadesse and Asmare, 2022). There are no clear differences in food insecurity levels by the choice of food security measurement indicator used, implying that no matter the choice of indicator, the evidence was consistent with associated food insecurity.

4.3. Representativeness of the studies

Regarding the level of representativeness, the studies are split almost equally with about half of the studies coming from nationally representative samples. For instance, all the studies coming from the World Bank High-Frequency Phone Surveys (Agamile, 2022; Bundervoet, Dávalos and Garcia, 2022; Dasgupta and Robinson, 2021, 2022; Rudin-Rush et al., 2022) are nationally representative as they are from previous nationally representative samples. Others utilised random digit dialling to construct national representative samples (Mueller et al., 2022). About half of the studies are however not nationally representative. These are sometimes following existing studies such as RCTs on other issues (Egger et al., 2021; Stein et al., 2022) or generally other previous studies that were only adapted to COVID-19 at the onset of the pandemic (Abay et al., 2021a, 2021b, 2022; Lasdun et al., 2022; Tabe-Ojong et al., 2022).

5. COVID-19 and food security

Earlier surveys at the onset of the pandemic reported associated increases in food insecurity as a result of COVID-19 in many African countries (Abay et al., 2021b; Adjognon, Bloem and Sanoh, 2021; Agamile, 2022; Amare et al., 2021; Bundervoet, Dávalos and Garcia, 2022; Dasgupta and Robinson, 2021, 2022; Egger et al., 2021; Ibukun and Adebayo, 2021; Kansiime et al., 2021; Mahmud and Riley, 2021; Mueller et al., 2022; Sassi and Trital, 2022; Tabe-Ojong et al., 2022). Most of these studies are based on the high-frequency surveys of the World Bank which follows up on some earlier pre-pandemic data as part of the Living Standard Measurement Surveys (LSMS). These surveys are high-frequency surveys that have been collected in several existing LSMS countries and beyond (Bundervoet, Dávalos and Garcia, 2022; Dasgupta and Robinson, 2021, 2022; Koos, Hangoma and Maestad, 2020; Rudin-Rush et al., 2022). Others are based on large baseline surveys which followed up with a phone survey at the onset of the pandemic (Lasdun et al., 2022; Tabe-Ojong et al., 2022). Several studies are also based on RCTs that were underway pre-COVID-19 (Aggarwal et al., 2022; Egger et al., 2021; Hirvonen, de Brauw and Abate, 2021; Huss et al., 2021; Lasdun et al., 2022; Stein et al., 2022). A handful of the studies focused on special vulnerable groups such as individuals living with the AIDS virus (Enane et al., 2021; Folayan et al., 2022; Kavanagh et al., 2021; Stein et al., 2022; Wagner et al., 2021).

There has been significant heterogeneity in the relationship between COVID-19 and food insecurity in various countries. Although it was projected that rural areas will be hit hard especially given their already high levels of food insecurity (Zidouemba, Kinda and Ouedraogo, 2020), one would expect that urban areas will be rather more affected since most of the containment measures were mostly implemented and strictly followed (if at all) in the urban centres. In line with this thought, households in urban areas in Mali were found to be more associated with food insecurity than households in rural areas which led to no gap in rural–urban food insecurity (Adjognon, Bloem and Sanoh, 2021). Contrasting evidence was reported in Burkina Faso, Ethiopia, Malawi and Nigeria where food insecurity was rather increasing in rural areas as compared to urban areas (Rudin-Rush et al., 2022). Maredia et al. (2022) further provide evidence of similar food insecurity associations in both rural and urban areas of Kenya, Zambia, Mali, Nigeria and Senegal. This points to the important role of context given that the underlying socio-economic and regulatory systems in all these countries may be different (Bundervoet, Dávalos and Garcia, 2022; Dasgupta and Robinson, 2022; Mueller et al., 2022; Tabe-Ojong et al., 2022). Context could also explain the difference in findings in the early period of the pandemic as opposed to many months into the pandemic (Aggarwal et al., 2022; Balana et al., 2020; Hirvonen et al., 2021; Maredia et al., 2022; Rudin-Rush et al., 2022; Tefera, Tadesse and Asmare, 2022). Overall, there seems to be limited geographical overlap in these studies which makes it hard to generalise findings.

While there is almost a unanimous consensus that COVID-19 is negatively associated with food security, one caveat to keep in mind is measurement error and precision. Hirvonen et al. (2021) and Abay et al. (2022) observed significant measurement errors in their studies in Ethiopia. Studying food consumption trajectories before and during COVID-19, Hirvonen et al. (2021) found that in Addis Ababa, Ethiopia, food consumption and household dietary diversity remained relatively unchanged and, to some extent, even improved compared to 2019. This could be due to a partial lockdown in Ethiopia as opposed to full lockdowns in other countries, which might have enabled food distribution channels to remain active and therefore pushed households to resilience. They contend that subjective well-being and food insecurity questions administered through phone surveys might not be precise enough to capture actual household experiences. This sentiment is echoed by Abay et al. (2022), who also highlight that the positioning of key modules such as food security influences the quality of data. They found that delaying the food consumption module by 15 minutes in a phone interview leads to a 17 per cent reduction in reporting of diet diversity scores, a 28 per cent reduction in the number of women reporting consuming four dietary diversity groups and a 40 per cent reduction in reporting of consumption of animal source foods. These findings cast doubt on the precision and correctness of phone surveys in measuring food insecurity. The authors related this imprecision to survey fatigue, but it can also be related to reduce respondent attention when there is no in-person respondent–interviewer interaction.

Finally, we highlight the emerging evidence of improvements in some countries. Aggarwal et al. (2022) report no associated declines in food insecurity in Liberia and Malawi at the onset of the pandemic (Aggarwal et al., 2022). Moreover, as countries reopened and people returned to work, food insecurity reduced as household incomes returned. In Nigeria, Balana et al. (2021) observed that as household income rebounded by 50 per cent (of pre-pandemic income levels), there were some associated reductions in severe food insecurity by about 8 percentage points. Household diet diversity scores were also observed to slightly increase by 5 percentage points. Improvements have also been reported in Burkina Faso, Ethiopia, Malawi and Nigeria (Rudin-Rush et al., 2022). This might suggest that a recovery is underway in several economies and that the food insecurity worries and experiences of 2020 are in remission.

6. Mechanism and channels through which COVID-19 affected food security

Two key mechanisms were found relevant in explaining the relationship between COVID-19 and food insecurity: (1) income and employment and (2) prices. As earlier mentioned, most of the studies proxied COVID-19 with the various containment measures put in place by many governments. These included lockdowns, social distancing and travel bans with lockdowns being the most used in many empirical studies.

6.1. Income channel

Income losses from COVID-19 have been the most reported cause of food insecurity. Most studies have established income losses as a result of various containment measures used in many African countries (Agamile, 2022; Balana et al., 2021; Bargain and Aminjonov, 2021; Bundervoet, Dávalos and Garcia, 2022; Egger et al., 2021; Kansiime et al., 2021; Mahmud and Riley, 2021; Maredia et al., 2022). Income losses have been argued to emerge from employment-related losses as a result of containment measures (Mahmud and Riley, 2021). However, these losses could also be due to price increases, especially for households that depend on markets for leveraging their food demands. Early in the pandemic, it was established that lockdowns may jeopardise the consumption and food security situation of households, especially those households who rely on labour income to finance food purchases (Arndt et al., 2020). Most of the abovementioned studies estimated income changes during the pandemic, but some studies also estimated the direct associations between income losses arising from the containment measures and food insecurity. Several studies use self-reported COVID-induced income shock as a proxy for COVID-19 and estimated its association with food insecurity and find that households exposed to COVID-19 via income losses experienced reductions in food consumption and an increase in food insecurity (Agamile, 2022; Hirvonen et al., 2021; Mahmud and Riley, 2021). For those with associated wage income losses, food insecurity could be lessened if employment-related losses push them to transition to the production of food crops and agricultural activities which they could use to balance up their food demands.

Akin to income losses, the containment measures equally created employment shocks through the closure of various businesses and associated job losses (Balana et al., 2020; Bundervoet, Dávalos and Garcia, 2022; Egger et al., 2021). Significant reductions in labour market participation and the probability of participation in non-farm business activities have been reported in Nigeria (Amare et al., 2021). Similarly, households in Uganda were found to report large wage income declines and reduced enterprise profits (Mahmud and Riley, 2021). These households were rather found to increase their labour supply to more farm production–oriented activities as a way of keeping up with wage income drops (Mahmud and Riley, 2021). Income losses have rightly been argued to emerge from employment losses (Mahmud and Riley, 2021). In Nigeria, households that experienced job losses as a result of business closures have been shown to likely report higher food insecurity (Akim, Ayivodji and Kouton, 2021), underscoring the link between job losses and food insecurity in the absence of safety nets.

6.2. Prices and supply chains

When it comes to the price effects, COVID-19 has been associated with various changes in market and food prices, especially for common staples and vegetable crops (Aggarwal et al., 2022; Dietrich et al., 2022; Hirvonen et al., 2021; Tabe-Ojong et al., 2022). These changes in consumer and farm prices have led to both winners and losers from the containment measures imposed by many governments (Hirvonen et al., 2021). COVID-19 led to large disruptions in food markets and market activity, which is reflected in price increases with some country heterogeneity (Adewopo et al., 2021; Aggarwal et al., 2022). For instance, Ethiopia, Kenya, Namibia, Liberia, Malawi and Tanzania all experienced increasing food prices (Aggarwal et al., 2022; Hirvonen et al., 2021; Tabe-Ojong et al., 2022). In Liberia, general food prices increased by up to 9 per cent and by 20 per cent for staple foods (Aggarwal et al., 2022). In Northern Nigeria, prices for the main staples increased from 30 to 50 per cent (Adewopo et al., 2021). In general, this was the same experience across several African countries for which combined analysis by Agyei et al. (2021) shows that prices for sorghum, maize and imported and local rice all increased compared to the pre-pandemic levels. An increase in food prices likely emanated from mobility restrictions and the disruptions of supply chains (Dietrich et al., 2022).

This increase in food prices can be linked to supply chains in multiple dimensions. On the one hand, many African countries depend to a substantial extent, on imported food and farm inputs. The disruption in shipping schedules and volumes likely caused price increases (Nkamleu, 2020). Country preventive mechanisms also meant that to some extent, farm inputs were in scarcity and production difficulties and lower yields were experienced (Middendorf et al., 2022; Nchanji et al., 2021; Vall et al., 2021). However, while most of this evidence points to weakening supply chains, there are indications that to some extent, supply chains were able to adapt and remain resilient. Van Hoyweghen et al. (2021) show that fresh foods and vegetable exports and imports in Senegal remained more or less the same in the early months of the pandemic (January to June 2020) as in the previous years (2016–2019), indicating robust resilience. However, evidence from a few more months into the pandemic showed a reduction in input access and farm yields with discernible income losses in Senegal (Middendorf et al., 2022). In Ethiopia, Hirvonen et al. (2021) also show that local logistical barriers created large and heterogeneous price changes favouring farm gate prices where markets were restricted. While there is no evidence assessing market structure especially with respect to the typology of food supply chains, it is likely that the level of resilience would have depended on the type of food supply chain (e.g. traditional, transitional or modern) (Reardon, Bellmare and Zilberman, 2020). Moreover, it also seems to matter the type of food studied. Studies assessing supply chain challenges (Hirvonen et al., 2021; Middendorf et al., 2022; Nchanji et al., 2021; Van Hoyweghen et al., 2021) all look at highly perishable fresh foods and vegetables, which as Hirvonen et al. (2021) highlighted are traded over short distances and less likely to be significantly affected by transportation and logistical bottlenecks. There is therefore a lack of evidence on other types of crops such as cereals and evidence on import-reliant countries.

7. Resilience and mitigating food insecurity

Few studies provide evidence on coping strategies employed by households. Using high-frequency phone surveys from six countries, Koos, Hangoma and Maestad (2020) provide some evidence on coping strategies suggesting that households employed various coping strategies such as reducing consumption, selling assets and borrowing. However, the evidence here is generally mixed and thin. For instance, while Middendorf et al. (2022) found that a majority of affected households did not implement any mitigation strategy, Koos, Hangoma and Maestad (2020) observed that in the first months of the pandemic, there was no reported increase in running out of food, implying households were resilient. Putting the timelines of the pandemic and the control mechanisms that followed, households responded to increasing food insecurity using a couple of strategies. While some of these strategies were external to the household, some were actions carried out by the households. In this section, we cover some of the coping strategies households are using while the more external ones will be captured in the following sections.

7.1. Stockpiling food

The early days of the pandemic were characterised by the implementation of prevention policies that did not have clear and definite timelines. Lockdowns were extended sometimes in various countries. As fear and uncertainty increased, many households resorted to stockpiling food as one of the coping mechanisms, especially as supply chains were expected to be adversely affected by the associated increase in food prices. In general, we do not find many studies that assessed the frequency of food stockpiling across many African countries. However, a few studies are worthy of mention here. Amuakwa-Mensah et al. (2022) examine stockpiling behaviour in 12 countries using two cross-sectional studies implemented between April and May 2020. They assess the change in purchases between the two timelines and find that overall, 41 per cent of the households report purchasing food packet sizes larger than usual. Their regression analysis also shows that concern about COVID-19 also increased the intensity of purchasing both smaller and larger packet sizes of food and household utilities. Although it could have been possible to conduct a country-level analysis of heterogeneity, they do not show this and therefore it was not possible to know how stockpiling could have been different across countries. Murendo et al. (2021) study food purchase behaviour in urban Zimbabwe in the early days of the pandemic, Fisher et al. (2022) study a South African township and Ben Hassen et al. (2022) study food consumption patterns among men and women in Tunisia, Morocco and Egypt. All find unanimous evidence that stockpiling increased at the onset of COVID-19 and the associated prevention policies. There is some heterogeneity in North African countries as Egypt has a lower prevalence of stockpiling (38 per cent) compared to Tunisia (59 per cent) and Morocco (53 per cent). In Zimbabwe, 60 per cent of the households reported stockpiling (Murendo et al., 2021) as were 50 per cent of households in Alexandria township (Fisher et al., 2022). We note, however, that stockpiling is likely related to other mechanisms such as drawing on savings and taking on credit. In addition, the actual net effect of stockpiling is not well known. Richer households or those that have some savings are the ones with the economic capability to stockpile (Amuakwa-Mensah et al., 2022). This in turn can create unfair food distribution and food access inequality and reduce food availability if stockpiling is not controlled through fair rations.

7.2. Social protection

As early as April 2020, many governments and various development and relief agencies began rolling out and augmenting relief support to households. This took the form of cash transfers, in-kind transfers (food support and school feeding), loans and credit schemes, public work programmes, and utility waivers mainly for electricity and water (Gentilini, 2022). These waivers were geared at easing financial burdens and supporting households. Cashless payments were also another strategy used to reduce the spread of COVID-19 by reducing contacts. While targeting and access to these social protection programmes were limited, it however reached some households in both rural and urban settings at different magnitudes (Maredia et al., 2022). Government’s social protection programmes are key in reducing food insecurity where they were available (Abay et al., 2021a; Dasgupta and Robinson, 2021, 2022; Gelo and Dikgang, 2022; Strupat and Nshakira-Rukundo, 2022). Gelo and Dikgang (2022) observe that households exposed to COVID-19-related job losses are more likely to report both adult and child food insecurity, but a government cash transfer and an old age pension reduced the likelihood of food insecurity by 16–24 per cent. Dasgupta and Robinson (2022) found that cash benefits are more effective than food assistance in reducing food insecurity during the pandemic. Cash transfers are also found to improve food insecurity and dietary diversity in Liberia and Malawi where households that received cash transfers performed better on all metrics of food security (Aggarwal et al., 2022). In an earlier analysis, Nechifor et al. (2021) show that governments can boost the recovery of food demand and the food sector through income support in Kenya (Arndt et al., 2020). However, in many instances, government support was insufficient to enable households to bounce back to the previous pre-pandemic levels (Balana et al., 2021; Tefera, Tadesse and Asmare, 2022).

7.3. Role of remittances

Like cash transfers and other social protection programmes which were rolled out by many governments and development agencies, households also received support from friends and relatives in the form of remittances. Remittances have been shown to mitigate the negative relationship between COVID-19-induced employment shock on food insecurity in Nigeria (Akim, Ayivodji and Kouton, 2021). Here, remittances from abroad are shown to depict larger associations than domestic remittances. Moreover, remittances have more pronounced effects in rural areas than in urban areas, a finding that furnishes and supports the income effect on food insecurity. In this light, financial inclusion and asset ownership may have a similar mitigating role to remittances. For instance, livestock accumulation and social capital are two important factors that cushioned households from falling into severe food insecurity in Nigeria (Balana et al., 2021).

8. Key messages and directions for future research

COVID-19 has been a global prolonged shock and will most likely leave economies and households in a depression that might last several years to recover from. Studies on poverty and consumption have estimated that under an extreme scenario of income and/or consumption drops of about 20 per cent, between 400 and 600 million people might fall back into poverty (Sumner et al., 2020). And yet, evidence from studies reviewed here (such as Egger et al. (2021)) shows that the magnitudes of consumption decline have been substantially higher than 20 per cent. This might imply that retaining the same assumptions, poverty will have increased more than the current estimates suggest. With increased poverty is increased food insecurity and reduced resilience capacity. In this study, we review the evidence of the associations between COVID-19 policies and food insecurity in Africa. We summarise the abovementioned findings in these key takeaways. For ease of expression, we combine these takeaways with opportunities for future research.

8.1. The role of emergency social protection

The COVID-19 pandemic was countered with what has been described as the largest social protection scale-up ever witnessed. Timely emergency social protection responses were key in preventing destitution. We observe that continuing, increasing and instituting emergency social protection interventions was strongly associated with reducing the likelihood of reporting food insecurity. This, therefore, calls for increases and continuous replenishment of disaster preparedness and emergency response budgets. For better disaster preparedness, governments should increase emergency food storage to enable emergency food distribution when needed and quick disbursement of cash support. This is even more relevant in low-income countries where only about 10 per cent of the population is reached by these emergency services.

8.2. Association between COVID-19 and food security

All our studies use various proxies for COVID-19 such as lockdowns, fear of COVID or the time between COVID-19 waves. These studies highlight the association between containment measures, job losses, economic slowdown, poverty and food insecurity. Most of the studies reviewed used pre-pandemic and pandemic data to assess changes in food insecurity at the onset of the pandemic. The real added value of the pre-pandemic data is that these studies can show more realistic correlations associated with pandemic policies. For instance, these studies provide information such as food prices or food consumption or food insecurity scores before and after (within) the pandemic. Moving ahead, it will be worthwhile to combine these pre-pandemic and pandemic data with other data sources on various socio-economic outcomes. One example is, linking health statistics such as health management information systems data with other epidemiological data that capture actual COVID-19 exposure to household data such as well-being, income and food security. This might be done through data combinations of health and census and survey data. The resources required might be a lot and the ethical issues of data access might also not be clear, but the benefits of these will be immense.

In the interim, where possible, for short- and medium-term analyses, researchers might consider continuous exploration of survey and non-survey data. Some authors have utilised Google mobility data to examine access to markets (Bundervoet, Dávalos and Garcia, 2022; Dietrich et al., 2022) and poverty in general (Bargain and Aminjonov, 2021). Adewopo et al. (2021) used crowdsourced price data to assess price changes in common staples in Northern Nigeria. In general, data like these are underused, yet they might be more readily available in real time and bring a lot of value, especially when combined with high-frequency phone surveys. One challenge with these data (Google mobility) is the lack of availability and granularity in many African settings. Studies that have used Google mobility data often have very few data points for African countries, further limiting what can be learnt. We can envisage that data such as satellite, night light activity and others of this dimension might be complementary in this regard.

8.3. Data quality challenges during the pandemic

As the number of phone surveys greatly increased during and post-pandemic, authors must understand some of the limitations of these phone surveys. Phone surveys have been shown to underestimate per capita consumption and overstate poverty headcount (Abate et al., 2022). These could arise because of survey fatigue which may occur earlier in phone surveys than in in-person surveys. There are also issues of representativeness as selection bias seems to be an important constraint in these methods of data collection (Brubaker, Kilic and Wollburg, 2021). These issues therefore should make readers and policy experts taking evidence from this recent evidence do so with some caution. Phone surveys might have been the most appropriate technology for data collection during the pandemic, but establishing the extent of the effects of pandemic prevention on well-being might require additional data.

8.4. Limitations in coverage across and within countries

Limited coverage in depth (within the country) and width (across countries) persists. We do not find many studies that revealed the effects of local variation in COVID-19 prevention policies. Metrics such as the stringency index are only at the country level and are only useful for cross-country comparisons. Cross-country comparisons are useful, and yet within-country differences also matter. In some countries, regions were exposed to varying levels of stringency based on whether they were considered hot spots or not. Future research using country-wide representative data might consider exploring these within-country differences and heterogeneity.

Yet, there are also limitations in terms of geographical coverage. Most of the studies in this review focus on a few countries. The map shown in Figure 3 indicated that the majority of the studies are from only four countries, namely Nigeria, Ethiopia, Kenya, and Uganda. There are very limited insights from countries in Central, North and most parts of Southern Africa. Given substantial within-country and cross-country heterogeneity in pandemic control policies, generalisations of single studies or even a group of studies like this should be made with caution. That our findings are concentrated in countries like Ethiopia, Kenya, Nigeria, and Uganda is not surprising given that these are already established as highly researched countries in Africa (Porteous, 2022). In addition, the availability of pre-pandemic data made follow-up studies assessing pandemic dynamics possible and easy to implement. While this sort of path dependency in research enables quick implementation of research during crisis times, it might inhibit learning from less researched countries. This is evident when we compare the number of studies per country with the average food insecurity/hunger index during the pandemic times (see Figure 3). It may thus be worthwhile for future studies to delve into the less studied regions and countries to improve learning and generalisations on the relationship between COVID-19 and food insecurity in Africa.

8.5. Compounded shocks and additional vulnerabilities

This review finds that there is very limited research in the understanding of how COVID-19 affected the food insecurity situation of already vulnerable groups such as those in conflict or displacement, those already under climate stress and the health-vulnerable subgroups. Only one study (Stein et al., 2022) assessed COVID-19 and food insecurity in the context of refugees in Uganda. Countries such as Ethiopia and the Democratic Republic of Congo were also going through conflict. Other shocks would include those already exposed to climate stress, droughts, floods and related shocks. Other vulnerable groups might include internally displaced people and individuals living with HIV/AIDS, among others. While there was some research on the latter, we suggest that more research should focus on these groups. As economies re-open, there will be feelings of lost opportunities to know how these specific subgroups fared. Indeed, their pre-existing vulnerability might imply that for these, the depth of poverty and vulnerability as a result of COVID-19 might be more far-reaching than the general population.

In some situations, households would have experienced compounded shocks. One example is the locust infestation in the East and Horn of Africa in 2020. While the locust threat was recognised in the literature (Griffith et al., 2020; Salih et al., 2020) as reports suggested over 44 million individuals in the Horn of Africa were at risk of acute food insecurity, we do not find any empirical evidence linking these two shocks. And yet, while rural households could have been somehow shielded from COVID-19 market disruptions, dependence on their farms for food supply would have been affected by locusts-related harvest losses. Evidence on locust coverage is thin. However, this can be assessed with, for instance, remote sensing and other vegetation data. This could likewise be combined with household survey data to assess the extent of food insecurity associated with such compounding shocks.

8.6. Looking into the long-term effects

In addition, there will be a need for more research on understanding the longer-term effects of the COVID-19 shock. These studies might exploit various identification strategies that can provide credible variation in exposure to COVID-19 across different regions and age groups. There is a large body of literature that shows for instance that early life exposure to famine and food insecurity is highly detrimental to child development and later life health and wellness (Cheng et al., 2020; Johnson and Markowitz, 2018; Oliveira et al., 2021), adulthood cognition (Arage et al., 2020a), adult anthropometry (Arage et al., 2021), and other aspects of health (Arage et al., 2020b). It might be fair to say that these shocks studied in say China or Ethiopia (the 1950s and 1980s famines in the respective countries) had far more reaching individual effects than COVID-19. Nonetheless, future interest remains in both the socio-economic and health dimensions of research frontiers.

9. Conclusion

This review sheds light on the association of COVID-19 with food security in Africa. The evidence strongly suggests that the pandemic is largely associated with food insecurity. With a few exceptions, studies reveal increasing insecurity during the pandemic compared to pre-pandemic levels. In the studies reviewed, we observe the mitigating effects of social protection interventions and social support through remittances. Food stockpiling was also reported by a few studies, but the questions on its net effect (protecting richer households while reducing food availability for households with limited resources) remain outstanding. The review concludes with six key messages and possible directions for future research.

1. COVID-19 is associated with higher levels of food insecurity, measured in multiple dimensions and indicators. The main pathways were income losses and increases in prices.

2. Social safety nets and remittances were key in reducing the food insecurity associated with COVID-19 by both replacing some income losses and providing food transfers.

3. Due to the nature of pandemic policies, it is not easy to identify within-country comparisons. There is also a lack of depth in cross-country comparisons given that only about half of the countries on the continent are represented in this review.

4. There is limited evidence of how COVID-19 might have interacted with other shocks such as locusts and weather variations.

5. Finally, while phone surveys came in appropriately to address data collection difficulties in times of limited mobility to collect data, phone survey data might have significant measurement errors. The majority of the literature here emanates from phone surveys. Therefore, readers and policymakers need to consider these issues while internalising these results.

Acknowledgements

The authors wish to thank the Editor, Timothy Richards, and two anonymous reviewers for providing constructive feedback on the previous versions of the manuscript. We are also grateful to Marilou Goussard Vincent for her research assistance. This work benefited from the support of the CGIAR research initiatives on National Policies and Strategies (NPS) and Fragility to Resilience in Central and West Asia and North Africa (F2R-CWANA). This study was also partly supported by the Deutsche Forschungsgemeinschaft (DFG) under the CRC/Transregio 228: Future Rural Africa: Future-making and social-ecological transformation (Project number: 328966760). All errors remain solely of the authors.

Supplementary data

Supplementary data are available at ERAE online.

Footnotes

References

Author notes