IMPACT OF COVID-19 CRISIS ON AGRICULTURE Evidence From Five Sub-Saharan African Countries Akuffo Amankwah and Sydney Gourlay BACKGROUND ing Standards Measurement Study (LSMS) Countries in Sub-Saharan Africa (SSA) have and the Poverty and Equity Global Practice not been spared from the negative impact teams.3 These five countries are part of the of the COVID-19 crisis. Though countries LSMS - Integrated Survey on Agriculture in the region have reported fewer cases (LSMS-ISA) project that fields multi-topic of COVID-19 than other parts of the world, household surveys with a focus on agricul- governments in these countries imple- ture. Thus, the households included in the mented various containment measures. HFPS are sub-samples of LSMS-ISA house- The containment measures implement- holds interviewed in the most recent face- ed by governments in the region varied to-face interviews in respective countries. across countries, but generally included A uniform methodology was adopted in nationwide or partial lockdowns, travel sampling, weighting, and implementing restrictions, schools and offices closures, the survey across the countries, making restrictions on social gathering, among cross-country comparison feasible. While others.1 Countries in the region were im- the phone surveys began after the onset pacted at the time of other shocks. For in- of the coronavirus pandemic, the timing of stance, Uganda and Ethiopia in Eastern Af- implementation varies across countries, rica were beset with locust invasion, while as does the intensity of the pandemic and the global fall in oil prices created a dual the local restrictions (see Annex I for an il- crisis for Nigeria as the country’s econ- lustration of survey timing and COVID-19 omy is heavily reliant on oil. Overall, the response). COVID-19 crisis, coupled with these oth- Davis et al (2017)4 report that more than er external shocks, is expected to impact 50% of households in Sub-Saharan Africa countries in the region negatively, and ar- generate their livelihoods from agriculture. ticulating a policy response requires un- In what follows, we explore the impact of derstanding how and which households COVID-19 on agriculture in SSA, looking at have been impacted and if households participation in agriculture before and af- may have been able to rely on or move ter the outbreak of the crisis. Agricultural into specific activities which may act as a shocks, changes in income, and expecta- buffer in times of crises. tions regarding harvests and revenue are The brief leverages COVID-19 high fre- also explored. Given that data collection quency phone survey (HFPS) data collect- coincided with the 2020/21 pre-harvest ed primarily by National Statistics Offices season, the brief focuses primarily on (NSO) 2 of five SSA countries (Burkina Faso, pre-harvesting and expectations. Follow- Ethiopia, Malawi, Nigeria, and Uganda), ing completion of the agricultural season with support from the World Bank’s Liv- and the related data collection, additional 3 This survey is part of the World Bank’s effort to support the collection of monthly high frequency phone surveys to monitor the 1 In Malawi, schools were closed but a planned nationwide impact of the COVID-19 crisis on households. lockdown was challenged in court, and ultimately was 4 Davis, Benjamin, Stefania Di Giuseppe, and Alberto Zezza. not implemented. 2017. “Are African Households (Not) Leaving Agriculture? Patterns 2 The Ethiopia HFPS was implemented by a private survey firm, not of Households’ Income Sources in Rural Sub Saharan Africa.” Food the national statistics office. Policy 67: 153–174 2 analysis will be presented on harvest ac- demic levels of 84% and 76% respectively.6 tivities and outcomes. The changes in Figure 1 are the product of the net effect of households moving PARTICIPATION IN AGRICUL- into and out of agriculture; in most cas- TURE BEFORE AND AFTER THE es, entries were larger than exits, with COVID-19 OUTBREAK the exception of livestock in Uganda.7 The data shows that agriculture contin- In general, the share of households that ues to be the main source of livelihood have entered into agriculture since the of Sub-Saharan African households, start of the pandemic is higher than those with the share of households involved exiting. For instance, in Malawi, about 9% in agriculture increasing since the start of households who were not involved in of the pandemic.5 Prior to the outbreak, agriculture (either crop or livestock farm- 76% of Nigerian households were involved ing) before the pandemic are doing so in agriculture (either crop or livestock now, compared to less than 2% that were farming), but the share has increased to involved in agriculture pre-pandemic who 84% since the start of the pandemic. We are not doing so since its onset. Similar- observed similar results in Malawi and ly, the share of Nigerian households who Uganda, where 91% and 79% of house- have gone into agriculture is higher (12%) holds respectively are involved in some than those exiting (4%) since the start of form of agriculture since the start of the the pandemic. COVID-19 crisis, compared to the pre-pan- FIGURE 1: PERCENTAGE OF HOUSEHOLDS INVOLVED IN AGRICULTURE BEFORE AND AFTER COVID-19 OUTBREAK, BY COUNTRY 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Before After Before After Before After Before After outbreak outbreak outbreak outbreak outbreak outbreak outbreak outbreak Ethiopia Malawi Nigeria Uganda Crop farming Livestock production Either crop or livestock 5 During the last post-harvest visit to the households in 2018/19, 6 As at the time of this report, Ethiopia was yet to collect households were asked if they are involved in crop or livestock information on post pandemic livestock production, hence we are farming activities. Similarly, during the 2020 phone interviews, not able to get a full picture of total share of households involved in respondents were asked if any member of their household has done agriculture (either livestock or crop farming). any crop farming or livestock production activities since the start of 7 Entry into agriculture is defined as those households who were the COVID-19 crisis (coinciding with the 2020/21 agricultural season not involved in agriculture pre pandemic, but are doing so since the in most of the study countries). We use these questions to construct outbreak, while exit from agriculture means those households who the before and after comparison and the churning in and out of were involved in agriculture pre pandemic but have not engaged in agriculture since the start of the crisis. any agricultural activities since the start of the pandemic. 3 FIGURE 2: CHURNING IN AND OUT OF AGRICULTURE SINCE THE START OF COVID-19 CRISIS 20% 18% 16% 14% 12% 10% 8% 6% 4% 2% 0% In Out In Out In Out In Out Ethiopia Malawi Nigeria Crop farming Livestock production Either crop or livestock We explore further how households however, we find more households go- in the study countries are moving in and ing out of livestock production (17%) than out of the different sub-sectors of agricul- those entering (10%) since the start of the ture by looking at crop farming and live- pandemic. Across countries, the per- stock production separately. The share centage of households going into live- of households that have gone into crop stock production appears higher than farming appears higher than those that those going into crop farming since the have exited. In Ethiopia, about 16% of start of the pandemic. This can possibly households that were not involved in crop be explained by the seasonal nature of farming before the pandemic are doing so crop production, compared to livestock now, compared to about 3% that were en- farming. gaged in crop farming before the pandem- Across countries, the movement of ic that did not cultivate crops during the households into crop farming since the 2020 agricultural season. Similarly, in Ma- outbreak seems to be more prevalent lawi, 11% of non-crop farming households in urban than in rural locations. Spe- are cultivating crops in the 2020 agricul- cifically, in Malawi, about 42% of urban tural season, compared to 2% that did so households were involved in crop farming in the last agricultural season but are not pre-pandemic, but the share increased to cultivating in the 2020 agricultural season. about 60% after the outbreak, compared In Nigeria, about 19% of households who to their counterparts in rural locations. did not own/raise livestock pre-pandemic We observe similar results for Nigeria and are doing so now, compared to about 15% Uganda. The high increase in urban dwell- that owned/raised livestock last year but ers participating in agriculture might be are not doing so after the outbreak. We the consequence of food security and em- find similar results in Malawi where the ployment challenges emanating from the entry into and exit out of livestock farming negative impact of the pandemic being are 16% and 13% respectively. In Uganda, higher in urban than in rural areas. 4 FIGURE 3: PERCENTAGE OF HOUSEHOLDS INVOLVED IN CROP FARMING, BY RURAL-URBAN 120% 100% 80% 60% 40% 20% 0% Urban Rural Urban Rural Urban Rural Malawi Nigeria Uganda Before outbreak After outbreak We observe further that more house- culture compared to their counterparts holds in urban Nigeria and Malawi are in rural areas. For instance, 21% of house- participating in livestock farming, while holds in urban Malawi who were not cul- the rural share involved in livestock pro- tivating crops pre-pandemic are doing so duction seems fairly constant in these two now, compared to 9% of their rural coun- countries. In Uganda, however, the share terparts. The data shows similar results of rural households involved in livestock for Nigeria and Uganda, where the share production decreased from 69% before of urban dwellers going into crop produc- the outbreak to 59% after the COVID-19 tion during the 2020 agricultural season outbreak, while the share of urban house- seems higher than the share transitioning holds decreased from 30% to 29%. from rural areas. In the case of livestock, Looking deeper at the transitioning in and 22% of urban Nigerian households who out of agriculture by rural-urban divide, were not owning livestock last year are we observe that, across countries, more doing so now, compared to 17% of their urban households are moving into agri- rural counterparts. In the case of Uganda, FIGURE 4: PERCENTAGE OF HOUSEHOLDS INVOLVED IN LIVESTOCK FARMING, BY RURAL-URBAN 80% 70% 60% 50% 40% 30% 20% 10% 0% Urban Rural Urban Rural Urban Rural Malawi Nigeria Uganda Before outbreak After outbreak 5 FIGURE 5: CHURNING IN AND OUT OF AGRICULTURE, BY COUNTRY AND LOCATION (% OF HOUSEHOLDS IN LOCATION) 30% 25% 20% 15% 10% 5% 0% Urban Rural Urban Rural Urban Rural Malawi Nigeria Uganda Crops in Crops out Livestock in Livestock out Either CF or LF in Either CF or LF out however, the share of urban households ture, non-farm family business, wage and going into livestock production after the remittances from abroad) and whether outbreak seems about the same as that the income from those sources increased, in rural areas, though the share of house- decreased or stayed the same since the holds in rural Uganda who have gone start of the pandemic. In April/May 2020, out of livestock production seems higher 41% of Ethiopian households who re- (20%) compared to 11% exiting in urban ceived income from agriculture in the last areas. Across countries, the data seem to 12 months, reported loss of income from suggest that rural households are exiting agriculture (i.e. agriculture income de- livestock production more than they are creased compared to before the pandem- entering (more pronounced in Uganda). ic), while 85% and 63% reported experi- This is probably due to the impact of the encing income loss from non-farm family pandemic on livestock production activi- businesses and remittances from abroad ties such as access to feed, animal health respectively. Similarly, in Malawi, 73% of services and markets.8 households who received income from agriculture in the last 12 months report- INCIDENCE OF INCOME LOSS ed loss of income from agriculture in May/ AND SHOCKS June 2020, while 84% and 58% reported While agriculture has been impact- loss of income from family business and ed by the pandemic, the effect seems wage work respectively, during the same less compared to other sectors. House- period. We observe similar results for Ni- holds were asked in different rounds of geria and Uganda. Across countries, the the phone survey if they received income share of households reporting income from specific sources (including agricul- loss from these sources, however, seems 8 In Nigeria, households were asked how the pandemic has affected their livestock activities; limited access to feed (89% of livestock households), animal health services/drugs (79%) and limited access to markets (82%) were reported. 6 FIGURE 6: INCIDENCE OF INCOME LOSS (% OF HOUSEHOLDS WITH INCOME FROM SOURCE IN LAST 12 MONTHS), BY COUNTRY AND INTERVIEW MONTH 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% April/May May/June May/June July/August April/May August June July Ethiopia Malawi Nigeria Uganda Agriculture income Non-farm business income Wage income Remittance from abroad to be reducing in the months following the households’ attention from the impact of first phone interviews. This might be at- the COVID-19 crisis, while in Malawi, there tributed to the easing of lockdown restric- was no government nationwide lockdown tions in the countries during subsequent implemented. interviews. Given the containment measures im- Households in SSA have also been plemented by governments in the study affected by price shocks in the form countries, it is important to understand of high prices of farming inputs and their impact on households day-to-day reduction in the prices of outputs, farming operations, in terms of being able though at different levels. In Malawi, to perform their farming activities. Thus, about 29% of farming households report- households were asked if they were able ed experiencing input price shocks, while to conduct their agricultural activities nor- 30% reported output price shocks. While mally despite the closures and restrictions the share of farming households report- in their countries. The fielding of this ques- ing input or output price shocks are low tion coincided with the planting seasons in in Uganda, the shares are unsurprisingly most countries except Malawi, where the high in Nigeria. These results may be ex- question was fielded during the harvest plained by the extent of lockdown restric- season. The data show that, except for Ni- tions implemented in the respective coun- geria, there is little evidence of house- tries and other shocks that were apparent hold’s having issues undertaking their during the period. The high percentage of crop farming activities, which corrob- households reporting shocks in Nigeria orates the churning in and out of agri- can be explained by the fact that Nigeria culture discussed earlier. For instance, experienced a dual crisis – COVID-19 and about 34% of Nigerian farming house- fall in oil prices – concurrently. The locust holds indicated that they were unable to outbreak in Uganda might have shifted perform their farming activities normally, 7 FIGURE 7: INCIDENCE OF PRICE SHOCKS, % OF AGRICULTURE HOUSEHOLDS Incidence of Price Shock (Decrease) Incidence of Price Shock (Increase) to Farming Outputs, % of Ag HHs to Farming Inputs, % of Ag HHs 35% 30% 60% 25% 50% 20% 40% 15% 30% 10% 20% 5% 10% 0% 0% Malawi Nigeria Uganda Malawi Nigeria Uganda (May/June) (April/May) (May/June) (May/June) (April/May) (May/June) while 10% of farming households in Burki- inter-state travel restrictions) were imple- na Faso were unable to do same, and mented for a longer period. nearly all Ethiopian farming households seem to have worked normally on their EXPECTATIONS REGARDING crop farms. These results can possibly be HARVEST AND SALES explained by the lockdown restrictions im- During the August phone interviews, agri- plemented in the countries. For instance, cultural households in Nigeria were asked we see that in Malawi, where strict lock- about their expectations concerning crop downs were not implemented, the major- harvests and revenue from crop and live- ity of farmers were able to go about their stock sales for the 2020/21 agricultural farming activities normally, compared to season. In order to track how these house- Nigeria, where strict lockdowns (including holds are updating their expectations giv- en the changes in the country, they were FIGURE 8: PERCENTAGE OF HOUSEHOLDS presented with the same set of questions UNABLE TO CONDUCT THEIR AGRICULTURAL in the September round of the survey. ACTIVITIES NORMALLY Overall, farming households in Nigeria 40% seem to update (change) their output 35% and sales expectations over time due 30% to the changes in the country. In August, 25% about 30% (54%) of current crop farming 20% households indicated that they expect 15% their harvest this year to be lower (high- 10% er) than what they harvested from similar 5% 0% planted area in the 2019/20 agricultur- Burkina Faso Ethiopia Malawi Nigeria al season, while the share of households (July) (April/May) (May/June) (April/May) who expect decline in output by the end 8 FIGURE 9: CROP HARVEST AND REVENUE EXPECTATIONS Expectations regarding revenue from Expectations regarding crop harvest crop sales (% of agriculture hhs who (% of agriculture households) normally sell) 60% 70% 50% 60% 50% 40% 40% 30% 30% 20% 20% 10% 10% 0% 0% August September August September Lower Same Higher Don’t know Lower Same Higher Don’t know of 2020/21 increased (decreased) to about FIGURE 10: EXPECTATIONS REGARDING REVENUE 36% (52%) during the September inter- FROM LIVESTOCK SALES (% OF LIVESTOCK FARM- ING HOUSEHOLDS) view. On expected revenue, we observed 70% that the share of households anticipating 60% reduction in 2020/21 agricultural season’s 50% sales revenue decreased from 29% in Au- 40% gust to 28% in September, while the share of those expecting increase in revenue 30% from sales rose from 56% to 62% between 20% August and September. The share of house- 10% holds who expect either their harvests or 0% August September sales revenue in 2020 to remain the same as Lower Same Higher that of 2019 agricultural season seem stable between August and September. Similarly, the share of livestock farming households expecting their 2020 sales revenue to be higher than that of 2019 decreased from 44% in August to 29% in September, while those anticipating sales to be higher increased from 49% to 60% between August and September respec- tively. The percentage of households who expect their livestock sales revenue to stay the same decreased from 7% to 6% be- tween August and September. 9 ANNEX I. COUNTRY-LEVEL Malawi COVID-19 RESPONSE AND 100 HFPS INTERVIEW TIMING The figures below illustrate the timing 80 of each HFPS survey round against the 60 COVID-19 Government Response Strin- 40 gency Index.9 Only HFPS survey rounds that are analyzed in this brief are included. 20 In all countries, subsequent survey rounds 0 have been or will be collected. The survey January 1 March 1 May 1 July 1 Sep 1 2020 2020 2020 2020 2020 round dates presented below are trimmed (5%) to eliminate outliers. HFPS COVID-19 Government Response Stringency Index Burkina Faso Nigeria 100 100 80 80 60 60 40 40 20 20 0 0 January 1 March 1 May 1 July 1 Sep 1 January 1 March 1 May 1 July 1 Sep 1 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 HFPS COVID-19 Government Response Stringency Index HFPS COVID-19 Government Response Stringency Index Ethiopia Uganda 100 100 80 80 60 60 40 40 20 20 0 0 January 1 March 1 May 1 July 1 Sep 1 January 1 March 1 May 1 July 1 Sep 1 2020 2020 2020 2020 2020 2020 2020 2020 2020 2020 HFPS COVID-19 Government Response Stringency Index HFPS COVID-19 Government Response Stringency Index 9 Thomas Hale, Sam Webster, Anna Petherick, Toby Phillips, and Beatriz Kira (2020). Oxford COVID-19 Government Response Tracker. Last updated Nov. 5, 2020. 10 IMPACT OF COVID-19 CRISIS ON AGRICULTURE EVIDENCE FROM FIVE SUB-SAHARAN AFRICAN COUNTRIES JANUARY 2021 WWW.WORLDBANK.ORG/LSMS