Food Security in Burkina Faso: Raising the Alarm1 04 April 2024 Summary Data from the Burkina Faso High Frequency Phone Survey suggest that the fraction of households that is moderately or severely food insecure has been rising steadily over the years. In January 2024 food insecurity was found to be particularly high: 38 percent; (59 percent in the Grand Sahel). Not only is this degree of food insecurity higher than what is expected for the time of year (shortly after the completion of the harvest), the January level is comparable to what is normally reported during the height of the lean season. These results which reflect experiences from of survey respondents, diverge from assessments prepared by FEWS NET and WFP based on market data. Yet in situations where purchasing power is limited and agricultural supply constrained, stable prices and famine have been found to co-exist. As the high frequency phone survey results are robust and based on direct observations, and as the reported levels are disconcertingly high, the possibility of an emerging food security crisis in Burkina Faso needs to be taken seriously. Figure S1: Evolution of the fraction of rural households reporting a lack of food 45.0 38.9 40.0 36.1 34.0 35.0 30.0 24.3 25.0 21.3 21.4 19.8 20.0 15.0 13.0 10.0 11.2 10.0 6.6 4.7 5.0 0.0 Meal skip Eating less Out of food Being hungry and not having to eat Jan/Feb 21 Jan/Feb 23 Jan/Feb 2024 Source: HFPS various rounds 1 This note was prepared by Macoura Doumbia, Johannes Hoogeveen, Elizabeth Taylor of the World Bank Poverty and Equity Practice and Marco Tiberti of the Work Bank’s LSMS team. 1 1. Introduction Households in Burkina Faso deal with major variations in welfare due to seasonality, but also caused by idiosyncratic and covariate shocks. Distinguishing between these shocks is critical for decision makers. When food insecurity is the result of bad luck, neighbors, friends and relatives may be able to help. But when food insecurity is high because of a common shock, social assistance is unlikely to be adequate and collective interventions are necessary. When data is collected regularly, household specific shocks can be identified by the share of households that ‘always’ reports food insecurity. Identifying seasonality is feasible as recognizable patterns occur throughout the year. Once idiosyncratic shocks and seasonality are accounted for, any remaining variability can be attributed to covariate shocks. In Burkina Faso the High Frequency Phone Survey (HFPS) provides the data needed to separate the different types of shocks. The HFPS is implemented by the Institut National de la Statistique et la Demographie (INSD) and is conducted using cell phone numbers of household members collected during the 2018/19 EHCVM survey. The HFPS was introduced during the COVID pandemic, and has been conducted regularly since (see annex 1 for more information on the data used in this note). Amongst the core questions the HFPS asks regularly is a module related to food security. Since July 2020, this module has been repeated 16 times (in 3 survey waves), allowing one to measure baseline levels of food security as well as seasonality. Results from the Jan/Feb 2024 round suggest an increase in moderate or severe food insecurity in rural areas over the baseline level of between 15 and 20 percentage points. On average 38 percent of Burkinabe households indicate to be moderately or severely food insecure (using the FIES score developed by FAO); in the Grand Sahel this is 59 percent! These levels are comparable to what was observed during the lean season of 2023. Extrapolating trends from previous years, it is projected that during the lean season of 2024, a majority of rural households can be expected to become moderately or severely food insecure. These survey results diverge from assessments by FEWS NET, and from price data projections of WFP. These organizations focus particularly on price movements as predictors for food security, and this note argues that under certain conditions price stability and extreme food insecurity can go together. Burkina Faso may well meet the conditions for such a situation. This note is organized as follows. Key results are presented in section 2, followed by a corroboration of these with data from FEWS NET and WFP. A discussion on how to reconcile the apparently contradictory findings follows in section 4. Conclusions are presented in section 5. 2. Rural food insecurity is alarmingly high for a non-lean season When considering food security in Burkina Faso, seasonality matters. With one rainy season per year, agricultural production is seasonal and so is the availability of food for consumption. The lean season occurs in the three months prior to the main harvest in September, during the months of June,-July and August (Figure 1). During these months average food consumption has been found to be 20 percent lower than during the non-lean season (Lain et al. 2021). 2 Figure 1: Crop calendar for a typical year in Burkina Faso Source : FEWS NET HFPS data illustrate the seasonal nature of food availability. Figure 2 presents food insecurity (FIES) scores along with the lean and non-lean seasons (colored bars). The results show how during the lean seasons food insecurity tends to be much higher. The importance of seasonality for food security is well illustrated by the first wave of the HFPS surveys: between July 2020 and July 2021 the HFPS fielded its food security module a total of 9 times. The graphs of the left illustrate how food insecurity fell during the harvest season between September to December 2020, reached a low around the end of the calendar year, after which food insecurity rose again to reach a peak during the months of July and August 2021. Figure 2: Food insecurity since June 2020 Source : High frequency surveys, various rounds 3 With the seasonal pattern of food security established, and having observed that January tends to be a time of relative food security, Figure 2 also demonstrates how in January 2024 the fraction of households recording moderate or severe food security is relatively high. Not only is the fraction high for a non-lean season, the fraction of food insecure household in January 2024 is almost as high as during the lean seasons of 2020 and 2022. The fraction of food insecure households tends to increase by some 10-20 percentage points between non-lean and lean seasons. Consequently, one could expect that a majority of households will be moderately or severely food insecure by July-August 2024. A breakdown of food security by location of the household is presented in Figure 3. Comparing the patters for rural respondents with those for Ouagadougou one observes how even in Burkina’s capital city households cannot avoid exposure to seasonal food insecurity. Yet the variation in insecurity and the degree of insecurity in rural areas is much higher. Typically, twice as many rural households report being moderately or severely food insecure as do households in Ouagadougou. Figure 3 demonstrates how in January 2024 and particularly households in rural areas are at risk of aggravated food insecurity. Not only is the fraction of rural households facing moderate or severe food insecurity much higher than what is found for Ouagadougou or other urban areas, one notes a consistent upward trend in rural food insecurity since June 2020. For Ouagadougou and other urban areas, this trend is much weaker or non-existent. Figure 3: Moderate or severe food insecurity by location Source : High frequency surveys, various rounds collected in three waves. A consideration of spatial patterns of food insecurity over time confirms the upward trend in food insecurity, as well as the elevated vulnerability of households in rural areas, and especially those in the north of the country. This is shown in Figure 4. The figure can be ‘read’ in various ways. Looking down the 4 columns one notes how across the country food security has gradually increased. It got worse between June 2022 and July 2023 (a comparison of lean seasons). It also got worse between March 2023 and Jan/Feb 2024 (a comparison of non-lean seasons). By considering March 2023 and July 2023 one observes how much food insecurity tends to worsen and to spread geographically between the non-lean and lean season. Given that in January 2024 the situation is already worse than it was during the lean season of June 2022, the situation is concerning. Finally, the maps show how northern Burkina Faso is particularly affected by food insecurity, and how this is spreading to the center and west of the country. Figure 4: Spatial pattern of extreme food insecurity. Source : High frequency surveys, various rounds 3. Comparison with market information A commonly used and trusted source for information on food security in Burkina Faso is FEWS NET which regularly produces assessments of the food security situation in the country. In its assessment of February 2024 FEWS NET is relatively positive. It writes: “At national level, the market for agricultural products remains characterized by a disruption of internal flows. The persistence of security risks on certain roads, although lower than in 2023, makes it difficult to access certain production areas and to transfer agricultural products to consumption areas. Overall, commodity price levels are similar to the same period last year, but remain above the five-year average. Across the country, prices are expected to rise above 5 their seasonal averages between March and September, putting pressure on the purchasing power of poor households.�2 That is not to say that FEWS NET is not concerned about the situation in the north of the country. It states: “the situation remains worrying in inaccessible communes in the north of the country�. But even for the north FEWS NET expresses optimism writing: � .. households [in the north] benefit from a wider safety radius than last year, allowing them to access marginal income opportunities from the collection and sale of firewood, the sale of water, market gardening and artisanal mining. These communes have also benefited from escort supplies over the past three months, and food assistance has also increased since December. This has improved food availability, food consumption and reduced the number of people exposed to the Disaster (IPC Phase 5). However, the IDPs and poor host households continue to limit the number of meals and the quantities consumed on a daily basis, exposing them to acute food insecurity Crisis! (IPC Phase 3!). In the absence of plans for humanitarian assistance in the coming months, Emergency (IPC Phase 4) outcomes are most likely between March and September.� Figure 5: FEWS NET food security assessments Source: FEWS NET. 2 https://fews.net/node/31744, accessed April 2024. 6 The qualified optimism of FEWS NET contrasts with the geographic spread of elevated levels of food insecurity found from the HFPS data. Data from on food prices also don’t suggest major concerns. According to respondents in the HFPS prices have eased (Figure 6). This is confirmed with market data (Figure 7). After a peak in 2022, prices for staples like sorghum and millet have come down and since January 2023 they been relatively stable. The last few data points in Figure 7 are price projections and these too do not give much reason for concern about projected price increase. The only markets where prices are very high (and rising) is in the Sahel region where a blockade prevents the movement of goods in and out of the area. Figure 6: how have rural prices changed over the past 12 months? 100 90 80 Percent of households 70 60 50 40 30 20 10 0 Jun-22 Sep-22 Dec-22 Jan-23 Jul-23 Sep-23 Nov/Dec 23 Jan/Feb 24 Much increased Slightly increased The same Decreased Don't know 7 Figure 7: Prices for millet and sorghum, by month and region Sorghum 550 500 450 400 FCFA per kg 350 300 250 200 150 100 Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May 2020 2021 2022 2023 2024 Millet 600 500 FCFA per kg 400 300 200 100 Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May Jul Sep Nov Jan Mar May 2020 2021 2022 2023 2024 BOUCLE DU MOUHOUN CASCADES CENTRE CENTRE-EST CENTRE-NORD CENTRE-OUEST CENTRE-SUD EST HAUTS-BASSINS NORD PLATEAU-CENTRAL SAHEL SUD-OUEST Source: World Food Program 2024. 4. Extreme food insecurity and price stability may co-exist Results from the HFPS on the one hand, and FEWS NET and WFP on the other appear contradictory, as the HFPS reports high levels of extreme food insecurity while FEWS NET and WFP data suggest market prices are relatively stable. Odd as this combination of results seems, they do not need to be contradictory. HFPS records food security as reported by households; FEWS NET and WFP base their assessments on market prices. Sen (1983) points out that both phenomena can go together. In his “Poverty and Famines� 8 Sen argues that in situations in which the purchasing power of households collapses it is possible to observe price stability and hungry households at the same time. There are reasons to believe Burkina Faso meets the conditions for a “Sen-type� famine. Purchasing power could be particularly low due to a combination of past shocks and stagnating rural incomes.3 Indeed poverty in Burkina Faso has been on the rise for some time. Between 2018/19 and 2020/21 poverty increased by two percentage points nationally. In rural areas, with the exception of the wealthiest, all households saw their consumption decline (see the growth incidence curve in Figure 6). Since 2021 the situation for rural households has not become any better, with negative or sluggish growth at best, high inflation and increasing insecurity. The graphs which show an upward trend in food insecurity bear this out. Figure 8: Evolution of poverty in Burkina Faso National poverty incidence Growth incidence curve 60 51.152.7 Poverty incidence (%) 50 41.443.2 40 30 20 16.6 13.1 10 0 National Urban Rural 2018/19 2020/21 Source: ECVHM surveys. Rural households, and certainly those in the arid zones who regularly experience weather shocks, rely on coping mechanisms to deal with shocks. They diversify their income sources, invest in self-insurance mechanisms like food storage, use livestock as buffer stock, and participate in social sharing mechanisms. These mechanisms are probably overstretched. Consumption is falling on average; insecurity is reducing the ability to diversify incomes through migratory labor, artisanal mining and other off-farm activities; buffer stocks become depleted one day and social sharing mechanisms are less suited for covariate shocks.4 3 High precautionary savings could also explain low purchasing power, but as annex 2 shows, rural households but the perception of risk about the likelihood of their incomes being affected by an adverse weather event or conflict was relatively low in January 2024. 4 According to January-February 2024 round of the HFPS, among rural households, the proportion expressing worry about food scarcity is notably higher for those who were unable to engage in farming during the 2023-24 agricultural season (89 percent) compared to those who did cultivate crops (48 percent). The primary reasons cited for the inability to cultivate include security concerns (49 percent) and insufficient rainfall (35 percent). Of the households 9 A combination of limited agricultural surplus production and declining purchasing power can create the conditions in which markets function nominally and prices are within ordinary ranges, but in which transaction volumes are low on account of limited supply as well as limited demand. These are the circumstances in which extreme food insecurity goes together with functional markets. They may well describe Burkina Faso. 5. Conclusion Famine is a grave a phenomenon, and credible indications that one might be developing should be taken seriously. Data from the HFPS show that in January 2024, during what is supposed the non-lean season, a large fraction of rural households faces moderate or extreme food insecurity. As food security outcomes usually deteriorate during the first half of the calendar year, there is a significant risk that a majority of rural households will be moderately or severely food insecure during the lean season of 2024 in July and August. The data from the HFPS sketch a food insecurity situation that is more alarming than that provided by FEWS NET or WFP. The approaches used by the HFPS and FEWS NET and WFP differ, however, with the first obtaining information directly from households and the latter two considering the evolution of prices. Extreme food insecurity and price stability can co-exist however, especially when household purchasing power is under strain. This may well describe the situation in Burkina Faso where the resilience of rural households has been tested by sluggish growth and a long series of shocks (adverse weather, COVID, inflation, conflict). As next steps, the team recommends first and foremost additional data gathering: a new round of the HFPS is in the field. Additional market data, particularly on the volume of transactions would help understand whether a Sen-type famine is developing. In addition, an early dialogue with WFP, FAO, FEWS NET and the authorities is critical as well. References Ritchie Hannah 2023. “How is food insecurity measured?� Published online at OurWorldInData.org. Accessed April 2024 from: 'https://ourworldindata.org/food-insecurity' Sen Amartya 1983. “Poverty and Famines: an essay on entitlement and deprivation�. Oxford University Press. Lain Jonathan; Stephanie Brunelin, and Sharad Tandon 2021. “Anticipating large and widespread seasonal deprivation in the Sahel�. Sahel Adaptive Social Protection Program Note 3. The World Bank. keeping livestock and expressing worries about food scarcity, 89 percent indicate a readiness to sell their livestock, whereas only 73 percent of those unconcerned about food availability stated the same intention. 10 Annex 1: Data The sample of the BFA Covid-19 HFPS is a subsample of the 2018/19 Harmonized Living Conditions Household Survey (EHCVM). EHCVM 2018/19 interviewed 7,010 households in urban and rural areas and asked households to provide phone numbers of the household head, or a household member, or a non- household member (e.g. friends or neighbors) so that they can be contacted for follow-up questions. At least one valid phone number was obtained for 6,877 households. These households established the sampling frame for the HFPS.5 To obtain representative strata at the national, capital (Ouagadougou), urban, and rural level, the target sample size for the HFPS was 1,800 households. To account for non-response and attrition, 2,500 households were called in baseline round (round 1) of the HFPS. 1,968 households were fully interviewed during the first round of interviews. Those 1,968 households constitute the final sample which is contacted in subsequent rounds of the survey (see Figure A1). Figure A1: Burkina Faso data collection program Each round of the HFPS has one respondent per household. The respondent is the household head or a knowledgeable adult household member. Interviewers are instructed to make every effort to reach the same respondent in subsequent rounds of the survey, in order to maintain the consistency of the information collected. However, in cases where the previous respondent is not available, interviewers identify another knowledgeable adult household member to interview. The HFPS is representative of households who have access to phones in Ouagadougou, urban and rural Burkina Faso. A key concern for any phone survey is the potential for sample bias as phone penetration in the most remoted areas is limited. Fortunately, the number of households without a valid phone number turns out to be limited (less than 2 percent of the EHCVM sample) and no great differences in main socio- economic outcomes are observed between the EHCVM sample and the HFPS final sample. In addition, the phone survey weights were calculated by adjusting the EHCVM 2018/19 household weights to reflect the selection and interviewing process. 6 5 The Burkina Faso HFPS data can be accessed here: https://microdata.worldbank.org/index.php/catalog/3768 6 Additional information about the sampling strategy, the weighting strategy and the implementation of the HFPS can be accessed here: https://microdata.worldbank.org/index.php/catalog/3768/download/57233 11 Being a phone survey, the HFPS has a light questionnaire and different modules are rotated in and out. The module on food security on which this note relies has been asked a total of 16 times since June 2020 and consists of questions developed by the FAO to measure the Food Insecurity Experience Scale (FIES) global reference scale. The FIES measures the share of the population that has experienced food insecurity using a list of eight questions such as: “During the last 12 months, was there a time when you (or any other adult in the household) were worried you would not have enough food to eat because of a lack of money or other resources?� and: “Was there a time when you (or any other adult in the household) had to skip a meal because there was not enough money or other resources to get food?� These eight questions increase in severity: the first question is about worrying that one might not have enough food to eat at some point over the year. This evolves to actually eating less than is sufficient, and finally to going a whole day or more without any food. Figure A2 shows the definitions of food insecurity, ranging from mild to severe. Moderate food insecurity is generally associated with the inability to regularly eat a healthy, nutritious diet and is an important indicator of poor dietary quality and a high risk of micronutrient deficiencies. Severe food insecurity is more strongly related to insufficient quantity of food (energy) and therefore strongly related to undernourishment and hunger. Figure A2: Levels of food insecurity according to the FIES scale Source: Hannah Ritchie 2023. 12 Annex 2 Precautionary savings are a less likely explanation for low purchasing power as households’ perceptions about the risk of their incomes being negatively affected by extreme weather or conflict are relatively optimistic. Figure A3: Likelihood that extreme weather events will negatively household income 100 90 80 Percent of households 70 60 50 40 30 20 10 0 Jun-22 Sep-22 Dec-22 Jan-23 Jul-23 Sep-23 Nov/Dec 23 Jan/Feb 24 Extremely likely Likely Neither likely nor unlikely Unlikely Extremely unlikely Don't know Source: HFPS, various rounds. Figure A4: Likelihood that conflict will negatively affect household income Rural 100 90 80 Percent of households 70 60 50 40 30 20 10 0 Jun-22 Sep-22 Dec-22 Jan-23 Jul-23 Sep-23 Nov/Dec 23 Jan/Feb 24 Extremely likely Likely Neither likely nor unlikely Unlikely Extremely unlikely Don't know Source: HFPS, various rounds. 13