The coronavirus pandemic and food security: Evidence from Mali

Abstract

This paper documents some of the first estimates of changes in experienced food insecurity associated with the coronavirus pandemic in a low-income country. It combines nationally representative pre-pandemic household survey data with follow-up phone survey data from Mali and examines sub-national variation in the intensity of pandemic-related disruptions between urban and rural areas. Although rural households are more likely to experience food insecurity prior to the pandemic, we find that food insecurity increased more in urban areas than in rural areas. Just three months after the onset of the pandemic, the rural-urban gap in experienced food insecurity completely vanished. These findings highlight that understanding effect heterogeneity is critically important to effectively designing and targeting post-pandemic humanitarian assistance.

Keywords: COVID-19; Food Security; Mali; Pandemic; Sub-Saharan Africa.

Fig. 1

COVID-19 Infections and Deaths. Notes: These figures come from the Humanitarian Data Exchange (HDX) COVID-19 sub-national case data, supported by the United Nations Office for the Coordination of Humanitarian Affairs. They show the 7-day moving average of new daily COVID-19 infections and deaths for Mali and the three most populous urban areas: Bamako, Sikasso, and Segou.

Fig. 2

Google Mobility Data. Notes: These figures come from Google’s COVID-19 Community Mobility Data. With the same kind of aggregated and anonymized data used in Google Maps, these data show changes for each day in time spent at specific types of places relative to the baseline period. The Google Mobility Data use a baseline of the median value for the corresponding day of the week during the 5-week period of January 3 through February 6, 2020.

Fig. 3

Coronavirus Pandemic Awareness, Beliefs, and Behavior. Notes: These descriptive statistics come from the World Bank’s COVID-19 high-frequency survey from Mali. Missing and refused responses are excluded from these statistics. Standard errors are clustered at the sampling cluster level. ∗∗∗, ∗∗, and ∗, in each graph’s label indicate statistical significance at the 1, 5, and 10 percent critical level, respectively. Table A1 in the Supplemental Appendix provides additional details on these statistics.

Fig. 4

Food Insecurity Classifications—Baseline Levels. Notes: These descriptive statistics come from the EHCVM sample and represents the baseline levels of each of the three food insecurity classifications.

Fig. 5

The Coronavirus Pandemic and Food Security Challenges—Descriptive Results. Notes: These descriptive statistics come from the World Bank’s COVID-19 high-frequency survey from Mali. Missing and refused responses are excluded from these statistics. Within each panel, the left-most two columns represent mean responses to each of the eight FIES questions (listed below). The right-most two columns represent mean responses to the question, “Was this due to the COVID-19 crisis?”. FS1 = “Household members have been worried that they will not have enough to eat?”. FS2 = “Household members have been worried that they can not eat nutritious foods?”. FS3 = “Household members had to eat always the same thing?”. FS4 = “Household members had to skip a meal?”. FS5 = “Household members had to eat less then they should?”. FS6 = “Household members found nothing to eat at home?”. FS7 = “Household members have been hungry but did not eat?”. FS8 = “Household members have not eaten all day?”. Standard errors are clustered at the sampling cluster level. ∗∗∗, ∗∗, and ∗ in each graph’s label indicate statistical significance at the 1, 5, and 10 percent critical level, respectively. Table A3 in the Supplemental Appendix provides additional details on these statistics.

Fig. 6

Food Insecurity Classifications—Follow-up Levels. Notes: These descriptive statistics come from the COVID-19 Panel Phone Survey sample and represents the follow-up levels of each of the three food insecurity classifications.

Fig. 7

Self-Reported Coronavirus Pandemic Impacts. Notes: These descriptive statistics come from the World Bank’s COVID-19 high-frequency survey from Mali and show unconditional differences in each of these variables between urban and rural areas. Missing and refused responses are excluded from these statistics. Standard errors are clustered at the sampling cluster level. ∗∗∗, ∗∗, and ∗, in the each graph’s label indicate statistical significance at the 1, 5, and 10 percent critical level. Table A2 in the Supplemental Appendix provides additional details on these statistics.