Intra-African Agricultural Trade and Food Availability: Product-Level Panel Evidence

Abstract

This paper asks whether a larger intra-African share of food imports is associated with
higher national food availability and with less volatile availability. We merge BACI
bilateral trade flows with FAO Food Balance Sheets in a panel of 43 African importers and 81 product
units (37,102 observations, 2010 to 2023). The outcome is ln(1 + food supply quantity in kg per
person per year); models use country-product and year fixed effects, country-clustered errors and
wild bootstrap inference; fixed-effects Poisson (PPML) and weighted estimates serve as sensitivity
checks. Comparisons across country-product pairs show a positive association (+1.8% in 1
+ availability per 10 percentage points), which vanishes within pairs (-0.11%; 95% CI -0.30% to
+0.08%, on the same scale), that is -0.002 kg per person per year (95% CI -0.005 to +0.002) at the
median availability of 0.73 kg. A zero-inclusive PPML estimator gives -0.41% in availability (95% CI
-0.84% to +0.02%). Slopes differ across categories (p = 0.018), but negative estimates for oils and
pulses do not survive multiplicity adjustment. Supplier concentration is not significantly related to
volatility. These null results do not show that supplier origin cannot matter. To our
knowledge, few studies relate the supplier composition of imports to product-level availability in
Africa.  Within pairs, availability is positively associated with import volumes, partly by
accounting construction, but not detectably with their supplier composition; integration may not be
well judged by its share alone, and monitoring could usefully track availability. The findings do not
evaluate trade facilitation, sourcing schemes or reserves.