A rainfall shortfall barely moved household calories in Malawi. What changed was the plate: nutrient-dense foods lost 0.7 percentage points and staples filled the space

Food-security surveys often ask whether a household obtained enough energy or ate often enough. Researchers asked whether that misses a quieter response to bad weather: keeping calories steady while replacing varied, nutrient-dense foods with staples. They analysed four waves of Malawi's household panel, covering 5,866 household-wave observations in 2010, 2013, 2016 and 2019.

A rainfall deficit of 20 percentage points was not robustly linked with lower energy intake. It was linked with a 0.7-percentage-point fall in the share of energy from nutrient-dense foods, about 4.1% of that share's mean. The staple share rose, and fruit and pulses bore the adjustment while cereals did not.

Households also consumed 0.7 fewer distinct food items, and the share from their own production fell by 1.9 percentage points. Real food expenditure did not rise, suggesting that market purchases did not fully replace what farms lost.

The authors give the uncertainty unusual prominence. The diet-composition measures are different summaries of the same food vector, not independent confirmations. With standard errors that allow nearby places to share the same weather shock, a joint test across four composition measures had p=0.10. The estimated direction was consistent but modest and imprecise. The practical lesson is therefore about measurement: calories alone may miss dietary thinning, while a single significant-looking result can overstate certainty if geography is ignored.