Misidentified household food insecurity: how relying on a single survey round produces targeting errors in food security programs

Abstract Humanitarian assistance programs in low- and middle- income countries often depend on household surveys conducted at a single point in the year to identify which households are food insecure and eligible for assistance. Yet household food security can fluctuate substantially across seasons and weeks. As a result, a one-time survey conducted during an unusually good or bad week may misidentify households’ actual food security status, leading to targeting errors, either incorrectly including or excluding households from assistance. Utilizing unique year-long weekly financial diary data and Monte Carlo simulations that replicate single-round household surveys, we quantify misclassification rates for three widely used food security indicators. Misclassification rates for a single indicator are substantial, ranging from 15 to 48 percent. Combining multiple indicators reduces misclassification rates to approximately 6 to 15 percent, using data and tools already available to program implementers.