Instability in Survey-Reported Farm Size: Evidence from Panel Data in Ethiopia and Malawi

Reliable measurement of farm size is central to empirical research on agricultural structure, land inequality, and land use efficiency in developing countries. Most studies rely on single-round household survey data and implicitly assume that reported farm size is stable and accurately measured. This paper questions that assumption using balanced panel data from Ethiopia and Malawi. We exploit within-household variation over time by comparing reported owned farm size in each survey round to the household-specific maximum observed across rounds, interpreted as an upper-envelope benchmark. We document large and widespread shortfalls from this benchmark that are frequently reversed across survey rounds, indicating episodic instability rather than monotonic landholding change. Instability is strongly associated with parcel attrition – captured by deviations from maximum plot counts and unmeasured parcels – while indicators of real landholding change explain little of the observed variation. These findings imply that instability in reported owned farm size can materially affect measured farm size distributions, land inequality, and inferences about land markets and allocative efficiency.