Are National Agricultural Panels Fit for Purpose? Parcel Instability, Land Rental Market Measurement, and Allocative Efficiency in Sub-Saharan Africa.

Abstract of the Paper

Land rental markets are often viewed as an important mechanism for improving allocative efficiency in smallholder agriculture. A growing literature has used nationally representative household panel data, particularly the LSMS–ISA surveys, to study land rental market participation and land allocation in Sub-Saharan Africa. However, the reliability of these analyses depends critically on how accurately land ownership and operational holdings and rental transactions are measured in the survey data.

This paper examines how measurement instability in reported land holdings affects empirical assessments of land rental markets and allocative efficiency. Using LSMS–ISA panel data from Ethiopia, Malawi, and Uganda, we document substantial instability in reported ownership holdings across survey rounds and show that this instability is systematically correlated with land rental participation. Rental land access statistics for landless and land-poor tenants are particularly sensitive to instability in reported ownership, with strong implications for the assessment of how pro-poor land rental markets are. We also find persistent large imbalances between reported tenant and landlord activity, suggesting substantial underreporting of rented out land.

We then assess allocative efficiency using three complementary approaches: the Bliss–Stern benchmark relating net land leasing to owned land, dynamic models of tenant participation, and conditional rental intensity models. The results indicate that ownership measurement instability can affect estimates of allocative efficiency, but tenant-side dynamics remain informative. Overall, the findings suggest that while survey-based estimates of landlord participation may be unreliable, nationally representative panel data can still provide useful insights into land rental market functioning and underlying reasons for limited adjustment when measurement constraints are explicitly addressed.