Rural land rental markets in developing countries: Can survey design innovations improve land market participation statistics?

A longstanding puzzle in the African land rental market literature is the often-observed discrepancy between the number of tenants (renters-in) and the much smaller number of landlords (renters-out) in survey data. If this discrepancy derives from systematic biases in survey data responses on rental market participation, then the existing body of survey-based empirical work on land rental markets impacts may be fundamentally flawed. To examine this issue, we implemented two survey experiments. First, we tested the hypothesis that some categories of rented land are underreported because enumerators and respondents focus primarily on parcels directly managed or cultivated by the household. A random subset of respondents received a priming nudge reminding them to account for all land, including rented- or sharecropped-in and rented- or sharecropped-out parcels. Second, we tested whether households underreport rented- or sharecropped-out land due to reluctance to disclose activities that may carry social or institutional repercussions, using a double-list experiment to infer true rates of participation. Interestingly, our results indicate a significant underreporting of both renting-in and renting-out land but arising through different mechanisms. The priming nudge increased reports of renting-/sharecropping-in by 4 percentage points (equivalent to 13% of landlords in the sample) but had negligible effects on reported renting-/sharecropping-out. By contrast, the list experiment suggests that the true share of renting-out households is about 15%: much higher than the 3% in parcel-roster responses. These results underscore the need for improved survey methods to accurately observe land rental market participation and evaluate its impact.