Climate Change Impact on Nigeria’s Agriculture: A comparison of Revenue Versus Self-Reported Land Value

Abstract

This study investigates the sensitivity of Ricardian climate impact estimates to different land value conceptualizations in Nigerian agriculture. Using farm-level data from the National Bureau of Statistics and climate projections from the WorldClim database, the paper compares two Ricardian model specifications: one based on farm revenue (Net Revenue, NR) and the other on Self-Assessed Value of Land (SAVL). The NR model captures current earnings from crop and livestock production, while SAVL reflects farmers’ perceived market value of land, incorporating future uncertainty. To ensure comparability, NR was adjusted using a 5% interest rate to approximate the discounted stream of future returns. A semi-log functional form was employed, with climate variables specified as seasonal averages and quadratic terms, while soil, altitude, and socio-economic indices were included as controls. Weighted least squares regression was used to address spatial aggregation bias. Results show that climate variables significantly influence both models, though the direction and magnitude of effects differ. Temperature generally had a detrimental effect on land value (−6.01%) but a modest positive effect on revenue (+1.45%), while precipitation increased land value (+6.50%) but reduced revenue (−1.85%). The NR model produced more extreme marginal estimates across seasons and overestimated climate-induced damage by 3.6% compared to the SAVL model, a difference found statistically significant at the 10% level. These findings support the hypothesis that unadjusted revenue models exaggerate climate impacts due to their failure to account for future uncertainty. The study underscores the importance of land value conceptualization in climate impact modeling and recommends caution in using revenue-based proxies for policy analysis in developing agricultural economies.