Estimating the effect of climate on agriculture with the productive value of agricultural land

Objective: the Ricardian model, in which market land values are modeled as a function of climate, has been estimated extensively in the context of developing countries where its core assumptions are likely to fail due to missing or incomplete markets. This article proposes a new measure of land valuation that better reflects agricultural productivity in such contexts: the Productive Value of Agricultural Land (PVAL). Methodology: a three-step empirical strategy is applied to data from a survey of Mexican rural households. The first step is the estimation of an agricultural production function. In the second step, parameter estimates are used to calculate PVAL. In the third step, PVAL is used as the dependent variable in a Ricardian regression. Results: suggests that PVAL increases with more precipitation and decreases with by extreme heat. When the Ricardian regression is estimated using market land values, the positive effect of precipitation is underestimated and the effect of extreme heat on land productivity is null. Limitations: omitted variables bias could still influence the Ricardian estimates obtained using PVAL. Originality: a novel version of the Ricardian model is estimated, one that does not rely on market values of land. Conclusions: failing to account for the market setting of agricultural producers, particularly in developing countries, may lead to an underestimation of the effects of climate change in agricultural productivity.