CLIMATE CHANGE AND AGRICULTURAL PRODUCTIVITY IN SUB-SAHARAN AFRICA: A SPATIAL SAMPLE SELECTION MODEL

Using data at a high spatial resolution, we estimate a cereal yield response function conditional upon climatological and topographical features using a recently developed estimator for spatial process models when sample selection is of concern. We control for localized spatial correlationin unobserved disturbances affecting both the selection to plant cereals as well as in the resulting conditional yield response. We find that cereal yields across Sub-Saharan Africa will decline with increasing temperatures resulting from global climate change, and that failing to control for sample selection leads to underestimation of these adverse effects.