Determinants of Land Allocation in a Multi-Crop Farming System: An Application of the Fractional Multinomial Logit Model to Agricultural Households in Mali

Effective food security work in developing countries, such as Mali, relies on a thorough understanding of the rural farming system. A common approach is to study land allocation decisions to specific crops. In accomplishing this, one challenge is to model all production outcomes in a multi-crop system. This paper attempts to overcome this challenge to study the determinants of household allocation to cotton, maize, sorghum, millet, and secondary crops. First, a reduced form of the agricultural household model helps to identify factors that explain land allocation to various crops. This framework is applied to survey data from six villages in Mali’s Koutiala Cercle. A fractional multinomial logit econometric model is used to estimate the effect of household and production attributes on shares of cotton, maize, sorghum, millet, and secondary crops simultaneously, the results of which are presented as average marginal effects. Among other results, the analysis shows that ethnic groups not native to the Koutiala Cercle are associated with significantly smaller shares of maize, and that villages with better market access are correlated with much higher shares of secondary crops and smaller shares of cotton. These results provide insights for policymakers on the role of cotton in farming system, the need to promote and develop better markets for coarse grains and secondary crops, and the importance of understanding the dynamic farming system in Mali’s Koutiala Cercle.