Essays on Technology Adoption in Senegal
African agriculture is characterized by very low average productivity. This results in a very high yield gap, i.e. the average yields achieved by farms are up to 90% below the yields that can be achieved by applying proven best-practice technologies. A central problem of low agricultural productivity is, therefore, technology adoption, i.e. the question of why farms do not apply available best-practice technologies. In this context, this dissertation investigates the mechanisms of technology adoption using a unique farm data set of more than 4000 farms in Senegal and innovative econometric methods. A first descriptive analysis reveals dual Senegalese agriculture with a small percentage of farms using modern technologies, i.e. irrigation, use of mineral fertilizers and pesticides and improved seeds, and a majority of farms using traditional extensive farming without the use of purchased inputs and irrigation. For example, the use of N-fertilizer in the majority of traditional farms is less than 30kg/ha while modern farms use more than 300kg/ha. While a shift from traditional to modern agriculture at the macro level has a clear positive effect on food security and rural development, the question arises as to the key micro-level barriers that prevent traditional farms from using modern technologies. While the potential obstacles have been identified from the theoretical literature, i.e. transaction costs in credit, labor, goods, and insurance markets as well as imperfect technological knowledge of farmers, for practical agricultural policy the question arises as to which are the central causes in a specific empirical case. This is particularly important because the efficient agricultural policy measures to reduce these obstacles differ significantly depending on the specific obstacle. In this interesting and highly relevant area of agricultural policy, the present study makes central contributions by applying innovative econometric methods for the microeconomic analysis of technology adaptation, i.e. the concrete obstacles to the application of modern agricultural technology at the farm level. In total, the dissertation comprises 4 contributions. In the first contribution, a flexible bivariate probit model is applied to analyze the joint use of certified seed and mineral fertilizer in rice and peanut production. While the flexible versus the standard probit model is theoretically and statistically preferable, both approaches lead to the same key policy implications. The second paper analyzes the impact of multiple technology decisions on technical efficiency and yield using rice production as an example. On a methodological level, the paper combines a metafrontier approach with a multinominal treatment-effects model to take into account the heterogeneity in rice production and potential selection bias in the choice of technologies. A remarkable result of the analyses is the identification of significant knowledge gaps as a central obstacle for the use of modern inputs. The third paper examines the importance of yield risk for the use of modern inputs and its significance for the income and food security of agricultural households. Methodologically, an endogenous switching regression model is used to adequately analyze the treatment effects of modern input use. In the fourth paper, an interdependent farm household model is used as a theoretical approach to analyzing participation in relevant agricultural input and output markets. Transaction costs are a central determinant of the market participation of agricultural enterprises. Since transaction costs can be specific for different input and output markets, different market regimes result, including complete self-sufficiency, selective participation in specific output or input markets, and complete market participation. Methodologically, a multinomial endogenous treatment effects model is applied to empirically analyze the market participation decisions of individual farm households. Interestingly, farms participate selectively in output and input markets. This implies market-specific transaction costs, which cannot be explained by general factors such as infrastructure and market distance, but rather, for example, by specific social network structures that determine selective access to markets.