Technology adoption and risk exposure among smallholder farmers: Panel data evidence from Tanzania and Uganda

Abstract

This paper investigates the empirical linkages between production risks and the adoption of modern inputs among smallholder farmers in Tanzania and Uganda using household panel LSMS-ISA datasets. Applying a moment-based approach and a Mundlak-Chamberlain IV fixed effects model to control for unobserved heterogeneity and endogeneity, the paper uses a translog production function to estimate the mean, variance, skewness, and kurtosis of production. These estimated moments of production are then included in a multivariate adoption model to assess their effects on input adoption decisions. Results reveal that the first four moments of production significantly explain changes in the probability of adopting chemical fertilizer, improved seeds, and pesticides. While the use of these modern inputs is found to be risk decreasing, estimates suggest that the higher their purchasing costs, the greater the cost of farmers’ private risk bearing. Under the assumption of a moderate risk aversion, the risk premium amounts to 8.7% and 13.7% of the expected production revenues in Tanzania and Uganda, respectively. This willingness-to-pay is largely explained by production volatility and downside risk aversion and to a small extent by kurtosis aversion. The findings underscore the need to account for farmers’ preferences towards higher order moments when designing and implementing modern inputs’ adoption policies in developing countries.