| This research develops a dynamic analysis of the role of risk and learning in technology adoption, with an empirical focus on the adoption of Genetically Modified (GM) corn in the U.S. Corn Belt. A conceptual structural dynamic programming (DP) model is developed to capture the relative roles of individual and social learning under uncertain profitability for both conventional and GM technology.;The DP model involves solving Bellman equation under imperfect state information. It relies on sufficient statistics given by the mean and variance of profit under normality assumption. Farmers' learning process is given by the evolution of the mean and variance of profit and represented by the Kalman filter algorithm, where degrees of individual learning and social learning are parameterized. And farmers' risk aversion is specified using an additive mean-variance utility function under normality and Constant Absolute Risk Aversion (CARA). Parameters are estimated by nesting the DP problem (Bellman equation with the Kalman filter) within a minimum-distance estimator.;The model is applied to a unique panel dataset of U.S. corn farmers collected by dmrkynetic (DMR). Four models of farmers' adoption pattern are developed. First, an aggregate model of a representative farmer is applied, covering the whole DMR panel dataset (a benchmark case). Second, three disaggregate models are applied to three sub-groups of farmers: the early-, the intermediate-, and the late- adopters of GM technology. The disaggregate models allow the investigation of parameter heterogeneity across groups of farmers.;Our empirical analysis provides new and useful information on the role of risk, risk aversion, and social learning in GM technology adoption. Our results confirm the followings: First, farmers are risk averse in adopting new technologies (GM seeds). Second, both individual and social learning play a key role in adopting GM technology in positive directions with high statistical significance. Third, the impacts of individual learning are shown to be larger than the impacts of social learning. Further, our empirical results across farm types demonstrate that farmers adopt GM technology later as they are more risk-averse and rely more on social interactions (as information externalities are stronger). |
