A Bayesian approach to German agricultural yield expectations

Agricultural yields depend on an encompassing set of technical and environmental factors. The development of such factors often is not fully observable, and their interactions and impacts on yields have not been completely understood. This paper proposes a Bayesian method for forming agricultural yeidl expectations based on past yields. With this method, the development of expectations on yield trend and variability over time is retrace, and expectations for the future are derived. German winter wheat, corn, and aggregated cereal yield data from 1950 through 2006 on the national scale are used for updating. It is shown that the expectation that yields follow a stable positive linear trend with increasing variance become the dominant hypothesis by 1990, and gains a final weight of more than 99 per cent for all crops considered.