Public Policies for Pest Control: Toward an Analytical Framework

In dryland wheat farming in Western Australia, the way in which a particular paddock responds to fertilizer application can vary markedly from year to year. Uncertainty about grain yield arises because inputs which can be controlled by the farmer, such as the rate of applied nitrogen, interact with exogenous factors such as climate, soil physical and chemical factors, and diseases, weeds and pests. Thus, fertilizer decisions have to be based on farmers' subjective beliefs about yields and its determinants. Farmers may have access to information which reduces production uncertainty before the time of fertilization decision. For instance, soil tests which measure soil nutrient levels have been available to farmers in Western Australia for several years. While they have been widely adopted in determining rates of phosphorus and potassium fertilizers, their usefulness in making nitrogen fertilizer recommendations is unclear. Recently, it has been claimed by CSBP that nitrogen applied in accordance with soil test results and the CSBP soil testing model was highly profitable, even in ares experiencing poor seasonal conditions (Wesfarmers Ltd Annual Report 1985). The aim of this paper is firstly to analyse existing opportunities for farmers to make decisions on fertilizer application rates contingent on available information about soil nutrients and seasonal conditions; and secondly to evaluate the potential value of further research on soil chemistry and plant nutrition to improve fertilizer decision making. To estimate the value of the information on soil nutrient status, it is necessary to compare the value of the optimal strategies, given utilization of the information with value of the optimal strategies if it is ignored. An appropriate framework for this is Bayesian decision theory (Anderson, Dillion and Hardaker 1985). This theory takes account of the precision of the information that is available, and is a useful basis for comparing costs and benefits of different sources of information, such as soil test data vis-a-vis paddock histories.