A decision-theoretic approach to the planning of agricultural extension : a thesis presented in partial fulfilment of the requirements for the degree of Master of Agricultural Science in Farm Management at Massey University
The extension agency, faced with the need to make more effective use of its resources, requires information about the value of the alternative extension messages which it expects will assist farmers to increase net income. It is hypothesised that Bernoullian decision theory is applicable to the extension agency's problem by helping it to assess the expected value of the increases in aggregate farm incomes following extension. An extension message is seen as assisting farmers to make decisions and thereby increasing expected income. Where the extension information is aimed at helping the farmer estimate the occurrence of the uncertain events in a decision problem, Bayes' theorem provides the basis for a method of obtaining the value of the information. An extension message can also assist by helping to analyse the decision problem or by providing information about some new or innovative course of action for solving the problem. The difficulty encountered by most published methods for evaluating agricultural extension is that of determining the proportion of the change in farm income due to extension and that due to other factors which are also affecting farm income. The method outlined in this thesis relies on a preposterior estimate of the value of an extension message which largely overcomes the problem of estimating the without-advice situation. A start was made on testing the proposed method by obtaining information from several dairy farmers about specific decision problems, the alternative courses of action and the other details that would enable a model of the decision problems to be synthesised. Because of the difficulty of obtaining that information, and of developing an adequate model of a problem, the attempted application was reduced to one farmer and the particular problem of summer-feeding of the herd. Summer rainfall, pasture growth, milkfat output and milkfat price were the sources of uncertainty which were incorporated into the decision model. The analysis indicated only limited potential for additional information to assist the farmer with the decision problem. The research provided some support for the hypothesis since it was found to be possible to simulate a farmer's decision problem under uncertainty and to obtain a pre-posterior estimate of the farmer's expected income without advice.