A decision-analytic approach is taken to the problem of assessing the economic value of imperfect weather forecasts. Emphasis is placed on measures of the quality of such information and on the relationship between quality and economic value. The fallowing/planting problem for a spring wheat farmer is examined in detail as a specific application. It is assumed that the farmer's goal is to maximize the total expected discounted return over an infinite horizon, which places this problem within the general framework of Markov decision processes. By means of stochastic dynamic programming, the economic value to the farmer of currently available seasonal precipitation forecasts, as well as of hypothetical improvements in the quality of such forecasts, is estimated. Because the relationship between the quality and value of forecasts is highly nonlinear, the need to explicitly determine valueof-information estimates, rather than relying on quality as a surrogate for value, is made clear.
KEY WORDS Dynamic programming Markov decision process
Value of information Weather Forecasts Consideration of the economic value of weather forecasts is appealing because such information is regularly provided to potential users and the corresponding observed weather events are routinely recorded as well, implying that the quality of the forecasts can be determined in a straightforward manner. In fact, in recognition of the uncertainty inherent in this information, many weather forecasts are expressed in the form of probabilities (Murphy, 1985). Further, because weather forecasts are actually employed in real decision-making situations (e.g. Katz et al., 1982), the application of decision-analytic models in this context is not restricted to purely hypothetical studies. Our interests lie primarily in quantifying the economic value of weather forecasts in specific decision-making situations, including the related problem of obtaining a relationship between improvements in the quality of such forecasts and concomitant changes in the value of this information.
* Also affiliated with National Center for Atmospheric Research, U.S.A.