Dynamic Cost-Loss Ratio Decision-making Model with an Autocorrelated Climate Variable

Abstract

A dynamic decision-making problem is considered involving the use of information about the autocorrelation of a climate variable. Specifically, an infinite horizon, discounted version of the dynamic cost-loss ratio model is treated, in which only two states of weather ("adverse? or ?not adverse") are possible and only two actions are permitted ("protect? or ?do not protect"). To account for the temporal dependence of the sequence of states of the occurrence (or nonoccurrence) of adverse weather, a Markov chain model is employed. It is shown that knowledge of this autocorrelation has potential economic value to a decision maker, even without any genuine forecasts being available. Numerical examples are presented to demonstrate that a decision maker who erroneously follows a suboptimal strategy based on the belief that the climate variable is temporally independent could incur unnecessary expense. This approach also provides a natural framework for extension to the situation in which forecasts are available for an autocorrelated climate variable.