Strategic Decisions under Nonstationary Conditions: A Stopping-Control Paradigm

Abstract

Long-range engineering planning involves multiple objectives and uncertainty about a future state of the environment that will evolve from the current state through nonstationary stochastic processes (climate change is one possible source of nonstationarity). The questions that often arise are: When is it optimal to stop waiting for a reduction of uncertainty and make a commitment to a plan ? What is the optimal magnitude of a control or design variable (size of the project) ? This decision problem is modeled as a finite horizon, discrete-time, continuous-state, nonstationary stopping-control process with Markovian forecasts of an uncertain state, Bayesian updating of distributions, and a multiattribute utility function representing the decision maker’s preference over outcomes. The model rests on Bayesian principles of rationality. Aspects arising uniquely from the nonstationarity of environmental processes are highlighted.