Un modèle canonique d'option réelle bayesienne et son application au principe de précaution

Uncertainty on effective stocks or damages is the core of most irreversible economic decisions involving natural resources and the environment and, more generally, of most investment decisions. The discussion on the relevance of the precautionary principle is illustrative of this role. Although there is an extensive literature dealing with the theoretical aspects of this problem, economists still lack an operational decision rule. The matter is that the real option theory, now widely used to deal with markovian uncertainty, i.e. independent exogenous repeated chocks affecting the expected return from an irreversible project as time goes, is not adequate to analyze problems characterized by Bayesian uncertainty, i.e. problems where uncertainty about the effective value of parameters is resolved with time by a learning process based on the acquisition of information in the form of quantitative as well as qualitative signals. This paper is aimed at remedying at these shortcomings. The real option theory is adapted to an explicit Bayesian learning process in continuous time. Application to Genetically Modified Organisms farming is used as an illustrative example.