Stochastic risk in forecasting vs. policy oriented uncertainty: the case of alpine crossings

This paper focuses on uncertainties in traffic forecasting. Three major sources of uncertainties are observed for freight demand models. The first one is the model specification itself. We are not interested by it. The second one concerns uncertainties over forecasting hypotheses. A mean to control such uncertainties lies in the introduction of risk in the Costs Benefits Analysis (CBA). Two directions have been taken by this research. The first one is the theoretical framework of CBA under uncertainty mainly developed after Dixit and Pindyck (1994). The second one is more empirical and uses Monte Carlo simulations. Major results of these researches are presented. Then, we apply them to a large transport investment simulation. These tools cannot be used for all kinds of uncertainties. The second part of this paper deals with the third source of uncertainties i. e. policy oriented uncertainties. For them, previous methods are useless. The current Alpine crossings context shows that transport policy is a major determinant of traffics. Furthermore, long term forecasting cannot exclude the possibility of changes in transport policy. This uncertainty should be controlled. It is the role of strategic modeling.