Stochastic dominance analysis for the assessment of intertemporal policies

In this dissertation I develop insights and tools concerning a sustainable evaluation of intertemporal policies. My study concentrates on the temporal dimension characterizing public projects, environmental policies whose effects will be spread out over a long number of years. The results render contributions to fields of intertemporal decision theory and environmental economics, as well as to them more specific areas of sustainability and cost benefit analysis. This section gives a brief overview of the key issues addressed in my dissertation. When addressing long term decision making problems, like for example the economic evaluation of climate policies, there is a lot of disagreement in evaluating the desirability of alternative policies. The core of these discussions about intertemporal choices, is based on frequently contested assumptions about time and risk preferences. This could lead to legitimate disagreement about which discount function, representing the temporal preferences, and which utility function, representing risk preferences, to use. Motivated by these observations, in the first chapter of the thesis we develop a method for finding spaces for agreement, giving an interconnected access to the domains of time and uncertainty when comparing alternative intertemporal, risky policies. We first formalize the Time-Stochastic dominance criteria, and then, wanting to exclude the extreme, pathological risk preferences, we introduce the Time-Almost Stochastic criteria. Using an integrated assessment model, we apply the dominance criteria developed and ask whether there are climate policies that everyone who shares a broad class of time and risk preferences would agree are better than business-as-usual. The use of a constant discount rate for the appraisal of very long term public projects has created a strong debate in the economic literature. Considering the difficulty of arguing persuasively for one given discount function, in the second chapter we concentrate on deriving intertemporal dominance conditions for ordering finite streams of unidimensional deterministic outcomes, distributed in discrete time, focusing on different classes of discount functions, subject to various restrictions. We extend the existing, standard time dominance results by considering complete monotone discount functions, and we investigate the possibility of increasing the comparability of intertemporal prospects by concentrating on the infinite order time dominance. We parallel our results with those available, in the continuous space, for infinite stochastic dominance and Laplace transforms dominance. Finally, the third chapter of this dissertation investigates intertemporal dominance conditions for comparisons of finite unidimensional streams of outcomes in discrete time. We follow Ekern's [1981, Time dominance efficiency analysis. Journal of Finance, 36, 1023-1034] approach based on unanimous net present value comparisons for classes of discount factors representing temporal preferences. In order to overcome the problem of dictatorship of the present in intertemporal evaluations we restrict the class of discount factors, by imposing a limit on the decrease of the weight attached in the current evaluation, between the outcomes of two future adjacent periods. The restricted time dominance theorems provide parametric dominance conditions that make explicit the policy maker's trade-off between current and future periods and the willingness to postpone her myopic judgement. We show that these conditions can be summarized by a single cutoff point that can be interpreted as the maximal decrease in the discount factor, which guarantees unanimous dominance for the class of intertemporal preferences considered