Using models under risk and uncertainty

Devajyoti Ghose

11.1 INTRODUCTION

Let us begin with a bold generalisation: most people are risk averse. Risk managers, almost by definition, are risk averse and, as we shall assert, they are also uncertainty averse. Uncertainty aversion is different from risk aversion, although risk managers may not always explicitly account for the distinction.¹¹ Recall that in the expected utility framework, ambiguity aversion and uncertainty aversion are treated as synonyms and used interchangeably.

In this chapter, we shall suggest simple rules, grounded in the theory of decision-making under uncertainty, for using models that are subject to risk as well as to uncertainty, assuming that decision makers are both risk and uncertainty averse. While decision theory is rather abstract and technical, with much of it stemming from normative axioms for consistent choices, we shall facilitate the transition from theory to practice by developing intuitive rules in the context of the choices available to a modeller or a model user who is interested in managing risk and uncertainty. We shall discuss the suitability of the alternative methods for estimating risk and uncertainty in prediction models, developed in Chapter 10