In an earlier entry, I started an examination of the benefits of conducting a risk analysis (or risk assessment). The focus of that discussion was the around the idea that a static Excel model is generally inherently biased, and that part of the role of risk analysis (in financial decision making) is to establish the average. In other words, part of the reason is to establish a base figure. In this blog we discuss the role of risk analysis and modelling in establishing variability of the output.
First, whereas financial theory often puts a lot of emphasis on the average of the outcomes, real-life decision-making involves consideration of output variability. For example, one may reject a project which is profitable on average (i.e. has a positive net present value, defined as the average of the discounted cash flows) because some of the possible outcomes are unacceptable to the decision maker (whereas core financial theory says that the average is the generally the relevant quantity). In other words, consideration of variability may allow a decision maker to bring in their own risk tolerance into the decision. A similar argument applies when calculation a required safety margin above a base case (e.g. in cost budgeting).
Secondly, there are situations where the consideration of variability is an inherent part of the modeling process. For example in the valuation or calculation of contingent claims (e.g. options, real options, profit share agreements), the variability needs to be considered in order to be able to perform the subsequent calculations (which typically involve calculating the average of the value of a non-linear payoff function).
Third, in non-financial contexts, the mean (average) may not be a relevant decision criteria. For example one’s average arrival time at work may not impact whether an employee is regarded as valuable or reliable, nor may the average level of toxins in a potentially contaminated product be relevant. Formally these situations are ones where the link between the modeled quantity and the truly relevant quantity is of a non-linear nature that is however not captured in the model (e.g. the bosses utility function relating to an unreliable employee); often the capturing of these effects is too complex or impractical and the translation of the model’s output (e.g. arrival time at work) into the true effect (whether the employee is fired, gets a raise etc) is made outside of the model.
Stay tuned: we will continue to examine the importance of risk analysis for decision making under uncertainty in a few entries to come.
Dr. Michael Rees
Director of Training and Consulting