Now that the American elections are over and once again the predictions made by the vast majority of pollsters have been shown to be inaccurate, it seems a good time to reflect on our habit of crystal ball gazing.
In The Black Swan, Nassim Taleb shows that we continually construct narratives to explain the past. Then we convince ourselves that we understand the past and therefore become overconfident in our ability to predict the future.
In Thinking, Fast and Slow, Daniel Kahneman points out that we can know something only if it is true and knowable. No one could possibly “know” the outcome of the American elections or the Brexit vote in advance. However we often hear experts or politicians discussing what “will” happen and how they “predicted” a certain outcome.
Kahneman cautions against about jumping to conclusions on the basis of flimsy evidence (WYSIATI or What You See Is All There Is). He gives many examples of over optimistic forecasting. For example, the Scottish Parliament building was completed in 2004 at a cost of approximately £431 million, having been estimated at £40 million in 1997.
So why is all this relevant to financial modelling?
Firstly, we need to remember that financial models are based on assumptions, some of which will be more reliable than others. Good contracts can reduce some of the risk. However, each model should also include a robust sensitivity analysis. Sometimes people seem to view the Base Case as exactly what will happen without much analysis of other potential options.
Secondly, in a financial model a crystal ball gazing formula can easily appear. This is when a particular cell (A) is calculated by looking at something as yet unknown in the future which itself depends on cell A: this will generate a circular reference. Why is this unnecessary and also fundamentally flawed? Because the calculation of any number can only ever depend on the information available at the time of the calculation (WYSIATI).
As Pliny the Elder said: “The only certainty is uncertainty”. He evidently knew about circular references.