Maybe I should make a Decision Making week around here. Because yesterday’s Financial Modeling post got me thinking about sensitivity modeling. It’s not just about financial modeling.
According to Wikipedia, Sensitivity Analysis is:
the study of how the uncertainty in the output of a mathematical model or system (numerical or otherwise) can be apportioned to different sources of uncertainty in its inputs.
You may be wondering what that means. It boils down to the fact that for every given input, you should understand how that value could change. Let’s take a simple example, inflation rates could govern everything from wage inflation to lease rate increases to revenue projections. But if you forecast a single value for inflation – let’s say 2% – your model will be exactly wrong for most years. Inflation is a variable that goes up or down annually. It’s relatively predictable but unknowable exactly until after the fact.
When making decisions, some may be more or less pegged to inflation. Maybe certain markets have rents that vary more in line with inflation while another runs independently. Therefore a point estimate of 2% for will not tell you the true expected value of your options. Because in lower inflation years the scenario most pegged to inflation may be the best choice but in higher inflation years the scenario less pegged would be the better bet. But you’ll never know unless you measure the sensitivity of the scenario outcomes around the inflation input.
And this is just around one relatively straightforward input. Now think about varying corporate growth projections, market adoption expectations, employee turnover rates (especially important for call center clients) and many others. Suddenly that point solution you thought gave you such a clear cut answer is actually misleading you to a very significant extent.