Modeling Inflation Uncertainty in Monte Carlos

Generally, when one performs a Monte Carlo simulation to model the distribution of uncertainty in a cost estimate, they rarely evaluate inflation risk. Costs are normalized into Constant Year (i.e. Base Year) dollars and the Monte Carlo is run without any uncertainty placed on the forecasted inflation index. I will show one way of incorporating inflation risk into your risk analysis which will tend to create a wider distribution of cost outcomes as well as higher most likely values.

Below show an example where an estimate with two different cost elements receive different triangular assumptions.

Inflation Risk in Monte Carlo Simulations; Green Cells Stochastically Determined

What makes this model unique is that the inflation and escalation indexes also have triangular assumptions. The costs are time-phased over five years with differing outlays (shown in column G), and each of these five years will have a stochastically (randomly) determined inflation value.

For each run, a randomly picked Base Year (Period 0) cost will have a weighted index that is also randomly determined applied to it. Performing thousands of these types of runs will provide one with a range of potential cost outcomes. It is important to realize that the output is in Then Year dollars and not Base Year dollars.

In this Monte Carlo, we now have incorporated technical risk (the distribution around the Base Year cost estimate) as well as inflation risk (different draws from a distribution for each year of the outlay).

One should certainly look into injecting some autocorrelation into the inflation rates by using a Markov chain. This is because a year's inflation rate is not independent of the rate observed in the previous year, but in fact is highly dependent upon it.