Practical Spreadsheet Risk Modeling for Management

Selecting Distributions 111

We caution readers that these comparisons are likely to vary across

data sets. Comparisons of the models will depend on the volatility of the

data, the actual form of the best tting distributions, and the amount of data

available. In particular, as more data are available, the differences between

these methodologies will decrease. It also seems that parameter uncertainty

will have more profound impacts on skewed distributions than on sym-

metric ones.

While ModelRisk includes powerful techniques for tting distributions to

data and the ability to include and quantify the uncertainty about the tted