Introduction to “Risk” Analysis

Summary

The previous chapters and Volume 2 deal with cost estimating of the “nominal” project which means the project as it can be represented:

• by a nominal description of what is to be realized,

• by a nominal description of how it will be done,

• assuming that everything is going to be all right.

We already saw that this cost has a certain level of inaccuracy, due to the fact it is computed using a sample, which bears on a limited set of values, and these values may be more or less scattered.

But no project is really nominal. This does not mean that the nominal project is purely ideal: it means that some realistic hypotheses were made and that it is assumed that these hypotheses are correct.

There are a lot of reasons for a project to be more or less different from what is expected. The consequences of the differences are generally called “risk”.

Cost estimates cannot be perfect. There are three reasons for this imperfection:

• 1. Inaccuracy due to the cost estimating tool which is used for preparing the estimate. This has already been dealt with when discussing estimating methods; it may come:

• from the data themselves (how scattered they are),

• from the methods used to extrapolate from the data.

• 2. Imprecision related to the parameters describing the nominal project.

• 3. Uncertainty about what is going to happen during the project life.

Dealing with risk can be, and is sometimes, done manually. This is quite possible for small projects but would be difficult for large projects.

As mathematical tools for carrying out this analysis are now well known and easy to implement even on desk computers,we will develop some mathematical concepts about it. The purpose is not to improve the “accuracy” of risk analysis (this word has little meaning here), but to carry it out very quickly letting the user think about the inputs and the outputs, forgetting the translation from the first ones to the second ones.

Both imprecision and uncertainty are dealt with in this chapter.