Reading: Coping with Measurement Error

The most prominent way to assess measurement error is to build scenarios. Based on the information provided, we could create the following three scenarios: Best case, expected case, worst case. The following table summarizes the scenarios.

For each variable, the best case takes on the most optimistic outcome whereas the worst case takes on the most pessimistic outcome. Realistically, the expected case corresponds to the assumptions we used for our base case valuation so far:

Conducting the scenario analysis is fairly simple. If we have set up the valuation model in Excel, all we have to do is change the cell which contains the respective assumption.In fact, Excel even contains a "Scenario manager,” which is listed under Data à What-if-analysis:

This scenario manager allows us to compile various scenarios in the same worksheet without having to change the content of individual cells. Therefore, it provides a relatively fool-proof approach to scenario analysis.

If we plug the various scenarios in our valuation model, we can find the resulting firm values. The last row of the above table summarizes the results: Under the best-case assumptions, the estimated firm value is 72'451 whereas the worst case assumptions yield a firm value of 46'000.Therefore, the valuation is fairly sensitive to our assumptions. The possible value range between the best and the worst case is 26'388, which corresponds to 46% of the base case value.

The main advantage of the scenario analysis is that it is simple and intuitive. Because people often think in scenarios, they are also fairly easy to communicate.

Remember from before that we wanted to answer the following questions with respect to potential measurement errors:

• Which are the main drivers of firm value?

• What can happen if our assumptions are wrong?

• How likely are the various outcomes?
  

The scenario analysis only provides partial answers:

• First, it does not identify individual value drivers. We see that the valuation is sensitive to our assumptions, but we do not see which assumptions are particularly critical. For example, it could be that measurement error in the credit spread has virtually no impact on the valuation whereas the impact of the long-term EBIT-margin is significant.

• Second, it provides some information about what can happen if our assumptions are wrong. The problem with our simple analysis above is that the world typically has more than 3 scenarios. It could be that some variables turn out worse than expected whereas others are better than expected. Put differently, it makes the very strong assumption that all variables are perfectly correlated.

• Finally, it does not tell us much about the probabilities of the various outcomes.
  

Therefore, the scenario analysis provides a first starting point for a careful analysis of measurement error. For a more profound analysis, we have to turn to additional techniques.