The Flaw of Averages: Why We Underestimate Risk in the Face of Uncertainty

CHAPTER 43

Scenario Libraries: The Power Grid

In the mid-1950s, in the spirit of Francis Galton’s dice, the Rand Corporation published a book containing 1 million random digits for use in simulating uncertainty.¹ I understand it was a great hit among insomniacs. Ironically, this simple idea is very close to the scenario libraries in use at Shell, Merck, and Olin today to maintain enterprisewide databases of uncertainties. Such libraries form the power grid of Probability Management.

Modular Risk Models and Age of the Scatter Plot

When Fred retrieves the number 2 from an information system and Joe retrieves the number 3, their results are modular (as in modular furniture) in that they can be consolidated. Just add them together, as in 2 + 3 = 5. But if Fred retrieves the distribution of petroleum prices and Joe retrieves the distribution of an airline stock, how do we consolidate these into a portfolio? Recall from the ladder analogy of Chapter 3 that the simulation of the sum is not the sum of the simulations.

In general, probability distributions don’t add up like numbers for two reasons. First, there is the diversification effect described in terms of the movie portfolios; that is, the sum of two dice or spinners goes up in the middle. Second, there are the interrelationships to worry about, as discussed in terms of the investment examples. When petroleum prices go up, airline stocks tend to go down.

When Markowitz added the distributions of a set of stocks to get the distribution ...