Utility Tree
A utility tree is a decision-making tool that multiplies possible outcomes by their probabilities to compute expected values, enabling comparison of options and selection of the one with the greatest expected payoff.
THE IDEA
EXERCISE 39 Japanese Goods
Your U.S. corporation is considering importing certain goods from Japan for sale in the corporation’s nationwide chain of general merchandise stores. The corporation has narrowed its interest to three items—a decorative vase, a battery-powered electric hand drill, and a combination stereo radio and disc player—each of which is manufactured by a different Japanese firm. Because these firms historically have refused to deal with non-Japanese-owned companies, persuading them to sell their product to the corporation will be expensive and time-consuming. But they are feeling the pinch of increased competition and have begun making noises about relaxing their self-imposed prohibition against dealing with foreign merchandisers.
The U.S. corporation's management has decided that, rather than pursue all three Japanese firms simultaneously, it will concentrate its lobbying effort on one firm. Your job is to analyze the problem and identify which of the three products will give the biggest return to the corporation during the coming sales year. The corporation will then lobby that firm.
According to a research report, the vase will earn the corporation a profit of $100,000, the hand drill $200,000, and the stereo $500,000. The probability of persuading the firms that produce these products to sell to the corporation is estimated to be .9 for the vase, .5 for the drill, and .1 for the stereo. On which item should the corporation concentrate its lobbying effort? Using a utility tree, work out the problem on a sheet of paper.
BEFORE CONTINUING, WORK THE PROBLEM USING A UTILITY TREE.
The solution to Exercise 39 (page 337) shows the completed tree, whose perspective is the U.S. corporation’s profits. Here again we could usefully dispense with the “No” outcomes.
The calculation assumes no profit if a firm declines and leaves out lobbying costs. We multiply the utilities ($100,000, $200,000, $500,000) by their respective probabilities (.9, .5, .1) to produce the expected values ($90,000, $100,000, $50,000). The drill option is the preferred choice because it offers the greatest expected value: $100,000. Does the drill’s being the preferred choice pass the sanity check? Does it make sense? I think so. While the probability of the U.S. corporation’s marketing the drill is 55 percent of the probability of its marketing the vase, the profit from the drill is 100 percent more than that from the vase. And although the profit from the stereo is 150 percent more than that from the drill, the probability of marketing the stereo is only 20 percent of the probability of marketing the drill.
Concept Summary
The utility tree (or expected value decision tree) helps decision-makers choose among alternatives by calculating the expected value of each option. The expected value is the sum of each possible outcome's utility multiplied by its probability. The option with the highest expected value is typically the best under uncertainty, assuming risk neutrality.
When to use it
Choosing which product to launch when each has a different probability of success and profit.
Selecting among R&D projects with uncertain outcomes and varying potential returns.
Allocating marketing budget across channels with different conversion rates and customer lifetime values.
Deciding which supplier to negotiate with when success probabilities and cost savings differ.
FROM THEORY TO EVERYDAY LIFE
In practice
Japanese Goods Import Decision
A U.S. corporation must choose which of three Japanese products to lobby for importing: a vase (profit $100,000, probability 0.9), a hand drill ($200,000, 0.5), or a stereo ($500,000, 0.1). Using a utility tree, expected values are calculated: vase $90,000, drill $100,000, stereo $50,000. The drill is the optimal choice.
Investment Portfolio Selection
An investor compares hypothetical payoffs: Stock A pays $10,000 with an 80% chance, Stock B $20,000 with a 40% chance, and Stock C $50,000 with a 10% chance. Assuming a zero payoff otherwise, expected payoffs are A $8,000, B $8,000 and C $5,000. A and B tie on expected payoff. Investment costs, possible losses and risk tolerance would also matter.
A MOMENT TO REFLECT
Put it to work.
How would the decision change if the probabilities or payoffs were slightly different? Could a different option become optimal?
What non-quantitative factors (e.g., strategic partnerships, brand reputation) might you consider alongside expected value in a real business decision?
In what situations might the expected value approach lead to a poor decision, and how could you mitigate that risk?
CONTINUE THE CONVERSATION
Source texts
- The Thinkers Toolkit by Morgan D Jones