Making decisions in a risky world

 

Senator Margaret Chase Smith stands up to Senator Joe McCarthy

People make choices among alternatives, and they can’t know the future. So they can anticipate that each option they face usually presents risks, but they can’t always anticipate those risks. Certainly they can’t guarantee that by pushing this button the desired outcome will happen; rather, there are likely to be several outcomes, each with its own value and probability. Rational choice theory offers a general theory of decision-making under conditions of risk and uncertainty, and this theory of choice underlies much of microeconomics. The assumption is that decision-makers are rational in calculating the “expected value” of each of the options they face (the value of a particular outcome times the probability of that outcome), and that they then choose the option with the greatest expected utility.

Aristotle, on the other hand, had a more nuanced conception of “rational deliberation” in action, which involve consideration of multiple factors in a situation of choice. Practical reason is more than a question of estimating probabilities and utilities; it is an extensive reflection upon the circumstances, values, outcomes, and meanings of the choices that we make. (Here is an earlier post on rational deliberation about important choices in life; link.)

Here is a simple example. It’s a considered risk to go on winter vacation for two months with the heat turned off — “save a little money and count on the weather not getting down to the frozen-pipes level”. You care about the $800 you might save in heating bills, and you judge that a disastrous freeze ($12,000 damage) is too unlikely to worry about. But managing risk is something we can think about analytically. We can assign value to the possible outcomes and we can make an estimate of the likelihood of success and failure, and then we can choose the plan or action with the greatest expected value. In the winter vacation example, we’ve already specified the value of the good and bad outcomes, so we just need to make an estimate of the likelihood of each scenario. If we leave the heat turned off, we have a 100% likelihood of a saving of $800. The best winter forecast of the likelihood of a deep freeze event in my neighborhood is only 1%. This means the expected value of the frozen pipes is -$120 and the net expected gain/loss is $680. On these estimates it is prudent to leave the heat turned off. But if the likelihood of a deep freeze event is greater — say 8% — then the situation changes. Now the expected loss is -$960, and the net loss is -$160. It is no longer prudent to leave the heat off during the winter vacation. I’m better off leaving the heat on for the duration and being sure of avoiding the disastrous freeze.

So what are some of the blindspots of the theory of expected utility as a general model for decision-making? On the rational-choice approach, prudence (wisdom in handling risky situations) means a choice among options ranked by the expected gain/loss of each option. This is one common way of thinking about choice under circumstances of risk and uncertainty. This what economists mean by the advice, “maximize expected utility!”. It is the scheme that underlies cost-benefit analysis. And within its domain it is guaranteed to produce the greatest overall utility when applied consistently over time. But as a rule of choice aimed at guiding a person’s life choices, it is plainly a loser’s maxim. It is appropriate only for “rational fools”, as Amartya Sen put the point in an article of the same name (link). Here are several problems.

First, many life choices involve incommensurable outcomes. A person choosing between career options illustrates this. Alice has a choice: to become an expert on financial systems, because this choice has a 95% probability of generating lifetime wealth holdings of $10 million; or to become a climate activist, because this choice will give meaning to her life and will lead to social relationships with people she can respect and trust. There is no common denominator here, and Alice’s choice cannot be guided by the expected utility rule. Alice simply has to decide: which eventual outcome will be most important to her when all is said and done, a large stock portfolio or a lifetime of meaningful and satisfying work?

Second, many important life choices cannot be fitted out with objective or numerical probabilities. We may be able to assign subjective estimates to outcomes — “It is more likely that my home will burn down as a result of electrical shorts than because it is struck by a small private plane” — but we can’t make a pretense of assigning a specific probability to each possible event. And this is even more of an obstacle when it comes to less tangible events. Imagine if Caesar responded to the soothsayer’s warning “Beware the Ides of March” by jotting down probabilities about these possible hazards: the legions in Britain may mutiny (5%); Mt. Vesuvius may erupt (1%); my friends may murder me (10%); I may get bad news from Gaul that gives me a migraine (15%); … Pointless!

Third, many choices involve values and normative commitments in their doing as well as tangible benefits in their outcomes. This is a point that Sen emphasizes in “Rational Fools”. Another example will help here. William the Haberdasher recognizes that he can make a few extra dollars in profits by stocking Chinese knockoff “Mani” suits where his clientele is expecting Italian “Armani” suits. But William believes he has a duty of honesty to his customers, so he refrains. This is a considered and deliberative choice on William’s part that does not depend upon a cost-benefit analysis. It is not “rational”, if the only relevant consideration is the amount of profit of one business practice versus another; but for William, this is not the case.

It also seems that the expected utility rule is questionable when the unlikely negative outcome is very, very bad. (This seems particularly relevant in the case of a decision to implement highly dangerous technologies like first-generation fission reactors.) Suppose there is a choice between action A and B. A has an 80% probable benefit of $1000 and a 20% possible loss of $1,500. The expected utility is $500. B has a 99.9% likelihood of a return of $2000 and a 0.1% likelihood of a return of -$10,000, with a discounted loss of -$10. The expected return of option B is $1,990 — apparently a no-brainer from the point of view of the expected utility rule. But here the choice depends upon more than the dollars and cents (or rupees), because the unlikely outcome is also extremely bad. In the context of an agrarian society, for a poor landless tenant a debt of $10,000 means debt bondage for life. The landless tenant would surely be prudent to avoid that risk altogether and go with option A. The return on this option will be much less, but it rules out lifetime debt bondage completely. (Some development economists argue that the relatively slow adoption of Green Revolution technologies in rice farming followed from just such a risk-averse strategy. It was called the maximin strategy: choose that option with the least-bad worst outcome. This is the decision rule that John Raws recommends in arguing for the Difference Principle in his two principles of justice.)

And what if the pending choice is a one-off situation — you will never be faced with this choice again? Now the rationale for the expected utility rule disappears, because that rule depends on repeated similar choices. But for a unique choice you will never get the chance to make up for the unlikely roll of the dice. There is only the action and the outcome. The actor needs to ask: is there a best choice here and now, that I will not regret even if the dice rolls against me?

So Aristotle was right: wisdom in action within a risky world of choice cannot be reduced to simple arithmetic. Rather, reflection and deliberation about means and ends, moral and personal commitments, and the context of action and its consequences are needed in order to make genuinely wise decisions. “Self-interest” is only one part of wisdom; character and commitments are also part of the process of choice in important decisions. Phronesis rather than calculation is unavoidable in life’s more complicated and multifaceted situations of choice. (Here is a post on the topic of character that is relevant to decision-making; link.)