02.03.2013 Views

Thinking and Deciding

Thinking and Deciding

Thinking and Deciding

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

INVESTORS AND ENTREPRENEURS 519<br />

ability, described in Chapter 11 (Einhorn <strong>and</strong> Hogarth, 1985). Specifically, when<br />

probabilities are ambiguous, people imagine that they could be higher or lower, <strong>and</strong><br />

they adjust the probabilities by forming an average of what they imagine. As a result,<br />

the probabilities are adjusted toward some central point, such as .5. When probabilities<br />

are extreme (near 1 or 0), it is easier to think of ways to adjust them toward the<br />

middle than ways to make them even more extreme.<br />

Insurance, by its nature, deals with low-probability events. When experts disagree<br />

about the probability of these events, or when the events are new <strong>and</strong> unfamiliar,<br />

they are ambiguous, <strong>and</strong> people adjust them. Moreover, buyers <strong>and</strong> sellers of<br />

insurance have different incentives to think of probabilities that are higher than what<br />

would otherwise be the best guess. The seller must worry about having to pay for<br />

a claim, so the seller sees the bad event (for example, a fire) as a loss. The buyer<br />

is compensated if the bad event happens, but the premium paid is a loss. Because<br />

losses loom larger than gains (Chapter 11), the sellers imagine more ways that a bad<br />

event can happen, so, when the probability is ambiguous, they adjust it upward more<br />

than the buyers do. As a result, they dem<strong>and</strong> very high premiums, much higher than<br />

expected value, whereas the buyers are willing to pay premiums that are only a little<br />

higher than expected value. The difference means that insurance will not be sold.<br />

The sellers, the insurers, will dem<strong>and</strong> a higher price than the buyers are willing to<br />

pay. This effect has been found in questionnaire studies (Hogarth <strong>and</strong> Kunreuther,<br />

1989; Kunreuther, Hogarth, <strong>and</strong> Meszaros, 1993).<br />

This effect may explain such real events as the breakdown of the market for<br />

environmental liability insurance in the United States in the 1980s, when many<br />

retroactive changes in the law had made the future uncertain for everyone (Huber,<br />

1988). People felt a sense of ambiguity because the missing information about future<br />

changes in the law was salient.<br />

Investors <strong>and</strong> entrepreneurs<br />

Normative theory argues that risky investments should pay a bit more than safe ones.<br />

Declining marginal utility makes people rationally risk averse, so they will dem<strong>and</strong><br />

greater expected return, a premium, for speculative investments. Some of this effect<br />

will be reduced if investments can be pooled in ways that reduce the risk. A fund<br />

made up of many different kinds of speculative investments might not itself be so<br />

speculative if some of its components make money while others crash. So this kind<br />

of pooling would make the expected premium smaller. It is thus a puzzle that the<br />

premium for stocks, equities, is huge. From 1889 to 1978, the annual real rate of<br />

return on stocks was about 7%, while it was about 1% on government bonds. (“Real”<br />

means after taking inflation into account.) Stocks did better over the long term (thirty<br />

years) even for people who bought stock before the 1929 stock market crash, so long<br />

as they held their stock <strong>and</strong> did not sell. This is called the “equity premium puzzle”<br />

(Mehra <strong>and</strong> Prescott, 1985). It is a puzzle because it cannot be explained in terms of<br />

any plausible story about declining marginal utility.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!