THE EXPECTED UTILITY MODEL REVISITED

The expected utility model provides the basis for most of the research on the economics of risk. It was the topic presented in Chapter 3. Under the expected utility model, individuals make decisions among alternative wealth levels x by maximizing EU( x) where E is the expectation operator. The utility function U( x) is defined up to a positive linear transformation. It is sometimes called a von Neumann Morgenstern utility function. We saw in Chapter 4 that risk aversion, risk neutrality, or risk loving preferences correspond to the function U( x) being respectively concave, linear, or convex.

One of the main advantages of the expected utility model is its empirical tractability. This is the reason why it is commonly used in risk analysis. But is the expected utility model a good predictor of human behavior? Sometimes, it is. And sometimes, it is not. This chapter evaluates some of the evidence against the expected utility model. It also reviews alternative models that have been proposed to explain behavior under risk.

The first challenge to the expected utility model is the following: Is it consistent with the fact that some individuals both insure and gamble at the same time? Friedman and Savage proposed to explain this by arguing that, for most individuals, the utility function U( x) is probably concave (corresponding to risk aversion and a positive willingness to insure) for low or moderate monetary rewards, but convex...