139
N O N E X P E C T E D - U T I L I T Y T H E O R Y
theory escapes their net: the decision-weighting models. It is easy to see why models
with probability transformations do not imply approximate risk neutrality for small
risks since risk averse behavior can be generated by nonlinear probability weighting
even where the utility function is linear. So, for example, aversion to probabilistic in-
surance is easily explained by overweighting of the small probability of nonpayment.
As such, decision-weighting models stand out as leading contenders to explain
aspects of insurance behavior that it has long been known standard theory cannot
handle. There is growing evidence that probability weighting may be an important
ingredient in explaining a variety of field data relating to gambling and insurance
behavior and several examples are discussed by Camerer (2000).
Another field phenomenon that has perplexed economists is the size and per-
sistence of the excess return on stocks over fixed income securities. This is the so
called
equity premium puzzle
and it is the economics equivalent of the crop circle:
we have seen it in the field, but we have real trouble explaining how it got there.
Since the return on stocks is more variable, standard theory is consistent with
some difference in the long-run rates of return, but since Mehra and Prescott
(1985) it has been recognized that the observed disparity implies implausibly
high degrees of risk-aversion in standard models of asset pricing. One possible
explanation for (part of) the equity premium has been suggested by Epstein and
Zin (1990). They show that a recursive utility model using rank-dependent prefer-
ences predicts an equity premium, though only about one third of the size that is
usually observed. A full, and in my view much more convincing, account has
been suggested by Benartzi and Thaler (1995) who show that the level of equity
premium is consistent with prospect theory, with the added assumption that
agents are myopic (i.e., they assess expected returns over “short” time horizons).
The crucial element of prospect theory for this explanation is loss-aversion. In the
short run, there is a significant chance that the return to stocks is negative so if, as
loss-aversion implies, investors are particularly sensitive to these possible nega-
tive returns, that would explain the equity premium for myopic investors. But just
how loss-averse and how myopic do agents have to be for this explanation to
work? Benartzi and Thaler show that, assuming people are roughly twice as sensi-
tive to small losses as to corresponding gains (which is broadly in line with exper-
imental data relating to loss-aversion), the observed equity premium is consistent
with the hypothesis that investments are evaluated annually. This is a very simple,
and to my mind, intuitively appealing account of another important field phenom-
enon which has defied explanation in standard theory.
Notice that while loss-aversion can be accommodated in conventional models
like the sign- and rank-dependent theories, the other ingredient in this explanation
of the equity premium—i.e., myopia—belongs in another tradition. This is essen-
tially a bounded rationality assumption, and while the one-year time horizon has
a nice ring of plausibility to it, it sits much more naturally alongside procedural
theories like the original version of prospect theory. Bounded rationality assump-
tions seem to be providing the missing links necessary to explain an increasing
range of economic phenomena (see Camerer 1998 for a recent review of applica-
tions in individual decision making).
Do'stlaringiz bilan baham: |