109
N O N E X P E C T E D - U T I L I T Y T H E O R Y
option gives one million U.S. dollars for sure; the second gives five million with a
probability of 0.1; one million with a probability of 0.89, otherwise nothing.
4
What would you choose? Now consider a second problem where you have to
choose between the two prospects:
s
2
5
($1M, 0.11; 0, 0.89) or
r
2
5
($5M, 0.1;
0, 0.9). What would you do if you really faced this choice?
Allais believed that EU was not an adequate characterization of individual risk
preferences and he designed these problems as a counterexample. As we shall
shortly see, a person with expected utility preferences would either choose both
“
s
” options, or choose both “
r
” options across this pair of problems. He expected
that people faced with these choices might opt for
s
1
in the first problem, lured by
the certainty of becoming a millionaire, and select
r
2
in the second choice, where
the odds of winning seem very similar, but the prizes very different. Evidence
quickly emerged that many people did respond to these problems as Allais had
predicted. This is the famous “Allais paradox” and it is one example of the more
general common consequence effect.
Most examples of the common consequence effect have involved choices be-
tween pairs of prospects of the following form:
s
*
5
(
y
,
p
;
c
, 1
2
p
) and
r
*
5
(
q
,
p
;
c
, 1
2
p
), where
q
5
(
x
, •; 0, 1
2
•) and 0
,
•
,
1.
5
The payoffs
c
,
x
, and
y
are
nonnegative (usually monetary) consequences such that
x x y
. Notice that both
prospects
s
* and
r
* give outcome
c
with probability 1
2
p
: this is the “common
consequence” and it is an obvious implication of the independence axiom of EU
that choices between
s
* and
r
* should be independent of the value of
c
.
6
Numer-
ous studies, however, have found that choices between prospects with this basic
structure are systematically influenced by the value of
c
. More specifically, a vari-
ety of experimental studies
7
reveal a tendency for individuals to choose
s
* when
c
5
y
, and
r
* when
c
5
0.
A closely related phenomenon, also discovered by Allais, is the so called
com-
mon ratio effect
. Suppose you had to make a choice between $3000 for sure, or
entering a gamble with an 80% chance of getting $4000 (otherwise nothing).
What would you choose? Now think about what you would do if you had to
choose either a 25% chance of gaining $3000 or a 20% chance of gaining $4000.
A good deal of evidence suggests that many people would opt for the certainty of
$3000 in the first choice and opt for the 20% chance of $4000 in the second. Such
a pattern of choice, however, is inconsistent with EU and would constitute one ex-
ample of the common ratio effect. More generally, this phenomenon is observed
in choices among pairs of problems with the following form:
s
**
5
(
y
,
p
; 0,
1
2
p
) and
r
**
5
(
x
, •
p
; 0, 1
2
•
p
) where
x
s
y
. Notice that the ratio of “win-
ning” probabilities (•) is constant, and for pairs of prospects of this structure, EU
4
In Allais’s original examples, consequences were French Francs.
5
It will be convenient to use a scaling factor
l
at several points in the paper, so to avoid repetition,
assume 0
,
l
,
1 throughout.
6
The original Allais problems are recovered from this generalization setting
x
5
$5M;
y
5
$1M,
p
5
0.11 and
l
5
10/11.
7
Examples include H. Moskowitz (1974), Paul Slovic and Amos Tversky (1974), and MacCrimmon
and Larsson (1979).
Do'stlaringiz bilan baham: |