today. This is a departure from traditional approaches, which hold that the
only thing people think about when choosing a portfolio is the future con-
sumption utility that their wealth will bring. While our preferences are non-
standard, this does not mean that they are irrational in any sense: it is not
irrational for people to get utility from sources other than consumption, nor
is it irrational for them to anticipate these feelings when making decisions.
Introducing utility over gains and losses in financial wealth raises a num-
ber of issues: (1) How does the investor measure his gains and losses Xt+ 1?
(2) How does zttrack prior gains and losses? (3) How does utility vdepend
on the gains and losses Xt+ 1 ?, and (4) How does ztchange over time? Sub-
sections A through D below tackle each of these questions in turn. Finally,
subsection E discusses the scaling factor bt.
A. Measuring Gains and LossesThe gains and losses in our model refer to changes in the value of the in-
vestor’s financial wealth, even if this is only one component of his overall
wealth. For simplicity, we go one step further. Even though there are two fi-
nancial assets, we suppose that the investor cares only about fluctuations in
the value of the risky asset.^6
Next, we need to specify the horizon over which gains and losses are
measured. Put differently, How often does the agent seriously evaluate his
investment performance? We follow the suggestion of Benartzi and Thaler
(1995) that the most natural evaluation period is a year. As they point out,
we file taxes once a year and receive our most comprehensive mutual fund
reports once a year; moreover, institutional investors scrutinize their money
managers’ performance most carefully on an annual basis. Since this is an
important assumption, we will investigate its impact on our results later in
the chapter.
Our investor therefore monitors fluctuations in the value of his stock
portfolio from year to year and gets utility from those fluctuations. To fix
ideas, suppose that St, the time tvalue of the investor’s holdings of the risky
asset, is $100. Imagine that by time t+1, this value has gone up to
StRt+ 1 =$120. The exact way the investor measures this gain depends on
the reference level with which $120 is compared. One possible reference
level is the status quo or initial value St=$100. The gain would then be
measured as $20, or more generally as Xt+ 1 =StRt+ 1 −St.
This is essentially our approach, but for one modification that we think is
realistic: we take the reference level to be the status quo scaled up by the
PROSPECT THEORY AND ASSET PRICES 229(^6) A simple justification for this is that since the return on the risk-free asset is known in ad-
vance, the investor does not get utility from changes in its value in the way that he does from
changes in risky asset value. We also show later that for one reasonable way of measuring
gains and losses, it makes no difference whether they are computed over total financial wealth
or over the risky asset alone.