We're using cookies, but you can turn them off in your browser settings. Otherwise, you are agreeing to our use of cookies. Learn more in our Privacy Policy

Bridge over ocean
1 November 2009 CFA Institute Journal Review

What Drives the Disposition Effect? An Analysis of a Long-Standing Preference-Based Explanation (Digest Summary)

  1. M.E. Ellis

The authors simulate investor trading behavior that follows prospect theory to
see if such behavior leads to a disposition effect for individual investors.
Simulations based on annual gains and losses show a disposition effect only when
risky investments have low expected returns or when many trading opportunities
exist. Simulations based on realized gains and losses demonstrate a disposition
effect over a wider range of inputs. These results indicate that realized gains
and losses may provide more insight into investor behavior than do returns based
on a fixed time frame.

A “disposition effect” occurs if investors are more likely to sell stock that
has increased in price relative to its initial cost than stock that has decreased in
price. This effect is well documented for individual investors, but modeling its cause
has been troublesome. The authors propose an alternative model that uses prospect
theory, which is based on differing utility for gains and losses rather than utility of
final wealth level. According to prospect theory, investors are risk averse for moderate
gains (a concave utility function) and risk seeking for moderate losses (a convex
utility function). Also, they react more negatively toward losses than they do
positively toward gains of an equal magnitude. Based on parameters developed in prior
finance literature, the authors use simulation analysis to see if prospect theory leads
to a disposition effect for investors. They use two approaches: First, investors receive
utility only at the end of each year, and second, investors receive immediate utility
when a trade is made and gains or losses are realized.

In the model, investors may invest in a risky asset or a risk-free asset. The potential
gain on the risky investment is larger than the potential loss, thus making the expected
return on the risky investment positive. Investors have T opportunities
to trade within the year. Each trading period, the price of the risky asset either
increases or decreases, implying t + 1 possible outcomes, where
t = time, at any point in time. At the end of each trading period,
investors make the trade decision based on a utility function that follows prospect
theory. Each period, the authors calculate the proportion of gains realized (PGR) ratio,
which is the number of stocks with a gain that are sold relative to the number of stocks
with a gain that could be sold. They also develop a similar proportion of losses
realized (PLR) ratio and then compare the two ratios. If the disposition effect holds,
the average PGR should be larger than the average PLR.

The authors predict, and the simulation results confirm, that expected returns on the
risky asset must exceed a “threshold” before investors are willing to invest
in the risky asset. The level of the threshold decreases as the number of trading
opportunities increases.

Results based on annual gains and losses do not show consistent evidence of a disposition
effect. The objective of this model is to maximize the gain at the end of the year, but
intermediate results have no utility and investors are averse to a loss at the end of
the year. Because the magnitude of a gain is greater than the magnitude of a loss for
any one period within the year, if investors have a gain in a given period, they can buy
more shares and have a year-end gain even if they lose in the next period. If investors
have a loss in a period, they can sell shares and reduce their year-end loss, but the
gain on the remaining shares may be sufficient to cause a year-end gain if the return is
positive in the following period. Therefore, investors buy shares following a gain and
sell shares following a loss, which is the opposite of a disposition effect. In this
model, the disposition effect holds only if many trading opportunities exist
(T is large) and the expected return on the risky investment is
low, thus making the magnitude of the gain and loss more similar.

The realized gains and losses simulation uses a three-period structure
(T = 2). Results based on this approach find the disposition effect
over a wider level of expected returns. Under this model, investors want to experience a
loss only once, so they do not sell losers at the intermediate period
(t = 1). Investors are more likely to spread the positive
experience of a gain over multiple periods and to sell at least some of the winners
immediately.

This research suggests that investors are more focused on realized gains and losses than
on a fixed-period time horizon. This approach differs from common finance paradigms.