Appendix: Experimental TestsIn our paper on myopic loss aversion reproduced here, we used the concept
to try to “explain” the equity premium puzzle. We did so by estimating the
evaluation period that would make investors indifferent between stocks
and bonds. The answer was one year. In that paper, we did not actually
“test” our explanation other than by asking whether this evaluation period
seemed plausible. Since this paper has been published, researchers have of-
fered several experimental tests of myopic loss aversion, which we summa-
rize in this appendix.
Thaler et al. (1997) explored the separate effects of myopia and loss
aversion in a setting that allows for learning. In particular, they asked eighty
undergraduate students to allocate an endowment between a generic stock
fund and a generic bond fund. The two funds, however, were unlabeled,
and the subjects had to learn about the risk and return profiles of the funds
through a simulated twenty-five-years of experience. The experiment in-
cluded four conditions, and the subjects were randomly assigned to one of
the conditions. In the “monthly” condition, the subjects made two hundred
investment decisions, each corresponding to (approximately) a one-month
period. In the “yearly condition,” the subjects made twenty-five annual de-
cisions, and in the “five-yearly” condition, they made five decisions. There
was also another condition, where losses were eliminated by introducing a
high level of inflation. However, the specific rate of inflation was unknown
to the subjects, so they were induced to think in nominal terms. Hereafter,
we refer to this condition as the “inflated monthly” condition, since it was
based on the monthly data. At the end of the twenty-five-years of learning
experience, the subjects made a final allocation decision for the following
fifty years.
The results are consistent with myopic behavior. In particular, as the ag-
gregation period lengthens, so does the allocation to stocks. The final allo-
cations to stocks in the monthly, yearly, and five-yearly conditions are 40.9
percent, 69.6 percent, and 66.2 percent, respectively. The small difference
between the yearly and five-yearly allocations is in the wrong direction and
could be attributed to the limited learning opportunities in the five-yearly
condition (five trials). The results are also consistent with loss aversion. The
final allocation to stocks in the inflated monthly condition is 72.4 percent,
versus 40.9 percent in the noninflated monthly condition.
Gneezy and Potters (1997) provided additional tests showing that the
thought of what the aggregate data looks like could by itself reduce myopic
behavior. In particular, they presented eighty-three undergraduate students
with a sequence of three independent lotteries. Each of the lotteries had a
probability of two-thirds to lose the amount invested, and a probability of
218 BENARTZI AND THALER