A loyal MR reader asks about prospect theory. I feel it is usually too experimental and too ad hoc, are those really the general biases out there in markets? I was heartened to see the following good paper (non-gated here) on prospect theory and the stock market. The abstract:
We study the asset pricing implications of Tversky and Kahneman’s
(1992) cumulative prospect theory, with particular focus on its
probability weighting component. Our main result, derived from a novel
equilibrium with non-unique global optima, is that, in contrast to the
prediction of a standard expected utility model, a security’s own
skewness can be priced: a positively skewed security can be
"overpriced," and can earn a negative average excess return. Our
results offer a unifying way of thinking about a number of seemingly
unrelated financial phenomena, such as the low average return on IPOs,
private equity, and distressed stocks; the diversification discount;
the low valuation of certain equity stubs; the pricing of
out-of-the-money options; and the lack of diversification in many
household portfolios.
In other words, we can think of stocks as a lottery ticket. They offer a chance at the thrill of victory, and not just a mean-variance pair; this may help explain various pricing and return anomalies. Am I convinced? No. Am I moved? Yes.
#16 out of 50.
I never understood prospect theory.
T&K had a number of experiments that showed that individuals would treat the very same lottery differently depending on whether losses or gains were emphasized in the problem description. I find these experiments compelling.
But what is not compelling is what T&K concluded from these experiments. They postulated a weird ranking of probabilities.
Whether such a function could exist is not relevant. The question is: how can experiments which seriously undermine the very idea of a mathetmatical form of expectation be support for prospect theory? Never understood it.
I have not read this paper, but I will note that using efficient markets
analysis with ratex one would expect skewness to be priced for assets
when it exists, which is often, even though most textbooks stick to mean-
variance presentations, while overwhelmingly most assets exhibit kurtosis,
if not skewness.
A long-known example of skewness comes from forex markets with the
so-called “peso problem” phenomenon, where the forward markets for Mexican
pesos have often regularly underpredicted the mean of actual future spot
exchange rates for pesos. This was first identified in the Ph.D. thesis
of Kenneth Rogoff at MIT back in 1979. The explanation is that the market
is rationally imputing a negative skewness due to the occasional appearance
of large devaluations of the peso, which are not offset by equivalent
revaluations.
CPT is interesting, but the matter of framing is broader than just CPT.
Comments on this entry are closed.