ABSTRACT: Two agents trade an indivisible
asset. They are risk neutral and share a common benchmark dividend
model.
However,
each has doubts about the specification of this model. These doubts
manifest themselves as a preference for
robustness
(Hansen and Sargent (2008)). Robust preferences introduce pessimistic
drift distortions into the benchmark
dividend
process. These distortions increase with the level of wealth,
and give rise to endogenous heterogeneous beliefs.
Belief
heterogeneity allows asset price bubbles to emerge, as in Scheinkman
and Xiong (2003). A novel implication of
our analysis
is that bubbles occur when wealth inequality increases. Empirical
evidence supports this prediction. Detection
error
probabilities suggest that the implied degree of belief heterogeneity
is empirically plausible.
Hope For The Best,
Plan For The Worst (with Xiaowen Lei)
Slides
NotebookLM
podcast
ABSTRACT: This paper studies asset
pricing when individuals struggle to strike a balance between doubt and
hope. We
argue this internal struggle is consistent with recent
evidence from neuroscience. We operationalize it using the robust control
and filtering approach of Hansen and Sargent (2008).
Our key innovation is to assume that filtering is optimistically
biased.
Investment decisions, however, reflect doubts about
model specification, and are pessimistically biased. We show that doubts
about model specification, combined with optimistically
distorted beliefs about dividend growth can not only explain low
average price/dividend ratios and high average returns,
but can also generate large procyclical swings in price/dividend
ratios. High and volatile returns occur despite the
fact that investors have low (approximately logarithmic) risk aversion.
Belief
distortions are empirically plausible, with detection
error probabilities in excess of 10%.
Competing Models With Feedback
(with In-Koo Cho)
ABSTRACT: This paper studies misspecified Bayesian
learning with endogenous data when models compete via model
averaging. A decisionmaker forecasts an endogenous
price sequence by averaging the forecasts from two models, one with
constant parameters and one with time-varying parameters.
The time-varying parameters model is misspecified. It excludes a
relevant explanatory variable and fails to recognize
the presence of model averaging. In contrast, the constant parameters model
includes all relevant explanatory variables and recognizes
that prices are generated by averaging the two forecasts. If expectational
feedback is weak, the correctly specified constant
parameters model prevails. However, if feedback is strong and the excluded
fundamentals are not too important, the under-parameterized
time-varying parameters model survives the competition, and prices
become endogenously nonstationary. Simple time-varying
parameter models do well because they better respond to time variation in
the data that their own use generates.
A Behavioral Defense of Rational Expectations
ABSTRACT: This paper studies decision
making by agents who value optimism, but are unsure of their
environment. As in
Brunnermeir
and Parker (2005), an agent's optimism is assumed to be tempered
by the decision costs it imposes. As in
Hansen and Sargent
(2008), an agent's uncertainty about his environment leads him
to formulate `robust' decision rules. It is
shown that when
combined, these two considerations can lead agents to adhere
to the Rational Expectations Hypothesis.
Rather than
being the outcome of the sophisticated statistical calculations
of an impassive expected utility maximizer, Rational
Expectations
can instead be viewed as a useful approximation in environments
where agents struggle to strike a balance
between doubt
and hope.