|
|
Research
Job Market
Paper
"Learning and the Yield Curve" (Download
PDF)
Abstract:
Two central implications of the Expectations Hypothesis under
rational expectations are inconsistent with yield curve data: (i)
future expected long yields fall, instead of rising, when the yield
spread rises, and (ii) long yields are excessively volatile with
respect to short yields. I document these puzzles in the U.S. and the
U.K. data, for different sub-samples, and for both real and nominal
yields. I then propose a micro-founded optimization framework in which
boundedly rational agents use adaptive learning to form expectations.
The model is successful on both dimensions. First, the belief structure
rationalizes the pattern of yields observed in the data so that the
first puzzle does not hold with subjective instead of rational
expectations. In particular, intertemporal income and substitution
effects are amplified relative to the rational expectations case,
causing expected long yields to rise when the yield spread falls. The
second puzzle is partly accounted for by the extra volatility due to
parameter uncertainty. These results suggest that it is the assumption
of rational expectations that is at odds with the data, not the
(subjective) Expectations Hypothesis. In addition, I find that: the
model generates systematic forecast errors in yields and inflation that
are consistent with survey data; higher yield volatilities during
different monetary policy regimes match the data; and fiscal policy
affects the yield spread because Ricardian Equivalence no longer holds.
Works in Progress
“Fiscal Policy and the Yield Curve”
Abstract:
I analyze the effects of fiscal policy on the yield curve. The
empirical literature has been inconclusive about how long and short
yields are affected by fiscal policy. For instance, while Engen and
Hubbard (2004) find insignificant effects of government deficits on
interest rates, Evans and Marshall (2002) find varying effects,
depending on how the identification of fiscal shocks is constructed.
These approaches, however, do not model the long end of the yield
curve. I first use a structural VAR, following the identification
scheme of Blanchard and Perotti (2004), to analyze the effects of
government spending and taxes on the short and long ends of the yield
curve for U.S. data. I then use a New Keynesian framework with adaptive
learning in which the fiscal authority issues riskless bonds of all
maturities in non-zero net supply. Ricardian Equivalence no longer
holds – out of the rational expectations equilibrium, an individual
household may not recognize that its tax obligations are the same as
all other households, and that the present value of government debt
must equal the discounted sum of tax obligations. I find that a
positive deficit shock increases the short yields, and as the rise in
long yields is smaller, the yield spread declines on impact.
“Is Ricardian Equivalence Learnable?”
Abstract:
In this paper, I ask the following question: when optimizing agents’
expectations deviate from rational expectations, does the canonical
model of fiscal policy, Ricardian equivalence, hold? This paper
illustrates that when expectations are formed using adaptive learning,
it does, for several fiscal policy specifications. In a real exchange
economy, as long as the steady state level of taxes is not too high
relative to output, Ricardian equivalence is obtained. When nominal
rigidities are introduced via Calvo pricing, Ricardian equivalence is
consistent with the standard monetary and fiscal policy specifications.
|
|
