Arturo Estrella
personal web site
|
Arturo Estrella Professor and Department Head |
 |
SELECTED TOPICS IN MACROECONOMICS:
Predicting recessions with the
term structure of interest rates: I. Empirical
In 1988-89, developed with Gikas Hardouvelis a probit model
of the probability of a future recession. The model has correctly
forecasted the last three
U.S.
recessions in real time with a lead time of about one year and no
false positives. See
original
1989
working paper, 2006 New
York Fed article, and yield curve FAQs
with extensive bibliography.
Predicting
recessions with the
term structure of interest rates: II. Theoretical
Extending
work of Svensson
and others, developed theoretical macroeconomic model that
explains the
ability of term interest rate spreads to predict future real activity
and
inflation. See article
in Economic
Journal 2005.
Macroeconomic
modeling
With
Jeff Fuhrer, showed that forward-looking macroeconomic models popular
in the
recent literature (i) are mathematically incapable of capturing robust
properties of
macroeconomic data (AER
2002) and (ii) fare worse than backward-looking models
in empirical tests of the Lucas critique (REStat
2003). Hybrid models are found to be more
suitable for empirical monetary policy analysis.
SELECTED TOPICS IN ECONOMETRICS:
Pseudo R-squared for probit
and other DDV models
Solving
a simple differential equation,
derived a pseudo
R-squared for dichotomous dependent variable models that outperforms
other
alternatives, including the classic McFadden measure. Published in JBES
1998,
the measure has become part of the standard output in RATS, TSP, SAS,
and LIMDEP.
Econometric breakpoint
testing
(i)
In empirical work with Jeff Fuhrer (see macroeconomic modeling above),
derived
and implemented a method for calculating exact p
values in breapoint tests based on GMM estimates. The
mathematical basis for the method and computational optimization are
discussed
in ET
2003, which also provides critical values for
single-breakpoint
tests.
(ii) With
Tony Rodrigues, developed a directional
breakpoint test for a single parameter.
Time series filtering and
the business cycle
Studied
the analytical effects of using various popular time series filters to
extract business cycle fluctuations from macroeconomic data. Results
are not always
what one might expect and guidelines for selecting specific filters are
provided. In an empirical application, used frequency domain filters to
study the
relationship between labor
productivity growth and the business cycle.
Generalized canonical
regression
Linear
combinations of explanatory variables with statistically optimal
weights are the
norm in
linear regression and many of its extensions. But, what happens when
the
dependent variable is also a linear combination of several variables
with
optimal
weights? This
article
shows how it can be done.
SELECTED TOPICS IN BANK CAPITAL AND REGULATION:
What
is the best general
approach to bank capital regulation?
A
series of articles discusses basic principles for bank capital
regulation using an
eclectic
approach that extends beyond the boundaries of finance to
include
philosophy, legal theory, and institutional economics. See Prolegomenon,
Formulas,
and Choices.
Are
complicated capital
requirements better than simple formulas?
It ain't
necessarily so, as this
article
with Sangkyun
Park
and Stavros Peristiani shows. Moreover, a regulatory
minimum requirement based on value-at-risk, if binding, is likely to be
procyclical.
Can investors rely on banks
to provide voluntary information about risk?
Market
forces help by inducing banks to reveal valuable information to
investors, but this
article shows that it also helps
to have independent bank examiners with access to
internal
bank information and the best interests of the public in mind.