|
Abstract: We
develop and estimate a model of term structure of interest within the context
a Dynamic Stochastic General Equilibrium (DSGE) model featuring multiple
(monetary) policy and volatility regimes. In this model, the pricing kernel
is thus fully endogenized in sharp contrast to
arbitrage-free term-structure models. The estimation of this model is by
tuned Bayesian MCMC methods based on the recent work of Chib
and Ergashev (2009) and Chib
and Ramamurthy (2010). Under a prior distribution that incorporates the view
of a positive term premimum, our estimation results
for U.S. quarterly data from 1986:Q4 to 2008:Q4 reveal that (a) that our
general equilibrium model performs favorably in terms of marginal likelihoods
and Bayes factors in relation to the arbitrage-free
term-structure model with change-points developed in Chib
and Kang (2009), (b) that the U.S. central bank has pursued a ``more active''
monetary policy since 1995:Q2, (c) that during the same period, both the
average term premium and its volatility have fallen, and and
(d) that consequently, whereas the term premium explains a significant
portion of the term spread in the (``less active'') first regime, the
relative share of the market expectations has increased in the second regime.
We take these findings as revealing the important role that general equilibrium
modeling of the term structure, estimated by Bayesian methods, can play in
understanding the complex dynamics of interest rates.
|