文档介绍：Bond risk premia
John H. Cochrane∗and Monika Piazzesi†
October 5, 2001
Abstract
This paper studies risk premia in the term structure. We start with regressions
of annual holding period returns on forward rates. We find that a single factor,
which is a tent-shaped function of forward rates, can predict one-year bond excess
returns with an R2 up to 45%. Though the return forecasting factor has a clear
business cycle correlation, it does not forecast output, and macro variables do not
forecast bond returns.
We relate these time-varying expected returns to covariances with various shocks,
which is the same as finding the market prices of risk that justify a yield VAR as
an affine term structure model. The time-varying expected return can be almost
entirely accounted for by a time-varying risk premium for level-shocks in yields, or
by a time-varying risk premium for ary policy shocks.
How could such an important factor have been missed? The return forecsating
factor does very little for understanding yields. The expectations hypothesis and
level ans slope factors are an excellent approximation for yields, though a poor
one for returns. Also, bond data do not follow a monthly AR(1), with a pattern
that suggests measurement error. Hence, if one followed the usual approach in term
structure analysis, starting with a monthly k-factor model chosen to minimize pric-
ing errors, and then finding implied annual return forecasts, one pletely
miss the forecastability of annual returns.
∗University of Chicago and NBER. Graduate School of Business, University of Chicago, 1101 E. 58th
St. Chicago IL 60637. john.******@. I gratefully acknowledge research support from
the Graduate School of Business and from an NSF grant administered by the NBER.
†UCLA and NBER. Anderson School, 110 Westwood Plaza, Los Angeles CA 90095, ******@
1
1Introduction
This paper studies risk premia in the term structure of interest ra