In 2004 all publications will carry a motif taken from the €100 banknote. This paper can be downloaded without charge from

Montreal, the CEPR meetings at Gerzensee, Econometric World Congress in London, EFA in Moscow, NYU Stern

monetary policy, interest rate risk We especially thank Bob Hodrick for providing detailed comments and valuable suggestions.

Models of the term structure based on only observable variables have had limited success in explaining movements in longer-term interest rates. A key assumption in much of this literature is that agents know all the parameters describing the model of the economy and that these parameters are fixed for all time. In this paper, we relax both of these assumptions and assume that agents regularly re-estimate the parameters of their models–both those determining the point forecasts and those describing economic volatility–based on incoming data. In this way, we allow for the real-time problem of pricing assets based on the information set available at the time. In addition, we allow for discounting of past data reflecting a concern on the part of agents for structural change in the economy. We find that the learning model with discounting does a much better job at explaining longer-term yields than an equivalent model with constant coefficients estimated over the full sample; in particular, the deviations from the expectations

From 2004 to 2006, the FOMC raised the target federal funds rate by 4.25%, yet long-maturity yields and forward rates fell. We consider several possible explanations for this “conundrum. ” The most likely, in our view, is a fall in the term premium, probably associated with some combination of diminished macroeconomic and financial market volatility, more predictable monetary policy, and the state of the business cycle.

Through explicitly incorporating analysts ’ forecasts as observable factors in a dynamic arbitragefree model of the yield curve, this paper proposes a framework for studying the impact of shifts in market sentiment on interest rates of all maturities. An empirical examination reveals that survey expectations about inflation, output growth and the anticipated path of monetary policy actions contain important information for explaining movements in bond yields. Although perceptions about inflation are largely responsible for movements in long-term interest rates, an explicit slope factor is necessary to adequately capture the dynamics of the yield curve. Macroeconomic forecasts play an important role in explaining time-variation in the market prices of risk, with forecasted GDP growth playing a dominant role. The estimated coefficients from a forward-looking monetary policy rule support the assertion that the central bank preemptively reacts to inflationary expectations while suggesting patience in accommodating real output growth expectations. Models of this type may provide traders and policymakers with a new set of tools for formally assessing the reaction of bond yields to shifts in market expectations due to the arrival of news or central bank statements and announcements.

This paper develops a new approach to identification and estimation of affine-termstructure models. We establish that three popular canonical representations are each, for di¤erent reasons, unidentified. We also demonstrate that a failure of local identification can complicate numerical search for the maximum-likelihood estimate when one uses conventional estimation methods. We propose minimum-chi-square estimation as an alternative to maximum-likelihood estimation, and show that, although it is asymptotically equivalent or sometimes even identical to MLE, it can be much easier to compute. Our approach can allow the researcher to recognize with certainty whether a given estimate represents a global maximum of the likelihood function and make feasible the computation of small-sample standard errors. We illustrate the benefits of these new methods by revisiting three separate applications of affine-term-structure models.

for helpful comments and suggestions, and my thesis advisors, William McCausland and René Garcia, for their insight, guidance and support over the past three years. Financial support from CDP Capital is gratefully acknowledged. Imposing identifying restrictions in empirical work is sometimes more difficult than is commonly appreciated. In the classical framework, they can introduce bias in the estimators and produce confidence intervals that are too conservative. In the Bayesian framework, as long as priors are proper, posteriors are proper and the model is identified in that specific sense, but they can introduce numerical difficulties. When the set of parameters maximizing the likelihood function is bounded, one does not need to impose such restrictions through priors to make valid inferences for the model considered. The problem of signing and labeling of factors in linear state-space models, which may be subject to weak-identification problems, falls into that category and is addressed in this paper. We propose and implement a new MCMC algorithm for latent-factor Gaussian affine term-structure models where we mix over all symmetric sub-models obtained by relabeling and changing signs. Our illustrative example on monthly zero-coupon bonds from the 1988-2004 period sheds new insights about the affine model’s ability to describe the term structure. Comparing two popular specifications of pricing errors, we show that errors from the state-space approach exhibit lower variances and autocorrelations than errors from models where the factors are recovered by inverting the pricing equations. Posterior mean pricing errors also suggest that the short rate might not be as good a proxying factor as longer term rates. 1

Abstract not found

This paper extends the benchmark Macro-Finance model by introducing, next to the standard macroeconomic factors, additional liquidity-related and return forecasting factors. Liquidity factors are obtained from a decomposition of the TED spread while the return-forecasting (risk premium) factor is extracted by imposing a single factor structure on the one-period expected excess holding returns. The model is estimated on US data using MCMC techniques. Two findings stand out. First, the model outperforms significantly most structural and non-structural Macro-Finance yield curve models in terms of cross-sectional fit of the yield curve. Second, we find that financial shocks, either in the form of liquidity or risk premium shocks, have a statistically and economically significant impact on the yield curve. The impact of financial shocks extends throughout the yield curve but is most pronounced at the high and intermediate frequencies.