Real interest rates, inflation and the term structure of interest rates.

Abstract

Chapter one proposes a new model for estimating economic agents' anticipation of the real rate of interest. It decomposes the nominal short term interest rate into an ex ante real interest rate and an expected inflation rate, according to Fisher's equation. Assume the ex ante real interest rate follows an autoregressive structure and inflation follows an IMA(1,1) process. Using the information in the nominal short term interest rate and the inflation series, the ex ante real interest rate is estimated by maximum likelihood using the Kalman filter to calculate the likelihood function. The results show that the time series of estimates of the ex ante real interest rate extracted from the model rejects the random-walk hypothesis at the 1% significance level. Chapter two proposes a model to extract the dynamic structure of inflation and the real interest rate contained in the term structure under the expectations hypothesis. Six different models are used to investigate this issue. This paper uses quarterly data from 1960:1 to 1991:1 for inflation, three month nominal short term interest rates and long term yields with maturities from one to five years. The models are estimated by maximum likelihood, using Kalman filtering. The results imply that the expectations hypothesis of the term structure of interest rates can only be reconciled with the data with a random-walk real interest rate and a time-varying term premium. Chapter three takes a different approach to explore the implications of the expectations hypothesis. It models inflation and short term real interest rates using a technique that allows for changes in regimes. This paper examines whether persistent shifts in the term premium can be eliminated after taking into account the regime shifts in the processes for inflation and real interest rates. The result shows that the regime-switching model does not reconcile the data with the expectations hypothesis. However it also suggests that persistence of the term premium obtained from a fixed-coefficient ARIMA model may be due in part to systematic forecast errors that are eliminated by allowing for regime changes.