Models which represent long-term interest rates as long averages of expected short-term interest rates imply, because of the smoothing implicit in the averaging, that long rates should not be too volatile. The volatility of actual long-term interest rates, as measured by the variance of short-term holding yields on long-term bonds, appears to exceed limits imposed by the models. Such excess volatility implies a kind of forecastability for long rates. Long rates show a slight tendency to fall when they are high relative to short rates rather than rise as predicted by expectations models.

Models which represent long-term interest rates as long averages of expected short-term interest rates imply, because of the smoothing implicit in the averaging, that long rates should not be too volatile. The volatility of actual long-term interest rates, as measured by the variance of short-term holding yields on long-term bonds, appears to exceed limits imposed by the models. Such excess volatility implies a kind of forecastability for long rates. Long rates show a slight tendency to fall when they are high relative to short rates rather than rise as predicted by expectations models.

Application of some advances in econometrics (in the theory of cointegrated vector autoregressive models) enables us to deal effectively with two problems in rational-expectations, present-valu e models: nonstationarity of time series and incomplete data on infor mation of market participants. With U.S. data, the authors find some relatively encouraging new results for the rational-expectations theo ry of the term structure and some puzzling results for the present-va lue model of stock prices.

This paper analyzes the correlation between interest rates and prices which as persisted for the past quarter of a millennium and has been termed the Gibson Paradox. Spectral techniques confirm the correlation between long-term interest rates and prices for very long-term swings (the Gibson Paradox), but indicate a significant short cycle correlation only for short-term interest rates, which we term the Kitchin Phenomenon. Past explanations of these correlations have often failed to distinguish cycle lengths and term of interest rates involved. Our analysis rejects Irving Fisher's "price expectation" explanation and the Sargent-Wicksell velocity of money explanations. We propose alternative explanations which in part relate to the characteristic behavior of governments during wartime and in part to distributional effects of unanticipated inflation. Our analysis strongly suggests that prior to World War I nominal long and short rates of interest can be regarded as real rates.

This paper analyzes the correlation between interest rates and prices which as persisted for the past quarter of a millennium and has been termed the Gibson Paradox. Spectral techniques confirm the correlation between long-term interest rates and prices for very long-term swings (the Gibson Paradox), but indicate a significant short cycle correlation only for short-term interest rates, which we term the Kitchin Phenomenon. Past explanations of these correlations have often failed to distinguish cycle lengths and term of interest rates involved. Our analysis rejects Irving Fisher's "price expectation" explanation and the Sargent-Wicksell velocity of money explanations. We propose alternative explanations which in part relate to the characteristic behavior of governments during wartime and in part to distributional effects of unanticipated inflation. Our analysis strongly suggests that prior to World War I nominal long and short rates of interest can be regarded as real rates.