Modelling and Forecasting Short-Term Interest Rate Volatility: A Semiparametric Approach

Abstract

This paper employs a semiparametric procedure to estimate the diffusion process of short-term interest rate. This method is compared in its ability to capture the dynamics of short rate volatility to a class of one-factor diffusion models where the conditional variance is serially correlated and levels dependent. The Monte Carlo study shows that the semiparametric approach produces more accurate volatility estimates than models that accommodate asymmetry, levels effect and serial dependence in the conditional variance. The empirical evidence based on U.S. three-month Treasury bill rates further indicates that the semiparametric procedure is better than the widely used single-factor diffusion models in forecasting the future volatility of interest rate changes. The improvement in modelling short rate volatility using the new procedure has implications for pricing interest rate derivatives.