Empirically Confronting Stochastic Singularity: An Application to the Cox, Ingersoll, and Ross Model

• Activity Number: 473
• Type: Contributed
• Date/Time: Wednesday, August 1, 2007 : 2:00 PM to 3:50 PM
• Sponsor: Business and Economics Statistics Section
• Author(s): Kenneth Roskelley*+ and Christopher Lamoureux
• Companies: The University of Arizona and The University of Arizona
• Keywords: Testing Models of Arbitrage ; Forecasting Interest Rates

Abstract

We use encompassing regressions to evaluate interest rate forecasts along the yield curve over the period 1989 through 2005. Inputs to the encompassing regression include the forecast from one- and two-factor Cox, Ingersoll, and Ross (CIR) models--obtained by estimating the model without adding pricing errors, the slope of the yield curve, and the change in Fed target rates. At the short end, we also include the forward rate and forecasts derived from popular non-parametric models. This empirical design has three advantages. First, we maintain the stochastic singularity common to all no-arbitrage models. Second, rather than accept or reject the model's cross-sectional restrictions, we evaluate if the model contains any useful information about yield dynamics. Third, we restrict our attention to a time period of transparent Federal Reserve policy.