A nonparametric dimension test of the term structure

Abstract

This paper addresses the problem of conducting a nonparametric test of the dimension of the state variable vector in a continuous-time term structure model. The paper shows that a bivariate diffusion function of the short rate process is a sufficient condition for the term structure to be driven by two stochastic factors. Using an easy-to-implement kernel smoothing method the number of state variables can be tested under very unrestrictive assumptions. The results suggest that continuous-time models for the US interest rates should contain at least two stochastic factors.