The Semiparametric Model of Interest Rate Term Structure Based on GCV Method and Its Empirical Comparison

Abstract

In order to improve the smoothness of curve fitted by the interest rate term structure model of polynomial spline functions, the adaptive semi parametric regression with a penalized item is introduced to estimate the unknown parameters. The generalized cross-validation method is discussed to select the smoothing parameter, and genetic algorithm is applied to search the optimal smoothing parameter. Then, the empirical results show that this model with penalty function is relatively effective in China. However, the curve fitting smoothness is improved to some extend at the expense of fitting accuracy.