Abstract: Establishing a sound and complete term structure of interest rates and reflecting the supply and demand relationship in the economic market is of great significance to the development of China′s market economy. The purpose of this study is to construct an econometric model that is suitable for China′s actual situation. And the model can more accurately describe the term structure of interest rates. Static estimation and dynamic estimation are two basic methods for estimating the term structure of interest rates. Static estimation is the basis for verifying the dynamic estimation model and performing dynamic analysis. By analyzing the important methods in static estimation-the basic model structure of the cubic spline function estimation method,and based on the disadvantages of market experience and basic genetic algorithm for the demarcation point estimation of the cubic spline function, improved Genetic Algorithm, which has a hierarchical mechanism and dynamic probability, is used to estimate the breakpoint of cubic spline function;comparing the estimation results of time points of multiple bond samples, it is proved that the cubic spline function based on improved genetic algorithm is better than market experience and simple genetic algorithm in estimating the term structure of interest rate both in-sample model estimation and out-of-sample model prediction.

Key words: Spline Function, Improved Genetic Algorithm, Dynamic Cross Probability, Interest Rate Term Structure