The Research Of Jump Models For Predicting Short-term Interest Rate And Jump Behavior Effect Factors

As one of economic variables, interest rate has very important functions in finance market. Particularly, short-term interest rate plays fairly important roles on the pricing of derivative products, the transmission of central bank monetary policy and risk management. Therefore, how to construct a suitable dynamic term structure of interest rate model which can offer a good description of short-term interest rate dynamic behaviors and predict the rate precisely has always been the core study on short-term interest rate.The paper is based on that short-term interest rate is current hot issues which national macro-control pays close attention to and its volatility is very high. The author evaluates and analyzes forecasting models such as Vasicek model. Point out deficiencies and defects in those models. Based on the optimal and improved Vasicek model, the paper constructs two jump models with constant and time-varying volatility. In the following, this paper makes use of maximum likelihood estimation, Wald test and likelihood ratio test and Monte Carlo simulation to put up an empirical analysis with the two jump models. The paper uses seven days bonds repurchase rate of Shanghai exchange market to substitute for short-term interest rate. The sample is from June05,2006to April20,2010. It describes the dynamic characteristics of short-term interest rate.The empirical results show that:Both the constant volatility jump model and the time-varying volatility jump model demonstrate that short-term interest rate has significant jump behavior. The predictive power and fitting result of jump models are better than the corresponding non-jump models. At the same time, the predictive power and fitting result of the time-varying volatility jump model is better than the constant volatility jump model. Besides the time-varying volatility jump model can completely describe the dynamic characteristics of short-term interest rate such as mean reversion, time-varying volatility and jump behavior.It shows that the jump models constructed in the paper are feasibility, validity and practicality. Based on the empirical results, the paper makes further analysis of specific factors such as week effect and common stock new issues effect which affect jump behavior. The jump model is linked with economic factors which can make the central bank macro-control better. The conclusions are drawn that Friday and the day before issues day have greater influences at the probability of jump. The paper is in order to solve the problems that forecasting tools are seldom and inaccurate.The paper is consisted of six chapters. The first chapter is introduction. The second chapter is relative theories of modles for predicting short-term interest rate and the simulation method. The third chapter is the choice and establishment of the jump models for predicting short-term interest rate. The fourth chapter is the empirical analysis of the jump models for predicting short-term interest rate. The fifth chapter is the analysis of factors affecting jump behavior on the base of the time-varying volatility jump model. The last chapter is the conclusions and the prospects of this paper.