The Study Of Dynamic Term Structure Models Of China

Term Structure of Interest Rates is a yield curve displaying the relationship between spot rates of zero-coupon securities and their term to maturity, and as the benchmark for the pricing of derivatives, risk management, arbitrage, speculate, and portfolio management it has always been a heated research topic in the field of . Along with the process of innovation in China's financial system, advancement of marketization of interest rate in China and the development of money market, the practical use of Term Structure of Interest Rates has become increasingly important.First this essay reviews the research on Term Structure of Interest Rates from home and abroad, including the formation theory, static approximation, and various dynamic models and their evaluation methods. Using Inter-bank Offered Rate theory, this essay comprehensively analyzes the dynamic changes of term Structure of Interest Rates in China's market. By comparing the performance of several models this essay analyzes which model would best suit for the Chinese market, and thus comes to the conclusion that: 1) there is appearent mean regression phenomenon and level phenomenon in Chinese market; 2) the existence of nonlinear drift in Chinese market cannot be proved; 3) extended CKLS model that combines level effect and heteroscedasticity achieves best performance. At last, this essay analyzes the possible future research topics in Term Structure of Interest Rates.The main conclusions of this dissertation are as follows: 1.Interest rates in Chinaet also show mean reversion and level effect,and the short-term rates have faster mean reversion speed and stronger volatility than long-term rates.2.The factor which affects the dynamic behavior of short and medium-term rate is level,and the factors which affect the dynamic behavior of long-term rates are level and slope;3.For modeling the dynamic of short and medium-term rates,it only need to build a one-factor model, but for long-term rates or term structure building a two-factors model is a more suitable choice.