Interest Behavior Research Based On Jump Diffusion Model

This paper builds up a basic frame of one factor jump diffusion model, and splits it into three parts——drift, diffusion and jump components. According to 4 dimen-sionalities, this paper sets up 8 models and discusses 18 conditions in order to compare strengths and weaknesses of objects in the 4 dimensionalities. The 4 dimensionalities include linear or nonlinear drift function, constant or time-dependent volatility param-eter (stochastic volatility), non-jump, one-jump or double-jump model of interest, and jump size of normal or double exponential distribution.According to principles of stationarity, correlation, and trading volume, this paper makes empirical analysis of Shibor(1W), and tries to estimate parameters by Markov Chain Monte Carlo method. Because all the MCerror/sd are small, it is regarded as effective convergence. Besides, analysis of β (mean of interest) and λ (rate of Poisson process) explains that strengths of jump model and double exponential distribution of jump size.At last, this paper compares the mean average error of simulations of models in order to get 7 conclusions.