中文摘要：

以中国上海证券交易所从债券价格导出的0．5年期利率的周度数据为分析对象，使用SNP法，估计出短期利率的条件密度函数，发现其条件分布具有明显的异方差性和非正态性．然后利用EMM法实证分析了常用的连续时间的单因子和两因子短期利率模型．单因子短期利率模型包括Vasicek模型，CIR模型，CKIS模型等，两因子利率模型包括Gallant，Tanchen给出的随机波动率模型和Balduzzi等人的随机均值回复模型．实证结果表明所有的单因子短期利率模型都不能很好地描述中国上海证券交易所债券市场上的短期利率变化，CKIS模型是它们中表现最好的单因子利率模型．随机均值回复模型也不能描述短期利率的变化，只有随机波动率模型可以描述上海证券交易所的短期利率的变化．

英文摘要：

According to weekly data of half-year interest rate derived from the trading prices of Government bonds in the Shanghai Stock Exchange, and by the SNP approach, conditional density of the half-year interest rate is estimated, to find that the density shows obvious heteroskedasity and no-normality. By the EMM approach, continuoustime interest rate models are estimated and tested. The single-factor models include the Vasieek model, the CIR model, and the CKLS model, eet. The two-factor models include the stochastic volatility model proposed by Gallant and Tanehen, and the stochastic mean model proposed by Balduzzi et al. The results show that all the single-factor models cannot match dynamic change of the short interest rate, and the CKLS model does the best among them. The stochastic mean model cannot catch dynamic change of the short interest rate eirther. Only the stochastic volatility model can fit the dynamic statistical behavior of the change in short interest rate.