高频环境下连续时间模型的计量检验及其在金融市场分析中的应用

本项目研究在高频数据环境下的连续时间模型的计量检验问题。我们将基于现有的连续时间模型计量检验方法的不足之处，通过引入一些高频数据的特征，提出一些更适合高频数据分析的计量方法，包括对漂移函数与波动函数的联合检验、跳稳健的波动函数估计与检验、噪声稳健的跳跃检验或波动函数检验、共同跳跃检验等。我们将依据经验过程理论去构造检验统计量，并推导这些检验统计量的大样本性质，包括相合性与局部功效特征。我们还将利用蒙特卡洛模拟方法去比较各类检验方法的小样本性质。本项目最后将研究的理论方法应用于我国金融现货市场与衍生品市场的实证分析。我们将探讨这些市场的微观结构特征，如跳跃与微观噪声对两个市场间对冲关系的影响。

This project studies the econometric testing for continuous time model under the environment of high frequency data. Based on the disadvantage of existing testing of continuous time models, we will provide some more satisfactory econometric methods for high frequency data by incorporating the characters of high frequency data. The methods being studied include the simultaneous test of drift function and volatility function, the jump robust estimation and test for volatility function, the noise robust test for jump or volatility function, and the cojump test. We will construct the test statistics based on empirical processes, and analysis their large sample properties, including consistency and local power. We will also study the finite sample properties by Monte Carlo simulations. Finally, we will apply our proposed methods to empirical analysis of financial spot market and derivatives market. We will studies the microstructure of the markets, such the effect of jump and market noise on the hedging relation between them.