基于高频数据的Levy跳跃扩散模型的计量检验和估计

本课题研究基于高频数据的Levy跳跃扩散模型的计量检验和估计，并将其方法应用于商品期货的风险管理、对冲和投资组合分析。相对于低频数据，高频数据能够更准确地估计资产的波动性、资产之间的相关性和相依性、以及对冲比例等，因此能更准确地度量投资组合的风险和更有效地进行资产配置。本课题将首先研究基于高频数据的参数和半参数Levy跳跃扩散模型的计量检验，跳跃是否存在，多资产是否同时发生了跳跃，波动跳跃模型等。资产组合的风险度量、资产之间的对冲比例等的计算取决于准确地估计资产收益率之间的协方差、相关系数或相依系数。本课题将发展新的高频数据的协方差和相依系数的估计，以及微观结构噪音对其估计的影响。

This research project studies the issues of econometric testing and estimation of Levy jump diffusion models based on high-frequency data, with applications to risk measurement, hedging, and portfolio analysis for commodity futures. Relative to low-frequency data, high-frequency data can be used to more accurately estimate volatilities of assets, correlations and dependences among assets, hedging ratios, etc.. So portfolio risk can be better measured and more efficient portfolio can be formed based on high-frequency data. This project will first study the issues of econometric testing of parametric and semi-parametric Levy jump diffusion models, testing whether jump exists, testing co-jumps among different assets, stochastic jump diffusion models,etc. Calculations of portfolio’s risk and hedging ratios among assets depend on accurate estimations of covariances, correlations or dependence measures among assets’ returns. We’ll develop new methods for estimating covariances and dependences based on high-frequency data and how these estimates are affected by microstructure noise.