金融时序数据具有高度非线性，导致它们在不同分位数下有完全不同的表现。因此，如何估计金融时序数据的动态条件分位数，逐渐成为计量经济学的前沿问题之一。本项目旨在以金融时序数据为研究对象，建立合理而且实用的动态分位数回归模型，发展模型的估计理论与方法。具体内容包括：研究门限自回归分位数回归模型；研究半参数和非参数分位数回归模型，研究基于copula的分位数回归模型，以及删失机制下动态分位数回归模型。本项目的研究意义在于利用金融时序数据特点，充分挖掘该类数据的暗含信息，并对所构建的不同非参数或者半参数模型，发展出新的实用统计推断方法。最后,本项目将通过大量的计算机模拟实验和实际数据分析来检验所提模型和方法的可行性和有效性，给金融时序分析提供重要的理论依据和实践指导。

Due to its the highly nonlinearity, financial time series may have a different behavior across quantiles. Thus, how to estimate the dynamic onditional quantiles in the financial time series data has become a cross-frontier issue in econometrics. Our project aims to establish some suitable and practical semiparametric dynamic quantile regression models for financial time series, and to develop new estimating methods and inferences. It includes threshold quantile autoregressive models, semiparametric/nonparametric quantile regression models, semiparametric conditional quantile estimation through copula-based models, and dynamic quantile regression models for censored data. This project is to develop some new practical methodologies for the different semiparametric models, by taking advantage of the nonlinearity of financial time series data and fully exploiting the potential information implied in such data. Furthermore, simulation studies and real data examples should be conducted to illustrate the theories and methods. Therefore, the proposed research aims to provide an important theoretical evidence and practical guidance for financial time series.