随着全球金融市场的不断发展和复杂化，金融机构和投资者面临着更多的风险挑战，更加深入地研究和应用风险度量方法，以应对日益复杂和多变的金融市场环境准确评估和量化风险变得尤为关键。在整个金融风险管理体系中，行业风险是一个重要的研究方向，并且尾部风险会带来较大的冲击。了解和度量行业尾部风险是有效风险管理的基础，可以更好地识别风险来源并采取相应的措施。行业风险的变化往往猝不及防，检验风险水平在何时发生了显著变化可以帮助更好的识别市场变化和优化风险策略。因此本文提出对行业尾部风险的度量研究以及变点检验。

在此基础上，本文提出使用期望分位数来作为风险测度方法，考虑到金融时间序列往往受到历史信息的影响，提出期望分位数自回归模型。本文选取沪深300的五个行业指数作为研究对象，采用期望分位数自回归模型对行业指数的日对数收益率进行实证分析。基于非对称最小二乘法估计模型参数并检验侧模型预测的有效性，同时根据模型计算出的风险值，对选取的五个行业的风险进行初步评估。最后利用变点检验找出风险显著变化的时间点并进行分析说明得出结论。

理论研究以及实证分析结果表明：在风险度量研究上，期望分位数是一种在学术原理上具有明显优势的方法，充分考虑了极端损失的影响，对金融数据厚尾特征敏感。期望分位数自回归模型在行业尾部风险度量上有着良好的度量和预测效果。通过风险平均值对行业风险进行初步评估，选取的五个行业风险值从小到大排序为：金融、医疗、工业、能源、材料。最后对计算出的风险值进行变点检验，并识别变点附近的重要事件，帮助更好理解市场变化对风险的影响。

With the continuous development and complexity of the global financial markets, financial institutions and investors are facing more risk challenges, and it has become especially critical to study and apply risk metrics more deeply to accurately assess and quantify risks in response to the increasingly complex and volatile financial market environment. Industry risk is an important research direction in the overall financial risk management system, and tail risk can be a big shock. Understanding and measuring industry risk is the foundation of effective risk management, allowing for better identification of risk sources and the adoption of appropriate measures. Changes in industry risk are often caught unawares, and examining when risk levels have changed significantly can help better identify market changes and optimise risk strategies. Therefore, this paper proposes a metric study of industry tail risk and a change-point test.

On this basis, this paper proposes the use of expectile as a risk measurement method, considering that financial time series are often affected by historical information, and proposes the expectile autoregressive model. In this paper, five industry indices of CSI 300 are selected as the research objects, and the daily logarithmic returns of the industry indices are empirically analysed using the expectile autoregressive model. Based on the asymmetric least squares method, the model parameters are estimated and the validity of the side model prediction is tested, and the risk of the selected five industries is initially assessed based on the value-at-risk calculated by the model. Finally, the variable point test is used to find out the time points of significant changes in risk and analyse and illustrate the conclusions.

The results of theoretical research as well as empirical analyses show that the expectile is a method with obvious advantages in terms of academic principles in the study of risk measurement, which fully takes into account the impact of extreme losses and is sensitive to the thick-tailed characteristics of financial data. Expectation quartile autoregressive model has good measurement and prediction effect on industry tail risk measure. Preliminary assessment of industry risk is carried out through the average value of risk, and the risk values of the five selected industries are ranked from smallest to largest: financial, medical, industrial, energy, and material. Finally, the calculated risk values are subjected to a change-point test and important events near the change points are identified to help better understand the impact of market changes on risk.