Abstract: With respect to risk of financial gain from different policies, a policy risk factor identification method based on Hilbert-Huang transform is proposed. To begin with, we decompose the financial time series into several intrinsic mode functions which have different time scale features. Secondly, we restructure the intrinsic mode functions and form three basic components which reflect short fluctuation, medium-term change form and general tendency of the original series by using spectrum, power spectrum and T-test methods. In terms of the short fluctuation component, we get the abnormal volatility by calculating the frequency and time domain features of the series through empirical mode decomposition and the Hilbert spectrum analysis; in medium-term component respect, we quantize and simulate the policies shock by event-study analysis, and finally match the abnormal volatility with the quantitative policies and recognize the policy risk factors. The method is of important reference significance for analyzing the effect of regulatory policies on financial returns fluctuation. Finally the high precision and wide application prospect are illustrated by a numerical example with national real estate regulation policies and the fluctuation of real estate indexes.

Key words: policy risk, Hilbert-Huang transform, financial benefits, identification method

摘要： 针对政策可能对金融收益产生风险问题,提出了基于Hilbert-Huang变换方法的政策风险因子识别检测方法。通过经验模态分解,Hilbert-Huang频谱分析得到金融时间序列的时域和频域特征,通过与量化处理后的政策进行匹配得到政策产生的异常波动情况,从而实现对政策因子风险的识别与处理。研究结果对于探究宏观政策对金融收益的影响具有重要参考意义。最后以国家房地产调控政策与地产指数为算例,发现本研究提出的方法识别精度高,具有非常好的应用前景。

关键词: 政策风险, Hilbert-Huang变换, 金融收益, 识别方法