【作者】 高飞；

【导师】 何勇；

【作者基本信息】 山东大学 ， 应用统计（专业学位）， 2022， 硕士

【副题名】基于椭球因子模型视角

【摘要】 2020年1月我国湖北省武汉市爆发了新型冠状病毒疫情,而后新冠病毒因其传染性强的特点迅速传播至我国各个地区,不仅严重威胁人民生命健康安全,随着国内防疫措施的实施,社会运行基本停滞,经济发展也遭受重创。新冠疫情作为2008年国际金融危机以来世界经济面临的最大风险,对我国乃至世界经济的冲击都是重大且深远的。在后疫情时代的背景下,如何对中国宏观经济的影响因素变化进行分析,并对其未来走势进行预测仍是当前宏观经济研究学者需要解决的重要课题。本文依据一定的指标选取原则选取了 51个宏观经济指标,构建宏观经济指标体系,针对宏观经济数据的厚尾性特点,创新性地将高维椭球近似因子模型稳健两步估计方法(RTS)应用于中国宏观经济结构的影响因素变化与趋势预测研究,具体从以下三个方面入手。首先以动态的视角利用 MKER(Multivariate Kendall’s tau Eigenvalue Ratio)方法和MKTCR(Multivariate Kendall’s tau Transformed Contribution Ratio)方法估计潜在影响宏观经济结构的公共因子个数,探索突发性重大经济事件的冲击和公共因子数之间存在的关系。研究发现,在宏观经济面临类似于新冠疫情这样的冲击时,潜在公共因子个数会发生明显变化。接着本文将高维椭球近似因子模型应用于我们搜集的宏观经济面板数据中,利用RTS方法提取前四个公共因子并根据载荷矩阵赋予每个公共因子实际意义,研究发现前四个公共因子的得分在2020年新冠疫情爆发后也发生了相应的变化,研究各个因子得分随疫情的变化趋势对应对新冠期间保证宏观经济各个领域正常运行有着不同寻常的意义。同时我们发现第一公共因子可以解释为统计区间内最能代表国民经济发展的因子,其包含的主要变量代表了当下引领经济发展的主旋律。接下来我们对第一公共因子做重点研究,以动态的角度,基于滚动窗口的方法,分析随着时间的推移,影响第一公共因子的主要经济变量发生了怎样的变化,得到了各个时期尤其是新冠期间影响宏观经济结构的主要因素的变化趋势。最后本文将椭球近似因子模型应用于宏观经济变量滚动预测,结果表明椭球近似因子模型相较于传统的因子模型能更稳健地提取宏观变量中的公共成分,在宏观经济面临新冠疫情冲击时具有更好的预测效果。更多还原

【Abstract】 In January 2020,the Novel Coronavirus outbreak broke out in Wuhan,Hubei Province of China,and then spread rapidly to various regions of China due to its strong infectivity,which not only seriously threatened people’s lives and health safety,but also brought social operation to a standstill and economic development suffered a heavy blow with the implementation of domestic epidemic prevention measures.As the biggest risk to the world economy since the 2008 international financial crisis,COVID-19 has had a significant and far-reaching impact on China’s economy and even the world economy.In the post-epidemic era,how to analyze the changes of the influencing factors of China’s macroeconomic structure and predict its future trend is still an important issue for macroeconomic researchers.According to certain principles of index selection chosen 51 macroeconomic indicators,build a macroeconomic indicators system,according to the characteristics of the thick tail of macroeconomic data,the innovation in the higher dimensional ellipsoid approximation factor model robust two-step estimation method(RTS)application in China’s macroeconomic structure factors influencing the change and trend prediction research,the concrete from the following three aspects.Firstly,MKER(Multivariate Kendall’s tau Eigenvalue Ratio)method and MKTCR(Multivariate Kendall’s tau Transformed Contribution Ratio)method are used to estimate the number of common factors potentially affecting macroeconomic structure from a dynamic perspective,and the relationship between the impact of sudden major economic events and the number of common factors is explored.The study found that the number of potential common factors changes significantly when the macro economy faces a shock like COVID-19.Then this article will high-dimensional ellipsoid approximation factor model is applied to the macroeconomic panel data we collected,the RTS method is utilized to extract the first four common factors and according to the actual load matrix gives each public factor,the study found that the first four common factor score after new crown outbreak in 2020 also made corresponding change,Studying the variation trend of each factor score with the epidemic situation is of great significance to ensure the normal operation of all areas of macro economy during the response to COVID-19.At the same time,we find that the first common factor can be interpreted as the factor that best represents national economic development within the statistical interval,and the main variable it contains represents the main melody leading economic development at present.Next we do key research of the first public factor,to the point of view of dynamic,based on the rolling window method,analysis with the passage of time,the effects of the first public what kind of changes have taken place in the main economic variables,obtained during various periods,especially the new champions of the main factors influencing the macro economic structure change tendency.Finally,this thesis applies the ellipsoid approximation factor model to the rolling prediction of macroeconomic variables,and the results show that compared with the traditional factor model,the ellipsoid approximation factor model can extract the common components of macroeconomic variables more robustly,and has a better prediction effect when the macro economy is faced with the impact of COVID-19.更多还原