基于Vine copula模型的最优宏观经济预警指数研究

How to accurately predict business cycle is an important research field in modern macroeconometrics, which provides helpful guiding information for the production decision of firms, the investment decision of consumers as well as the policy decision of governments. This research attempts to aggregate a series of leading indicators that have predictive power of business cycle, and construct an optimal warning index for macroeconomics using the recent vine copula model in statistics. By construction, this index is supposed to maximize the predictive capacity as measured by the Receiver Operating Characteristic curve. Depending on the complexity of vine copula, this research will discuss specification and estimation of the static vine copula models with data of the same frequency and data of mixed frequency, as well as the dynamic vine copula model. These three models would be implemented on the real-life macroeconomic and financial data to develop two macroeconomic warning indices for China and the United States respectively. Furthermore, we will evaluate the quality of these indices in terms of both the in-sample and out-of-sample performances.

如何精准地预测宏观经济的周期性波动是现代宏观计量经济学中一个重要的研究领域，对企业的生产决策、消费者的投资决策以及政府的政策制定都具有前瞻性的指导意义。本研究利用统计学中最新的Vine copula模型，将一组对宏观经济走势有预测能力的领先指标进行合理加总，构建最优的宏观经济预警指数。该指数能够最大化由Receiver Operating Characteristic曲线衡量的预测能力。本研究将根据模型的复杂程度，分别讨论常系数同频数据Vine copula模型的设定和估计、常系数混频数据Vine copula模型的设定和估计、以及变系数Vine copula模型的设定和估计。在此基础上，结合现实中观察到的一组宏观经济和金融数据，运用上述三种Vine copula模型，分别构建中美两国的宏观经济预警指数，并从样本内和样本外两个视角，评估该指数预测宏观经济波动的绩效。

如何精准地预测宏观经济的周期性波动是现代宏观计量经济学中一个重要的研究领域，对企业的生产决策、消费者的投资决策以及政府的政策制定都具有前瞻性的指导意义。本研究利用统计学中最新的Vine copula模型，将一组对宏观经济走势有预测能力的领先指标进行合理加总，构建最优的宏观经济预警指数。该指数能够最大化由Receiver Operating Characteristic曲线衡量的预测能力。本研究将根据模型的复杂程度，分别讨论常系数同频数据Vine copula模型的设定和估计、常系数混频数据Vine copula模型的设定和估计、以及变系数Vine copula模型的设定和估计。在此基础上，结合现实中观察到的一组宏观经济和金融数据，运用上述三种Vine copula模型，分别构建中美两国的宏观经济预警指数，并从样本内和样本外两个视角，评估该指数预测宏观经济波动的绩效。