A Text Convolution Network Based Policy Uncertainty Index : An Application For China

Accurate measurements of economic policy uncertainty (EPU) are essential for understanding and predicting economic dynamics. In this study, we use text convolution network, an advanced natural language processing method, to develop an innovative monthly index of EPU. It is explored that this method can better analyze complex semantics and subtle language in textual data and accurately measure policy uncertainty. We apply this method to the Chinese real estate market based on the texts of 60 Chinese newspapers from 2005 to 2019. It is proved that the index effectively captures the characteristics of policy uncertainty in China’s real estate market. This new approach is not limited to the study of policy uncertainty but can be widely applied to other concept-based text analysis