A Method of Economic Indicator Nowcasting Using Baidu Searches

Abstract

Macroeconomic indicators reflect status of economic entities in different domains, especially some leading economic indicators, which are helpful for nowcasting consumption trends, like CCI and CPI, therefore are important for economic trend prediction, policy making and strategy decision. As the largest provider of search engine services in China, Baidu has massive time-series data that uncovers user search behaviors, which have some relationship with economic activities. In this paper, we propose a method called PBS (Predictable Baidu Searches), automatically mining correlated information in Baidu's massive search query data to nowcast leading economic indicators. PBS helps us understand our economic and nowcast macroeconomic indicators. The nowcasting of CPI and CCI in China demonstrates the validity of PBS.