Data-Driven Estimation of Economic Indicators with Search Big Data in Discontinuous Situation

Economic indicators play a crucial role in shaping policy and strategy decisions in both the public and private sectors. However, as government indicators based on macroeconomic data are typically released one to two months later, there is a high demand for timely estimates of these values, known as "nowcasting," which has thus far been unsuccessful. In this report, we present the development of the first statistical model for predicting government economic indicators using only non-preset search engine query data (Search Big Data). While traditional models relying on macro indicators struggled to accurately forecast economic outcomes during the disruption caused by the COVID-19 pandemic, the new model successfully nowcasts economic indicators prior to official government announcements. Our analysis of the key queries used in the model also reveals that libido and desire for laughter are key drivers of economic indicators in the time period and region analyzed