Motivation
The motivation behind this study is to explore the potential of using concert-related data, combined with Google Trends data, to predict Gross Domestic Product (GDP) growth in local state economies. This approach seeks to address the limitations of traditional economic forecasting methods, which often suffer from delays in data availability and lack frequent updates. The study examines whether the economic activity generated by major concert tours, such as those of Taylor Swift and Beyoncé, can be a viable indicator for current economic trends in the cities visited.
Data
The data extraction process for the study involved gathering information from three primary sources. The first source was the Bureau of Economic Analysis, which provided state-level quarterly GDP data. This data was used to calculate quarter-over-quarter GDP growth for each state. The second source was Wikipedia, from which concert attendance and revenue data were scraped, focusing on concert tours with crowd sizes and durations similar to the Taylor Swift and Beyoncé tours. The third source was Google Trends, offering state-by-state search trends data. This data was used to create a "Google Trend Index" to compare the popularity of each artist by normalizing against a common search term. The combination of these data sources enabled the analysis of relationships between concert activities, Google Trends data, and economic growth.
Results
The study's results showed that while there are some positive correlations between concert activities and GDP growth, these correlations are relatively weak. Machine learning models, specifically a bagging regressor and a neural network, were tested for predicting GDP growth using concert attendance and Google Trend Index data. The models achieved some success in predicting GDP growth from concert attendance data but were less effective when using Google Trend Index alone. The findings suggest that concert-related data might have predictive value for economic forecasting, but cannot be solely relied upon due to limitations like data granularity and the presence of confounding factors.
