Analyzing Twitter Sentiment and Hype on Real Estate Market: A Topic Modeling Approach

Abstract

This study examines the relation between sentiment and hype (intensity of coverage) on Twitter and the local housing market prices across 10 U.S. cities of the S&P/Shiller-Case Composite Home Price index from 2010 to 2021. Using Latent Dirichlet Allocation (LDA) topic modeling algorithm, we identify seven unique topics related to the housing market based on people's tweets: Households, Economic policy, Commercial real estate, Price and rate, Residential housing, Investing, and Future trends. We gather and analyze data on house price indexes, fundamental economic factors, and sentiment and hype scores for the discovered topics. The study finds that the sentiment of Price and rate, Residential housing, and Future trends are significantly and positively related to future house price changes. In contrast, the lags of sentiment of Commercial real estate and Investing have a negative relation with house price. Moreover, we document that hype scores not only have a positive relation with house price changes for all topics but also outperform sentiment scores for forecasting housing market prices. Overall, the study highlights the potential benefits of integrating social media data into existing economic models to gain a more comprehensive understanding of the factors driving fluctuations in the housing market.