Social media sentiment and house prices: Evidence from 35 Chinese cities

We develop a new social media sentiment index by quantifying the tone of posts about housing on Weibo between 2010 and 2020 in 35 largest cities in China. We find that the social media sentiment index significantly predicts house price changes for up to six quarters ahead, after controlling for the economic fundamentals. A 1% increase in an accumulated social media sentiment index results in an 0.81% increase in the house price inflation the following quarter, ceteris paribus. Our results cannot be explained by changes to policy, unobserved fundamentals, or censorship bias, and survive a battery of robustness checks. We show they support theories where disperse information has direct economic effects by facilitating social learning as in Burnside et al. (2016); Bailey et al. (2018); Bayer et al. (2021)