Research On House Price Expectations Based On The Big Data From Social Media

The house price experienced persistent and rapid appreciation during recent years in China.In addition to the support of economic fundamentals in this period,people's overoptimistic expectations of future house price growth were also one of the essential reasons of such price apprecitation.Hence,when building the so-called Long-Term Mechanism system for housing market,the Chinise government pays more attention to expectation management,anticipating to effectively guide and stabilize the expectations,and thus stabilize the housing market.However,the existing research on house price expectation in China is substantially lagging behind that in more developed countries,and thus cannot support those expectation-related policy making decisions and implementations.Based on a systematic review of the foreign and domestic existing literature,a theoretical framework of house price expectation research is established,and then the expectation-related economic patterns are theretically analyzed under this framework.With the help of a deep-learning-based text classification model as well as emerging technologies of big data analysis and text mining,the price expectation of the Chinese housing market is measured via the social media data from a microblogging platform.Based on the above theoretical analysis and measurement,the empirical analysis is conducted from two perspectives—the first is the contributory factors in expectation formation and the direct effects of the expectation on the housing market,and the second is the expectation diffusion patterns and the indirect effects of the expectation on the housing market.This dissertation concludes in the following findings:(1)It is effective to measure the house price expectation by constructing a deep learning model to classify microblogging messages according to their expectation tendencies,and the house price expectation index constructed by this text-based approach shows advantages over the expectation indicators built by the results from survey questionaires.(2)City-level economic fundamentals have significant effects on people's house price expectations,and the effect patterns are consistent to theoretical results;however,the past house price changes have no positive effects on house price expectations,which is inconsistent to the short-term momentum of house price changes.(3)People's house price expectations have positive and significant effects on the future resale house price changes,the future newlybuilt house price changes,and the future transaction volume of newly built houses;such effects show heterogeneities across different cities and market conditions.(4)The expectation tendencies expressed in the microblogging messages posted by social media users have significant effects on other users' expectations,which means that the expectations can diffuse across different individuals;such diffusion pattern shows temporal persistency and various heterogeneities.(5)People's house price expectations can diffuse across different cities,positively and significantly affecting the expectations of citizens in other cities,and thus indirectly affect the future housing market performances in other cities.Some policy implications can be derived from the above findings.The government should incorporate house price expectation indicators into the real estate market monitoring system.Hence,the government can not only utilize such indicators to forecast future housing market trends and thus optimize relevant market policies and planning,but also conduct expectation guidance more effectively and thus stabilize future house price trends.In particular,the government may impose some necessary interventions on the house price expectation diffusion on social media platforms,avoiding the potential overoptimistic tendencies of house price expectation caused by expectation diffusion.