Research On Network News Media Sentiment On House Prices Fluctuations Based On Sentiment Analysis

Since entering the 21 st century,China’s real estate market has experienced rapid development,and housing prices have generally risen at a high rate and fluctuated more frequently.Soaring or plummeting housing prices exacerbate the instability of the real estate market,which in turn poses a threat to the macroeconomy.The problem has not only aroused widespread concern from all walks of life,but also has triggered extensive discussions in academia.In recent years,with the rapid development of behavioral finance research,the relationship between media coverage and asset pricing has gradually become a new research hotspot.Internet news media play an important role in the real estate market,not only as information intermediary,but also the subjective emotional tendency of media reports can affect the cognitive sentiment and investment decisions of real estate market participants,and ultimately influence house prices.Based on the background,Firstly,the long-term trend of house prices are decomposes using HP filtering method.The components of house price fluctuations are striped and used to analyze the fluctuation characteristics of national housing prices during the period from January 2016 to December 2021.Secondly,the research data sets of real estate media reports are obtained using web crawler technology from several authoritative news websites.The sentiment pre-training model SKEP is used to judge the headlines of real estate online news media reports sentiment.And the emotional results are constructed a monthly real estate online news media sentiment time series index.Finally,considering the time-varying economic structure and variable external environment,a TVP-VAR model is constructed to examine the time-varying nonlinear effects of real estate network news media sentiment on house price fluctuations from three dimensions of shock time,response degree,and duration with the perspective of quantitative analysis.Through the above study,the research conclusions are as follows:(1)There are frequent fluctuations in house prices nationwide during the sample period,but they do not affect their overall upward trend.House prices fluctuate are highly at the point of strict policy regulation and at the point of epidemic shock.(2)The monthly real estate network news media sentiment time series index can well portray the state of media coverage of China’s real estate market,and the SKEP sentiment analysis model can accurately determine the sentiment tendency of news title.(3)There are significant dynamic time-varying characteristics of the influence of ntework news media sentiment on house price fluctuations.In the short term,media sentiment can cause sharp fluctuations in house prices,but in the medium and long term,the positive effect of the "invisible hand" on house price fluctuations is smaller,and house price fluctuations are more moderate.(4)There are differences in the duration and intensity of the impact of network news media sentiment on house price fluctuations at different points in time.The duration of the impact on house price fluctuations is shorter at the point of strict policy regulation and at the point when the real estate market is hit by unexpected events,and the intensity of the impact is higher at the point of strict policy regulation.The empirical results confirm the existence of the "media effect" in the Chinese real estate market.Based on this,this paper proposes the use of online news media sentiment indicators for market monitoring and early warning,and the improvement of the property market information mechanism.