| As one of the important engines of China’s economic development,the real estate industry is characterized by its large scale,long chain and extensive involvement.The real estate industry,as one of the important engines of China’s economic development,is characterized by its large scale,long chain and extensive involvement,and has a certain pioneering and systematic influence on the financial stability of the national economy and risk prevention.It should not be overlooked that 2021 is a turning year for the real estate industry.After repeated regulation and control by many parties,the“three highs” development model of “high leverage,high debt and high growth” in the real estate industry has come to an end and returned to a rational development channel.However,when the industry enters a new stage of high-quality benign development and faces the challenge of transformation,individual real estate companies have “burst”,the industry is facing an unprecedented crisis of confidence,and the volume of transactions in some cities has even gone to freezing point.Most home buyers are extremely pessimistic about the real estate market,and whether housing prices will fall has become an issue of great concern for people’s livelihood.Based on this,it is extremely important to study the factors that affect real estate prices and construct real estate price prediction models.By reviewing and summarizing the literature related to the real estate market at home and abroad,this paper selects the real estate prices in Zhengzhou which is the capital city of Henan Province,as the research data.Firstly,the basic trend of real estate prices in Zhengzhou City is analyzed through the H-P filter method.Then,the study finds that the real estate prices in Zhengzhou City have roughly gone through three stages between 2013 and 2021 but still have an overall upward trend.The first stage was before 2016,when house prices rose steadily but grew relatively slowly;the second stage was after the second half of 2016,when house prices soared and grew significantly faster;and the third stage was after the second half of 2019,when house prices began to drop steadily and gradually stabilize.After analyzing the trend of house prices,based on the existing literature base,firstly,this paper summarized the influencing factors of the real estate price in Zhengzhou from four aspects,which includes macroeconomic factors(gross regional product,interest rate,exchange rate),real estate development statistical factors(real estate development investment amount,completed building area of houses),public purchasing power factors(urban residents’ per capita disposable income,consumer price index)and supply and demand factors(local financial general budget expenditure,money supply,land transaction).Then,based on the gray correlation analysis,the correlation magnitude of each influencing factor was calculated.Based on the calculation results,a VAR model was constructed and analyzed by selecting one influence indicator from each of the four factors.Finally,the paper mainly used the GM(1,1)model and the ARIMA model which in the single-factor forecasting model,and the PCA-BP neural network model in the multi-factor forecasting model to forecast and analyze the real estate prices in Zhengzhou City,and compared the forecasting effects of different models through model evaluation indexes.In terms of the analysis of influencing factors,the study shows that real estate prices in Zhengzhou are influenced by a variety of factors together,among which the general budget expenditure of local finance has the greatest influence,urban residents’ per capita disposable income,GDP and the amount of investment in real estate development rank in the top four,while the positive influence of medium and long-term loan interest rates over five years on real estate prices in Zhengzhou is less.The results of the variance decomposition in the VAR model showed that in the short-term forecast,real estate price data and local general budget expenditure can basically explain most of the changes in real estate prices,with the relative contribution of real estate price data being as high as 87%;while in the medium-and long-term forecast,GDP and local general budget expenditure have a more significant impact on real estate prices.The relative contributions of GDP and local general budget expenditures are 21% and 56%,respectively.In terms of model prediction,by comparing the prediction accuracy and prediction precision of ARIMA model,GM(1,1)model and PCA-BP neural network model,it was found that the prediction result of PCA-BP neural network model was closer to the real value with MAPE value of 0.08% and RMSE value of 12.5363,and its prediction effect was significantly better than the other two single-factor prediction models.The model was next used to forecast the real estate prices in Zhengzhou City for the next six months(i.e.,from October 2021 to March 2022),and the prediction results indicated a downward trend in house prices.To address this situation,relevant recommendations are made to promote the healthy development of the real estate market in Zhengzhou based on the findings of the study. |
