Research On The Spatial Effect Of Chinese House Price Based On Semiparametric Spatial Regression Model

Since the marketization of housing,China's real estate industry has made great progress.At the same time,our country's housing prices have been rising for a long time.The long-term rise in house prices has attracted widespread attention from people across the country.There have been many studies to explain and analyze the long-term rise in China's housing prices.Research shows that China's housing prices are mainly affected by related factors such as land supply policies,real estate related tax policies,urbanization process and domestic monetary policies.In recent years,with the rapid integration of Beijing-Tianjin-Hebei,Hong Kong-Zhuhai-Macao and the Yangtze River Delta Metropolitan and the rapid construction of domestic high-speed rail and highways,more and more people are starting to do business and study across provinces.Moreover,this year,Chengdu and Chongqing began to accept each other's housing provident fund.These factors have made the spatial correlation of housing prices between cities significantly strengthened.It is foreseeable that the factors that determine the trend of housing prices in the future will not be limited to traditional factors such as land supply and urbanization,and various interactive activities between cities will also have a non-negligible effect on the trend of housing prices.In recent years,more and more studies have begun to pay attention to the spatial interaction effect of house prices,and more literatures use spatial autoregressive models.Based on combing the relevant research of house prices and the theory of spatial econometrics,this paper proposes to introduce non-parametric statistical analysis methods into the spatial econometric model,replace the spatial weight matrix of the traditional spatial autoregressive model with a spatial weight function,and then use B-Spline to approximate the unknown spatial weight function,Then,I use the GMM method is used for model estimation.Through empirical research,it can be found that using the unknown spatial weight function to represent the spatial interaction effect of house prices has the following advantages over the spatial weight matrix set using prior knowledge.First,the estimation result of the spatial weighting function depends entirely on the information of the data.The method of setting the spatial weight matrix must be somewhat different from the real spatial interaction in data.Experimental results show that the semi-parametric model of the spatial weight function will achieve better model results than the parametric model.Second,the spatial weight function can mine more information about spatial interaction effects,because the spatial weight function is a function of distance.Spatial weighting function can reflect the spatial interaction effect of house prices that change with distance at different distances.However,the coefficient of the spatial weight matrix is a scalar,which only reflects the spatial interaction effect in the average sense of all data.The results of empirical research show that the fitting effect of the semi-parametric space model on house price data is better than the parameter space model.The model selection indicators BIC and AIC indicate that the overall performance of the semi-parametric space model is better than the parametric space model.From the results of the semi-parametric spatial model,we can see that China's housing prices have a negative spatial interaction effect at the level of population distance,and the interaction effect generally weakens as the population distance increases.It shows that there is obvious population-based competition in housing prices between cities.The competitive relationship will gradually disappear as the populations distance increases.In addition,China's housing prices have a positive spatial interaction effect at the geographic distance level,and the aggregation effect generally weakens as the geographic distance increases.It shows that there is obvious aggregation of house prices between cities in the geographical area close to each other,and the degree of aggregation will gradually disappear as the geographical distance increases.Based on the results of empirical analysis,this article suggests strengthening the monitoring of house prices in key central cities,and paying attention to the leading role of national central cities in regional house prices.When formulating the coordinated development plan of the city circle and the local talent policy,government should focus on the interactive effect of housing prices to ensure the long-term healthy and stable development of housing prices.The innovations of this article are as follows:First,based on the traditional spatial econometric model,this paper uses non-parametric methods to deal with spatial interaction terms,and establishes a semi-parametric spatial model,which is more flexible in studying the spatial interaction effects of house prices.Then,this article use of population distance to analyze the spatial interaction effect of housing prices and expands the scope of application of spatial models,proposes new perspectives on the impact of demographic factors on housing prices,and facilitates the formulation of policies such as talent introduction and settlement.