Research On The Model For Compiling Price Index Of Ordinary Residential Houses In Chinese Cities

With the increase of state regulation and control on the newly-built ordinary houses market and the gradual expansion of the second-hand ordinary houses market,the transactions of the second-hand houses in some Chinese cities have surpassed the newly-built ordinary houses.At present,the compiling method of the house price index published in China is backward and the house price index of newly-built ordinary houses and second-hand ordinary houses is released separately,which leads to a certain deviation between the house price index and the real market situation.An accurate house price index is an important data basis for measuring the stability and health of the real estate market,so it is of great theoretical and practical significance to compile the house price index scientifically and conveniently.In this thesis,we take ordinary houses in Chinese cities as the research object,and systematically study the compilation theory and method of its price index.Firstly,we build a longitudinal unbalanced panel model for the house price index of newly-built ordinary houses.In the first step,we analyze the factors that affect housing prices and prove that the selection of matching pairs in the matching model constructed by the set of upright adjacent floors as the matching space will lead to different parameter estimation results.In order to overcome this problem,we make a more precise definition of the set of upright adjacent floors.Then,we establish the longitudinal unbalanced panel model including the property sales time,standard height and the interaction effect of the building type and standard height by considering the newly-built ordinary houses in the same set of upright adjacent floors as a whole.Through the discussion of the individual effect of the set of upright adjacent floors,the two-stage least square method and two-stage composite quantile regression method are designed to estimate the parameters of the longitudinal unbalanced panel fixed effect model.Secondly,we compare and analyze the accuracy and robustness of the house price index compilation model and the comprehensive quality of the house price index from the three aspects: the back-test error and robustness of model and the economic significance of the index.The research shows that the longitudinal unbalanced panel fixed effect model,which uses the two-stage composite quantile regression method to estimate parameters,has stronger robustness and higher accuracy.Moreover,the house price index compiled by this model has more obvious trend and better comprehensive quality.At the same time,this model is not sensitive to outliers,so it has a certain applicability to the compilation of the house price index in the case of abnormal housing price data.Therefore,it can be used as the main method for the compilation of the house price index,which lays a foundation for the subsequent research on the house price index compilation model of the unification of newly-built ordinary houses and second-hand ordinary houses.Thirdly,based on the longitudinal unbalanced panel model of newly-built ordinary houses,we establish a longitudinal unbalanced panel model for the house price index of ordinary residential houses in Chinese cities that unifies the newly-built ordinary houses and second-hand ordinary houses by introducing the monthly depreciation coefficient and the depreciation coefficient of the number of repeated transactions.By introducing dummy variables that characterize the type of real estate transaction,the corresponding method to estimate parameters is designed.The empirical results show that under the condition of controlling “group effect” such as housing age,housing price decreases by 3.6 percent for each year of increasing housing age,and housing price decreases by 2 percent for each repeat sale,which are consistent with the common sense of house price in Chinese cities.