Research On Credit Risk Assessment Of Real Estate Listed Companies In China

The real estate industry is a pillar industry that is related to the national economy and the people's livelihood.The report of the 19th National Congress pointed out that"the house is used for living,not for speculation",the government's series of regulation and control policies have made real estate companies face greater competitive pressures and funding needs.Therefore,the credit status of real estate listed companies is a very important issue for investors and financial intermediaries,and also a barometer of the development trend of the real estate industry.Based on this,the research question is:how to scientifically and accurately assess the credit risk level of China's real estate listed companies.This paper uses the combination of qualitative analysis and quantitative analysis to analyze the main sources of credit risk of real estate companies.On this basis,it designs an evaluation index system with the characteristics of real estate industry.Firstly,according to the operation and financial characteristics of real estate enterprises,nine financial indicators were selected and their calculation methods were revised.Secondly,nine non-financial indicators that can reflect soft power such as company size and governance structure were also included in the evaluation index system.In the empirical analysis,the Shanghai and Shenzhen A-share real estate companies were selected as samples,mainly based on whether they were specifically processed by the stock exchange,and the samples were classified into high credit risk group and low credit risk group.The relevant data from 2016 to 2018 is preprocessed,and the Logistic-based credit evaluation model is established by using the forward stepping method.The goodness of fit is tested.The five indicators that enter the Logistic model are used as input layers to establish neural network model.Because the two single models have poor discriminative ability for high credit risk group companies,this paper uses the combined evaluation method to input the predicted probability generated by the Logistic model as the sixth explanatory variable into the neural network model to establish a combined evaluation model.The accuracy rate is 99.1%,and the accuracy rate for the high credit risk group is 80%.In order to further improve the discriminative accuracy of the credit evaluation model for high credit risk group companies,a stochastic forest model based on SMOTE algorithm was established,which improved the discriminative accuracy of the two groups of companies to over 98%.Finally,using comparative analysis method,the characteristics,advantages and precision of the four credit evaluation models are compared and analyzed.The conclusions are as follows:(1)Compared with all A-share listed companies,the overall credit risk level of China's real estate listed companies in the past three years is relatively high.Among them,the credit level of Shandong Tianye Henderson Co.,Ltd.deteriorated sharply in 2016-2018.(2)Cash net asset turnover rate,land reserve,asset and debt guarantee multiples,and the top ten shareholders' shareholding ratio are important factors affecting the credit status of China's real estate listed companies.(3)The combination evaluation model based on Logistic and neural network and the random forest model are effective methods for the credit evaluation of real estate listed companies in China.(4)Applying the SMOTE algorithm to the credit risk assessment of real estate listed companies in China can effectively solve the problem of uneven distribution of sample data.Based on the characteristics of the real estate industry,this paper improves the comprehensiveness of credit evaluation indicators and the accuracy and robustness of the credit evaluation model.It scientifically and effectively evaluates the credit risk level of China's real estate listed companies in the past three years and its main influencing factors.It can also provides ideas for listed companies in other industries.