Real Estate Enterprise Bond Default Risk Warning Under The "Three Red Lines" Policy

Due to the new regulations issued by the Central Bank and the Ministry of Housing and Construction for real estate,the "three red lines" and the new crown epidemic,the cost of corporate financing has risen significantly,which has led to frequent defaults in the bond market and greatly affected investors’ confidence.On the one hand,as an important asset allocation in the portfolios of domestic commercial banks and other institutional investors,the default of bonds is not only manifested in the failure to pay principal and interest on time,but will also be transmitted to the stock market and derivatives market,which will greatly affect the stability of the capital market;On the other hand,the existence of adverse selection makes investors usually need higher yield to compensate for the risk they take when investing in bonds,which further aggravates the problem of difficult and expensive financing for SMEs.Therefore,in response to the national policy of "no speculation in housing",the central bank and the Ministry of Housing and Construction have taken the initiative to meet with large enterprises to ask them to reduce leverage and improve the soundness of real estate enterprises.As one of the pillar industries of the country,the real estate industry is closely related to the people’s well-being,so the bond mines of real estate enterprises will greatly affect the stability of society.Therefore,it is especially important to construct bond default risk early warning models in advance and identify the issuing entities that may default on bonds proactively in advance,so as to optimize the investment decisions of institutional investors,minimize the losses of individual investors and strengthen the risk prevention awareness of bond issuing entities.In this paper,the random forest algorithm is used to construct a model to warn the bond default risk of real estate companies,transforming the qualitative bond default risk into a more intuitive dichotomous problem,and obtaining a better-performing model through model setting and parameter optimization.The study shows that the random forest algorithm is suitable for prediction and evaluation of default risk because of its high accuracy and recognition ability,and it is less prone to overfitting.This paper designs a random forest algorithm-based bond default risk warning model with an AUC index as high as 0.983658 and a total sample accuracy of 94.89% in the test set.Through sample analysis,this paper considers the profitability index represented by the three red lines indicator,the net asset return ROE and the solvency represented by the interest earning multiple as the Important indicators in early warning models..Therefore,profitability and solvency based on specific indicators,rather than on the three red line indicators can be used as key points for risk monitoring.The construction and application of bond default risk early warning models for real estate enterprises use various publicly available financial data and external macro data.In order to make the model more accurate,the regulator should accelerate the reform of the information disclosure mechanism in the financial capital market;real estate enterprises should regularly use algorithmic models to check their bond default risk status,and individual Investors should use a more scientific approach to identifying the risk of bond defaults.Moreover,the "three red lines" policy is a bit harsh for most enterprises,and policy makers should make appropriate optimization on the basis of the existing policy,so as to better promote the healthy and sustainable development of real estate enterprises.