Research On Influencing Factors Of Credit Spread In China’s Bond Market

In recent years,China’s bond market has experienced frequent defaults,which has gradually broken rigid payments,and the issue of bond credit risk has become more serious.The credit spread represents the credit risk of the bond market,so the research on bond credit spread is of great significance.And in recent years,China’s bond market has also shown new characteristics.The government is paying more and more attention to the development of private enterprises and enterprises in less developed regions and the situation of bond financing.For example,under the policy guidance of targeted poverty alleviation,since 2017,local governments have issued special poverty alleviation bonds for poverty alleviation projects.At the same time,in order to further support private enterprise financing,private enterprises have successfully issued corporate bonds since 2018.Therefore,this paper selects 2017-2019 credit debt data for empirical research,with a total of 10890 samples.This paper combines the theoretical model of credit spreads and previous research on the factors that influence credit spreads,and constructs the factors that affect credit spreads from the macroeconomic level and the micro level of the enterprise.In the empirical study,this paper chose a random forest model,which has good prediction accuracy,is not easy to overfit,can effectively process high-dimensional data,and avoid multi-collinearity.This paper draws the following conclusions through empirical analysis:First of all,the evaluation of micro-level data indicators of credit spreads is more explanatory than the macro-level data indicators.In this paper,the importance of all variables in the model is measured.From the perspective of the importance of the overall variables,credit rating has the largest impact on credit spreads among all macro and micro variables.The other top influencing factors are the macroeconomic prosperity index,the property rights of the enterprise,the general budgetary revenue of the local finance of the province,and the asset-liability ratio.As a whole,large state-owned enterprise bonds with developed regions,high credit ratings,and low leverage have lower credit spreads.At the same time,this paper innovatively uses the random forest algorithm to reduce the dimensionality of the explanatory variable of the credit spread.In the case of reducing one-third of the variables,a more practical simplified model of random forest is obtained,and the model interpretation degree and accuracy are very highSecondly,this paper conducts a comparative study of non-state-owned enterprises and state-owned enterprise bonds.It is found that the importance and ranking of the factors affecting the credit spreads of the two types of bonds are different.In non-state-owned enterprise bonds,regional factors are no longer the main influencing factors of credit spreads.Credit spreads are mainly affected by the subject ’s credit rating,macroeconomic operating conditions and corporate financial conditions.At the same time,whether it is a listed company or industry attributes,two variables that are not important in state-owned enterprises,have greatly increased in the importance of non-state-owned enterprise bondsFinally,this paper makes a comparative study of less developed regions and developed regions,and finds that the main factors affecting the credit spread of these two types of bonds are different.Compared with bonds in developed regions,credit spreads in underdeveloped regions are more susceptible to macroeconomic impacts The importance of the three variables representing short-term debt-servicing ability,business operation ability,and local-enterprise relationship has greatly increased.The credit rating still has an important impact,but its importance is greatly reduced.At the same time,the nature of corporate property rights is no longer the main factor affecting bond credit spreads in less developed regions.