Reserch On Local Government Contingent Debt Scale Measurement And Risk Early-warning

Local government contingent liabilities are the "gray rhinoceros" of systemic risk in China.The lack of data on local government contingent debt and its concealment have prevented the current local government debt regulation model from effectively regulating it,providing conditions for the rapid expansion of local government contingent debt.It also leads to the inability to effectively carry out theoretical research,because the scale of local government contingent debt is the cornerstone of the study of contingent debt and its derivative issues.According to IMF’s l arger debt statistics caliber,the size of Chinese government contingent debt may reach 42 trillion,and the overall debt burden ratio has exceeded the EU’s alert line,and the local government contingent debts measured by different calibers all keep expanding at a rate of more than 20%,indicating that contingent debt risks are gathering.On the other hand,since there are few cases of local government contingent debt defaults in China,there are fewer applications of early warning for contingent debt risk in practice,while early warning for contingent debt risk can provide preventive plans based on the monitoring and warning results before the outbreak of debt crisis to achieve the role of preventing and defusing the occurrence and spread of local government contingent debt risk.Therefore,it is urgent to grasp the scale and substantial risk of contingent debt,so as to develop an appropriate risk warning mechanism to reduce the potential risk of contingent debt.This study considers that local government contingent debt is the use of financial resources by local governments to assume off-budget expenditure responsibilities if an uncertain event occurs,due to factors such as statutory responsibilities,explicit and implicit guarantee responsibilities,moral obligations or expected responsibilities of local governments.Based on this,a statistical analysis framework of contingent debt size is proposed that determines the source of contingent debt based on government expenditure responsibilities and obligations,determines the statistical caliber of contingent debt based on the principle of financial fund repayment,and measures the conversion rate of contingent debt based on the uncertainty trigger mechanism.Based on the analysis framework,it is determined that the local government contingent debts mainly come from local state-owned enterprise debts,financing platform debts and PPP projects,and at the same time,a contingent debt scale measurement model of state-owned enterprises based on KMV debt default risk,a contingent debt scale measurement model of financing platform based on ROA operational risk;and a contingent debt scale measurement model of PPP projects based on the government’s repayment ratio is constructed.It is estimated that the overall expansion trend of the national local government contingent debt scale from 2010-2019,the scale rose from 2.32 trillion yuan to 15.05 trillion yuan,an increase of 5.47 times,while the GDP only increased by 1.46 times in the same period.2019,the scale of state-owned enterprises contingent debt is 5.23 trillion,the scale of financing platform contingent debt is 7.56 trillion,the scale of PPP projects The expected default probability of contingent debt is between 0.00%and 0.4%between 2010 and 2015,and less than 0.5%means no default risk,but after 2016,the expected default probability of contingent debt of local governments nationwide plunged to 84.95%and reached 100%in 2019.The regional structure shows that the default risk of contingent debt in different regions is climbing,and does not show obvious regional characteristics.The scenario simulation analysis finds that the new crown epidemic shock will accelerate the exposure of local government contingent debt risk in China;paying off local government contingent debt by disposing of the stock assets of state-owned enterprises can effectively reduce the contingent debt risk.Finally,this study constructs 41 indicators in six dimensions from two perspectives of contingent debt solvency and contingent debt scale risk,and finally obtains 13 early warning indicators through indicator screening and factor analysis.The accuracy of the contingent debt risk early warning models trained using machine learning methods ranged from 71.19%to 85.00%,with the random forest method having the highest accuracy and the strongest early warning ability of the early warning indicators related to the risk of debt of Chinese owned enterprises,the risk of debt of PPP projects and the risk of debt of financing platforms.The main contributions of this study are as follows:(1)This paper expands the connotation of public debt theory,while proposing a statistical analysis framework for the scale of contingent debt,laying a theoretical foundation for the application of public debt theory in local government debt management in China.Most of the existing public debt theories use cash basis and accrual basis as accounting standards for financial reporting and do not include contingent liabilities in public debt statistics.This study unifies contingent debt into public debt theory based on the principle of financial resources repayment,and constructs a statistical analysis framework of contingent debt scale from government expenditure responsibility and obligation,financial resources repayment principle and uncertainty trigger mechanism.(2)A model of contingent debt size classification and measurement based on the uncertainty trigger mechanism of contingent debt is proposed,which provides a new method for contingent debt size measurement and new data for the related research on contingent debt.Compared with the existing studies,this paper not only takes into account the risk characteristics of different types of contingent debt sources,but also constructs a statistical model of contingent debt size based on different risk characteristics that are compatible with each other.Based on the risk of debt default of local state-owned enterprises,the risk of sustainable operation of financing platforms and the contingent debt triggering mechanism of the risk of operation of PPP projects and the risk of project violation,the corresponding contingent debt scale measurement model is constructed.(3)Research on local government contingent debt risk has been promoted.At present,there are few studies on local government contingent debt risk literature,and most studies focus on local government debt risk research.This study takes the repayment priority of direct and contingent debts as the entry point,strips the secured fiscal revenue of local government contingent debts from the overall secured fiscal revenue of local governments,and also constructs a contingent debt default risk measurement model based on the results of contingent debt size measurement.(4)An early warning model of local government contingent debt default risk was constructed.Due to the small number of contingent debt default cases and lack of data,there is almost no research on local government contingent debt default risk early warning.This paper not only systematically analyzes the contingent debt early warning input index system,but also designs the contingent debt risk early warning output index,and constructs a local government contingent debt risk early warning model using five machine learning methods.Compared with the traditional linear early warning model,the early warning model constructed using machine learning methods shows a higher accuracy.