The CS model---a flow of funds based framework for analyzing sovereign credit risks in an international credit chain economy: An adaptation, extension and empirical analysis

The Cossin and Schellhorn (CS) corporate credit risk structural model is applied, adapted and extended for analyzing credit risks pertaining to U.S. dollar denominated sovereign debt in an international credit chain economy, where any country can lend to any other country, so that each country is subject to counterparty risk either from direct or remote borrower countries. By taking into account the role of each government's treasury account and using queuing theory, a semi-closed form formula is obtained for pricing foreign and domestic debt where a country's strategic default decision drives credit spreads. In this framework, the liabilities consist of senior claims, such as foreign debt, which play the role of "debt" in the corporate model, and of junior claims, such as domestic debt, which play the role of "equity" in the corporate model. An empirical analysis is performed on three focus countries; Chile, Russia and Spain. A sensitivity analysis is also performed in order to analyze the interactions and interrelationships of the countries that construct each credit chain taking advantage of the fact that the CS model is analytically tractable. The empirical results show the superiority of the CS sovereign model that outperforms the single-agent model in all instances, and to a higher degree when evaluating riskier bonds. However, in general the CS sovereign model underestimates observed credit spreads, and to a higher degree when evaluating riskier bonds. A cross-section time series regression analysis is performed where observed sovereign bond prices were regressed against the CS sovereign and single-agent model derived credit spreads, as well as the observed credit spreads in three separate equations. The theoretical, single-agent and observed credit spreads were found to be significant at a 5% level, however only the marginal effects of the observed credit spreads were congruent with financial literature. The CS sovereign model derived credit spreads were also analyzed as a function of several economic and index variables to explore the determinants of credit risk. Results show that the equation that incorporates political risk ratings as the independent variables performs better than the equations that incorporate composite, economic and financial risk ratings. In this specification, the marginal effects of all the risk rating coefficients were congruent with the financial literature; however the S&P coefficient and the market index coefficient were not. Further research will be needed to investigate the influence of a larger counterparty chain or network on credit spreads, including stochastic interest rates in the CS model, as well as improving fit.