Estimation and Forecasting of Sovereign Credit Rating Migration Based on Regime Switching Markov Chain

Abstract

Our research aims to develop the regime switching Markov chain (RSMC), a discrete time Markov chain whose underlying regime is depending on a hidden Markov model, which express the dynamics of sovereign credit rating migration. Estimated based on a version of the Expectation-Maximization algorithm, the regime in RSMC indicates either economic expansion or contraction. Then, we apply RSMC to the monthly time series of the sovereign credit rating of 41 nations from January 1994 to December 2018. At first, we confirm that the estimation of RSMC is superior to a homogeneous Markov chain. It implies that the credit rating dynamics are subject to the underlying economic condition. Secondly, we observe that the second tier and non-investment credit ratings in economic contractions are likely to be downgraded. We also detect the continental clustering of economic contractions for the Asian currency and European sovereign debt crises. Lastly, we discover that the forecasting performance of RSMC is superior to that of the benchmark, especially for the second tier and non-investment credit ratings. In conclusion, we claim that RSMC can improve the management of sovereign credit risk exposures.

Keywords

• Credit migration
• economic forecasting
• hidden Markov models
• Markov processes
• regime switching
• sovereign credit rating