Credit risk management has long been considered an important issue since the world financial markets started to grow and integrate. Investigation of credit quality is based on inspection of the credit or finance history from financial statements and historical data. Furthermore, because of the improvement of information transmission and closer trading relationship between countries, the contagion effect between financial markets has become worldwide and significant over time. That is, shock originating from one market actually transmits to other markets, which makes the credit management more complicated. Therefore, the credit risk correlations, which are the dependence among risks, become important issues in the research of risk management. The purpose of this study is to explore the credit risk relations among debt markets and to explain how the relations vary over time.
The sovereign debt markets provide high-quality data sources for the measurement of credit risk. Investigating price and yields involves a more quantitative analysis of credit quality characteristics. According to Bank of International Settlement, the total size of government bond markets had grown from 24,154.2 billion in 2006 to 26,200.7 billion dollars in September, 2007. Particularly, the Asia local currency government bond markets had grown rapidly in past years with 21% growth rate in
2
2007. Although debt credit contagion has a great impact on construction of credit-sensitive portfolios for the banking and investment management (see Zhou, 2001), it is a source of substantial instability. The observation of price and credit spread change of bonds directly recovers the credit variation. Theoretically, credit spread is always viewed as a measurement of credit risk for debt (Manzoni, 2002).
There have been a lot of researches on the determinants of bond price and yield, especially on bond return and credit spread. Taking credit spread for example, macroeconomic and financial variables, such as GDP growth, inflation and stock market return, are considered crucial factors of credit spread change (see Collin-Dufresne, Goldstein and Martin, 2001).
The objectives of this study are to investigate the credit risk correlations among sovereign debts and the factors associated with the correlation of debts issued by sovereigns. To our knowledge, the phenomenon of contagion among financial markets is obvious that the crisis from one country may be triggered or extends to others. Suggested by existing literatures, credit risk is closely connected with the financial market uncertainty which could be reflected by the implied volatility of stock. Also, the importance of stock volatility on credit spread at the aggregate level has also been discussed extensively in the literature (see Cambell and Ammer, 1993).
To proxy uncertainty, economic variables from the US market are widely taken as the
3
world factors because the US financial markets have great influence on emerging countries. For our purpose of analysis credit risk correlation, the financial markets factors from the US are adopted. Thus, the implied volatility from equity index options, specifically the Chicago Board Options Exchange Volatility Index (henceforth VIX), is used to measure the uncertainty of financial market and its influence on the credit spreads correlations.
Since the credit spread, like default rate, shows to be not constant over time (see Dungey, Martin and Pagan, 2000), it requires an approach which is capable of capturing time-varying behavior of credit spread as well as the exogenous variables.
For the past few years, the GARCH family models have been used widely in time-series studies. Engle (2002) proposed the dynamic conditional correlation model (the DCC model), which is modification of multivariate GARCH model, to study dynamic correlation between assets. In our study, we also adopt the modified DCC model, called DCCX model, proposed by Liao (2008). This model allows adding an exogenous variable while we estimate conditional correlation. As we described in previous paragraph, the exogenous variable is VIX as measurement of market uncertainty to investigate the variation of correlations (see Connolly, Stivers and Sun, 2005).
4
This paper provides a new perspective and methodology to analyze credit spread.
Credit spread is always considered a proxy for credit risk, which had been use to estimation the credit correlation between markets. However, this is the first research to investigate the dynamic relation between credit spreads and also the determinants.
The empirical results of our study present that there are dynamic correlations among Asian debts markets and among those with the US. The standard and modified DCC models provide estimation of dynamic correlations and help explain for the major economic issues. The results from the DCCX model show that the credit correlation tends to be positively related with VIX. That is, the debts markets connect closer when crisis occurs and economic events happen. Works on credit correlation are beneficial from preventing asset value loss on one side and from well diversifying risk of portfolio on the other side. The results will assist those responsible for managing the risks of sovereign debt as well as diversifying portfolio assets allocation. Besides, investors would benefit from consideration of market uncertainty while they analyze credit risk and make decisions.
The remainder of this paper proceeds as follows. The relevant literatures about credit risk, cross-market hedge and methods are reviewed in section II. In section III, we provide the methodology applied in our study. It includes DCC and DCCX approaches associating with relevant literatures and models. All the empirical results
5
and discussion of estimations are listed in section IV. The conclusion is given in the last part of Section V.
6