Forecasting credit spread volatility: Evidence from the Japanese Eurobond Market

Different approaches to forecasting the volatility associated with the credit spreads on Yen Eurobonds are investigated. The actual volatility, historical volatility and estimated conditional volatility on spreads derived from a regression-based model with a GARCH and ARMA specification are compared within an adaptation of Black's (J. Finance, 31, 1976, 361-367) option-pricing model. Surprisingly, the regression forecast over a medium forecasting horizon suggests that historic volatility provides the better forecast. The implications of these results for volatility forecasting and credit spread modelling are also discussed.