Wai-Sum Chan
Abstract
A multiple time series approach is used to forecast the short-term U.S. corporate bond default level. These time series have two auxiliary economic variables: U.S. price inflation and U.S. GNP growth rate. Actual U.S. data from the turn of the century to the present are used to estimate the parameters of multivariate time series model. Diagnostic checks are performed to examine adequacy of the model. The model's forecast for the aggregate U.S. bond default level in 2000-2001 are 0.42% and 0.56%, respectively, while the forecast for the speculative-grade default rate in 2000 is 3.6%, which is more pessimistic than some other forecasts available in the market.
Key words and phrases: autoregressive, moving average, stationary, forecasting, high-yield bonds, vector time series.
Wai-Sum Chan
Dept of Statistics & Actuarial
Science
The University of Hong Kong, Pokfulam Road
HONG KONG
© Copyright Absalom Press, Inc.