Credit Index Forecasting: Stability of an Autoregressive Model

Abstract: This thesis investigates the robustness and stability of total return series for credit bond index investments. Dueto the challenges which arise for financial institutes and investors in achieving these objectives, we aim to createa forecasting model which matches the statistical properties of historical data, while remaining robust, stable andeasy to calibrate. To reach this goal, we implement autoregressive time-series models for credit spreads, a Vasicekmodel for the interest rate and use transformations to create total return series. We find that our autoregressivemodel performs well in terms of robustness and stability, while being statistically accurate for the Investment GradeIndex. The High Yield model has good statistical accuracy, but is lacking in stability and robustness.