This paper examines the daily volatility of changes in the 10-year Treasury note utilizing the iterated cumulative sums of squares algorithm [C. Inclan, G. Tiao, Use of cumulative sums of squares for retrospective detection of changes of variance, J. Am. Stat. Assoc. 89 (1994) 913-923]. The ICSS algorithm can detect regime shifts in the volatility of the interest rate changes. A general model allows for endogenously determined changes in variance while the more restrictive model forces the variance to follow the same process throughout the sample period. A comparison of the out-of-sample volatility forecasting performance of two competing models is made using asymmetric error measures. The asymmetric error statistics penalize models for under- or over-predicting volatility. The results shed light on the importance of ignoring volatility regime shifts when performing out-of-sample forecasts. The findings are important to financial market participants who require accurate forecasts of future volatility in order to implement and evaluate asset performance.
|Number of pages||8|
|Journal||Physica A: Statistical Mechanics and its Applications|
|State||Published - Sep 15 2006|
- Asymmetric forecast evaluation
- Interest rate
- Regime shifts