Rapach et al. (2013) have recently shown that U.S. equity market returns carry valuable information to improve return forecasts in a large cross-section of international equity markets. In this study, we extend the work of Rapach et al. (2013) and examine if U.S. based equity market information can be used to improve realized volatility forecasts in international equity markets. For that purpose, we obtain volatility data for the U.S. and 17 international equity markets from the Oxford Man Institute's realized library and augment for each foreign equity market the benchmark HAR model with lagged U.S. equity market volatility information. In-sample as well as out-of-sample evaluation results suggest a strong role for U.S. based volatility information. More specifically, apart from standard in-sample tests, which find U.S. volatility information to be highly significant, we show that this information can be used to substantially improve out-of-sample forecasts of realized volatility. Using large out-of-sample evaluation periods containing at least 2500 observations, we find that forecast improvements, as measured by the out-of-sample R 2 (relative to a model that does not include U.S. based volatility information), can be as high as 12.83, 10.43 and 9.41 percent for the All Ordinaries, the Euro STOXX 50 and the CAC 40 at the one-step-ahead horizon. Moreover, forecast improvements are highly significant at the one-step-ahead horizon for all 17 equity markets that we consider, yielding Clark-West adjusted t statistics of over 7. We show further that the improvements from including U.S. based volatility information are consistently experienced over the entire out-of-sample period that we consider, and hold for forecast horizons of up to 22 days ahead.

Rapach et al. (2013) have recently shown that U.S. equity market returns carry valuable information to improve return forecasts in a large cross-section of international equity markets. In this study, we extend the work of Rapach et al. (2013) and examine if U.S. based equity market information can be used to improve realized volatility forecasts in international equity markets. For that purpose, we obtain volatility data for the U.S. and 17 international equity markets from the Oxford Man Institute's realized library and augment for each foreign equity market the benchmark HAR model with lagged U.S. equity market volatility information. In-sample as well as out-of-sample evaluation results suggest a strong role for U.S. based volatility information. More specifically, apart from standard in-sample tests, which find U.S. volatility information to be highly significant, we show that this information can be used to substantially improve out-of-sample forecasts of realized volatility. Using large out-of-sample evaluation periods containing at least 2500 observations, we find that forecast improvements, as measured by the out-of-sample R 2 (relative to a model that does not include U.S. based volatility information), can be as high as 12.83, 10.43 and 9.41 percent for the All Ordinaries, the Euro STOXX 50 and the CAC 40 at the one-step-ahead horizon. Moreover, forecast improvements are highly significant at the one-step-ahead horizon for all 17 equity markets that we consider, yielding Clark-West adjusted t statistics of over 7. We show further that the improvements from including U.S. based volatility information are consistently experienced over the entire out-of-sample period that we consider, and hold for forecast horizons of up to 22 days ahead.

No abstract

We evaluate the relevance of covariances in the transmission mechanism of variance spillovers across the US stock, US bond and gold markets from July 2003 to December 2012. For that purpose, we perform a comparative spillover analysis between a model that considers covariances and a model that considers only variances. Our results emphasise the importance of covariances. Including covariances leads to an overall increase of the spillover level and detects the beginnings of the financial crisis and of the US debt ceiling crisis earlier than the spillover measure that considers only variances. Even for the low-dimensional system that we consider, one misses important variance spillover channels when covariances are excluded.

We evaluate the relevance of covariances in the transmission mechanism of variance spillovers across the US stock, US bond and gold markets from July 2003 to December 2012. For that purpose, we perform a comparative spillover analysis between a model that considers covariances and a model that considers only variances. Our results emphasise the importance of covariances. Including covariances leads to an overall increase of the spillover level and detects the beginnings of the financial crisis and of the US debt ceiling crisis earlier than the spillover measure that considers only variances. Even for the low-dimensional system that we consider, one misses important variance spillover channels when covariances are excluded.