Dynamic hedging in the presence of volatility transmission and time varying basis risk

Abstract

The analysis undertaken in this research is a first attempt to comprehensively model all four S&P500 markets simultaneously. Special classes of Simultaneous Volatility (SVL) structures compete both within and out of sample with separate dynamic estimators, including GARCH. Synchronously sampled half-hourly observations are sampled from transaction data for these four financial assets. Out of sample hedge ratios are projected by dynamically updating the estimating equations one observation at a time and substituting in current and lagged values. These full system volatility models substantially dominate all remaining (7) competing models and the outperformance is substantial. Given that international portfolio managers and speculators hedge domestic spot market-risk in the very liquid U.S. spot and derivative markets, then these results should be of considerable interest.