Stock index hedge using trend and volatility regime switch model considering hedging cost

This paper studies the risk hedging between stock index and underlying futures. The hedging ratios are optimized using the mean-variance utility function as considering the hedging cost. The trend of returns and variance are estimated by the model of regime switch on both vector autoregression (VAR) and GARCH(1,1) compared to three restricted models: VAR switch only, GARCH(1,1) switch only, and no switch. The hedge portfolio is constructed by Morgan Stanley Taiwan Index (MSTI) and Singapore Traded MSTI futures. The hedge horizon is set as a week to reduce the hedging cost and the weekly in-sample data cover from 08/09/2001 to 05/31/2007. The rolling window technique is used to evaluate the hedge performances of out-of-sample period spanning subprime, Greek debt, and post-risk durations. The subprime period indeed is evidenced very vital to achieve the hedge performance. All models perform surprisingly far above average during subprime period. The hedge ratios indeed are the tradeoff between maximum expected return and minimum variance. It is demonstrated challenging for all models to increase returns and reduce risk together. The hedge context is further classified into four hedge states: uu, ud, du, and dd (u and d denote respectively usual and down) using the state probabilities of series. The regime switch models are found to have much greater wealth increase when in dd state. It is decisive to hedge risk in dd state when volatility is extensively higher as observed recurrently in subprime period. Remarkably, the trend switch is found having larger wealth increase while the volatility switch is not found prominent between models. While the no switch model has larger utility increase in uu state as most observed in Greek debt or post risk period, its performance is far below average like other models.