Hedging performance of Chinese stock index futures: An empirical analysis using wavelet analysis and flexible bivariate GARCH approaches

• We assess hedging performance of Chinese stock index futures.
• We employ multiple hedging models to calculate optimal hedge ratios.
• Wavelet analysis yields best in-sample hedging effectiveness in constant models.
• Short horizons favour DCC BGARCH model while long horizons favour CCC BGARCH model.
• Short horizons prefer time-varying whereas long horizons favour constant hedge.

In this paper, we assess the hedging performance of the newly established CSI 300 stock index futures over some short hedging horizons. We use wavelet analysis as well as conventional models (naïve, ordinary least squares, and error-correction) to compute the constant hedge ratios. The constant conditional correlation (CCC) and dynamic conditional correlation (DCC) bivariate generalised autoregressive conditional heteroskedasticity (BGARCH) specifications are employed to calculate the time-varying hedge ratios. Overall, we find that the CSI 300 stock index futures can be an effective hedging tool. Among the constant hedge ratio models, the wavelet analysis yields the best in-sample hedging performance, though its out-of-sample hedging performance is similar to other models. Comparing the time-varying ratio models, the CCC BGARCH model is better in terms of in-sample hedging effectiveness while for out-of-sample hedging performance, the DCC model is better with short hedging horizons and CCC model is more favourable with long hedging horizons. Finally, the question whether time-varying ratios outperform constant ratios depends on the length of the hedging horizon. Short horizons favour BGARCH hedging models while long horizons favour constant hedging ratio models.