本研究使用已實現雙變量GARCH(Bi-RGARCH)模型來檢驗是否能透過考慮台灣現貨與期貨市場之最小變異避險策略中的盤後交易資訊，藉此提高避險績效表現。研究期間為2017年5月15日至2019年12月31日，其中，樣本內避險期間為2017年5月15日至2018年12月28日，剩餘期間用於評估樣本外避險績效。而為了穩健性，樣本內避險期間自2017年5月15日延長至2019年6月28日，樣本外避險期間則是最後半年。避險績效分別透過變異數、二次效用函數以及半變異數進行評估。
實證結果顯示，無論採用何種評估方式，在樣本內、外期間，Bi-RGARCH-AHT模型績效表現均優於Bi-RGARCH模型。另外，已實現雙冪次變異的結果與已實現變異結果一致。SPA檢定結果顯示，通過盤後交易資訊揭示對避險模型的統計意義。上述結果表明，若將盤後交易資訊納入避險策略中，則避險行為將受益於盤後交易。
This study uses the bivariate realized GARCH (Bi-RGARCH) model to examine whether the hedging performance can be improved by considering information of after-hour trading in minimum variance hedging strategy for Taiwan spot and futures markets. The research period starts from May 15, 2017 to Dec 31, 2019, in which the in-sample hedging period spans from May 15, 2017 to Dec 28, 2018, and the remaining period is left for evaluation of out-of-sample hedging. For robustness, in-sample hedging period is extended from May 15, 2017 to Jun 28, 2019, and out-of-sample period is the final half year. The hedging performance is respectively evaluated by variance, quadratic utility and semi-variance functions.
The empirical results that, irrespective of evaluation methods, the Bi-RGARCH-AHT model outperforms the Bi-RGARCH model for both in- and out-of-sample periods. In addition, the results of realized bi-power variance are consistent with those of realized variance. The SPA test reveals the statistical significance of the hedging model with after-hour trading information. These results imply hedging activity would benefit from after-hour trading system if considering information of after-hour trading into hedging strategy.
