多變量偏態分配之投資組合：台灣上市之中國槓桿ETF研究

全球ETF近10年的交易規模及投資報酬率大幅成長，中國更適時開放滬港通、深港通股票市場，之後，MSCI也積極加入中國成份股，2019年更宣布每季增加成份股比重，可見國際投資機構都看好中國股票市場。本論文將利用多變量GARCH、狹峰、厚尾及偏態T分配的模型，實證2015年以來在台灣上市的跨境滬深300正向2倍及反向1倍槓桿型ETF投資組合及上證50ETF、A50期貨投資組合的避險效益及報酬率。本文以Bauwens and Laurent (2005)所提出之多變量偏態T分配結合多變量GARCH模型，應用於股票收益的投資組合模型，尤其是預測投資組合的風險價值，期以最佳權重分配獲得投資報酬。實證結果顯示多變量GARCH模型融合多變量偏態T分配，確實可捕捉到波動度較大的金融商品，具有顯著的解釋能力，其中現貨ETF與期貨的避險績效最佳，而滬深300槓桿/反向ETF的投資組合報酬相對較高且為正，最後本文融合模擬實務投資方式，每日調整最適投資比重，驗證可獲取的報酬率則更高。

關鍵字

槓桿型ETF ；反向型ETF ；多變量；偏態T分配

並列摘要

In recent years, the scale and transaction volume of global ETFs have grown substantially. After China opened Shanghai-Hong Kong Stock Connect and Shenzhen-Hong Kong Stock Connect, MSCI also joined China's constituent stocks. In 2019, MSCI also announced the proportion of cost-increasing stocks per quarter. It is clear to see that the international market is optimistic about the Chinese market. This paper will use the multivariate GARCH skew-T distribution model to observe the portfolio of Yuanta Daily CSI 300 Bull 2X ETF , Yuanta Daily CSI 300 Bear -1X ETF and A50 china index ETF , A50 index futures hedging benefits and return rate. Using Bauwens and Laurent (2005) with the multivariate GARCH model, we can analyse stock portfolios to assess their level of risk and predict their renumeration. The empirical results show that the multivariate GARCH model integrates multivariate skewed -T distribution, which can capture financial products with large volatility and has significant explanatory power. The spot ETF and futures have the best risk-avoiding performance while the CSI 300 leverage ETFs' portfolio return is relatively high. Finally, the study integrates the simulation of practical investment methods, adjusts the proportion of the most suitable capital daily and verifies that the renumeration could be higher.