探討金融海嘯前後波動率指數、黃金價格與不同市場股價指數關連性研究

Study on Relationship among Volatility Index,Gold Price,and Various Market Indices-Before and after Financial Tsunami

孫健銘(Chien-Ming Sun)

指導教授 ： 胡為善

中原大學/商學院/企業管理研究所/碩士(2015年)

https://doi.org/10.6840/cycu201500519

摘要

過去七年，全球經濟先後遭逢了2008年起的金融海嘯、2010年起之歐債危機以及2011年起的美國財政懸崖及其他國際重大金融危機，使得VIX指數及黃金價格屢創新高。然而自2012年起，當全球經濟狀態仍處於低迷時，被視為傳統避險工具的黃金，卻逆勢下跌，使投資人血本無歸，也讓分析師跌破眼鏡，紛紛詢問是否黃金所處之避險地位已受到動搖了嗎？本研究以VIX指數與黃金價格為主軸，探討在歐美與新興市場中，VIX指數、黃金價格、大盤指數三者間的因果關係，以及探討在經歷金融危機之後，VIX指數與黃金價格間之關聯性是否有所改變。本研究亦期望能幫助投資人當面對金融危機發生時，得採用適當之避險與預測方法，以達其避險之目的。研究期間自2005年1月1日至2014年12月31日止，共十年計120筆月資料，並以2008年9月作為金融海嘯之分界，將資料分為海嘯之前期與後期。本研究採用共整合、向量自我回歸(VAR)、向量誤差修正(VECM)、Granger因果關係檢定及變異數分解等方法進行實證分析。實證結果歸納如下：
1.本研究發現在全樣本期間，VIX、黃金價格與S&P500指數間並未存在共整合關係。但是VIX、黃金價格與歐洲MSCI指數以及VIX、黃金價格、與新興市場MSCI指數，兩組內的各變數皆具有共整合關係。而在金融海嘯的前期及後期期間，VIX、黃金價格及S&P500指數之間均存在共整合關係。
2.本研究透過因果關係之檢定結果發現，在全樣本期間，黃金價格單向領先且影響S&P500指數、VIX指數及歐洲MSCI指數，但在新興市場MSCI指數的各變數則不具因果關係；而在金融海嘯前期，黃金價格單向領先且影響VIX指數；但在金融海嘯後期，黃金價格單向領先且影響S&P500指數與VIX指數。本研究亦發現在前期與後期內，歐洲MSCI指數與新興市場MSCI指數兩組間的各變數皆不具因果關係。
3.依據誤差修正之實證結果顯示，在全樣本期間，VIX指數與歐洲MSCI指數及新興市場MSCI指數間皆呈現顯著負相關。此外，在金融海嘯前期，S&P500指數與VIX指數呈現顯著負相關，但黃金價格與VIX指數則呈顯著正相關；而在金融海嘯後期，VIX指數與S&P500指數及黃金價格皆呈現顯著正相關；此外，在誤差修正調整速度方面，當變數脫離長期均衡關係時，在全樣本期間，VIX指數會向下修正，黃金價格與新興市場MSCI指數則會向上修正，以回復長期均衡。而在海嘯前期時，黃金價格與VIX指數會向下修正；但在海嘯後期，只有VIX為顯著，且向下修正，以回復均衡。
4.經由衝擊反應檢定的結果，本研究發現，在全樣本期間之第一期，黃金價格與S&P500指數均受本身之衝擊效果最為明顯，但VIX指數卻受S&P500指數之衝擊效果最為明顯。
5.透過預測誤差變異數分解結果，本研究發現，在全樣本期間的第一期，依據各變數之分解值，最能解釋VIX指數者為S&P500，其解釋力為62.75%，而VIX本身的解釋力卻只有35.72%，係因VIX係由S&P500指數的選擇權價格之波動性計算出來所致。此外，最能解釋S&P500指數及黃金價格者均為其本身，其解釋力分別高達99.5%及100%。

關鍵字

全球金融海嘯 ； 因果關係 ； 共整合關係 ； 波動率指數 ； 黃金價格 ； 預測誤差變異數分解 ； 衝擊反應函數 ； S&P500指數 ； VAR模型 ； VECM模型

並列摘要

Over the past seven years, the global economy experienced the financial tsunami since 2008, the European debt
crisis since 2010 and the US fiscal cliff and other major international financial crisis since 2011.
Subsequentially, VIX index and gold prices hit record highs during these periods. However, when the global economy is still in the doldrums state, gold price dropped dramatically since 2012, causing investors huge losses, and made financial analysts doubt whether gold is still a safegrand for investors. This study examines the relationship among the VIX index, gold price and the stock indices in the emerging markets and European markets before and after financial tsunami in September 2008 employing the Co-integration test, Vector Auto-regressive model (VAR), Vector error correction model (VECM), Granger causality test, and Forecasting-error variance decomposition (FEVD). This investigation also attempts to assist the investors to use the appropriate forecasting and hedging methods when they face against financial crisis. The sample period runs from January 1 2005 to December 31 2014, with a total of 120 monthly data. This study then divideds the full sample period into two sub-periods: one runs from January 1 2005 to September 2008, and another runs from September 2008 to December 2014. Empirical results are summarized below:
1.During the full sample period, there is no co-integration among the VIX index, gold price and the S&P500 index. However, this study finds that the co-integration exists among VIX index, the gold price, and the MSCI European Index, as well as that among the gold price, the VIX, and the MSCI emerging markets. Additionally, VIX index, the gold price and the S&P500 index have co-integration for both sub-periods.
2.Granger causality test results show that, during the full sample period, the gold price unilaterally affects the S&P500 index, VIX index and MSCI European markets index. However, there is no causal relationships exists between MSCI emerging Index and other parameters. Meanwhile, the gold price unilaterally impacts the VIX index during the frist sub-period; while the gold price unilaterally influences the S&P 500 and the VIX indices for the second sub-period. Empirical results also indicate that neither European nor emerging markets have casual relationships during the first and second sub-periods.
3.The error correction results show that the VIX and the European and emerging market MSCI indices are negatively correlated for the full sample period. Meanwhile, the S&P 500 index is negatively related with VIX during the first sub-period; while the gold price is positively related with the VIX index for the second sub-period. Additionally, both the S&P 500 index and the VIX index, and gold price and the VIX index are also positively related during the second sub-period.
4.The impulse response function test results indicate that the gold price is 100% shocked by the gold price itself in the first term, while the S&P500 index is 99.5% shocked by the S&P500 index itself in the first term. However, the VIX index is influenced most significantly by the shock of S&P500 index in the first term.
5.The forecast-error variance decomposition (FEVD) results show that, for the full sample period, the most significantly explanatory power of the VIX index arises from the S&P500 index (i.e., 62.75%), and the 2nd most significantly explanatory power arises from the VIX index itself (i.e., 35.72%). The possible explanation is that the VIX is a key measure of market expectation of near-term volatility conveyed by the S&P 500 index’s option price. However, the only significantly explanatory power on the shock to gold price and to the S&P 500 index arising from the gold price itself and the S&P 500 index itself, respectively.

並列關鍵字

Co-integration ； Forecast-error variance decomposition ； Global financial crisis ； Gold price ； Granger causality test ； Impulse response function ； S&P500 index ； Volatility index ； VAR model ； VECM model

參考文獻

張欣怡(2012)，「波動率指數、美元指數與石油價格關聯性之研究」，中原大學企業管理研究所碩士論文。

王秀香(2012)，「金融危機下影響黃金現貨價格變動因素之探討」，淡江大學財務金融學系碩士論文。

駱武昌、吳明珊、吳斯偉（2011），「台灣、歐洲與美國股市間波動外溢效果」，會計與財金研究，4(2)：29-50。

陳志杰(2012)，「台灣大型權值股股價報酬與VIX指數、黃金報酬之關聯性分析」，國際財務金融研究所碩士論文。

陳惶博(2009)，「波動率指數與總體經濟指標關連預測之研究-以灰色關聯分析及類神經網路模型系統作探討」，中原大學企業管理研究所碩士論文。