製造業採購經理人指數、安碩道瓊能源基金指數與 S&P500 指數連動性之探討

Elucidating the Relationship among PMI Index, iShare US Energy Fund IYE and S&P500 stock Index

林莉芳(Li-Fang Lin)

指導教授 ： 胡為善

中原大學/商學院/企業管理研究所/碩士(2016年)

https://doi.org/10.6840/cycu201600634

摘要

近三年來，全球經濟已逐漸從2007年開始發生的金融海嘯後趨於穩定。美國股市雖為全球發展最成熟之金融市場，但投資人仍然深恐金融危機再一次降臨，因而購買各種ETF之基金，以期達到避險之目的。另一方面，由於能源短缺為近半世紀來眾所關注之重大議題，投資人擔心若再度發生能源短缺，又會對全球經濟造成重大影響。本研究特以安碩能源基金指數代表上游原料部分，並以製造業採購經理人指數代表下游部分，至於大環境面，則以S&P500指數來代表美國股市的重要指標。研究期間自2000年9月至2015年8月，整個期間共分為三段，第一段期間自2000年9月1日至2005年8月31日，第二段期間自2005年9月1日至2010年8月31日，第三段期間則自2010年9月1日至2015年8月31日止。本研究係採用共整合檢定、Granger因果檢定、向量自我回歸模型(VAR)、向量修正模型(VECM)、衝擊反應函數及預測誤差變異數分解模型等，以進行驗證探討S&P500指數，採購經理人指數及安碩能源基金三者間之關係。本研究實證結果彙總於下：
1. 本研究發現，在全期間樣本內，S&P500指數、製造業採購經理人指數與安碩能源基金指數三者具有共整合關係。但在第一段期間內，雖然S&P500指數，製造業採購經理人指數與安碩能源基金指數三者間並未呈現共整合關係，但S&P500指數與製造業採購經理人指數兩者之間仍存在共整合關係；而在第二段期間內，本研究證實S&P500指數、採購經理人指數與安碩能源基金指數之間具有共整合關係；而在第三段期間內，三個變數皆不具有共整合關係。
2. 本研究經由修正誤差模型得知，在全樣本期間內，採購經理人指數之誤差修正調整速度，呈現顯著負向效果，表示採購經理人指數可迅速調整至均衡狀態;在第一段期間內，S&P500指數與採購經理人指數之誤差修正項對採購經理人指數具有顯著正向效果，其表示採購經理人指數須經較長之誤差修正調整時間，才能調整至均衡狀態。而在第二段期間內，S&P500指數、採購經理人指數與安碩能源基金指數間之誤差修正項只對採購經理人指數具有顯著正向效果，表示採購經理人指數不易調整至均衡狀態。
3. 本研究透過Granger因果檢定結果得知，在全樣本期間內，S&P500指數與採購經理人指數彼此間呈現雙向回饋之因果關係。而安碩能源基金指數對採購經理人指數及S&P500指數均呈單向因果關係；但在第一段期間內， S&P500指數會單向影響採購經理人指數及安碩能源基金指數。但在其他指數間並無因果關係，其原因可能係受到2005年國際原油價格屢創新高之影響；本研究在第二段期間內發現，S&P500指數與採購經理人指數彼此間及S&P500指數與安碩能源基金指數互相間均呈現雙向回饋因果關係。而安碩能源基金指數對採購經理人指數則呈現單向因果關係；但在第三段期間內，S&P500指數對安碩能源基金指數呈現單向因果關係，而安碩能源基金指數對採購經理人指數也呈現單向因果關係。
4. 本研究經由衝擊反應函數分析發現，S&P500指數、採購經理人指數與安碩能源基金指數皆受其自身衝擊影響最為明顯，且均至第7至9期才逐漸收斂，回到均衡狀態。
5. 本研究從預測誤差變異數分解之結果發現，在全樣本期間，第一段期間及第二段期間中，S&P500指數、採購經理人指數與安碩能源基金指數最具解釋力的來源均來自三個變數本身，但S&P500指數對採購經理人指數及安碩能源基金指數的解釋力。在第一段期間內，最高分別逾20%及逾30%。而在第二段期間內，S&P500指數對自身的解釋力最低時則降至55%。反之，採購經理人指數及安碩能源基金指數對S&P500指數的解釋力，則均可逾20%。另外，採購經理人指數對自己的解釋力在同期間內亦降至為62.78%。反之，安碩能源基金指數及S&P500指數對其解釋力則分別達23.42%及13.79%。另外在第三段期間中，S&P500指數及採購經理人指數對自身的解釋力，仍然分別維持在98%及85%以上。但安碩能源基金指數對自身的解釋力，自第二期起，竟然降至43%。反之，S&P500指數則對安碩能源基金指數卻更具解釋力，自第二期起，竟高達55%以上。

關鍵字

Granger 因果關係、S&P500指數、安碩能源基金、向量自我迴歸模型、向量自我修正模型、採購經理人指數

並列摘要

For the past three years, the global economy has been gradually stabilized from the Financial Tsunami since 2007. Although the US stock markets are the most matured ones in the global financial markets, various investors still worry about the history repeats itself, the financial crisis can happen again, the investors thus are seeking for trading various instruments to hedge their risks. On the other hand, the energy shortage has still been one of the popular subjects over the past half a century. If the energy shortage occurs again, it will also have a significant impact on the global economy. This study elucidates the relationship among the S&P500 index (S&P500), the manufacturing purchasing managers index (PMI) and the iShares energy index fund (iShare). The study period runs from Setember 2000 to August 2015, and that is classified into three sub-periods. The first sub-period runs from September 1, 2000 to August 31, 2005; the second sub-period runs from September 1, 2005 to August 31, 2010, and the third sub-period runs from September 1, 2010 to August 31, 2015. This investigation employs the Co-integration test, the Granger causality test, the Vector Autoregression (VAR), the Vector Error Correction Model (VECM), the impulse response function (IRF) and the forecast error variance decompostions (FEVD) method. The empirical results are summarized below:
1. For the full sample period, this study finds that the co-integrated relationship exists among the S&P500, the PMI and iShares. During the first sub-period, although S&P500, the PMI and iShares do not have co-intrgrated relationship, yet the S&P500 and the PMI have co-integrated relationship. During the second sub-period, the S&P500, the PMI and iShares have co-integrated relationship, indicating that the long-run equilibrium exists among these three parameters. However, during the third sub-period, no co-intergrated relationship exists among these three variables.
2. The error correction model results indicate that, during the full sample period, the error correction to adjust the speed of the PMI is significantly negative, suggesting that the PMI is easily adjusted to the equilibrium through the error correction term. During the first sub-period, the PMI’s adjustment speeds are significantly positive, suggesting that the PMI is not easy to adjust to equilibrium.
3. The Granger causality test results indicate that the S&P500 and the PMI have mutually causal effects during the full sample period. However, the iShare unilaterally affects the S&P500 and the PMI. During the first sub-period, the S&P500 unilaterally affects the PMI and iShares; while no causal relationship exists for other variables. The possible explanation is that the international crude oil prices hit the record high during this period. During the second sub-period, both the S&P500 and the PMI and the S&P 500 and iShares have mutual causalities. However, the iShares unilaterally affects the PMI. During the third sub-period, the S&P500 unilaterally affects the iShares, and the iShares unilaterally affects the PMI.
4. This study finds that, through the impulse response function, each of the S&P500, the PMI and iShares has the most significant shock arising from itself, and it converges to the equilibrium after the 7th to 9th terms.
5. The forecasting error variance decomposition results indicate that, during the full sample period, the first sub-period and the second sub-period, the S&P500, the PMI and iShares have the most explanatory power arising from themselves. However, during the third sub-period, although the most explanatory power for the S&P500 (98%) and the PMI (85%) are still arising from themselves since the first term thereafter. Yet, the explanatory power of the iShares less than 43% arising from itself and that of the S&P500 to the iShares rises to 55% since the second term thereafter. During the first sub-priod, although the most explanatory power for these three parameters are still arising from themselves, yet the most explanatory power of the S&P 500 to the PMI rise to 22.37% at the tenth term and that of the S&P 500 to the iShares rises to more than 30% since the second term thereafter. During the second sub-period, although the most explanatory power for these three parameters are still arising from themselves, yet the most explanatory power of the S&P500 arising from itself fell to 55%, and that of the PMI and iShares to the S&P 500 rises to 20% and above from the third term thereafter. This study also finds that the explanatory power of the S&P 500 and the PMI to the iShares rises to 10% and above since the third term thereafter. Additionally, the explanatory power of the PMI itself fell to 62%, and that of the iShares and the S&P500 to the PMI rises to 23% and 13% since the seventh term thereafter.

並列關鍵字

Forecasting error variane decomposition, Granger causality test, ImpulseResponse Function, iShare US Energy Fund Index, Manufaturing Purchasing Management Index, S&P500 index, VAR, VECM

參考文獻

劉苑伶 (2010)，「三個能源期貨價格預測模型比較分析及匯率關聯性之研究－以NYMEX與ICE為例」，中原大學，企業管理研究所。

洪惠娟 (2004)，「S＆P500指數、期貨與ETF價格發現之研究」，淡江大學，財務金融

林政勳 (2006)，「隱含波動率與波動率隨機過程模型之分析-S&P500指數選擇權市場之

郭姿儀 (2012)，「探討Black-Scholes模型中的Delta是否隱含S&P500指數之預測訊息」，

林煜城 (2008)，「S&P500指數期現貨與NASDAQ指數期貨之關聯性及波動外溢與跳躍