新冠肺炎期間應用分量迴歸模型
探討VIX波動對美股及台股之報酬影響

本研究使用TEJ及R統計軟體，整理樣本期間2019 年 1 月2日至 2022 年 2 月11日之日資料，蒐集各個變數786 筆日資料作為研究樣本，變數包括台灣加權股票指數、美國納斯達克、S&P 500、道瓊工業平均指數、費城半導體指數，探討各個指數對於VIX指數之反應，其並利用分量迴歸模型觀察VIX指數對美股與台股有何不一樣之影響。透過模型可得以下結果：一、利用PP單根檢定法，VIX指數、台灣股市及美國股市之股價報酬皆顯著拒絕單根的虛無假設，顯示研究變數報酬皆為穩定時間序列。二、採用OLS方程式估計結果看出，各項變數報酬(R)受VIX波動(RVIX)整體皆為負向顯著影響效果，而QR方程式估計結果看出，各項變數在不同的分量下，則會呈現正向關係，表示QR及OLS有顯著差異。

關鍵字

分量回歸模型； VIX指數；新冠肺炎

並列摘要

This research uses TEJ and R statistical software to organize the daily data from 2 January 2019 to 11 February 2022 and collects 786 daily data for each variable as the research sample. The variables include Taiwan Capitalization Weighted Stock Index, Nasdaq Composite Index, The Standard and Poor's 500, Dow Jones Industrial Average, PHLX Semiconductor Sector, to explore the response of each index to the VIX index and use the quantile regression model to observe the difference between the VIX index on US stocks and Taiwan stocks. The following results can be obtained through the model: First, using thePP unit root test, the stock price returns of the VIX index, the Taiwan stock market, and the US stock market all significantly reject the null hypothesis of a single root, indicating that the research variable returns are all stable time series. Second, using ordinary least squares mothed, the overall relationship between stock price index (R) and VIX index volatility (RVIX) is negative, while the QR equation estimation results show that the variables under different quantile will be a positive relationship was shown, indicating that QR and OLS were significantly different.

並列關鍵字

Quantile Regression Model ； VIX ； Covid-19

參考文獻

11.Phillips, P. C. B. ; Perron, P. (1988). "Testing for a Unit Root in Time Series Regression" (PDF). Biometrika. 75 (2): 335–346. doi:10.1093/biomet/75.2.335.

1. 黃韋中 (2021). 利用VIX 指數和ARMA-GARCH 模型預測波動度之目標波動度策略績效分析. 金融學系, 國立政治大學: 1-47.