總體經濟數據公布對台股的報酬、交易量、波動性分析：以日內資料分析

Abstract

總體經濟數據的統計與公布，不僅描繪了一個國家的實際經濟表現，還時刻左右著資本市場的情緒和資產的漲跌表現。尤其是那些對全球經濟具有重大影響力的國家，在公布經濟數據時，更常引發外溢效應，牽一髮而動全身。因此，總體經濟數據的公布對股市的影響一直是學術界的熱門研究領域。本研究透過彭博統計數據公佈前市場對該數據的預測，計算出公佈後未被市場預期的衝擊程度，並採用ARCH-GARCH模型探討該衝擊對股市報酬率、交易量、波動性的影響。此外，本研究納入美國與台灣的總經數據，採用日內高頻交易資料捕捉更即時與單純的影響。研究發現超預期的美國的Core CPI, Core PCE, CPI , PCE, PPI會負向影響台股1分K與5分K的報酬率，超預期的Initial Jobless Claim, Nonfarm Payroll數據則會負向影響台股報酬率，台灣的經濟數據則僅有超預期的進口量負向影響報酬率，以及超預期工業生產數值正向影響報酬率。大於5分鐘的時間區段則大部分的經濟數據皆不影響報酬率。交易量則是大部分的美國數據公布時均會負向影響當下的交易量。且各個時間段下均存在不對稱的波動性。

The statistics and release of macroeconomic data not only depict the actual economic performance of a country but also constantly influence the sentiment of the capital markets and the performance of asset prices. Especially for those countries with significant impact on the global economy, the release of economic data often triggers spillover effects, affecting the global market. Therefore, the impact of macroeconomic data releases on the stock market has always been a popular research topic in academia. This study calculates the unexpected shock of data releases by using market forecasts from Bloomberg before the release and employs the ARCH-GARCH model to explore the impact of these shocks on stock market returns, trading volume, and volatility. Additionally, this study includes macroeconomic data from both the United States and Taiwan, using intraday high-frequency trading data to capture more immediate and pure effects. The research finds that unexpectedly high U.S. Core CPI, Core PCE, CPI, PCE, and PPI negatively affect the 1-minute and 5-minute returns of Taiwan stocks, while unexpectedly high Initial Jobless Claims and Nonfarm Payroll data negatively affect the returns of Taiwan stocks. For Taiwan's economic data, only unexpectedly high import volumes negatively affect returns, and unexpectedly high industrial production positively affects returns. For time periods longer than 5 minutes, most economic data do not affect returns. In terms of trading volume, most U.S. data releases negatively affect the immediate trading volume. Furthermore, there exists asymmetric volatility across different time periods.