新聞及投資人情緒:股票報酬與波動之衝擊

Abstract

本文探討新聞情緒與投資人情緒對市場交易行為之衝擊及程度差異，並將檢視其運用在投資組合構建之表現。研究分別應用文辭探勘解析技術分析臺灣每日財經新聞之隱含資訊內涵並建構市場新聞情緒指標，以及參考Baker and Wurgler (2006)建構投資人情緒指標。實證發現市場新聞情緒相較投資人情緒指標對市場交易行為有較強烈的影響，其效果特別是在景氣擴張期間更為明顯。橫斷面分析結果顯示，當前一期市場新聞情緒較低時，規模小、波動高且獲利波動低的公司會有較高的報酬表現。研究進一步探討情緒指標對機構投資人交易行為的影響，結果發現外資及自營商的交易行為較易受市場新聞情緒影響，而投信的交易行為則較明顯受投資人情緒指標影響。實證分析結果顯示新聞情緒可為市場報酬及波動度的有效衡量指標，並可運用在投資組合的構建。

This study examines and compares the sentiment effects driven by trading and news information on market returns and volatility as well as the applicability of news sentiment in portfolio construction. We employ linguistic analysis to construct the market aggregate news sentiment index (MANSI), drawn from daily financial news about Taiwanese listed firms, and form the investor sentiment index (SI) by following Baker and Wurgler (2006). Evidence shows that MANSI has a stronger influence on market stock returns and volatility than investor sentiment. The results also reveal that the news sentiment effect is more profound during market expansion periods. The cross-sectional portfolio analyses conditionally suggest that lower news sentiment on smaller, more volatile, and less profitable volatile stocks generate higher returns. In addition, we develop long-short portfolios based on size, volatility, and profitability to show that the previous MANSI state is a significant determinant of portfolio returns. Additionally, MANSI has more influence on the trading behavior of foreign institutional investors and dealers, whereas investor sentiment reveals a stronger influence on domestic institutional traders. The results suggest that news sentiment can be an effective proxy for market returns and volatility, which can be applied in portfolio construction.