News Sensitivity and the Cross - Section of Stock Returns

Abstract:     
The paper is the first one outside the high-frequency domain to use sentiment - signed news to directly compare news and no - news stock returns. This is done by estimating whether returns on positive, neutral and negative news days are significantly different from the average daily return for a large sample of US stocks over the period from January 2003 to August 2010. The general results show that positive news days indeed have above-average returns and negative news days returns are below average, while the neutral news days are economically barely distinguishable from the average. The market also proves to be fast and accurate at pricing new information, as there are no signs of drift shortly after news days. On the contrary, a directionally correct and statistically significant movement can be found on the day before the news day. The cross-sectional analysis reveals significant differences in the strength of market reactions between stocks ranked on size, book-to-market or news coverage. The general results however hold across all sub-samples and are also not driven by earnings announcements or past stock returns. Moreover, the average news sensitivity is itself a priced source of risk. A portfolio of stocks with high sensitivity to news outperforms a portfolio of stocks with low sensitivity by a statistically and economically significant 0.84% per month. This news premium seems to primarily relate to the high impact of news in situations of general uncertainty.

Number of Pages in PDF File: 47

Keywords: company news, sentiment, news analytics, public information

JEL Classification: C33, G12

working papers series