Ninety Years Of Media Coverage And The Cross-Section Of Stock Returns
University of Mannheim – Department of International Finance
University of Mannheim – Department of International Finance
November 12, 2015
Abstract:
Using a novel dataset on New York Times coverage of U.S. firms from 1924 to 2013, we re-examine the relation between media coverage and stock returns. The relation of changes in media coverage and returns is consistent with an attention-driven price pressure effect: Top-quintile outperform bottom-quintile coverage-change stocks by 10.68% during the formation year. Over the next two years, these stocks underperform their counterparts by 5.04%. In contrast to previous findings, the level of media coverage positively predicts stock returns. Top-quintile outperform bottom-quintile coverage stocks by 2.76% per year. This strategy is investable, exhibits an annual portfolio turnover of only 33%, does not depend on illiquid stocks, and attains a Sharpe Ratio of 0.48 (Momentum: 0.49).
Ninety Years Of Media Coverage And The Cross-Section Of Stock Returns – Introduction
We analyze the relation between media coverage and the cross-section of stocks returns. Our novel dataset on New York Times coverage of U.S. firms reaches back to 1924. This uniquely long panel of media coverage allows us to analyze how changes of visibility and the persistent level of visibility are related to stock returns. Until the 1950s, newspaper articles were the major channel of information dissemination to the general public. With the introduction of television to private homes in the 1960s, and internet in the 1990s, the dominance of newspapers in informing a broad audience has diminished. As an illustration, the circulation of daily newspapers, normalized by the U.S. population has declined steadily from 35% in 1950 to 13% in 2013, while the number of TVs per capita has increased from less than 3% to over 36% over this period.1 Hence, the unique 90-year length of our dataset is valuable, not only because it increases the power of our statistical tests, but also because it covers the earlier, newspaper-dominated decades of the 20th century.
We contribute to the literature in two main ways. First, we analyze the relation between year-to-year changes in media coverage and the cross-section of stock returns. We test whether increases in media coverage go along with impact-reversal patterns in stock returns as suggested by Barber and Odean (2008). To the best of our knowledge, we are the first to analyze the relation between low frequency changes in media coverage and stock returns. We find that stocks with a strong increase in media coverage outperform stocks with a strong decrease by 10:68% in the formation year. Subsequently, these stocks underperform their coverage decrease counterparts by 5:04% over the next two years. Second, we exploit our unique and comprehensive media coverage panel, starting in the early 20th century, to re-analyze the relation between the level of media coverage and the cross-section of stock returns. This research question was previously analyzed by Fang and Peress (2009) for a subsample of mostly large cap U.S. stocks. They find that during the ten-year period from 1993 to 2002 higher levels of media coverage are associated with lower stock returns. Relying on the entire cross-section of U.S. stocks and adequately controlling for firm size we obtain the opposite result. Stocks with high size-adjusted media coverage outperform stocks with low size-adjusted media coverage by 2:76% per year.
The relation between media coverage and stock markets has become a popular topic in empirical finance. Tetlock (2007) analyzes the relation between the tone of the Wall Street Journal’s ‘Abreast of the Market’ column and daily stock market activity for 1984 to 1999. He finds that more pessimistic tone of this column predicts negative stock market returns on the next day, and a reversal over the following week. Garcia (2013) extends Tetlock’s study by analyzing the impact of the tone of two New York Times columns on daily stock market returns for 1905 to 2005.2 He confirms the findings of Tetlock (2007) and additionally finds that return predictability is much stronger in recessions. Hillert, Jacobs, and Muller (2014) use a 1989 to 2010 panel of media coverage and article tone on the firm level to empirically test competing behavioral theories of momentum. To the best of our knowledge, theirs is the longest comprehensive dataset of newspaper coverage on the firm level used in empirical finance up to this study.
Does an increase in media coverage lead to temporary overvaluation? Barber and Odean (2008) hypothesize and find that increases in attention lead to more retail trading, and| due to short sale constraints of retail investors|more buying than selling. They argue that these order-imbalances should then lead to an impact-reversal pattern in stock prices. The idea that short sale constraints cause higher prices and lower expected returns goes back to Miller (1977), who models asset prices under short sale constraints with heterogenous-beliefs. Empirical evidence of the impact of changes in visibility (or ‘attention’) on stock returns is generally in line with the Barber and Odean (2008) model. Da, Engelberg, and Gao (2011) use Google Search Volume to measure attention on the stock-level and find a short-term, intra-annual impact-reversal pattern. In a detailed case study of attention-driven overreaction, Huberman and Regev (2001) show that the price of EntreMed’s stock sharply increased after a May 3, 1998 Times frontpage article whose content had been published months before, on November 27, 1997 in Nature. In the months subsequent to the initial increase from $12 to $85, the stock price reverted to a price level of approximately $25. Using a more general approach, Tetlock (2011) measures the textual similarity of firm-specific articles and finds an overreaction of prices to stale news. Other studies analyze product advertising as a stimulus for investor attention. Grullon, Kanatas, and Weston (2004) find that higher levels of advertising are associated with higher liquidity and a larger number of shareholders. Lou (2014) finds that advertising leads to temporarily higher price levels, consistent with an attention-driven impact-reversal pattern. Focke, Ruenzi, and Ungeheuer (2014) shed doubt on the conclusions of Grullon, Kanatas, and Weston (2004) and Lou (2014). Using a high frequency dataset of advertising, they find evidence against strong liquidity and return effects of advertising.
See full PDF below.