Rankings Of Published Price-Earnings Ratios And Value Investor Attention
University of Rochester, Simon Business School, Students
June 22, 2016
Price-earnings (P/E) ratios are the most popular proxy for fundamental value and are widely published using a common methodology. This paper explores whether stocks with high P/E rankings are especially salient to individual investors with attention constraints. Consistent with the role of attention, P/E rankings predict the returns of strategies based on size, liquidity, and short-term reversals as well as increases in trading volume and liquidity. Financial data providers publish P/E ratios for stocks with positive earnings, but do not publish P/E ratios for stocks with negative earnings. P/E rankings predict returns and changes in trading volumes for stocks with positive earnings, but not for stocks with negative earnings. In an event study of all stocks which cross the 0 P/E threshold, the value of a positive P/E ratio is around 2.25%. Stocks which cross into positive P/E territory trade more actively than stocks which cross into negative P/E territory. Overall, a value-weighted extreme decile P/E attention strategy earns average monthly returns of 119 basis points from 1973 to 2015 with an annual Sharpe ratio of 0.94. These strategy returns are robust to fundamental factors and momentum in prices and earnings individually.
Rankings Of Published Price-Earnings Ratios And Value Investor Attention – Introduction
The relation between attention and stock returns is well documented. Kahneman (1973) establishes the importance of attention as a scarce cognitive resource which influences the decision-making process of all individual investors. Merton (1987) hypothesizes that for any asset, a broader investor base is associated with more diversified private information, a lower required discount rate, and a higher price. Gervais et al. (2001) and Kaniel et al. (2012) document a positive relation between current relative volume and future returns in equity markets. Barber and Odean (2008) relate high volumes, extreme returns, and news coverage to net purchases by individual investors. I extend this literature by considering published price-earnings (P/E) ratios as another important attention-grabbing characteristic. If enough value investors with attention constraints search for stocks in a list ranked by published P/E ratios, then these rankings can influence stock returns, trading volumes, and liquidity.
Empirical results support the P/E attention hypothesis. In a monthly time-series regression from 1973 to 2015, a long-short decile strategy based on P/E rankings and changes in rankings earns an average value-weighted monthly return of 119 basis points with an annual Sharpe ratio of 0.94. Strategies earn significant alphas in the Fama and French (1993) three-factor model, the Carhart (1997) four-factor model, and the Fama and French (2015) five-factor model. These factor models include controls for exposure to market beta, size, value, profitability, investment, and price momentum. P/E rankings can only change because of relative changes in price or relatively strong or weak year-over-year quarterly earnings. Nevertheless, P/E rankings still predict strong variation in returns within portfolios which are already sorted on past returns or earnings momentum.
Stoll (1978) proposes a model in which individuals act as dealers willing to provide liquidity when institutional participants require immediate execution. Kaniel et al. (2008) show evidence that individuals do in fact provide liquidity by purchasing stocks following recent declines. If individual investor attention is positively correlated with P/E rankings, then these rankings should predict the returns of trading strategies which attempt to profit from short-term reversals. Likewise, the added visibility of high P/E rankings should predict the profitability of strategies which involve buying small or illiquid stocks. Furthermore, if stocks with high P/E rankings attract additional investor attention, then the added participation of these investors should predict higher trading volumes. The data support all of these predictions.
I use data from Google Trends to establish the unrivaled popularity of P/E ratios as measures of fundamental value. Google Trends has a new feature providing users a search volume index (SVI) for a “topic” rather than a very specific “search term.” This SVI is a standardized time series of internet searches for search terms which Google attributes to the broader topic. Da et al. (2011) show a positive relation between the SVI for individual ticker symbols and subsequent cross-sectional variation in abnormal turnover and future returns. Figure 1 shows the SVI time series for topics that could plausibly proxy for fundamental value. Specifically, I consider the sorting variables for the 12 value anomalies in Hou et al. (2014). There are only four variables with non-trivial search volume: P/E ratios, P/B ratios, market leverage, and dividend yield.2 P/E ratios have the highest SVI in every single week of the sample, which extends from January 2004 to April 2016. The average SVI for P/E ratios is 6.9 times larger than the average SVI for P/B ratios.
It is straightforward to estimate the P/E rankings at the end of any trading day. Major financial data providers publish P/E ratios using the most recent intraday or closing price and the four most recent quarters of earnings per share (EPS). Table 1 shows P/E rankings for S&P 400 Midcap stocks on two dates: August 31, 2015 and November 30, 2015. Between August and November, four stocks enter the list of 20 stocks with the highest P/E rankings : CYH, CAA, TEX, and RCII. These four stocks are especially salient to a value investor who regularly evaluates S&P 400 stocks with the highest P/E rankings and who has the resources to pay attention to about 20 stocks. The expected aggregate effect of this individual investor attention is a higher equilibrium price and an increase in trading volume for any of these four stocks.
P/E rankings could proxy for fundamental risk rather than value investor attention. To address this objection, I rely on an important institutional detail. P/E ratios for stocks with negative earnings are not published. Figure 2 shows Google Finance stock quotations for Ford Motor Company (F) and Tesla Motors Inc (TSLA). Ford has positive earnings and a published P/E ratio while Tesla has negative earnings and does not have a published P/E ratio. For stocks with positive earnings, P/E rankings convey information about both visibility and fundamentals. For stocks with negative earnings, P/E rankings only convey information about fundamentals. Strategies based on P/E rankings or changes in rankings only earn significant returns and predict changes in trading volumes in the sample of stocks with positive earnings. I also conduct an event study of all stocks crossing the 0 P/E threshold and estimate the value of a positive P/E ratio at around 2.25%. Stocks which cross into positive P/E territory have greater abnormal trading volume than stocks which cross into negative P/E territory.
See full PDF below.