Swedroe: Low Priced Stocks No Bargain
As I wrote about last week, the absolute level of a firm’s stock price is arbitrary, as it can be easily manipulated by the firm through altering the number of shares outstanding (for example, by splitting the stock). Despite this obvious fact, the research into investor behavior has found a strong preference among individuals for low-priced stocks.
For instance, the research shows that individual investors tend to hold lower-priced stocks than institutions. And there’s also evidence demonstrating that the number of small shareholders in a stock increases following a split to a lower price level.
Research demonstrates this preference for low-priced stocks by individual investors is motivated by an irrational belief that stocks with lower nominal prices possess greater upside potential (more room to grow and less to lose) and the lottery-ticket effect. As with lotteries, investors are searching for “cheap bets” with large upside potential. They prefer positive skewness even if it means lower average expected returns.
The overly optimistic expectations among investors regarding low-priced stocks predict that low-priced stocks will likely be overpriced relative to high-priced stocks, thereby delivering lower future risk-adjusted stock returns.
More Evidence On Low-Priced Stocks
Justin Birru and Baolian Wang contribute to the literature on low-priced stocks through their March 2016 paper, “The Nominal Price Premium.” The authors begin the paper by noting that “the cross-sectional relationship between raw nominal prices and future returns is likely to underestimate the real economic magnitude of the nominal price premium.” They explain that is because a sort on raw nominal price is confounded by the mechanical relationship between raw nominal price and expected returns.
Any model of prices (such as the Gordon Constant Growth Dividend Discount model) inversely links prices and expected returns—stocks with higher risk will have higher expected returns, causing future cash flows to be discounted at a higher rate, leading to a lower current valuation.
Therefore, it should not be surprising that past research has failed to document a relationship between nominal price and future returns. A sort on raw nominal price combines two countervailing forces: a nominal price premium (predicting that low-priced stocks should exhibit low future returns), and a mechanical discount-rate effect (predicting that low-priced stocks should exhibit high future returns). For this reason, using nominal price as a sorting variable will underestimate the nominal price premium.
To address this problem, Birru and Wang employed a set of nominal variables from accounting statements: assets per share, book value per share, earnings per share and dividends per share. This set of variables, while highly correlated with nominal prices, is unlikely to suffer from the mechanical relationship with expected returns discussed previously. Their accounting data set covered the period 1967 through 2012, while their returns data set covered the period 1968 through 2013.
Using these variables, Birru and Wang found “strong evidence in support of a nominal price premium for low-priced stocks.” They found that a strategy holding a long position in high “fitted” price stocks and a short position in low “fitted” price stocks generates a value-weighted (equal-weighted) four-factor alpha of more than 85 (88) basis points per month.
Consistent with the idea of overpricing in low-priced stocks, they found most of the abnormal returns accrue to the short leg of the strategy rather than to the long leg, which is invested in high-priced stocks. Specifically, they found the short leg of the strategy earned -77.9 basis points per month and the long leg of the strategy earned a statistically insignificant 7.5 basis points.
Impact Of Arbitrage Limits
Further consistent with a mispricing story, the authors found that the results were stronger in the presence of greater limits to arbitrage (small stocks have greater limits to arbitrage because they are typically more difficult to short, have less institutional ownership, and greater illiquidity and idiosyncratic volatility).
Low-priced stocks become most overpriced relative to high-priced stocks during high-sentiment periods, meaning the bulk of the strategy returns occur following times of high investor sentiment and, additionally, as sentiment decreases, high-priced stocks attain high returns relative to low-priced stocks.
In other words, the returns to a portfolio that is long high “fitted” price stocks and short low “fitted” price stocks are negatively related to contemporaneous changes in sentiment. Their findings offer further support for the concept that even after publication of research exposing a pricing anomaly, limits to arbitrage can prevent sophisticated investors from fully correcting instances of that mispricing.
For example, Birru and Wang write: “After splitting to a lower price level, stocks exhibit positive abnormal returns in the short run, followed by negative abnormal returns in the long-run. This is consistent with investors bidding up the prices of low-priced stocks to overvalued levels, and this being followed by the slow correction of mispricing.”
They also write that “low-priced stocks have systematically negative earnings surprises, while high-priced stocks experience systematically positive earnings surprises.”
And perhaps surprisingly, the authors also found similar evidence when examining analyst price forecasts—low-priced stocks have substantially higher analyst price appreciation expectations than high-priced stocks.
Specifically, they found that, over the entire sample period, for a three-day period after earnings announcements, the market-adjusted abnormal return is -79 basis points for the lowest decile of “fitted” price stocks, while the three-day market-adjusted abnormal return is 23 basis points for the highest decile. Both numbers are statistically significant at the 1% confidence level.
What’s more, the figures were even larger over the more recent subperiod from 1995 through 2013. The three-day periods following earnings announcements (which occur four times per year at quarterly earnings release dates) accounted for a significant portion (30-60%) of the long/short premium.
Birru and Wang conclude: “The results lend strong support to the hypothesis that investors are overly optimistic regarding low-priced stocks and overly pessimistic regarding high-priced stocks.”
Summarizing, Birru and Wang state: “These findings on investors’ expectational errors cannot be explained by purely a preference for skew channel (the preference for lottery-like distributions) and therefore provide supporting evidence that the results are (at least partially) driven by a belief-based channel.”
This led the authors to conclude that it’s the combination of the irrational belief that stocks with lower nominal prices have greater upside potential and the preference for skewness that results in the overpricing among low-priced stocks, and the asset pricing outcomes they observed.
On an interesting side note, firms that rely on the evidence from academic research when designing portfolio construction rules (such as Bridgeway and Dimensional Fund Advisors) have long screened out low-priced stocks. (Full disclosure: My firm, Buckingham, recommends both Bridgeway and Dimensional funds in constructing client portfolios.)
Larry Swedroe is the director of research for The BAM Alliance, a community of more than 140 independent registered investment advisors throughout the country.