Swedroe On Large-Stock Value Premium

Taking a look from another angle.

Recently, we have seen a rise in the level of discussion about whether there is a significant value premium in large-cap stocks. The value premium is the tendency of stocks with low prices relative to measures of their value to outperform stocks with high relative prices. Since large-cap stocks make up about 90 percent of the total global market capitalization, this is an important issue.

Professors Eugene Fama and Kenneth French examined this issue in their 2012 study, “Size, Value, and Momentum in International Stock Returns,” and found that there was a value premium in large-cap stocks of 0.17 percent per month. And while it wasn’t statistically significant—the t-stat was 1.09—a 2 percent per year premium is certainly economically significant. Other studies have found similar results, at least when ranking by price-to-book (P/B) ratio.

It’s also worth noting that the average return spread among large U.S. stocks sorted on P/B ratios was 38 basis points per month from July 1926 to June 1990 (the t-stat was 2.43), but shrank to just 5 basis points per month from July 1990 to June 2013 (the t-stat was 0.26).

That’s consistent with a situation in which arbitrageurs trade away the value premium among large U.S. stocks following the publication of Fama and French’s research prior to (and during) that period. In a 2012 study, however, Fama and French did find an economically large and statistically significant global value premium among small stocks.

Sandro C. Andrade and Vidhi Chhaochharia—authors of the April 2014 study “Is There a Value Premium Among Large Stocks?”—took another, and somewhat different, look at this issue.

While Fama and French use the book-to-market ratio to determine value, Andrade and Chhaochharia use the price-to-earnings (P/E) ratio. And while Fama and French used regional breakdowns (North America, Europe, Japan and Asia-Pacific ex-Japan), Andrade and Chhaochharia used both regional breakdowns and a global pooling approach. Their study covered the 24-year period from July 1990 through June 2013. Following is a summary of their findings:

• Using a global pooling approach, a value-weighted and long-short HmL portfolio (meaning it’s long the lowest 30 percent of stocks ranked by P/E ratio and short the highest 30 percent) earns, on average, a statistically significant premium of 0.64 percent per month (the t-stat was 2.61). That’s almost four times what Fama and French found.
• Using regional breakpoints, there still is an economically large and statistically significant value premium of 0.37 percent per month (the t-stat was 2.38) among large stocks when stocks are sorted by P/E ratios. While less than the 0.64 percent per month premium found when using global breakpoints, it’s still more than twice as much as Fama and French found. And it’s statistically significant.
• Using global breakpoints and P/B ratios, there was a statistically insignificant value premium of 0.15 percent per month (the t-stat was 0.86). This finding is very similar to that of Fama and French. Again, it’s worth noting that even this premium is economically significant.

The authors drew the following conclusions: First, using the P/E metric restores the statistical significance of a large stock value premium. Second, using the P/E value metric is more important than using regional versus global breakdowns.

Third, although the ultimate source(s) of the value premium remains elusive, the results suggest that risk-based explanations may be less vulnerable to criticism than previously thought: “If the global value premium was indeed due to mispricing, and sustained because of high trading/shorting costs, then the global value premium should be negligible among large stocks. This is not what we find.”While many value index funds use the P/B metric to set their fund construction rules, there are others that use multiple valuation screens (such as P/E, price-to-cash flow, price-to-sale and price-to-dividend ratios). The evidence from this paper provides support to the multiple-screen approach.

Interestingly, in their 2005 study, “The Value Premium and the CAPM,” Fama and French also found that when they used earnings-to-price (E/P) ratios rather than book-to-market ratios to separate value and growth stocks for the period 1963-2004, there was little difference between value premiums for small and big U.S. stocks.

They also found that in 14 international developed markets, when ranking stocks by E/P, there was little difference between the value premium in large and small stocks during the period from 1975 through 2004.

It shouldn’t come as a surprise that value provided a larger premium when sorted globally rather than regionally. Cheaper countries in the aggregate tend to do better, and sorting globally gives greater allotment to stocks in those nations.

Larry Swedroe is the director of research for the BAM Alliance, a community of more than 140 independent registered investment advisors throughout the country.