Does the value premium for U.S. stocks, as measured by book-to-market ratio, persist after its initial discovery/publication in 1992? In their January 2020 paper entitled “The Value Premium”, Eugene Fama and Kenneth French assess whether the value premium in the U.S. declines or disappears in a post-publication sample that is as long as the discovery sample. Unlike many researchers, they focus on difference in returns between high book-to-market (value) stocks and the value-weighted market, not the return spread between value and low book-to-market (growth) stocks. To control for firm market capitalization effects, they consider separately stocks with capitalizations above (big) and below (small) the NYSE median. They specify value (growth) stocks as those at or above the 70th (below the 30th) percentile of book-to-market values of NYSE stocks. They re-sort stocks at the end of each June, with book-to-market ratio measured at the end of the fiscal year during the prior calendar year. The overall value premium is the capitalization-weighted combination of big value and small value. Using annual book-to-market ratios and market capitalizations, and monthly returns, for all NYSE, AMEX and NASDAQ stocks during July 1963 through June 2019, they find that:
- Average value premiums are much lower after 1991.
- For the full sample period, capitalization-weighted big (small) value stocks beat the market by an average 0.21% (0.45%) per month. The overall average monthly value premium is 0.26%.
- During 1963-1991, big (small) value stocks beat the market by an average 0.36% (0.58%) per month, with average overall value premium 0.42%.
- During 1963-1991, big (small) value stocks beat the market by an average 0.05% (0.33%) per month, with average overall value premium just 0.11%.
- Capitalization-weighted big (small) growth stocks have approximately zero average premiums relative to the market over the full sample period and both subperiods. For example, during 1963-2019, the average big (small) monthly growth premium is -0.02% (0.06%).
- With the assumption that regression coefficients are constant over the sample period, regressions that forecast value premiums for stocks with book-to-market ratios in excess of the market book-to-market ratio reliably confirm a post-publication decline in the expected value premium.
In summary, evidence indicates that the value premium for U.S. stocks materially declines post-publication, but the high volatility of monthly value premiums limits confidence in the decline.
Cautions regarding findings include:
- Reported returns and premiums are gross, not net. Accounting for frictions associated with portfolio reformation would lower value portfolio returns. These frictions are higher during the first subperiod than the second, suggesting exploitation barriers play a role in value premium decline. The low frequency of portfolio reformation (annual) mitigates this concern.
- Investors often make decisions on evidence that falls short of strong statistical reliability.
See also “Value Investing Dead?” and “Value Investing Not Dead?”.
