David Blitz, the head of quant research at Robeco, and Robeco researcher Matthias Hanauer have posted a commentary on the value factor. This is one of the classic Fama-French factors. It began life in 1992 as a simple “high minus low” (HML) calculation; the postulation that stocks with high book-to-market ratios (value stocks) outperform those with low book-to-market ratios (growth stocks).
But investors who relied on this factor have paid a price for it in recent years. Value has underperformed consistently since 2007. So, is it time to retire that factor entirely? Hanging over the whole idea of a finite list of isolatable investing “factors” is always the prospect that each of them represents an inefficiency, and that markets regularly move toward efficiencies, closing the windows represented by certain factors. Can we say, pursuing that line of thought, that a factor that was readily observable when Glengarry Glen Ross was in theaters (1992) has had its day and left us? Coffee, after all, is for closers.
The question has received a lot of attention of late, most recently from Blitz and Hanauer. In their paper they propose both to bury and to resurrect the value premium.
A Short Sad History of the Deceased
As Blitz and Hanauer remind us, the “first serious doubts about the value premium arose during the tech bubble in the late 1990s.” After all, the premise of the factor is that value stocks outperform growth stocks. During the tech bubble of the 1990s growth stocks regularly outperformed value. During a bubble most people don’t see it as a bubble—people tend to see it as a new normal, which is why it keeps on bubbling. This new normal was actually a bubble and burst early in this millennium.
So, does that mean Fama and French were right after all, and the appearance that they were out of date was itself a passing one? Well, no. Houge and Loughran studied the matter afresh in 2006 and found no evidence that value is superior to growth for equity indices, mutual funds, and large-cap stocks.
Indeed, in 2015 Fama and French themselves revised the HML nature of the value factor, suggesting that the real point is that the securities of firms that invest conservatively are superior to those of firms that invest aggressively (CMA).
But Blitz and Hanauer look at the evidence and conclude that neither HML nor CMA carries its own weight, theoretically or practically. If the value factor is to survive at all, it will have to be resurrected as something more complicated, as a package of four “value signals.”
The book-to-market ratio is just one of these signals. The second is the EBITDA/EV ratio, an enhancement of the earnings-to-price ratio. The third is cash-flow to price, and the fourth is the net payout yield (NPY). NPY is “essentially dividend yield, plus share buybacks, minus share issuance.”
Blitz and Hanauer explain how they determine what counts as a value stock given this whole package. They then crunch the numbers to determine how the more sophisticated notion of value performs as a fact for a sample period that “starts in January 1986 for the US and Developed-ex-US and January 1996 for Emerging Markets,” and continues until June 2020, presuming portfolios rebalanced monthly, and looking to total returns in US dollars.
They find that this notion of value can lead to much better investment results than the older notions. Furthermore, the backtesting indicates that value redefined has a good long-term track record that does not warrant the existential concerns that have dogged value as defined by Fama and French via the corporate balance sheet. It has a utility than cannot be explained by implicit exposures to the (other) Fama-French factors.
A Final Thought
The difference between the older notion of value and the newer understanding Blitz and Hanauer urge upon us is much like the difference automobiles manufactured in the golden age of the Detroit Big Three and automobiles manufactured today. The older autos didn’t have computer systems under the hood. Their systems were all accessible to shade-tree mechanics. Today, the mechanic under the shade tree needs a degree in data science. Likewise, the Fama-French model encourages back-of-the-envelope calculations by human beings. New additions and corrections to the model, of which this is but one example, discourage any such thing, and suggest one must leave the model to the computers and those who tend them.
Interested in contributing to AllAboutAlpha? Drop us a line at firstname.lastname@example.org