The Fallacy of Fama-French

Originally published in the Brandywine Asset Management Monthly Report.

In 1992 Eugene Fama and Kenneth French published a paper titled “The Cross-Section of Expected Stock Returns.” In that paper they introduced what has since become known as the “Fama-French three-factor model.” What they showed was that stock prices were driven not solely by a mystical “risk premium,” but that, in addition to beta, there were other factors that explained the movement of stock prices. In particular, the additional two “factors” they identified in their paper were that small capitalization and high value stocks outperformed other stocks. Over the years other academics expanded on the three-factor model by discovering that factors such as momentum and liquidity also affected equity prices.

The most shocking part of these discoveries was that they were considered “discoveries” at all. People have been using momentum to profitably trade markets for decades. Brandywine is aware of this first-hand. In the early 1980s Brandywine’s founder, Michael Dever, set up a fund to be traded by Richard Donchian, who first used momentum strategies in a fund he launched in 1949. (Mr. Donchian became known as the “father of trend following.”) And of course everyone knows about value investing, not the least because of Warren Buffett’s success in following the concepts first presented in 1934 by Benjamin Graham and David Dodd in their book Security Analysis.

In 1991, a year prior to the publication of the Fama-French three-factor model, Brandywine launched our Brandywine Benchmark program, which incorporated more than two-dozen return drivers (a return driver is the underlying condition that drives the price of a market, similar to a “factor”). These included some of the return drivers later discovered by academics, such as momentum and value. But rather than publish a paper and consider that to be the epitome of our research, Brandywine understood these were just two out of potentially hundreds of return drivers that could be exploited to profit from market movements. So Brandywine didn’t stop there. We not only developed numerous trading strategies based on additional distinct return drivers, but we also applied them to a diverse range of relevant global financial and commodity markets, not just stocks. As described in previous monthly reports, this approach has multiple benefits, including returns that are unrelated to those provided by “conventional” investments such as stocks and bonds (i), lower event risk (ii), and a greater likelihood that future returns will approximate past returns (iii).

Unfortunately for individual investors, however, the mutual fund industry has seized on the academic research, using it to provide cover for the creation of numerous mutual funds designed to capture specific factors. You can now put money into “smart beta” funds that pick stocks by value, size, momentum, liquidity, dividend history, earnings surprises, insider activity, sales growth, and a myriad of other specific factors. In an industry focused on selling the most recent “hot” fund, this serves the mutual fund companies well. But pumping out trivial “discoveries” in the form of alternative mutual funds and ETFs is a major disservice to investors.

Return Drivers & Performance Predictability
If there is one major takeaway from Brandywine’s research and trading over the past 35 years, it is that no single return driver (factor) such as these will provide either consistent or predictable returns. We don’t know of any legitimate trading strategy that will not have extended periods of significant drawdowns (relative to its prospective return). We realize people show performance of specific approaches that appear to be consistent and predictable, but at some point they will suffer multi-year losing spells. We also realize that people have back-tests of trading strategies that show consistent performance, but we can also assure you that they manufactured those results by sacrificing the future predictability of returns for that trading strategy.

But don’t confuse that with us saying that those return drivers are not worthy of serving as the basis for trading strategies. It just means that those return drivers should not be utilized on a stand-alone basis. Unfortunately, the investment industry does just that. It promotes the latest academic research as the “Holy Grail” of investing and people flood it with money.

We see the same focus on specific return drivers among institutional investors and their consultants. Recently for example, a large pension plan issued a mandate specifically for trend following CTAs. We agree that trend following is a valid return driver, but we take issue with the thinking behind the mandate. Their consultant had read research that showed the tail risk mitigation of having trend following CTAs in a portfolio. But that is a very simplistic conclusion. Trend following is just one return driver. There are many other return drivers that can provide that same benefit. And by combining trading strategies based on those other return drivers into a more diversified portfolio, they would get even better tail risk protection but with substantially greater predictability of performance.

We also see many institutional investors or funds-of-funds that are fixated on finding “specialist” managers that are uncorrelated with the core managers in their funds. The problem with this approach is that the past performance of a specialist manager has almost no predictive ability. Again, it doesn’t mean the return driver being captured by the specialist manager isn’t valid. It just means that it is not predictable.

Predictability is everything. There is no benefit in a great past performance record if it is not representative of future performance. In recognition of this, prior to the launch of the Brandywine Benchmark program in 1991, Brandywine conducted an extensive research project that focused on discovering the characteristics of a portfolio that contributed to producing predictable performance. This was a significantly different approach from that taken by virtually all other researchers then and now, who focus on developing portfolio allocation models that produce “optimal” performance. This research resulted in Brandywine’s “Predictive Diversification” portfolio allocation model. We came to realize during our research that to produce the most predictable returns, a portfolio must be balanced across a wide range of unrelated return drivers and markets.

Brandywine’s Symphony & Predictive Diversification
Brandywine expanded on this philosophy with the launch of Brandywine’s Symphony Program in 2011. Not only does Brandywine’s Symphony Program exploit many of the same return drivers first developed into trading strategies and incorporated into the Brandywine Benchmark program more than two decades ago, but it also takes advantage of more recent research to add trading strategies that were unavailable at that time. Mr. Dever discusses one of these strategies in chapter three of his best-seller and specifically in the Action Section for that chapter (pages 19-26) on the book’s web site.

Despite the continued proven success of Brandywine’s approach, we see little evidence in a stampede of like-minded followers committed to providing investment programs that are broadly diversified across both return drivers and markets. In fact, as described above, the opposite is occurring. Interest in specialist mutual funds and managers appears to be increasing.

While we at Brandywine are often frustrated with the archaic state of the investment industry and what passes for “research,” we are most concerned with the disservice it provides to individuals who are looking for answers on how to invest their money. That said, we also recognize that it is precisely this archaic state that provides us with our “edge.” To the extent that other managers dismiss return drivers, ignore opportunities in markets that are not featured regularly in the popular press or financial TV shows and instead focus on the “usual” suspects (stocks, bonds, gold), pursue specialty strategies to the exclusion of others that can improve the consistency and predictability of their returns, don’t understand the significance of predictive diversification, or are simply unable to devote the thought cycles to understanding and researching an approach that is so different from the conventional investment industry, we will benefit.

(i) Unrelated Returns:
Versus CTAs:
Versus Stocks:

(ii) Lower Event Risk:

(iii) Predictability:
Diversification Leads to Predictability:
Fatal Flaw in Mean-Variance Optimization:
Predictive Diversification:

Share this post:
  • Facebook
  • Twitter
  • email