Should you ignore financial academic research?
Marketers like nothing more than to pitch a product “grounded in decades of academic research.” The implication is that objective academics have dispassionately studied market phenomena to uncover canonical truths. In theory, this makes the investment decision easy – think “value outperforms growth” or, more recently, “quality stocks outperform.”
Don’t be fooled. Academic papers are subject to five very serious limitations:
1. Publishing Bias. You only see papers with “interesting” results. No one gets tenure with a synopsis that says “we looked hard and found nothing.” There’s a great temptation to turn the statistical dials until you find something compelling. Even raise this as a question and academics react in shock and horror – how could you possibly question their objectivity? This is complete nonsense – they’re as self-interested as everyone else.
2. The Big Splash Phenomenon. The initial results get the press; well founded critiques don’t. Occasionally, there’s brouhaha about a statistical error (see Reinhart/Rogoff or Piketty), but ordinarily the synopsis becomes part of the investment canon.
3. The Assumptions Are Everything. The one paragraph synopsis is the fun part – slogging through seventy pages of assumptions and statistics clearly isn’t. 99% of people stop at page one and lose sight of all the myriad assumptions that underscore the results.
4. The World Changes – A Lot. Conclusions sound a lot more convincing if something as been true for “decades.” But, seriously: who actually thinks that what was relevant in, say, the 1960s has much bearing on market dynamics or investor behavior today?
5. Business and Academia Overlap. The real money is made when academic ideas are commercialized. Once big asset managers get behind an idea, it takes a life of its own. Soon everyone in the chain has a direct economic incentive to keep pushing a strategy, even when the evidence mounts that it might be wrong.
To take a real life example, consider the “value” factor (that cheap stocks outperform expensive ones). Let’s give Fama-French the benefit of the doubt and ignore publishing bias. The Sharpe ratio of the value factor was approximately 0.55 during 1963-1990 (the window studied), but declined almost immediately and has been roughly zero over the past decade. Fama-French arguably were correct that phenomenon was there for almost three decades, but there are many non-statistical reasons for this: a dearth of asset-based value investors, phone-based retail brokerage that pushed story stocks, etc. Guess what? The world changed and a simple, book value based test stopped working almost as soon as it was “discovered.”
Or look at the “size” factor, whereby small capitalization stocks outperformed large ones. Here the problem is the reliability of the data fed into the statistics machine. Even today, a shockingly large percentage of the universe of “small” stocks would have a market capitalization below $100 million – as we say, these stocks trade by appointment only. Imagine what they must have looked like in 1963. The “last price” has little bearing on whether you could actually buy the stock there. Garbage in, garbage out.
These are cautionary tales for investors who are piling into strategies built around the “quality” factor. It’s important to remember that the first paper on this was published in 2012, so almost all the “historical” data is back-tested. As soon as the first paper was published, the author was hired as a consultant by a large asset management firm, which was then followed by dozens of smart beta proponents – irrespective of whether the data continues to look good, the marketing train has left the station.
And missing is serious debate of whether the “phenomenon” really is likely to persist or, like the value factor, it reflected a temporary anomaly. Consider this: in the 1980s, value investors were predominantly focused on asset values (think Seth Klarman’s Margin of Safety); by the mid-1990s value investors shifted to using discounted cash flow models to assess “intrinsic value.” Hence the stability and quality of a company’s cash flows started to matter a lot more, and money flowed into those stocks accordingly. As one clever commentator noted after the publication of Stocks for the Long Run in the early 1990s (which “proved” that stocks always outperform bonds), “of course it’s true until everyone knows it’s true.”
So, the simple advice to advisors and investors is to take any investment themes supported by academic research with a huge grain of salt. When the world is awash in value investors who like quality stocks, do we really care that if you’d bought quality stocks fifteen years ago you’d have outperformed the rest of the market? Look at another way, would you base a decision to buy Apple today on a research report from 2002?
Andrew Beer is managing partner and co-portfolio manager, dynamic beta, at Beachhead Capital Management, New York.