The episode is the financial world’s version of what’s known as the replication crisis, in which many well-known academic and scientific studies in medicine, psychology and other fields have proved impossible to reproduce. Earlier this year, Francesca Gino, a star Harvard Business School professor known for her research on honesty, was put on leave after being accused of falsifying data. (She’s denied the allegations and is suing Harvard and her accusers for defamation.)
A band of researchers led by Campbell Harvey, a professor at Duke University and a partner at quant shop Research Affiliates, has claimed a similar reckoning is long overdue in finance. Against this backdrop, hundreds of factors discovered in stocks—so many it’s now dubbed a “factor zoo”—have also now come into question, which in turn has fueled a string of defenses from other quants.
In the bond world, Dickerson and his co-authors didn’t stop with BBW. Along with tearing up the predictive factors that paper proposed, they took aim at a slew of other patterns spotted in the literature, including ones based on return reversals, momentum and default risk. Some were wrongly constructed, others were riddled with data errors, and many couldn’t beat a broad bond benchmark, they argued.
In a separate paper, researchers including a quant at AQR Capital Management, the asset management company led by Cliff Asness, also found that only about a fifth of bond-factor research could be replicated with clean data. In the context of stocks, AQR has been a strong proponent of factors. The latest drama underscores just how much messier the bond market still is for quants. Every company has multiple securities, and each one has different terms, including maturity and creditor status. Meanwhile, swings in Treasury prices influence returns, and the market’s liquidity can vary massively. Many bonds can go days without trading.
Despite all the uncertainties, systematic credit investing is a burgeoning business. Man Group, Robeco Institutional Asset Management and BlackRock are among its long-standing proponents. Multimanager hedge funds such as Citadel and Millennium Management are edging in, and Acadian Asset Management, a $100 billion quant shop, is the latest to plant its flag.
Many fiercely contest the assertion that most factors are likely to be false. At Robeco, a Dutch manager overseeing almost $200 billion, Patrick Houweling has also failed to replicate the BBW paper. But the co-head of quant fixed income takes issue with the conclusion that no factors have been definitively proven in the asset class. He says practicing quants like him, compared with academics, have better data and understanding of how to make these strategies work in real life. Even before BBW, along with former colleague Jeroen van Zundert, who’s now a portfolio manager at Cubist Systematic Strategies LLC, Houweling had written about the versions of equity factors such as low risk or relatively cheap pricing that work in bonds, too. “Similar factors work in the corporate bond markets when defined properly and also when tested properly,” he says.
Still, the retraction of the BBW paper has few precedents. Compared with other disciplines, top finance journals have had shockingly few retractions—only two so far, and the other was in 2021.
Dodgy findings should now be less likely, with the top three finance journals all requiring published papers to share the programming code underlying their work. (BBW came out before this policy was implemented at the JFE.) Journals are also taking a greater interest in studies that refute others; the rebuttal paper by Dickerson, Mueller and Robotti, for instance, is in November’s issue of the JFE.
Meanwhile, Dickerson himself is still threatening to stir things up. In his new paper, he and his co-authors parse 563 trillion possible models to find that most proposed bond factors add little value. Next he plans to show how trading costs erode profits from running quant strategies in bonds. Spoiler alert: It doesn’t look good.
“It’s really great because now we’re going to write papers that are still inherently very interesting, even if we don’t get this magical ‘x causes y,’ ” he says. “We try to re-lay the groundwork for corporate bonds, and I feel as if we’ve kind of achieved that.”