Can shrewd investment managers consistently outperform the market, or are their successes largely the product of luck? For decades, that question has been at the center of a spirited debate in the financial world. If markets are truly efficient — that is, if they fully reflect all available information — conventional wisdom says active management should be largely futile, and investors would be better off buying a passive index fund. Yet active funds continue to attract trillions of dollars from investors betting that some managers really do possess a repeatable edge.
“This is a crucial question in finance,” says Stijn Van Nieuwerburgh, the Earle W. Kazis and Benjamin Schore Professor of Real Estate at Columbia Business School. “There are thousands of funds out there whose managers claim they can outperform the market meaningfully and consistently.”
A new study from Van Nieuwerburgh — co-authored by Ron Kaniel of Rochester University, Markus Pelger of Stanford, and Luofeng Zhou of New York University — suggests those claims aren't entirely misplaced. Analyzing nearly 30 years of data from more than 3,000 U.S. bond mutual funds, the researchers found that some managers do generate superior risk-adjusted returns over a significant time horizon. What’s more, those managers can be identified in advance using modern machine learning techniques.
Separating skill from luck
The study builds on an earlier project in which the researchers used machine learning to identify skilled equity fund managers. This time, they turned to bond funds, assembling a massive dataset that included fund characteristics, fund family information, detailed portfolio holdings, and macroeconomic variables, creating a far richer picture than previous studies had been able to assemble.
"It's one of the most comprehensive datasets on bonds and bond funds that has ever been put together," Van Nieuwerburgh says.
Machine learning made it possible to analyze this enormous volume of information, uncovering relationships that would have been difficult to detect using traditional statistical techniques. The researchers designed the study to test whether those relationships held up on data the model had never seen before — a crucial safeguard against simply finding patterns that exist only in hindsight.
The results were striking. The researchers found that roughly the top 10% of bond fund managers consistently outperformed after adjusting for market risk, while the bottom 10% consistently underperformed. In fact, the weakest managers proved even easier to identify than the strongest ones. A strategy that invested in the predicted top-performing funds while avoiding the predicted worst performers generated meaningful excess returns throughout the nearly 30-year sample. And those performance differences persisted for years rather than disappearing after a few months.
Perhaps the study's biggest surprise was what didn't matter. The characteristics of the underlying bonds — their maturities, yields, liquidity and other features — turned out to provide relatively little information about manager skill once broader market movements had been accounted for. In other words, says Van Nieuwerburgh, “You can’t predict the funds by the assets they hold.”
Instead, the strongest predictors came from the funds themselves, particularly their history of generating superior risk-adjusted performance.
Rethinking active management
The findings don't overturn decades of research suggesting that consistently beating the market is extraordinarily difficult. Most active bond managers still fail to generate persistent outperformance.
But they do suggest the picture is more nuanced than an either-or debate between active and passive investing. Genuine skill appears to exist, even if it's rare, and advances in data science are making it easier to distinguish that skill from chance.
For investors, that means evaluating an active manager may be less about scrutinizing today's portfolio and more about identifying evidence of a repeatable investment process. That said, according to Van Nieuwerburgh, “Many investors might be better off with cheap, passive alternatives like index funds or ETFs. That way, they avoid the 10% of funds that underperform, and they don’t have to pay fees typically associated with active management.”
The findings arrive as millions of Americans approach retirement and begin shifting a larger share of their portfolios from stocks into bonds in search of income and stability. For many of those investors, choosing between passive and active bond funds will become an increasingly important decision. While the study doesn't settle that debate, it does suggest the question may no longer be whether skill exists, but how to recognize it before it shows up in the returns.
