Why Factor Investing Isn’t Working

Pension funds and endowments aren’t getting the factor premiums they desire — just uncompensated risk.

Illustration by II

Illustration by II

Asset owners have increasingly incorporated factor-based funds into their portfolios in an attempt to earn equity premiums at a low cost. But they are ending up bearing currency, country, sector, and other risks that are going uncompensated, according to an analysis of 500 complex institutional portfolios by Northern Trust Asset Management.

NTAM has analyzed more than 500 portfolios over the last five years and has evaluated more than 180 portfolios since 2017 to determine whether institutions were earning sufficient excess returns for the active risks they were taking and the fees they were paying. Public and private pensions represented about two-thirds of the portfolios, with the remainder being superannuation funds and endowments.

The asset manager found that institutions weren’t getting the results they wanted for a number of reasons. First, targeted factor exposures are being offset by the other managers held in asset owner portfolios, creating index-like outcomes at higher fees. Second, many endowments and retirement plans have invested in poorly designed factor-based funds, which don’t efficiently deliver the premiums that investors are counting on.

NTAM also found that portfolios are often stocked with so many managers with similar holdings that the potential for outperformance is essentially diversified away.

“The whole exercise of factor investing is delineating between the risks you get paid for and the risks those you don’t,” said Michael Hunstad, head of quantitative strategies at NTAM. “The challenge is you want to define the value, quality, or other factor in such a way that you get the good stuff and eliminate all the uncompensated risks. That’s really hard to do.”

[II Deep Dive: Smart Beta Is Coming for Active Bond Fund Managers, Study Shows]

Sponsored

Hunstad noted that there are decades of academic research that offers evidence that a handful of factors can generate higher risk-adjusted returns than the market. However, this research “doesn’t tell you how to extract that premium in the best possible manner,” he said. “No way.”

For example, famous research on the sources of stock returns by University of Chicago Professor Eugene Fama and Dartmouth’s Kenneth French in the early 1990s used the price-to-book measure to define a value stock. If an allocator today used that definition, however, the portfolio would have be significantly overweight financial stocks, said Hunstad, who added that the risk of financials in certain periods would overshadow the value premium.

“I believe in factors, but the implementation is horrid throughout the industry,” Hunstad said.

However, eliminating uncompensated risks is not easy, according to Hunstad. Such hidden risks are one of the reasons why systematic funds that are all called “value” or “momentum” can vary widely in terms of risks and performance, despite targeting the same factor premiums.

NTAM has developed what it calls a factor efficiency ratio to measure a factor fund’s exposure to uncompensated risks, which it believes are behind the wide disparity in performance in the smart beta world. A fund with a high factor efficiency ratio would give an investor high exposure to the factor they want and would include fewer uncompensated risks.

In separate research that was published in the Journal of Index Investing in 2015, Hunstad and another Northern Trust quant researcher evaluated 21 of the largest factor-based strategies available at the time, and analyzed how much of these strategies’ active risk was being used to target the particular factor — the compensated risk — and how much was uncompensated risk. “Our analysis suggests the vast majority — 83 percent — of active risk taken by smart beta strategies or funds is unintended/extraneous; therefore, investors are only getting compensated for 17 percent of the risk they are exposed to,” said Hunstad. He noted that while the research is a few years old, the point is still valid.

Related