David Nanigian, director of the Personal Financial Planning Program at Cal State Fullerton's Mihaylo College of Business and Economics

David Nanigian

David Nanigian, director of Cal State Fullerton’s Professional Certificate in Personal Financial Planning Program and an associate professor of finance at Mihaylo College of Business and Economics, recently presented a paper, “Rating a Robo-Rater,” at CFP Board’s Academic Research Colloquium. The paper looks at whether or not a machine-learning model designed to emulate the decision-making processes of human analysts can do as good a job as humans when picking actively managed mutual funds for investment.

The study is specific to products provided by financial services company Morningstar Inc., which in 2017 launched its “robo-rater,” because it is not feasible for Morningstar to assign a human analyst rating to every mutual fund in existence.

“This provides a laboratory to gauge the value of ‘soft information’ in analyzing mutual funds,” says Nanigian in a recent article about his study published on Machine Lawyering, a blog of the Chinese University of Hong Kong.

Nanigian shares the significance of his paper and his findings.

What is the significance of your paper, “Rating a Robo-Rater”?

There are two practical implications of my working paper for mainstream investors. First, “soft information” about mutual funds matters when selecting an actively managed mutual fund. Examples of “soft information” include the theoretical soundness of the “investment thesis” that guides a fund’s portfolio management strategy and the health of a fund’s workplace culture. Second, a reasonable expense ratio should be one of the main characteristics to look for when searching for a mutual fund.

What were your findings?

There were two main findings from my research on Morningstar Quantitative Ratings. The first main finding is that the artificial intelligence (AI) rating is a much weaker predictor of risk-adjusted mutual fund performance than the human-based rating. Both ratings are designed to help investors select mutual funds.

The Morningstar Quantitative Ratings are generated by a machine-learning model, which I refer to as a “robo-rater,” while the Morningstar Analyst Ratings are assigned by human analysts employed by Morningstar.

The reason the Morningstar Quantitative Rating is a weaker predictor of performance than the Morningstar Analyst Rating is because “soft information” is not factored into the assignment of Morningstar Quantitative Ratings as the robo-rater, by its nature, cannot analyze “soft information.”

The second main finding is that all of the limited predictive ability of the Morningstar Quantitative Rating is attributable to the robo-rater’s assessment of fees. This confirms the findings of a large body of scholarly research that expense ratios matter in mutual fund selection. Therefore, investors should have an eye toward mutual funds with reasonable expense ratios. Morningstar’s free Basic Fund Screener is a simple tool that investors can use to quickly identify such funds.

More on Mihaylo Research

Mihaylo College is a leader in business-related research, in addition to its role as a teaching institution. Read more of our articles on the latest Mihaylo research.