The Promises and Pitfalls of Robo-Advising

GAICIL Insights

Every technology has a customer it helps and a customer it merely entertains.¹ The Promises and Pitfalls of Robo-Advising, by Francesco D'Acunto, Nagpurnanand Prabhala and Alberto G. Rossi, identifies both for robo-advice. Robo-advisers are becoming ubiquitous, and their effects on investors are not uniform. Using a unique sample of individual brokerage accounts, the authors study the introduction of a wealth-management robo-adviser that constructs portfolios tailored to investors' holdings and preferences.

Something for wealth platforms to sit with: segment the marketing by mistake, with net worth a secondary concern. The investor who needs the robot most is the one holding a poorly diversified portfolio, and that investor may be the hardest to reach.

The picture is nuanced. Adopters resemble non-adopters in demographics and prior interactions with human advisers, yet tend to be more active and hold greater assets. Adopters experience diversification benefits. Ex ante undiversified investors increase stock holdings and hold portfolios with less volatility and better returns. Already well-diversified investors hold fewer stocks, see some reduction in volatility and trade more after adoption. All investors increase attention, measured by online account logins.

The conclusion sharpens the contrast. Underdiversified investors improve diversification in both number of stocks and market-adjusted volatility, with higher market-adjusted trade and portfolio returns. Highly diversified investors keep diversification unchanged and trade more, and the extra trading does not translate into better performance. Behavioral biases, including the disposition effect, trend chasing and the rank effect, are the only outcome that improves for all investors, although they do not disappear.

The promise is real. A portfolio optimizer that makes acting on advice simple and immediate helped underdiversified investors achieve better outcomes and reduced well-documented biases across the board. That is precisely what artificial intelligence tools in finance are meant to deliver.

The pitfall is the interesting part. Sophisticated, well-diversified investors gained little and may even show lower fee-adjusted performance after using robo-advisers. Algorithmic advice is most valuable where it corrects clear mistakes. For investors near the frontier, it may add costs or encourage trading without benefit. For the expert, the tool risks becoming a slot machine with a dashboard.

The design brief writes itself. Financial institutions should target underdiversified investors with robo-advising tools. Future research should explore optimal design of interventions tailored to different categories of investors. Algorithms that adapt recommendations to each investor's situation could capture the benefits and avoid the pitfalls, a natural application for more sophisticated AI that can personalize advice using detailed information on holdings and behavior. The best robo-adviser may sometimes tell an investor to do nothing.

The universal reduction in biases is striking. Even investors who gained nothing in performance became less prone to the disposition effect, trend chasing and the rank effect. Robo-advice appears to discipline behavior, an effect that may have long-run value as investors learn better habits.

Attention is a double-edged result. All investors paid more attention to their accounts after adoption. Engagement with one's finances is often a precondition for better decisions, and a tool raising attention may yield benefits beyond those captured in performance measures. For the already sophisticated, more logins seem to have meant more trades; for the novice, more logins meant learning.

The authors' verdict is balanced: despite its promises, robo-advising's pitfalls mean it is no panacea for individual investors' underperformance. That is the right stance for evaluating AI in finance broadly. The technology can deliver meaningful gains, and its effects depend on who uses it and how it is designed.

For regulators, the benefits of robo-advice are substantial for many investors, and disclosure of fees and expected benefits matters. Policies encouraging the spread of robo-advice to underdiversified households, while ensuring transparency for sophisticated users, would capture the gains documented here.

At bottom, algorithmic advice works best matched to need. For investors making clear mistakes, robo-advice improves diversification, returns and behavior. For those already well positioned, it offers less. As AI advisers grow more capable, tailoring will become easier and the gains could widen.