The fintech revolution is gathering pace, driven by improvements in efficiency, decision-making, and access to financial services. But fast finance also has many dangers, including security concerns and the risk of aggravating extreme market events. As a program leader at TSE Sustainable Finance Center, Milo Bianchi seeks to reveal the unintended consequences of algorithmic trading and other digital advances, and to harness their power to encourage responsible choices. His new working paper with Marie Brière, Head of Investor Intelligence & Academic Partnerships at Amundi Institute, examines the impact of using robo-advisors to provide investment recommendations. There is much to be gained, their study shows, if humans and robots work together.
Is it wise to rely on algorithms for financial advice?
Robo-advisors have low operating costs, which may allow them to reach a broader set of investors; they also use verifiable procedures, which may reduce biased advice. A fundamental question is whether these robots tend to complement, or rather to replace, investors' reasoning and actions. The extent to which investors keep an active role in their decisions appears to be a fundamental dimension when assessing whether and how robo-advisors can improve investors’ choices and promote financial capability.
However, defining the optimal degree of automation is not obvious. While reliance on algorithms seems particularly delicate in the context of financial services, evidence from other domains suggests that algo-aversion can be partly overcome by letting humans and robots interact. These interactions can also be useful for helping investors to learn how to manage their portfolios. At the same time, letting the human interfere may limit the effectiveness of the robot, undermining investors’ performance.
How do you study human-robot investment decisions?
We exploit the introduction of a robo-advising service by a large French asset manager in 2017. The robot starts by eliciting information on the client’s characteristics to build a profile and propose a portfolio allocation. Unlike many services that fully automate portfolio decisions, investors are free to ignore the advice. The robot also proposes rebalancing strategies over time, sending email alerts when the investor’s portfolio gets too far from the target allocation. These features allow us to study how investors’ responses to the robo-service evolve, as they experience market shocks or new investment opportunities arise. Such interactions are key to understanding the ultimate effects of the robot on financial outcomes.
What impact does the robo-advisor have on investor behavior?
Investors who use the robot do not view it as a substitute for their own participation. Instead, we observe an increase in investors’ attention to their portfolio, as measured by the time spent on the dedicated website. Robo-takers increase the number of connections on the platform by 0.3 per month, compared to the average of 0.8. Importantly, this effect persists beyond the time of the robo-subscription.
This increase in attention is associated with an increase in trading activities, and specifically in rebalancing activities that occur after the robo-subscription. In particular, our analysis highlights the effectiveness of the robot’s alerts as a tool to induce investors to rebalance their portfolio and stay closer to the target. We show that the alert increases investors’ attention and propensity to rebalance, and reduces the gap between current and target equity exposure by 4%, compared to an average gap of 11%.
How does use of the robot affect financial returns?
Investors who subscribe to the robo-service earn significantly higher risk-adjusted returns of between 2% and 4% per year. Changes in trading behavior at the time of subscription account for about 30% of this increase, but the most important part comes from dynamic changes in subsequent rebalancing over time.
We also investigate the financial costs of letting investors decide whether to follow the robot. Comparing the returns experienced by robo-takers and the counterfactual returns they would have experienced with fully automated trading, we observe a small average loss of between 0.04% and 0.1% per year. These costs depend to a large extent on rebalancing decisions. In particular, investors were significantly less likely to follow the robot’s recommendations during the bear markets of October-December 2018 and February-March 2020.
Is it time for robots to take over?
Our research offers grounds for optimism about human-robot interactions. We find that access to a robo-advisor induces investors to pay more attention to portfolios, increase trading activities, and results in higher risk-adjusted returns. These results are surprising in light of existing research documenting investors’ inattention to rebalancing opportunities, especially in long-term investment plans. Given that investors with less wealth and financial literacy tend to suffer more from such tendencies, the potential for robo-advisors to reduce portfolio inertia is encouraging.
Our effects are largely driven by the way investors change their behavior over time, suggesting that the robot can be used to improve investment decisions – by providing alerts, for example – while leaving investors as the ultimate decision-maker. With active human engagement, robo-advisors can be a powerful tool for improving education and financial capabilities.
FURTHER READING
Milo and Marie Brière’s paper ‘Human-Robot Interactions in Investment Decisions’ is available on the TSE website.
Article published in TSE Reflect, June 2023
