How investors learn from experience

by Philip Newall & Leonardo Weiss-Cohen
Posted on May 28, 2019

How can personal investors achieve the best possible long-term returns given the amount of risk they can tolerate? Financial theory dictates clearly that they should diversify broadly, keep their investment costs low, and avoid reacting too frequently to recent market events.

It has, for instance, been calculated that a reduction of 0.1% in fees paid by US investors on their mutual fund assets would lead to collective savings of $8 billion in a single year. This advice has been provided to investors in relatively jargon-free form for many years, for example across 12 editions of Burton Malkiel’s book A random walk down Wall Street, originally published in 1973.

No guarantee for the future

And yet only few investors follow the recommended strategy for achieving the best possible long-term returns. Many investors mistakenly extrapolate high past returns into the future. Think of the infamous Dot.Com stocks, subprime real estate, and cryptocurrencies – all of which led to spectacular losses after high initial returns over the last two decades. The question, then, is how to help investors care less about past returns, and get them to focus on buying-and-holding a low-fee portfolio instead?

You might think that helping investors improve their financial literacy is one obvious way of reducing the frequency of any financial error. But the evidence refutes this idea. One large meta analysis (a study of 201 prior studies) found that interventions to improve financial literacy had very little positive effect on downstream financial behaviours. Another study found that many financial advisers gave clients bad recommendations to invest in assets with high past returns – not out of self-serving financial interests, but because they really believed in their own advice, following the same strategies in their personal accounts!

“To help investors make better decisions, we should first understand how investors learn from experience.”

Diving into our psyche

The same study also found that more experienced advisers tended to advise to invest in assets with high past returns more frequently than their less experienced colleagues. This last study also leads on to the unique hypothesis underlying our current research: to help investors make better decisions, we should first try to understand how investors learn from experience.

This point might seem obvious, but it has actually been ignored by the majority of academic studies that aim to help investors achieve better returns. Many experiments aiming to test better ways of presenting useful information to investors use one-off hypothetical scenarios. These one-off scenarios cannot shed light on how investors learn with experience, because investments are repeated ongoing decisions. One-off scenarios also ignore a large psychological literature showing that repeated and on-off choices can yield surprisingly different patterns of behaviour.

Presenting our experiment

Our first experiment is now ready for take-off. In line with current best practices in psychology, we are recording how we will analyse the data before any participants have taken part. Individual US investors will take part in our study. They’ll get paid a basic amount for taking part, and can also earn extra money based on the returns their investments yield.

Unlike most previous research, participants will make repeated investment decisions across 60 months of simulated historical market returns. Each simulated month’s returns will be presented to participants, allowing us to analyse how they learn to invest, as their returns vary over time. The level of investment fees is the only other factor varying across the two funds presented to participants, providing a rational factor to base their investments on (lower fees are better).

In total we will run two similar experiments. The first experiment will test the current advice given to investors: “Past performance does not guarantee future results.” An earlier one-off scenario study found that this advice did not help, compared to giving no advice at all. Our large sample of participants performing an arguably more realistic task (involving repeated investment decisions) will allow us to better test the advice which is currently given to investors. In the follow-up study we will test the best-performing advice found in another one-off scenario study: “Some people invest based on past performance, but funds with low fees have the highest future results.”

How will our participants learn over time? And what will we, in turn, learn from them? We’ll keep you posted.

Philip Newall is a postdoctoral researcher at the University of Warwick.

Leonardo Weiss-Cohen is a postdoctoral researcher at City University London.