The future of robo-advisors, part 2

I have recently written a post on new research that shows that investors are perfectly willing to adopt robo-advisors over human advisors. This willingness to adopt robo-advisors seems largely driven by the expectation that algorithms are more objective and should deliver higher returns. My caveat, though, was that the study had been done with students at a university and the average age of the participants was 22.8 years. 

Back then I said:

“In my experience, younger investors are much more open to machine-based investment technologies than older ones.”

It seems that guess was mostly, but not entirely correct. A couple of weeks ago a new research paper by Daniel Ben-David and Orly Sade from the Hebrew University in Jerusalem landed in my inbox. I owe these two guys beer because they addressed exactly the question I had about the adoption of robo-advisors: Are younger investors more willing to adopt robo-advisors than older ones?

They looked at both the readiness to adopt either a robo-advisor, a human advisor, or a robo-advisor paired with a human for three different age brackets. They also looked at the willingness to pay for these different forms of advice as a measure of how much investors trust these three models.

And their results are somewhat surprising. First, to the non-surprising part. Their research found that investors aged between 20 and 30 are somewhat more willing to use a robo-advisor than a human advisor. They are also more willing to use a combination of robo-advisor and human advisor than a human advisor. Investors in my age bracket (31 to 44 years old), on the other hand, are much less willing to adopt a robo-advisor than a human advisor or a robo-advisor paired with a human. For investors aged 45 and above, there was no significant difference between robos and humans.

Of course, the willingness to adopt one form of advice or another is only part of the equation for a business. It provides an indication of how many clients one can attract with either a robo-advisor or a human advisor business model. The other part of the equation is the willingness to pay for the advice. And here the unsurprising outcome is that younger investors, aged 20 to 30, are not only more willing to adopt a robo-advisor but also willing to pay more for the advice. The difference in willingness to pay for robo-advisors and human advisors in that age group is statistically significant. Similarly, there is a statistically significant difference in willingness to pay for human advisors by investors aged 30 to 44. In the eyes of those investors, humans provide a more valuable service than algorithms.

But here is what surprises me. According to this study, the oldest investors aged 45 and above in this study were willing to pay about as much for robo-advice as for human advice. Only the combination of robo and human seemed worth less to them. This doesn’t seem to make sense. The younger age group in the study clearly trusted algorithms more than humans, while older investors clearly trusted humans more than algorithms. This can be explained by the fact that investors aged between 20 or 30 are digitally natives and grew up in a world that was already dominated by the internet, computers and all forms of digital information. Older investors, on the other hand, are immigrants to the digital world of today. We still remember the days before the internet. But so are investors aged 45 and older. Why should the older investors trust an algorithm the same as a human? And if we accept that, then why do they trust a combination of algorithm and human significantly less than each one alone? 

To me the results shown in the chart below for investors aged 45 and above seem intrinsically contradictory. Either they trust algorithms more than humans, in which case the willingness to pay for human advice should be much lower and we need to explain why this generation of investors resembles more the 20-somethings than the demographically more similar 30-somethings. Or they trust humans more than algorithms, in which case they should be willing to pay the algorithm much less than the human advisor and this age group would basically resemble the age group of the 30-somethings. My own biases lead me to doubt the results of the oldest investors in this study. It seems that either the willingness to pay for robo-advice or for human advice is a spurious result. But I cannot prove it and thus, all I can do is hope that someone will do additional research on the relationship between age and the willingness to pay for robo-advice and human advice.

Willingness to pay for advice by age group

Source: Ben-David and Sade (2019).