There’s been much talk about the growth of automatic financial advice, or robo-advisors, but we now have scientific evidence of how real investors actually use it. This shows that it has both benefits and drawbacks.
Economists at the University of Maryland studied the investment performance of clients of an Indian stockbroker which offered a tool which used mean-variance portfolio optimisation to recommend shares which investors could buy simply by pressing a key. They found that clients who held only a few shares before using the tool diversified better and so made higher risk-adjusted returns. They also found that diversified investors who used the product went on to hold fewer shares with no effect upon returns.
This tells us that robo-advice can reduce both under- and over-diversification.
What’s more, such advice also reduced – although did not eliminate – some of the more common investment errors such as the disposition effect, the tendency to sell winning stocks too soon while holding onto losers in the hope they’ll break even.
Robo-advice, then, has obvious benefits.
But it also has a drawback. The economists found that sophisticated investors who used it tended to trade more often to no useful effect, thereby incurring excessive dealing costs.
This is consistent with the idea that there can be an illusion of knowledge. Having what seems to be a clever advisor can give people the impression that they know more than they in fact do, which emboldens them to trade more often.
Economists at the University of Mannheim have found a similar thing in a different context. They’ve found that knowledgeable investors buy poorly performing expensive actively managed funds because they believe – wrongly – that their expertise enables them to spot good fund managers.
What both groups of investors forget is that even the best knowledge gives us only very limited foresight simply because the future is to a large extent inherently unknowable.
All this suggests that robo-advice is most useful for less knowledgeable investors or potential investors, not least perhaps because it might give them confidence to invest in equities in the first place. It gives them real knowledge, but to sophisticated investors it gives only an illusion of knowledge. Robo-advice has diminishing returns.
Here, though, lies an awkward paradox.
On the one hand, there’s a potentially big market here. A recent study by Deloitte says there’s “a rich opportunity for automated advice” created by the high cost of human financial advice and a mistrust of it caused by past mis-selling scandals.
On the other hand, though, it’s difficult for robo-advisors to reach less knowledgeable investors and to profit from them. For one thing, they pay little attention to financial affairs which means they don’t respond much to advertising. And for another they have only small sums to invest which means that even cheap automatic financial advice might not be profitable. For these reasons, robo-advisor Nutmeg is still making a loss and the actual market for robo-advice in the UK is still small.
This raises an important but under-appreciated problem in economics. Markets do not emerge spontaneously. It takes effort and money to bring buyers and sellers together even when the buyers have a good product and the sellers need it. This is why platform businesses such as Uber and Twitter spent years losing money: Richard Schmalensee’s and David Evans’ book, Matchmakers is good on this problem.
In his book, An Engine Not a Camera Donald MacKenzie quotes Leo Melamed, the chairman of the Chicago Mercantile Exchange in the 1970s and 80s describing the creation of equity index futures. A market, says Mr Melamed, “is more than a bright idea. It takes planning, calculation, arm-twisting and tenacity to get a market up and going”.
Even if robo-advice is a good thing for thousands of new smaller investors, therefore, it might well not spread as quickly or widely as it should.
In this sense, robo-advice suffers from the same problem as many other potentially good financial innovations such as the macro markets proposed 25 years ago by the Nobel laureate Robert Shiller. Good products might not find a market simply because of the difficulties of creating such a market. (The converse is also true: bad products such as fixed-odds betting terminals can become pervasive if they are cheap to introduce). Robo-advice, like macro markets, might generate a lot of consumer surplus for thousands of people. But that’s not enough to bring them into being; what matters is that the consumer surplus be monetisable by firms and this is not always the case. .
Healthy markets in good products are in part a public good: private incentives alone might be insufficient to get them going. In theory, this problem could be overcome by intelligent government support. In practice, of course, this won’t happen.
