Effecting behavior change in a world of automated financial advisors

Many judgments and decisions today are increasingly being made by (or with the help of) smart machines—computer algorithms that employ codified knowledge, understand and generate language, and learn over time. This development offers great potential advantages in terms of decision quality, productivity, and other benefits, although it has created considerable anxiety about potential job loss. One aspect of more automated decisions that few observers have addressed is the potential for human workers to focus on behavioral and change issues.

In a variety of different domains, “getting the right answer” is an objective that will often be best obtained with the help of computers. Data, analytics, and heuristic rules can often produce a high-quality answer to any question for which there is codified knowledge. Indeed, if there is considerable data and expertise to be absorbed, and time pressure to make a decision, it is often impossible for humans—even the most expert ones—to make the best possible decision. There is simply more data and analysis required than the human brain can muster on short notice.

However, even if judgments and decisions are being made by machines, it is usually humans who have to act on them—or not. You may have heard about “behavioral economics,” which is based on the revolutionary (if seemingly obvious) principle that humans routinely deviate from the principles of economic rationality that until recently were considered sacrosanct.¹

We humans, equipped with less-than-rational mental capabilities, have to decide whether to seek out machine advice in the first place, and to accept it once provided. We may have available, for example, high-quality driving direction advice from GPS mapping devices and apps, but many of us fail to consult these maps in the first place, or override their recommendations and become lost. We also have less-than-ideal self-control. For example, being provided with personalized diet and exercise recommendations is typically not enough to prompt the needed actions. Analytics and algorithms will often provide the best answer, but humans still must act upon it and change their behavior. One of us has referred to this as the “last mile problem” of predictive analytics.²

One area in which both automated advice and poor human behavior are both present in spades is personal financial investing. In the remainder of this essay we’ll use that important industry to illustrate both the challenges that human behavior presents to personal investing (often referred to as “behavioral finance” in the investment industry) and the opportunity for humans to address the problems that smart machines alone can’t address.

The rise of the robo-advisor

The recommendations traditionally made by personal financial planners and brokers—specifying what financial assets a client should invest in—are increasingly being made by computers. This trend even has a name—the “robo-advisor.” These tools, which are offered by both start-ups and well-established asset management firms, provide basic personal investing advice with regard to asset allocation, matching portfolios to risk preferences, and end-of-year rebalancing and tax loss harvesting. In most cases they recommend not individual stocks and bonds, but broad-segment (for example, S&P 500) mutual funds and exchange traded funds ETFs.

These decisions aren’t rocket science, and they are well within the capabilities of intelligent machines today. They can certainly handle more data and can provide more personalized recommendations than the typical human brain. They are also less expensive than the typical human brain; robo-advisors typically cost about 0.25 percent of assets, whereas a human advisor can often charge 1 percent or more of assets.³

Most robo-advisors only recommend investing actions, however, and don’t carry them out in an automated fashion. Even assuming that the investor has decided to engage with a robo-advisor in the first place, this leaves the “last mile” to be conquered: persuading said investor to take the recommended actions.

That is a considerably more difficult problem than determining what stocks and bonds to buy. The area of personal financial investing has long been known as one that is rife with problematic behaviors; it is, according to the behavioral insights classic Nudge,⁴ an example of a “fraught choice.”  Fraught choices are complex and difficult for humans in that they require specialist knowledge, are made infrequently, do not have immediate feedback, and have important effects that are only experienced in the distant future. It’s easy to see why investors so often make bad decisions like buying high and selling low, or underinvesting for retirement.

Robo-advisors and behavioral interventions

Even if some financial decisions are made well by robo-advisors, investors will almost certainly still need some help in adopting and maintaining responsible investing behaviors over time. What’s to keep them, for example, from panicky selling when there is a substantial drop in the stock market? What will prevent them from investing when the media (and taxi or Uber drivers) are talking up a rising market? Some other common ways that investors make poor decisions include “loss aversion”—caring more about not losing a dollar than gaining a dollar—and “familiarity bias”—being more willing to invest in familiar assets, like the stocks of companies in their home country, than those in companies they’ve never heard of.

The good news is that several of the firms that have adopted robo-advice have realized that there is a need for behavioral change. Some of the impetus for correct investing behavior is built into the automated advice itself. At Betterment, for example, one of the larger and more successful start-ups in the robo-advisor space, there is a “behavioral finance and investing” department comprising five experts focusing on how to improve their system’s investment advice, determining the right asset allocation, changing investment management strategies over time, and “behavioral design”—trying to ensure that Betterment customers display rational economic behaviors with their investments. For example, the company’s algorithms attempt to discourage such irrational behaviors as active trading and market timing. Betterment has a substantial amount of advice on its website as well, as does Wealthfront, another robo start-up. In one blog post, the Wealthfront chairman notes: “Despite how much we focus on fees in this blog, bad behavior is the single biggest destroyer of long-term returns for the average investor.⁵

A role for humans?

It’s difficult to truly influence behavior with blog posts alone, however, and some traditional investment firms are both adopting robo-advice and preserving a human role for behavioral interventions. Vanguard Group, for example, traditionally offered human advisors in its asset management business. Now, however, it has added some automated advice to these human capabilities in a “hybrid” offering called Personal Advisor Services. Not only does the Personal Advisor Services arrangement equip advisors to give better advice, it also gives them the capacity to serve more clients. By removing the burden of manual calculations, Vanguard enables them to focus more of their time and attention on the empathetic coaching that is their forte—and the lower fees charged for partially automated advice means more clients can have access to it.

With a lot of the basic investing decisions and information transmittal tasks now being handled by a machine, Vanguard executives felt that human advisors would have more time to work with clients on important financial behavior issues. The company thus embarked upon a strategy to equip its advisors with more behavioral coaching abilities. According to Karin Risi, Vanguard’s head of Personal Advisor Services,

The new system gives advisors more freedom to interact with their clients. Many of them are now using face-to-face video for these meetings, since all of the informational details are in the system. The advisors are also doing behavioral coaching when they interact with their clients—it’s very common, for example, for them to be a voice of reason when clients want to get out of the market in a downturn. Some of our clients turn to advisors for help because they know they lack the discipline to contribute steadily and take a long-term approach. It’s not unlike using a personal trainer to help you exercise.⁶

Like the robo-advisor start-ups, Vanguard also incorporates behavioral finance approaches, whenever possible, into the system itself. It tries to gently nudge clients, for example, into increasing their 401K contributions.

These hybrid offerings suggest new roles for humans in a world in which many decisions and judgments are made by machine. This could apply to a wide variety of individual and organizational decisions and actions. As in personal financial investing, perhaps humans can focus on the psychology and psychiatry of behavioral change—understanding customers’ (sometimes irrational) behaviors, persuading them to stay the course, and even reconciling the diverse orientations of a married couple. It may be that machines will never be able to take on such roles simply because they are not irrational enough to understand our behaviors. Finally an area in which humans have a competitive advantage!