story

Robo-Advisors and Household Finance: Promises and Risks

Read more about this CEPR-TFI event and watch the videos
Posted on October 18, 2018

A substantial part of household financial decision making is driven by factors that deviate from the standard textbook models. In the traditional world, when individuals had the levers to not only choose how much they spend and save or how much they invest in markets but also make active decisions regarding asset allocation and rebalancing, two important problems emerge. First, households systematically disregard the need for diversification and thus make suboptimal decisions such as holding concentrated portfolios. Second, households also tend to trade too much and not follow any rational rules of rebalancing, thus losing money to transaction costs.

A growing body of evidence has suggested that households may not be equipped with the necessary financial skills to make optimal decisions. Attempts to improve financial literacy have only given rise to mixed results; and letting households learn by “trial and error” have only added to the welfare costs on households in the long run. Thus, the world of financial advice was born. The consensus was that advice from skilled professionals who are appropriately equipped with the skills to make more sophisticated decisions on behalf of households will improve household financial outcomes.

“Money Doctors”

The revolutionary step to create “money doctors” who advise households and help with deciding their finances has not been without its own problems. A large and growing body of evidence dampens the promises of financial advice with a two-fold demand and supply problem. Those households who typically need financial advisors do not seek them, and even when they seek them, they do not always follow the advice given to them. This is akin to people avoiding going to a doctor — either because they do not know they need to, or because they consciously choose not to — or a doctor's prescription going unheeded by not purchasing the medicines from a pharmacy.

On the supply side, financial advice added additional layers of complexity in retail consumer finance. Those who provided advice were also those who were agents of firms that produce financial products, thus presenting a conflict of interest. When advisors were only paid by households who they render advice to, prices for a professional financial advisor exclude a large fraction of non-wealthy households who need such support. Adding to these concerns, advisors have their own biases, tend to provide canned solutions, not provide sufficient customized recommendations to households, thus, raising concerns about their true value to households.

One consensus approach to solving a lot of these problems is financial technology. Financial technology, in the form of automated financial advisory services or known popularly as “robo-advisors”, allows for economies of scale and customization, does not have supply constraints in the number of households it can reach, reduces costs and prices to reach more number of households, and importantly can help address behavioral biases on the demand side by automating decisions on household finances.

In this setting, the Think Forward Initiative and CEPR invited an eminent group of thinkers on robo-advice to discuss the promises and risks of such an approach, from academia, policy institutions and the industry - Christopher Carroll, Professor of Economics, John Hopkins University, Dan Egan, CEO at Betterment, Francesco D'Acunto, Assistant Professor of Finance at Boston College, and Karen Croxson, Head of Research & Deputy Chief Economist at Financial Conduct Authority (UK).

“A human advisor can reach about 10,000 people a year, but automated advice services have the potential to reach many more individuals.”

Promises

The first promise of robo-advisors, as Dan Egan called it, is the offer of low costs. This is important, especially in a context where many households cannot afford to save large sums of money in one go and yet need support in their financial lives. A related potential benefit is that of scale. If a human advisor takes 15 minutes per client and works 10 hours a day, and 5 days a week, that is still reaching out to only about 10,000 people a year. However, automated services have the potential to reach out to many more individuals, thus improving access to professional support to several more households than the human advisor framework may possibly manage.

The second promise of robo-advisors is that they can be without any conflict of interest, and be solely an agent of their clients and not those who manufacture financial products. This technology-driven approach allows for complete audit trails, with no black-box type approach while generating advise for households. Retrospectively, Dan argued, this allows for complete regulatory compliance that can be audited in real time and thus improve the decision-making environment.

The third promise of robo-advisors is no small feat. Often, the problem of generating customized advice relies heavily on generating optimal solutions to problems with hundreds of different constraints that households face in their financial lives. In Dan's world at Betterment, this would be a “300-dimensional optimization problem” that no human advisor can achieve easily. Such an approach is important as the goal often is, for instance, to keep people in the market long enough for them to reap the benefits of investing in financial markets.

In adding the researchers’ view, Francisco D’Acunto of Boston College argued that an additional promise of robo-advisors is a marked reduction in the behavioral bias driven financial decisions in markets. In summarizing his research contributions to the field, he argued that households can achieve high diversification and may be encouraged to think before they act, thus reducing the behavioral traits otherwise observed in their investment decisions.

While talking about the potential gains from robo-advice from the point of view of regulators, Karen Croxson of the Financial Conduct Authority argued that robo-advice could close the “advice gap” in the market by lowering the cost of advice, be available all day, every day, provide fast advice and solutions, provide tailored support for the financially vulnerable, and present much better audit trails for easier monitoring for the regulator. These benefits, coupled with the fact that regulators are motivated by technology agnostic principles that envelope the whole market for advice, have the potential to create a transparent market for better household decision-making.

“"Algorithms need to encode the principles of consumer protection in its functioning"”

Karen Croxson, Head of Research & Deputy Chief Economist at the FCA

Risks

These promises do not exist without challenges with an increasingly algorithm driven advice market. In highlighting the potential risks due to robo-advice, Karen Croxson suggested that algorithms need to encode the principles of consumer protection in its functioning. For instance, the principle of suitability states that any resulting financial decision from advice provided must be “suitable” to the household — a check that algorithms can naturally attempt to address. However, such assessments require a broad, complete information set about households, which are harder to obtain, thus, making algorithms work on partial information sets that may lead to algorithmic bias. Karen also highlighted the potential threat of systematic mis-selling, system glitches and its effect of trust in financial markets, and data breaches and violation of privacy as some additional concerns that need to be addressed to bolster the gains from robo-advice.

The role of the robo-advice industry, in the long run, may result in making it a systematically critical infrastructure in retail financial markets. Dan argued that there is a need to understand whether firms in this industry are providing independent assessments to households, or whether they are digitized platforms for specific brokerage houses. The former has important value addition to households, whereas the latter may lead to exploitation of behavioral biases to generate revenue. The principle of independence — just as with the detailed regulatory framework for differentiating marketing and agent sales from human financial advice — may be critical for the quality of this industry going forward, and regulators may need to assess how to differentiate market players who are digitized algorithmic platforms for specific brokerages, as opposed to independent advisors to households unrelated to any other manufacturer of financial products. The promises of robo-advice, while plenty, need to be tempered with a dose of realistic solutions to problems that may potentially come alongside these benefits.

“Curious to what Dan Egan, Karen Croxson and Francesco D'Acunto said? Watch the videos below:”