|  
  |  

Evaluation White Paper

Financial Advising

Financial literacy is attained a number of ways during an individual’s lifetime. Many adults rely on learning-by-doing (and paying for mistakes made along the way) while others seek out financial education and advice from other non-professional sources. For example, some adults turn to social networks (Chang, 2005) for financial advice and information. While friends in social networks can provide some information, many people may not fully trust this information or are uncomfortable sharing financial data with others.

An alternative to using social networks for financial information is to consult with a professional advisor. Approximately 28 percent of adults have used a financial planner according to the Certified Financial Planner Board of Standards (PSB LLC, 2010), which means planners can be an important source of financial education. As described by Collins (2012a), a financial planner or advisor can be helpful in many ways, such as providing information, defusing biases to avoid mistakes, helping clients think about issues, and dealing with emotional concerns. Advising, of course, is an educational process so some advisors will be better than others at providing sound advice, and an assessment of the quality of advisors can be difficult.

Financial advising can be relatively labor-intensive, making it costly for many consumers. To reduce this cost, many financial firms use automated systems to provide financial education and advice. These programs often include rudimentary financial training, followed by advice which is generated by an algorithm. The client is encouraged to act on the advice. Because the education is closely linked to financial decisions, the quality and delivery of the education can significantly affect the lives of the consumer.

10A. Key Programs and Resources

According to the U.S. Bureau of Labor Statistics (2016), approximately 250,000 people were employed as personal financial advisors in 2014, with growth of 30 percent expected by 2024. The financial education and advice provided by these individuals is theoretically tailored to the needs of the individual clients. In that regard, it is difficult to identify a specific curriculum. The CFP Board outlines eight categories of knowledge needed for certification of its members: professional conduct and regulation, general financial planning principles, education planning, risk management and insurance planning, investment planning, tax planning, retirement savings and income planning, and estate planning (CFPBS, 2015).

For investment decisions, an alternative to personal contact is some type of automated advice, which is provided by a large number of firms. Most of the automated firms use the terms “automated” or “algorithms” in the description of their services. Some services are fully automated, while others include periodic human review. The following are some the larger firms, with some firms offering both automated as well as personal advisor services: (1) Wealthfront (“an automated investment service”); (2) Betterment (“letting our advanced algorithms do all the work”); (3) WiseBanyan (“we use the power of algorithms and software”); (4) Charles Schwab Intelligent Portfolios (“service that builds, monitors, and rebalances your portfolio”); (5) FutureAdvisor (“diligent algorithmic monitoring that auto-rebalances your accounts”); and Blooom (“we will use our algorithm to…identify the right investments”).

10B. Major Topics and Literature Review

Financial Advisors

The literature on financial advice includes both theoretical and empirical studies (Inderst & Ottaviani, 2012). One topic for study is the characteristics of people who seek financial advice. Calcagno & Monticone (2015) provide a theoretical model in which consumers choose between delegating the choice of investment to an advisor, seeking advice from the advisor, or selecting their own portfolio choice. Consumers seek advice from an advisor who has better knowledge on the assets but who has incentive to sell their services. The model shows that more knowledgeable consumers receive more financial information from advisors. They concluded that financial knowledge and good advice are complements, not substitutes. The model also suggested that those with less financial knowledge are more likely to delegate their choices to an advisor. They concluded that “there is a scope for various types of financial education initiatives, such as financial education initiatives targeted at the population groups with the highest private costs for accessing financial knowledge, for rules that reduce the conflicts of interests between clients and intermediaries, as well as the subsidization of independent advisors” (p. 364).

Empirical evidence confirms the theoretical work. Calcagno and Monticone (2015) found that higher financial literacy (measured by eight questions) increases the demand for advice and the demand for holding risky assets. Collins (2012) also found that financial knowledge increases the demand for advice on investing, saving, mortgage, insurance and tax planning. However, financial knowledge was inversely related to debt planning. He concluded that financial literacy and advice are complements as opposed to substitutes. Finally, Robb et al. (2012) used the FINRA National Financial Capability Study dataset to confirm Collins results. They also note the fact that more research is needed to “assess to what degree the positive effect of financial knowledge reflects a cause or consequence of advice on saving, insurance or tax.”

Mullainathan et al. (2012) provided the most comprehensive evaluative study of the information given to consumers by professional advisors. The study used an audit-style methodology in which trained auditors were sent to financial advisors with one of four portfolios that were randomly assigned. The study sought to see to what extent low fee diversified portfolios were recommended. The study found that advisors did not correct inaccurate biases that their clients had. Advisors would often encourage biases that were favorable to the advisor (in terms of fees collected) and tended to push clients to actively managed accounts with higher fees. Gine et al. (2014) also conducted an audit-style study of bank branches near Mexico City. While the auditors sought loan or savings products, not advice on investment portfolios, the study also found that consumers with more financial knowledge were given more information about the products offered.

Automated or Robo-Advising

Recently, automated advising has come under increased scrutiny by regulators (SEC, 2015). At the core of these concerns is whether the consumer understands the underlying assumptions behind the model and the implications for the consumer if the assumptions of the model are incorrect. Another issue of concern is whether consumers understand the reasons for information sought by an investment tool and how that information will impact the output of the investment tool.

Consumers may have too much confidence in advice provided by automated services. They may incorrectly assume that the use of algorithms will (a) automatically generate higher returns (b) yield results that are consistent from one tool to another and (c) offer recommendations that are free from bias. A recent study by FINRA (2016a) documents these concerns numerically. The study reports the results from seven tools with the optimal percentage of equity assets recommended for a hypothetical consumer ranging from 51 to 91 percent. The report notes many consumers may not realize that bias exists in that the tools do not necessarily choose from the entire universe of investment opportunities, but are generally limited.

The FINRA report highlights the sophisticated financial knowledge required of consumers to evaluate the advice given by automated advising programs. Requirements include an understanding of what should be asked by the program in order to be properly advised (it is not clear how the naïve consumer is to know if something is omitted), understanding how rebalancing and tax strategies are performed, as well as knowing when conflicts of interest might arise. In sum, the need for financial knowledge is increased with the use of these programs. The degree to which the automated programs provide explanations on why specific information is gathered, the theoretical basis of the algorithms, and why the algorithms recommend a particular strategy has not been well documented. As with human advisors, more information is needed on how financial knowledge drives the demand for these products and what consumers are taught (explicitly or implicitly) through the programs and their use.

Little empirical evidence exists in the academic literature (the term “robo advisor” returns only one result on EconLit, for example) in terms of examining the educational experience provided by the automated advisors. The popular press has provided more comparisons (see Nerd Wallet, 2016; Moyer 2015a; Moyer 2015b). These comparisons focus primarily on features and little on the information required to open an account or the information shared with the consumer.

10C. Evaluation Practices, Strengths and Limitations

The evaluation of the relationship between financial advising and educational outcomes has been examined with two primary methodologies, surveys and audits. Surveys assess consumers who have used an advising service to see if any changes in financial knowledge, attitudes or competencies are measured. Surveys are relatively easy to administer and ultimately provide information on the correlation between professional advice (automated or otherwise) and financial knowledge. The inherent problem in survey studies is that causality is difficult to show. In general, most surveys are not longitudinal, so whether financial education drove the consumer to the advisor or whether the advisor educated the consumer is difficult to ascertain.

An ideal survey study would follow a cohort of consumers over time and examine those who consulted with advisors over the period of the study and those who did not. Of course such a study still faces the selection problem of people choosing to go to an advisor. A more controlled study could randomly assign consumers seeking advice to either a seminar experience or sending them to financial advisors. Measuring learning and behavior before and after different types of interventions could provide the data needed to compare the two financial education delivery methods. Measurable outputs could include responses to financial literacy questions or decisions made after the intervention. Of course, this style of study would be relatively expensive to administer.

The auditing style of evaluation, where an auditor plays the role of an individual seeking advice, while also costly, can provide a richer understanding of the educational process itself. For example, an audit targeted more toward assessing financial education could assess whether consumers were told about trade-offs between risk and return or the benefits of diversification. The audit can measure the time spent by advisors educating their clients and whether the advisor checks for understanding.

Although auditing is costly in evaluating human advisors, it is considerably easy with robo-advising because auditors do not have to physically travel to the agencies. Another advantage is that the “visit” is identical, so multiple auditors can be used to control for reviewer reliability. A rubric can be easily created for these audits as compared to live visits, and records of the visits could be easily kept. For example, the rubric could include: (1) the information requested from the consumer; (2) the type of basic financial education that is provided; (3) whether an explanation of algorithms and how they work is provided to the consumer; and (4) the recommendations that are given and the rationale for the recommendations.

10D. Public Communication

The issue of the quality of the education and information provided by financial advising are extremely important because most people receiving advice frequently make important financial decisions. Despite a fair amount of research on financial advising, the empirical question of the extent that financial advising – robo or otherwise – results in the acquisition of financial literacy is still an open question. The question is an important one. If consumers are supposed to learn from their financial experiences, the quality of the information they receive needs to be examined. If the results of these investigations find that little good knowledge is acquired, then this problem needs to be addressed. This is particularly true as consumers move toward automated products that may not fully explain their features and limitations.

Individuals seeking advice from professional advisors or through automated advising websites is not as well defined as other groups in this paper, but a growing literature exists about how professional advice is given to consumers. Focused research, however, on what people actually learn from the information received from professional or automated advisors is limited, offering opportunities for further research.