John Chalmers is Abbott Keller Professor of Finance at the University of Oregon Lundquist College of Business, and Jonathan Reuter is associate professor of finance at Boston College Carroll School of Management. This post is based on their recent paper, forthcoming in the Journal of Financial Economics.
Studies of financial services industries have consistently found evidence of conflicted advice. Financial advisors, in a variety of settings, including mutual funds, insurance, and brokerage accounts, have been found to recommend higher-commission products. One implication is that the quality of financial advice that investors receive has room for improvement, perhaps through increased standards of care. In our paper, we ask a different question: when are investors better off, even if they bought a relatively expensive product, than they would have been otherwise? Answering this question requires us to determine what an advice-seeking investor would have purchased in the absence of conflicted advice. While this counterfactual investment is inherently difficult to identify, we show that the institutional setting is likely to play an important role in determining the relevant comparator.
The counterfactual investment is critical to understanding the value of advice. In Gennaioli, Shleifer, and Vishny’s (2015) model, trusted financial advisors (“money doctors”) increase clients’ equity allocations above counterfactual levels of zero, thereby allowing clients to earn the equity risk premium. Because the fees charged by financial advisors are set to split the gains from trade, each party benefits. When the relevant counterfactual portfolio is a money market fund, potential gains from trade are large and advice seekers are likely to benefit, even if the recommendation is to invest in high-commission equity funds. In contrast, if the counterfactual investment also provides considerable allocations to equities, the value of the conflicted advice must be higher to provide benefit to the investor.
Studying the value of advice is difficult because those who seek advice are not randomly selected. An ideal experiment would focus on a large sample of real-world investors seeking to invest with advice, and randomly withhold advice from part of the sample. The causal impact of advice within this setting could be estimated by comparing the portfolios of broker-advised participants to the portfolios of reluctantly self-directed investors. Although the ideal experiment is infeasible in real-world settings, we exploit time-series variation in access to brokers in a quasi-natural experiment that addresses some of the selection and reverse causality challenges that vex other studies of advice.
We study participants in the Oregon University System (OUS) Optional Retirement Plan (ORP), a defined contribution (DC) retirement plan introduced in October 1996. Participants who choose to invest through ORP must select a single investment provider to receive retirement contributions made on their behalf. At OUS’ request, we refer to the three main ORP providers as HIGH (advice), LOW (advice), and NEW. HIGH employs a network of brokers to provide face-to-face recommendations, while LOW requires participants to select from an investment menu. Prior to November 2007 (Regime 1), ORP participants could choose to invest through HIGH or LOW. Effective November 2007 (Regime 2), new ORP participants were limited to LOW or NEW, while legacy participants in HIGH were allowed to continue directing contributions to HIGH. We analyze anonymized data that match administrative data on ORP participants with account-level data from HIGH, LOW, and NEW through the end of December 2009.
During Regime 1, demand for HIGH is negatively correlated with age, salary, and educational attainment and is significantly lower among participants working in an economics department or business school. These patterns suggest that ORP participants are more likely to seek advice when they have lower levels of financial sophistication, perhaps due to less investment experience. In predicting demand for HIGH, we find that 39.2% of participants with predicted demand for advice in the top quartile choose to invest through HIGH versus 14.8% of those in the bottom quartile of demand for advice. We administer an online survey to ORP participants which confirms that demand for advice on asset allocation and fund selection was an important factor in choosing HIGH, raising questions about how these participants would have invested in the absence of broker recommendations. Importantly, this form of sample selection argues against constructing counterfactual portfolios for broker clients from either the actual portfolios of self-directed investors or commonly used academic benchmarks, such as low-cost index funds.
How do reluctantly self-directed investors invest in the absence of broker recommendations? We hypothesize that removing access to broker recommendations will increase demand for default investment options among Regime 2 participants with high predicted demand for HIGH. Exploiting the introduction of target date funds (TDFs), we further hypothesize that the substitution of default investment options for broker recommendations will be strongest when the default investment option is a TDF, because TDFs allow participants to invest in a single fund that bundles asset allocation with portfolio management. When new participants lose access to brokers, overall demand for default options by NEW participants jumps to 65%, where the default option is a TDF, and remains at 22% for LOW, where the default option is a money market fund. Furthermore, we find that 47.5% of new participants with top-quartile predicted demand for HIGH choose to invest in a TDF versus 28.7% of new participants with bottom-quartile predicted demand for HIGH. In other words, many of the Regime 2 investors who we classify as reluctantly self-directed investors respond to the plan redesign by investing in TDFs.
Building on the finding that many ORP participants would have held TDFs in the absence of brokers, we compare broker clients’ actual portfolios with counterfactual portfolios based on TDFs over the complete sample period. We find that broker clients earn significantly lower after-fee annual returns and Sharpe ratios than they would have earned if they had been invested in age-specific Fidelity Freedom TDFs. We also find that the Sharpe ratios of the high-broker-demand investors who joined during the first year of Regime 2 are both higher and less variable than the Sharpe ratios of the high-broker-demand participants who joined during the last year of Regime 1, even when we exclude Regime 2 portfolios invested in TDFs. We conclude that few if any participants with high predicted demand for broker recommendations would have been harmed if ORP had removed HIGH and added NEW even earlier than 2007. In other words, when advice seekers are given access to a default investment option that imbeds advice on asset allocation and fund selection, this option can dominate conflicted advice.
Finally, to test Gennaioli, Shleifer, and Vishny’s (2015) prediction that actual portfolios of broker clients will have greater allocations to equity than counterfactual portfolios constructed without access to brokers (or TDFs), we focus on those participants who had the option to invest through HIGH during Regime 1. We compare the portfolio characteristics of participants who are predicted to invest through brokers and do with portfolios of participants who are predicted to invest through brokers but do not. Our estimated differences in risk-taking are striking. Participants who are predicted to invest through a broker and do, hold portfolios with higher total risk (the volatility of monthly returns is 0.6 percentage points higher) and higher systematic risk (the Capital Asset Pricing Model beta is 0.20 higher) than participants who are predicted to invest through a broker but do not. In other words, when the default investment option is a money market fund, as it is during Regime 1, brokers potentially can add value by increasing participants’ equity allocations. However, we find that exposure to higher levels of market risk during Regime 1 comes at the cost of higher commissions (Christoffersen, Evans, and Musto, 2013) and lower risk-adjusted returns.
The complete paper is available for download here.