Investing “for the long run” is an increasingly prominent phenomenon — one that has been made more pressing by the demise of the defined-benefits system and population aging. But these longer investment horizons may come with unintended consequences.
A new industry has arisen to address investors’ long-term investment needs, managed mainly by large investment groups, or fund families, that invest clients’ assets in underlying funds that are either part of the same family or a few very large families. This has induced significant concentration in the creation of investment vehicles. Such concentration raises the critical question of whether a closed architecture — i.e., a fund family investing in its own managed funds —provides the client with the potential benefits of “synergies” related to investing in one’s own products, such as informational advantage and related economies of scale, or whether the potential conflicts of interest swamp such benefits.
Another way of posing this question is to ask whether investors have a long time horizon that prevents them from adequately monitoring their asset managers. Indeed, a longer investment horizon will potentially increase the agency problems, providing the manager with more opportunities to favor its family’s funds at the expense of its own performance.
To investigate this issue, we focus on the the target-date fund industry. Target-date funds were created specifically to invest for the long run by gradually rebalancing portfolios to safer conditions as investors approach retirement — that is, moving investments away from equity and into bonds. More specifically, when the target date is far away, a TDF allocates most of the portfolio to equity investments. As the target date approaches, the TDF decreases the portion allocated to equity and shifts to bonds, reducing the fund’s overall risk exposure. This rebalancing is almost automatic and takes the form of a “glide path.” In this way, the very same fund with the very same clientele transits through the years over several different allocations.
The simplicity of the approach has been a critical factor in the success of these funds. Their number grew from just 63 in 2000 to 2,778 in 2019, with total assets under management surpassing $1.4 trillion by the end of 2019. This growth was stimulated by the U.S. Department of Labor’s decision to make TDFs one of the “qualified default investment alternatives” in employer-sponsored deﬁned-contribution retirement plans under the 2006 Pension Protection Act.
TDFs typically do not invest directly in equity or bonds, but in underlying pooled investment vehicles such as index mutual funds, exchange-traded funds, and actively managed mutual funds. The majority of TDFs invest most of their portfolios in actively managed mutual funds. These can be managed by the same fund family or by another family. In 2019 about 58 percent of TDFs invested only in their families’ funds. An additional 20 percent invested between 50 percent and 99 percent of their portfolios in their families’ funds. Only 10.6 percent of TDFs did not invest in their own families.
Because they invest in other investment vehicles, the majority of TDFs are closed-architecture funds, and about 20 percent of these TDFs have an open-architecture structure for the majority of their portfolios.
We test our hypotheses using a sample of 4,084 different TDF share classes that operated between 2000 and 2019. We first document that a longer horizon implies less attention to performance, with flow-performance sensitivity decreasing as the fund horizon lengthens. For long-horizon funds, flow-performance sensitivity is almost flat, which reduces the disciplinary power of investors’ flows. Outflows are always less sensitive to performance than in the case of non-TDFs, whereas inflow sensitivity increases as the horizon lengthens. When the funds are very far away from maturity, new flows are way less sensitive to performance.
This lower sensitivity can be explained by the change in regulation that made it possible for TDFs to exploit investor inattention deliberately. Specifically, the Pension Protection Act allowed pension plan sponsors to oﬀer TDFs as default investment options within 401(k) retirement plans. This “default” investment of new flows resulted in TDF flows becoming largely independent of a fund’s performance.
As investor sensitivity to performance decreased, the percentage of TDF portfolios invested in equity, given the same time horizon, has steadily increased. Investment in equity for the short (long) horizon increased by about 20 percent (35 percent) between 2008 and 2013 and by an additional 9 percent (10 percent) after 2013. Given the relative inattention of the long-term investor, the fund manager finds itself in the privileged position of investing in the long run with limited pervasive short-term scrutiny from the investor, which it exploits by tilting more toward risky equity.
We also note the existence of a strong link between horizon and performance. For the average investor who invests for 50 (20) years, we find that the drag on performance resulting from the behavior of TDFs reduces the investor’s cumulative performance by 20.6 percent (5.9 percent). Moreover, there is a negative correlation between performance and horizon: A ten-year increase in the horizon translates on average into a decline of 29 basis points in net-of-fees annual performance.
Next, we explain the source of that lower performance. We find that the negative relationship between performance and horizon is reinforced if the family funds in which the TDF invests have high outflows. In particular, each additional year of the horizon translates on average into 1.1bp to 1.2bp of lower net-of-fees annual performance for each standard deviation of outflows from the family’s funds. Moreover, the negative relationship between performance and horizon is amplified by the tendency of TDFs to invest in their own families’ funds. A standard deviation increase in both outflows from and exposure to those family funds leads to annual underperformance of between 4.7 and 5.8 basis points for each year of the horizon.
Overall, these results suggest that TDFs deliver worse performance on a longer horizon because they invest in the family funds that face outflows, likely in order to buffer the liquidity shocks. These funds are likely to be actively managed — i.e., funds for which the effect of performance on outflows is more critical and for which the buffer against liquidity shocks helps deliver better performance.
We also document the fact that TDF managers engage in fee skimming by charging higher fees on the underlying funds — the less observable ones. The longer the horizon, the higher the total fees. In particular, a horizon that is ten years longer results in 4.2bp to 4.8bp in increased total fees. Almost all of the effect comes from “underlying” fees — that is, those charged by the underlying investment vehicles. In contrast, in the case of directly charged fees, the horizon plays no role. This suggests that TDFs are used to subsidize the underlying funds managed by the same family. The average amount of (underlying) profit pumping — quantified as the difference between the expense ratio of each underlying fund and the cheapest share class in the same fund portfolio the TDF could have chosen instead — ranged from 3bp in 2005 to 1.9bp in 2018 and is higher for longer horizons.
Last, we see the risk implications of this behavior. There is a robust positive correlation between fund risk-taking and deliberately investing in family funds when their volatility of flows is high. An increase (with respect to the unconditional mean) of ten years in the horizon translated on average to a 1.9 percent higher drawdown for TDFs with a one-standard-deviation-higher tilt toward family funds and whose invested family funds displayed same-family outflows that were one standard deviation higher. Under this same scenario, a ten-year increase in horizon was also linked to 2.3 percent higher volatility and 1.6 percent higher beta.
Overall, we believe that TDFs are an essential innovation that aims to improve the welfare of investors in retirement accounts, but our results underline the importance of an open-architecture structure at the fund level. In 2017, Andrew Arnott, president and CEO of John Hancock Investment Management, called for an open architecture for TDFs: “Today plan-level best practices call for an open-architecture, or multimanager, lineup of investment offerings, but that line of thinking rarely extends to target-date portfolio construction. If open architecture is important, then perhaps more target-date funds should be open, incorporating a variety of specialized teams based on their merits.”
We can only second this sentiment. Another takeaway is the importance of transparency in the level of risk that TDFs are taking, both for short and long horizons.
Massimo Massa is a professor of finance and the Rothschild Chaired Professor of Banking at INSEAD. Rabih Moussawi is an associate professor of finance at Villanova School of Business. Andrei Simonov is a professor of finance at Michigan State University’s Eli Broad Graduate School of Management.