The value of this approach was underscored by the Department of Labor in 1996. Its Interpretive Bulletin 96-1Footnote1 stressed to plan sponsors the importance of investment education with respect to investment offerings on the DC core menu. It notes that “participants should consider their other assets, income and investments (e.g., equity in a home, IRA investments, savings accounts, and interests in other qualified and non-qualified plans) in addition to their interests in the plan” when making decisions about plan investments.
The importance of considering factors beyond age led to the advent of managed account programs in 401(k) plans, which allow participants to develop more personalized asset allocations based on their specific circumstance. However, advice can be personalized only if the plan participant actively engages with the managed account provider. The employee must supply information about themselves – not just at enrollment, but also on an ongoing basis.
Unfortunately, while nearly all participants in a 401(k) plan could benefit from asset allocation advice tailored to their unique circumstance – as could their plan sponsor fiduciaries – most are reluctant to spend the time and resources to provide the necessary information. They are in what we refer to as the “do-it-for-me” camp, for whom plan fiduciaries typically default into a traditional age-based TDF. Indeed, the requirement for participants to engage has been one of the key hurdles for adoption in the managed account space, where total assets are only about 10% of that of traditional TDFs.Footnote2 The problem, of course, is that traditional TDFs are personalized almost exclusively on a single variable: age.
Hence, as Figure 1 shows, there has historically been a personalization-engagement trade-off. Plan participants have effectively been limited to choosing between age-only personalization (with no engagement) and full personalization (with significant engagement). However, with today’s technology, auto-personalization is on the horizon.
Auto personalization provides the hands-off automation of traditional TDFs, while offering personalized asset allocation of the sort historically associated with managed accounts. By leveraging participant-specific information available on most recordkeeper platforms, plan sponsors can offer more personalized allocations tailored to an employee’s unique circumstance – while requiring no more participant engagement than a traditional TDF.Footnote3
How will it work?
Whether one is talking about traditional TDFs or a full managed account solution, numerous inputs go into the generation of an optimal long-term asset allocation. With traditional TDFs, all inputs except age are typically fixed at an “average” or “representative” value, whereas in managed accounts, all inputs are effectively “variable.”
However, despite the seemingly limitless number of inputs that can inform an asset allocation, the reality is that only a relatively small number are critical. These include a participant’s salary, account balance, and their contribution and employer-match rate – in addition to the most essential input of all: age. Importantly, these key ingredients are almost always available to plan fiduciaries, typically vis-à-vis the plan sponsor’s recordkeeper. It is this data availability that enables personalized asset allocation without active engagement required by the participant.Footnote4
Figure 2 shows an example of the type of asset allocation differentiation that a plan sponsor might expect should their participants go down the path of auto-personalization. The blue lines show the equity allocation at different ages for the industry average glide path, while the green dots show the personalized equity allocation for each individual in a plan.Footnote5 We see two key takeaways from this graph: 1) age-based target date funds are only “quasi-personalized” by age because everyone within a five-year band gets the same asset allocation and, 2) there can be significant differences in personalized equity allocations relative to a traditional TDF.
Some participants may require more equities – and some less – than that provided by a traditional TDF. And, of course, for some participants the equity exposure in a traditional TDF may be reasonably close to where it should be.
What types of plan sponsors is auto- personalization most appropriate for?
When considering the benefits of auto-personalization, it is useful to think about the distribution of a sponsor’s employee demographics. Specifically, the extent to which a plan sponsor’s employee base will benefit from personalized asset allocation is a function of both the average employee demographics as well as the diversity in those same measures.
While diversity may be somewhat more difficult to conceptualize, it’s actually the easiest to tie directly to the benefits of personalization. To the extent a participant population is highly diverse – meaning there is wide variation in variables such as salary and account balances – personalization may be highly beneficial because it can consider a wide array of employee characteristics and produce a differentiated asset allocation for each individual. This can be seen easily in Figure 2, which shows a wide range in the stock-bond mixes across this sample plan. In short, the greater the degree of diversity within a plan, the greater the benefits of personalization.
The average, on the other hand, defines the general demographic characteristics of a set participants. Traditional off-the-shelf TDFs usually assume national averages or representative values for the key demographic assumptions that go into glide path design. Therefore, employees whose salary, balance, and contribution rate are all similar to these averages may find the benefits of personalization to be somewhat low. If average characteristics differ meaningfully from representative values, however, then personalization can be beneficial because it can directly account for the average differences. Indeed, this is precisely why some plan sponsors adopt custom TDFs tailored to their average employee demographics.
However, while custom TDFs can help address situations in which average plan characteristics differ from assumptions underlying an off-the-shelf TDF, they will still not be able to account for the diversity in the employee base. If, for example, an organization is comprised of low-wage and high-wage workers, a custom TDF solution can address one of the groups, but not both. In fact, if averages are used, a custom solution likely addresses neither group. Hence, only a fully personalized solution can account for differences in both the average level of employee demographics and diversity across employees.
Figure 3 summarizes how different types of target date solutions can best address varying cohorts of employees.
Auto-personalization for the masses
In short, diversity in participant demographics should imply a corresponding diversity in asset allocation. However, given the large number of plan participants in the “do-it-for-me” camp, the reality is that most participants in traditional TDFs are personalized only along the dimension of age.
Fortunately, many of the key inputs that go into glidepath design are now available to plan sponsors without the need for direct employee engagement. Hence, participants and their plan sponsors may no longer need to choose between zero-engagement automation and full-engagement personalization. Auto-personalization is on the horizon.
1 https://www.federalregister.gov/documents/1996/06/11/96-14093/interpretive-bulletin-96-1-participant-investment-education Return to content
2 Total managed account assets were estimated to be $348 billion in 2019, according to Pension & Investments and Cerulli Associates., April 2020. Return to content 3 No engagement may be required on behalf of the participant when auto-personalization is the QDIA, and auto-enrollment is utilized. Different providers may require some level of participant engagement. Return to content 5 Sample plan. Industry average glide path data was provided by Morningstar as of 31 December 2020, the latest data available. Return to content