Morningstar - Q1 2020 - 41

EXHIBIT 1

of options leads to choice overload, thereby
dissuading the participant from self-directing.

Group Attributes In our study, we grouped participants by five different attributes,
dividing each one by five breakpoints.

When faced with a slate of investment alternatives
in a core menu, participants may adopt a strategy
of spreading their savings across offerings.
Benartzi and Thaler (2001) find that participants
appear to divide their contributions evenly across
the funds offered in the defined-contribution
plan (which the authors call the 1/n heuristic) in a
manner that is inconsistent with diversification
across asset classes. Huberman and Jiang (2006)
find that participants spread the bulk of their
portfolio among a greater number of funds in plans
with more options but find little evidence that
equity allocations are sensitive to the fraction of
stock funds on the core menu.

Attribute

Morrin et al. (2012) note that when the definedcontribution plan menu in the Oregon University
System was expanded to 19 options from 10 in
2007, the percentage of participants electing the
default increased to 33.5% from 21.2% and the
number of funds held by self-directing participants
increased to 5.3 funds from 3.7. The number of
funds held by participants increases when core
menus expand and declines when funds are cut
from the core menu (Keim and Mitchell, 2018).
Our Data Set
To explore the relation between core menu
sizes and participant-investment behaviors, we
analyzed data from a U.S.-based definedcontribution recordkeeper. The initial data set (raw
data) comprises 644,707 participant account
allocations across 545 401(k) plans. All data
is as of Dec. 31, 2018. We apply several filters
to the raw data to ensure we have timely
demographic data. We also include filters for age,
plan tenure, and salary. We also must be able
to identify 100% of funds held in the respective
portfolios. These initial filters reduce our
base data set to 439,859 participants. The actual
test groups vary based on the respective test
and are noted.

We group participants based on five attributes: the
number of funds in the core menu (excluding
target-date funds), age, deferral rate, salary, and
balance for comparison purposes. The breakpoints
for the respective attributes are included in

Number of Core Menu Funds

<12

12-14

15-19

20-24

≥25

Age

<30

30-39

40-49

50-59

≥60

5-7

8-11

12-14

≥15

Salary ($K)

<50

50-75

75-100

100-150

≥150

Balance ($K)

<50

50-100

100-250

250-500

≥500

Deferral Rate (%)

Source: Morningstar.

E X H I BIT 1 . We use these same thresholds for the
entire analysis.

Default Acceptance
The best investment portfolio for most participants
is the one they will not manage-such as a
target-date fund. Therefore, one potential objective
of the core menu is to maximize the acceptance of
the plan's default investment or any type of
professionally managed portfolio. In theory, core
menu size could affect default acceptance in
different ways. A small menu might nudge more
participants to use the default because they would
be less able to construct their own diversified
portfolio. Alternatively, a larger menu might scare
some participants away from building their
own portfolio and would result in them delegating
the task to the professionally managed option.

to including participants who are noted as
accepting the default (default investors), we also
include participants who have more than
95% of their portfolio in some type of allocation
fund (such as a balanced fund or target-date
fund). These participants effectively would
be considered to have fully delegated portfolioconstruction responsibilities, and while they
might not be coded as being defaulted into
the respective portfolio, the outcome is essentially
the same (they have delegated responsibility
of investment management). Default acceptance
rates for the respective attribute groups are
included in EXHIB IT 2 . The average default
acceptance rate among all participants in this
analysis was 80.6%.

To determine the relation between core menu size
and default acceptance, we filter the initial
data set to isolate those participants (and plans)
that have recently made a default election. Only
participants who have a plan tenure of less
than three years and a default election date on
file are included in this analysis. Among these
remaining participants, only those plans where at
least 50% of the participants were using
the default investment (a target-date fund) are
included. Default acceptance is typically high (for
example, 84% at Vanguard after one year,
according to Young and Young, 2019), and the
plan-level filter ensures we are only considering
plans with representative acceptance rates.

The demographic variables have a notable impact
on default acceptance. For example, younger
participants-with lower deferral rates,
lower salaries, and lower balances-have higher
default acceptance rates. These relations
are relatively monotonic and are obviously related.
(Higher deferral rates result in higher balances;
people with higher salaries tend to be
older; and so on.) From a demographic variable
perspective, we would likely describe those
investors who are not using the default
(self-directors) as being more sophisticated
than those who are using the default investment.
This has important implications for the next
analysis, which focuses on the quality of
the portfolios constructed by participants selfdirecting their accounts.

These filters result in a reduced data set of 73,537
participants across 213 401(k) plans. In addition

Those plans with fewer funds have significantly
lower default acceptance rates than plans with

morningstar.com/lp/magazine

http://www.morningstar.com/lp/magazine

Morningstar - Q1 2020

Table of Contents for the Digital Edition of Morningstar - Q1 2020