ICI’s Response to May 13 Entry on BrightScope Blog
We appreciate the opportunity to respond in detail to the recent blog entry on BrightScope about ICI’s research on 401(k) fees. The BrightScope blog critiques the sampling used in the Defined Contribution/401(k) Fee Study published by Deloitte Consulting LLP and the Investment Company Institute, and attempts to “reconcile” the Deloitte/ICI report with estimates of 401(k) fees made by others, particularly pension consultant Matthew Hutcheson.
In summary, the blog entry suffers from several basic flaws:
- The BrightScope entry attempts to reconcile our study with a commentary by Matthew Hutcheson (Elder Law Journal, “Uncovering and Understanding Hidden Fees in Qualified Retirement Plans”). It’s impossible to compare the two, because Hutcheson’s commentary does not describe or provide his methodology, sample size, or detail on anything other than a “conventional” plan based on his own experience. The Deloitte/ICI study uses an original survey that gathered more than 1,000 data points from each of the 130 plans included and includes comprehensive fee and other data on plans of all sizes.
- The BrightScope entry urges “a dose of reality about what the real DC plan population looks like,” and then proceeds to base its calculations on a market segment that includes 10 percent of participants (and 4 percent of assets) in 401(k) plans. This approach ignores the 90 percent of participants who tend to pay lower fees. If a clear understanding of what participants pay is the goal, the detailed data in the Deloitte/ICI study can be used to calculate a participant-weighted measure of fees. That figure is 0.86 percent of assets, far less than the median fee for “micro” plans (1.89 percent) that BrightScope uses.
- The BrightScope entry adds to 401(k) fees a crude estimate of the trading costs incurred by funds in 401(k) plans. While transaction costs are important and should be taken into account in selecting investment options, the rule-of-thumb approach that this entry and other critics take is not applicable to funds in 401(k) plans and exaggerates the impact of fees on participants’ returns.
These issues are discussed in detail below.
About the Deloitte/ICI Report and Its Sampling
In April 2009, Deloitte Consulting LLP and the Investment Company Institute (ICI) released a comprehensive study of the economics of 401(k) and defined contribution retirement plans, the Defined Contribution/401(k) Fee Study. As the report indicates, it is a study to understand the fee structures of 401(k) plans and how they work, using a sample of plans ranging from very small plans to very large plans. The survey captured fee information from a variety of service providers including mutual fund companies, insurance companies, banks, and third-party administrators. Indeed, the report provides voluminous information about differences in fees and fee ranges for five plan asset sizes:
- less than $1 million;
- $1 million to less than $10 million;
- $10 million to less than $100 million;
- $100 million to $500 million; and
- more than $500 million.
Exhibits throughout the report clearly indicate the size segments associated with various data points.
Section I of the report provides an extensive background on how Deloitte conducted the survey. The survey used a sampling technique known as nonproportional quotas. Knowing that the universe of 401(k) plans includes more than 450,000 plans, and that smaller plans are harder to find, the survey was specifically targeted across the spectrum of asset sizes and stayed in the field until specific quotas for plans of different sizes were filled. This survey design ensures that all segments of the population are included and resulted in the extensive, comprehensive data presented.
What ICI Has Said About the Survey
The Deloitte/ICI study clearly spells out how it uses its data. The report states that, "Although Deloitte and ICI believe the Survey results are representative, they cannot be projected to the entire population of U.S. 401(k) plans" (page 4).
In Section II, the report states: "Totaling all administration, recordkeeping and investment fees, the median 'all-in' fee for the plans in the Survey was 0.72 percent of assets ....The data show 10 percent of the plans in the Study had an 'all-in' fee of 0.35 percent of assets or less, while 10 percent of the plans had an 'all-in' fee of 1.72 percent of assets or more" (page 6).
Whenever ICI has discussed this 0.72 percent figure, it has specified that this is the median fee of the 130 plans in the study-not a median fee for all plans. In his speech at the ICI General Membership Meeting on May 6, ICI Chairman John Murphy described the study as follows: "Deloitte surveyed 130 plans, of all sizes and types, and calculated an 'all-in fee' that covered all the services, investment and administrative, for each plan. The bottom line? The median all-in fee for these plans was 72 basis points. Less than three-quarters of 1 percent-quite a bargain compared to the 3 percent that some critics cite." (Emphasis added.)
(See Murphy's full remarks.)
Murphy's remarks stated his opinion, backed by data from our survey, that the typical 401(k) participant receives an excellent value from his or her 401(k) plan.
Survey Methods in Hutcheson's Estimates of 401(k) Fees
The BrightScope blog entry attempts to "reconcile" the Deloitte/ICI results with estimates by others. It's interesting to note that the BrightScope author found all the data he needed to make this calculation in the very detailed breakdown of fees by plan size contained within the Deloitte/ICI study-evidence, it would seem, of the thorough and comprehensive nature of the study he is critiquing.
It is doubly interesting that the author would choose as his benchmark for this "reconciliation" a commentary by BrightScope advisory board member Matthew Hutcheson, published in Elder Law Journal. In that commentary, Hutcheson provides an example for what he calls a "conventional" plan (Table 1). "This is a simple example of the costs for a 'conventional' plan. Plans that utilize higher-cost funds will obviously cost more ... it is fair and reasonable to conclude that the average plan has total, 'all-in' costs that exceed the example above." In other words, Hutcheson appears to be arguing that typical plan costs are greater than the 3.37 percent shown in Table 1.
We would like to compare the Deloitte/ICI results to Hutcheson's commentary. Unfortunately, Hutcheson does not describe how he arrives at his "conventional" plan. Indeed, the "averages" that he cites are largely attributed as "consistent with first-hand knowledge of the author" (see footnotes 75-78). Hutcheson's article does not describe any survey methodology. He does not specify the number of plans in his sample of "first-hand knowledge," nor those plans' participant counts, asset size, or other characteristics or demographics. A reader cannot understand what segment of the plan population his "conventional" plan purports to represent.
For other data points, Hutcheson offers simple averages of mutual fund expense ratios and estimated trading costs. This ignores two important facts. First, only about half of all 401(k) assets are invested in mutual funds. And even for those 401(k) assets that are invested in mutual funds, previous ICI research casts doubt on the use of industry-wide averages as proxies for 401(k) fees. That research has shown that 401(k) plan assets that are invested in mutual funds tend to be concentrated in funds that charge below-average fees and that exhibit below-average portfolio turnover.
What Do 401(k) Participants Pay?
The BrightScope blog entry calls for "a dose of reality about what the real DC plan population looks like." Yet the BrightScope blog entry focuses its attention on a segment of that population, in a manner that tends to exaggerate the fees that the bulk of actual 401(k) participants pay.
It is true-and the Deloitte/ICI report spells out clearly-that most 401(k) plans are small and that small plans have higher fees measured as a percentage of assets. As the Deloitte/ICI study details, small plans tend to have higher costs because small plans have fewer assets and participants over which to spread certain fixed costs that must be incurred because of government regulations or the operational features of managing a retirement plan. (Our survey also found, however, that sponsors of such plans are more likely to pay a higher portion of these costs than sponsors in larger plans [Exhibit 24], which helps to reduce the costs to participants in these small plans.)
While it is true that most plans are small, it is also true that most 401(k) participants and most 401(k) assets are in larger plans. Small plans (less than $1 million in assets) account for 62 percent of the plans in the market, but only 10 percent of active 401(k) participants and 4 percent of 401(k) assets. Using the median fee for that "micro" segment, as BrightScope does, ignores the fact that 90 percent of participants are in plans where fees tend to be lower than in small plans. That would hardly seem to provide "a dose of reality" about the full population of 401(k) participants.
If policymakers or other interested readers want a sense of what 401(k) participants might actually be paying, the Deloitte/ICI study provides information to help in the analysis. Weighting the median fees by the number of participants in the plans results in an all-in fee of 0.86 percent-not much more than the 0.72 percent median for all 130 plans in the Deloitte/ICI survey, and far short of the 1.89 percent that BrightScope highlights.
For those who want more analysis, the Deloitte/ICI study provides econometric estimates of the fees by average account balance and plan size (Exhibit 3), as well as ranges of fees by plan size (Exhibit 32).
The BrightScope blog entry also considers the issue of trading costs. Trading costs are the inescapable expenses of buying, holding, and selling investments. Investors in every form of investment-direct investments, commingled or collective trusts, exchange-traded funds, separate accounts, or mutual funds-incur trading costs, and 401(k) participants are no exception.
Trading costs are important, and fiduciaries should consider such costs as one factor in selecting investments. The Department of Labor's guidance on fees makes clear, and we agree, that in considering the appropriateness of a pooled product, the fiduciary should consider the costs associated with a higher level of trading activity.
But contrary to the contention of the BrightScope blog entry, the Labor Department does not require that fiduciaries "measure" transaction costs. Nor does it specify, in the case of mutual funds, that those costs must be "measured" in some way other than the disclosures mutual funds already provide under SEC rules. By contrast, DOL's Fee Disclosure Form uses a fund's expense ratio and front-end load as relevant measures of fees.
ICI does not add estimates of trading costs to 401(k) fees because there is no agreed, reliable method of measuring such costs. The Securities and Exchange Commission, which regulates mutual funds, has wrestled with disclosure of trading costs for decades, and has never been able to devise a workable measure.
Trading costs include brokerage commissions, broker bid-ask spreads, market impact costs, and other costs. Only one of these costs-brokerage commissions-is easily measured, yet none can be ignored: An institution that focuses solely on commissions, for example, can find itself paying a high price in market impact and bid-ask spreads.
The BrightScope blog entry attempts to estimate trading costs using an assumption that a fund's trading costs are roughly equal to its investment expenses. This is based on a well-regarded paper by Edelin, Evans, and Kadlec (EEK). The authors of that paper write that their "estimates of annual trading costs are comparable in magnitude to the expense ratio" for domestic stock funds. However, this is a statement about the particular type of funds that they examined, and should not be taken as a rule of thumb for any particular fund or group of funds-particularly in the case of 401(k) plans.
The reasons why the EEK finding does not apply to 401(k)s include:
- While the EEK paper is based on trading costs for domestic equity funds, about one-quarter of the assets in 401(k) plans are invested in some type of fixed-income product, such as guaranteed investment contracts and money market funds. These funds tend to rely heavily on liquid cash instruments and other investment strategies, rather than on trading securities, to meet investor flows. Thus, they incur lower trading costs than do the funds in the EEK paper.
- The EEK findings are based on the average, or mean, trading cost for domestic equity funds of 1.44 percent of assets. A better comparison for 401(k) plans, however, would be the median trading cost, which EEK estimate at 0.89 percent. Previous ICI research has demonstrated that equity funds used by 401(k) plans tend to have lower-than-median levels of securities trading, and thus are likely to bear lower-than-median trading costs.
- Within the domestic equity category, 401(k) investments are concentrated in large-cap funds. As EEK note, these funds incur lower trading costs than small- and mid-cap funds. EEK estimate that the average trading costs on large-cap domestic equity mutual funds is 0.77 percent of assets, with a median of 0.55 percent.
Determining the true trading costs of the actual funds employed by 401(k) plans of different sizes would require a detailed econometric survey and study. What is clear, however, is that the BrightScope blog entry's crude estimate exaggerates the impact of trading costs on 401(k) investors' returns.
We welcome scrutiny of our research and an open dialogue on these important issues. We hope that BrightScope's readers will take the time to review our research themselves and draw their own conclusions about its thoroughness, rigor, and accuracy.