Analysis of Ayres and Curtis critique of 401(k) plans
In our previous article we reviewed [Professors Ayres and Curtis’s paper Beyond Diversification: The Pervasive Problem of Excessive Fees and ‘Dominated Funds’ in 401(k) Plans] (John M. Olin Center for Studies in Law, Economics, and Public Policy Research Paper No. 493). Our purpose in that article was simply to describe what Professors Ayres and Curtis are saying. In this article we evaluate their findings and proposals, discussing the limits of and possible objections to their conclusions.
Since a superficial reading of Ayres and Curtis’s paper might be ‘active management is bad/passive management is good,’ we begin with a quote from Alicia Munnell (Peter F. Drucker Professor of Management Sciences at Boston College’s Carroll School of Management; director of the Center for Retirement Research at Boston College and a well-known critic of high 401(k) plan fees) on that issue in her blog Encore, A Blog About Living In And Planning For Retirement:
In what follows, we begin with a discussion of issues with Ayres and Curtis’s methodology and then go on to discuss their policy proposals (we refer you to our earlier article for a detailed discussion of both).
Methodology – limits to the application of Ayres and Curtis’s findings
Do Ayres and Curtis ‘prove’ that 401(k) plan participants are on average losing 156 basis points annually to bad fiduciary decisions and participant asset allocation mistakes? Let’s begin with some issues with their methodology that Ayres and Curtis would concede:
- Their ‘ideal portfolio’ consists of three assets: the Russell 3000, the Barclay’s Aggregate and the MSCI EAFE. Other asset classes are translated to this standard using Sharpe Ratios. This method works well for large or (arguably) small cap equity, but what about illiquid asset classes like real estate, alternatives, or (arguably) emerging markets? For those asset classes there is often not enough data to produce Sharpe Ratios.
- They use a single ideal portfolio regardless of time horizon. They defend this approach in their 2010 paper (Measuring Fiduciary and Investor Losses in 401(k) Plans, Quinn Curtis and Ian Ayres) as follows:
Plan participants are often advised to shift their investments over their lifecycle to reduce risk as they approach retirement, but our methodology yields a single optimum allocation. Though the use of a single optimum is a simplification, it is nevertheless useful for our purposes. Our data includes only plan aggregate portfolios, and each observed portfolio reflects choices by all participants in the plan, including individuals of different ages. It is therefore reasonable to measure losses at the plan level relative to a single aggregate optimum portfolio, even as individuals in plans may shift their allocations over time.
While all of that may be true, if the sum of participant time horizons in a given plan does not match the “ideal” time horizon they are using to determine their ideal portfolio, non-mistaken asset allocation decisions by participants will show up as ‘losses’ in their numbers.
- In that regard, because they are unable to look at individual asset allocation decisions, their number for asset allocation losses – 65 basis points – is nearly useless. Here is their explanation of this problem:
Our measure of investor diversification loss likely understates diversification losses. Because we are only able to measure plan-level allocations, individual investors may have off-setting over-allocation in individual menu offerings.
Thus, Ayres and Curtis believe their number understates diversification losses, although it is conceivable (although perhaps less likely) that (e.g., when the issue raised in 2 is taken into account) it overstates those losses. 65 basis points is a big enough number that someone ought to undertake the effort to determine the actual cost of participant diversification mistakes.
- Ayres and Curtis’s analysis looks at plans that offer only publicly listed mutual funds – they exclude 71% of the plans in the Brightscope database because they (the excluded plans) use collective trusts or separate accounts.
An average of 156 basis points lost to bad fiduciary decisions and participant asset allocation mistakes is pretty scary. But if plans using collective trusts and separate accounts had been included that number would almost certainly go down, probably by a lot.This is because (1) the fees on collective trusts and separate accounts are generally lower than the fees on mutual funds and (2) large plan investment education/default policies are likely to be better at steering participants to better asset allocations.
Moreover, the data that Ayres and Curtis used was as of 2009. The trend toward using collective trusts and separate accounts has accelerated since then.
Ayres and Curtis’s decision to exclude plans with collective trusts and separate accounts was necessary because fund-fee information on those plans is not publicly available. But this feature of their methodology limits their findings, like much of the 401(k) fee controversy, largely to the medium- and small-plan universe. Their conclusions cannot be said to describe the 401(k) system as a whole, since they did not look at it.,/p>
Methodology – challenges to Ayres and Curtis’s findings
Now let’s turn to objections with which Ayres and Curtis might disagree:
1. Ayres and Curtis’s baseline for determining losses is a plan with around 25 basis points in investment management fees and 8 basis points in administrative costs. In calculating ‘headline’ losses to fees they look at the combination of investment management and administrative fees. This makes sense given the prevalence of revenue sharing arrangements in which investment management fees are used to subsidize the cost of plan administration.
Many would argue that both these baseline numbers are low. While it is certainly possible to get an equity index fund for 25 basis points (probably significantly less), is it clear that the bond market and the EAFE universe are so efficient that appropriate funds can be had for that price? Moreover, for a medium- or small-sized plan at least, 8 basis points seems like a very low cost-of-administration. Thus, one could argue that this is a ‘low-ball’ baseline. This is, however, at most a quibble – even if you were to add, say, another 20 basis points to baseline plan costs, Ayres and Curtis’s findings would be disturbing.
2. The throw-down question raised by Ayres and Curtis’s work is: participants are losing 156 basis points compared to a retail index fund and bare bones administration – can the 156 basis points be somehow justified based on either (1) the value of added diversification or (2) returns from some other source (a unique strategy or the superior intuition of a given manager)?
Let’s take those issues in reverse order.
In their 2010 paper Ayres and Curtis state: “we estimate returns using a factor model with no alpha.” Investopedia defines alpha as:
- A measure of performance on a risk-adjusted basis. Alpha takes the volatility (price risk) of a mutual fund and compares its risk-adjusted performance to a benchmark index. The excess return of the fund relative to the return of the benchmark index is a fund’s alpha.
- The abnormal rate of return on a security or portfolio in excess of what would be predicted by an equilibrium model like the capital asset pricing model (CAPM).
Ayres and Curtis qualify their findings as follows: “To the extent actively managed funds produce persistent positive alpha, this model may overstate the costs of holding such funds.” Whether there exists, with respect to actively managed large cap at least, ‘persistent positive alpha’ is the $64,000 question.
Generally, and along with some other researchers, Ayres and Curtis believe that, even if it exists, 401(k) plans generally do not capture persistent positive alpha. Some evidence indicates “abnormally high historical performance for funds included in 401(k) plans and show[s] that these returns don’t persist” – in other words shows that 401(k) plan fiduciaries are prone to ‘chase returns.’ Most would agree that is a bad thing.
Ayres and Curtis generally believe that “research suggests that low-cost index funds are likely to outperform actively managed funds.” And they believe that their results (e.g., the 156 basis points average loss) are consistent with consideration of potential alpha; their discussion is technical, and we refer you to their (2010) paper for details.
There is an ongoing, decades-old debate about ability of active management to add value in the large cap asset class. We are not going to try to resolve that debate here. We simply note that Ayres and Curtis generally favor passive/index funds.
As noted above, Ayres and Curtis use a limited set of asset classes – a broad equity index (the Russell 3000), a bond index (the Barclay’s Aggregate) and an international equity index (the MSCI EAFE). The passive-is-better-than-active argument is relatively easy to make inside such a limited and (relatively) market-efficient universe.
Is additional diversification – beyond those asset classes – worth the cost? Certainly the managers of large defined benefit plans – who most would concede are sophisticated investors – believe it is. Those plans typically include significant allocations to, e.g., alternative investments.
This argument cuts both ways – the typical 401(k) plan investor or, even, the typical small- or medium-sized 401(k) plan sponsor-fiduciary may not have the sophistication to pick an appropriate fund in these ‘less market-efficient’ asset classes. But packaging these asset classes, e.g., into a 401(k) plan target date fund is considered by many a viable strategy to increase diversification and returns. Management fees for these asset classes are necessarily higher and might, e.g., push fund fees above Ayres and Curtis’s 49 basis point limit on their proposed Enhanced Qualified Default Investment Alternative (EQDIA). That is, arguably at least, a problem.
3. Ayres and Curtis’s methodology highlights the problem of overpriced actively managed mutual funds. But what about overpriced index funds? They certainly, as Alicia Munnell notes, exist. They are also causing losses in the system. If we are going to go on a crusade against overpriced actively managed funds, shouldn’t we target overpriced passive funds as well?
A new, substantive fiduciary standard
Ayres and Curtis’s distinction between the current law’s focus on process and their proposed focus on substance may be more semantic than real. Theoretically a prudent process would prevent a fiduciary from selecting a fund that no prudent person would reasonably believe ought to be held by investors.
But their basic point is pretty persuasive: based on their findings (and the work of Benartzi and Thaler and others), the Hecker line of cases, which exonerates the inclusion of a ‘bad’ fund where there are sufficient good funds available, seems to be mistaken. Some may differ, but Ayres and Curtis muster compelling evidence for their claim that the inclusion of a ‘bad’ fund in a menu of good funds will necessarily result in participant losses and should therefore not be permitted. At a minimum, the burden of the argument has shifted to the high-cost fund industry to make a case that, on balance, the inclusion of high cost funds adds rather than subtracts from net 401(k) plan returns.
And they are right in pointing to the agency problems that make fund menu construction rules particularly problematic. Those problems are the main issue behind the Department of Labor’s ongoing ‘redefinition of ERISA fiduciary’ project.
One obvious problem, however, with their “no prudent person could reasonably believe that the fund in question ought to be held by investors” standard: as some courts have pointed out, investors do in fact buy these funds – someone is investing in them. Are all those investors unreasonable or do they just have a different view of how the market works? Can their proposed standard be administered? That is, will courts, and sponsor-fiduciaries, be able to consistently identify which are the ‘bad’ funds?
Mandated EQDIA + and investor sophistication test
Should we lock all 401(k) plan participants into an under-50-basis points EQDIA and not let them out unless they pass an investor sophistication test? There are quibbles here: Is 50 basis points the right number? What about bad 49 basis point QDIAs?
But what about the bigger question – is this sort of ‘hyper-nudge’ appropriate? One of the original selling points of the mutual fund-based 401(k) plan was that it made participants feel more like ‘owners.’ Choice – about how much to contribute and about how to invest – was one of the key features of the system. Since those (original) days, a lot of people – participant advocates, policymakers, sponsors and even participants themselves – have become disillusioned with choice.
Sponsors now typically default both contributions (automatic enrollment) and investments (QDIAs). Ayres and Curtis want to make the default cheaper – by mandating that the default have low fees – and stickier – by requiring a test to get out of it. Let’s take those issues one at a time.
Cheaper. There doesn’t appear to be a compelling reason why participants – who are not doing any choosing in the matter – should be defaulted into an expensive fund. It would seem that the main virtue of defaulting into such a fund would be to ‘seed’ revenue sharing. Remember the double agency problem that exists here – a (sometimes) biased consultant and a sponsor-fiduciary with no direct stake in the outcome.
Stickier. Are we at the point, with ‘nudge’ and ‘soft paternalism,’ that we are going to start deciding what people should do and put up hurdles to their doing something else that they (in our view, irrationally) want to do? Someone else will have to answer that question – it seems like a much broader issue than 401(k) policy.
Require ‘high cost’ plans to allow in-service IRA rollovers
In contrast to the ‘401(k) investor sophistication test’ proposal, the proposal to allow participants in ‘high cost’ plans to ‘roll out’ in-service to an IRA is ‘pro-choice.’ Like all Ayres and Curtis’s other proposals, adopting such a rule would disrupt the economics of 401(k) plans. But it aims, simply, to give participants in certain (high-cost) 401(k) plans the same investment rights/access to the market they would have if they were not in the plan. And it recognizes a reality that most would acknowledge – that in the small plan world sponsor-fiduciaries are often no better informed (or more ‘prudent’) than their employees.
In closing, let’s consider another quote from Alicia Munnell (Will Regulations To Reduce Ira Fees Work? Alicia H. Munnell, Anthony Webb, and Francis M. Vitagliano, 2013):
At its most basic, Ayres and Curtis’s work shows that, compared to a super-cheap index fund-based plan, 401(k) plan participants in medium- and small-sized (and even in some large-sized) plans may be ‘losing’ on average 156 basis points annually to ‘excess’ fees (relative to Ayres and Curtis’s super-cheap plan) and asset allocation mistakes.
This is not the 1990s – consistent double digit returns are, in the view of most, unlikely. In that context is there a justification – persistent positive alpha? diversification? participant freedom of choice? better service? – that makes such a cost spread reasonable? That’s not a rhetorical question – it is the challenge that Ayres and Curtis’s work presents.