There is a little noticed change in the IRS’s recent update of Publication 571 (which is the IRS’s 403(b) technical guide). In a highlighted box on page 4 is the following, under how the limits on “annual additions”-otherwise known as the 415 limits- apply:

More than one 403(b) account. If you contributed to more than one 403(b) account, you must combine the contributions made to all 403(b) accounts maintained by your employer. If you participate in more than one 403(b) plan maintained by different employers, you don’t need to aggregate for annual addition limits.

This innocuous seeming statement is actually pretty outstanding, finalizing a quiet morphing over a generation of the way the IRS applies a regulation in a way we rarely see.  The change: the employee is effectively no longer considered the employer for 415/403(b) purposes, but the rules have not really changed. In Gemology, according to Evan Giller-who first put me on notice to this “gem” of a change in 571- this type of change would be considered a “pseudomorph.” Pseudomorph is the name given to a crystal (or other gem) where one mineral replaces the mineral which actually formed the crystal without actually changing the the form or shape of the crystal.

Okay, I admit to taking some literary license with the analogy, but it seems appropriate here.

1.415(f)-(1)(f) of the tax regulations reflects its history: “the participant, and not the participant’s employer who purchased the section 403(b) annuity contract, is deemed to maintain the annuity contract…” This technically means that the employee is the employer for 415 purposes, and that all contributions to all 403(b) contracts of an individual count against the 415 limit-regardless of which employer plan it came from. For those who were around back then, I seem to recall that this was also the way we would run the 403(b) Maximum Exclusion Allowance (the now extinct “MEA”).

With the demise of the MEA, and with the growing focus of the IRS on employer accountability, we saw a shift over the years to treating the 415 limit as an employer-based limit, not an individual limit. This actually makes a lot of sense, as this limit is virtually impossible to enforce if applied beyond an employer.

The 2007 403(b) regulations actually were a bit ambiguous on this point, as they never touched on that pesky 415 reg. It only mandated the aggregation of 403(b) contracts of the same employer (with the exception of other companies under the participant’s control).  There has always been this uncertainty since then, because if the employee was the employer (as per 415), it still was an individual limit. The question was whether we could now effectively read the new 403(b) reg together with the old 415 reg to apply the 415 limit on an employer by employer basis, rather than treating the employee as the employer. The language under 415 could actually be read that way, if looked at generously (I guess).

Publication 571 seems to have solved this for us. It appears that the transition is now complete. There is no 415 aggregation of contributions to 403(b) contracts from unrelated employers (except, of course, for that weird “control” rule).  A Pseudomorph….





There has been increasing interest in the market to put together Multiple Employer Plans for 403(b) plans, and with good reasons. Tax exempt entities are really well suited to the sorts of economies of scale that a MEP can bring, and they often organize well around common associations.

But a 403(b) MEP is really complicated when you get down to it because-like anything 403(b), it seems-of the devil that exists in the details.  For example, besides  having to deal with the DOL 2012 Advisory Opinion on all MEPs; and  the fact that the Tax Code Section governing 401(a) MEPS does not apply to 403(b) plans (which means you can never really have a 403(b) MEP for tax purposes); and that the ERISA Section governing all ERISA MEPs (including 403(b) MEPs) requires compliance with that Tax Code Section which doesn’t cover 403(b) MEPS; you still have to deal with the complications of dealing with those legacy contracts of the various participating employers in the MEP. Then there is the issue of dealing with those combination of ERISA and non-ERISA 403(b) plans commonly sponsored by a 403(b)participating employer. These require use of traditional state law agency rules to make them work.

Yes, this is all manageable if you can have a MEP in the first place. The point here is that there are a lot of moving pieces in a 403(b) MEP that you don’t have to deal with in a 401(a) MEP. Using the aggregation model (see, for example, Section 202 of the recently introduced Retirement Enhancement and Savings Act of 2018 (“RESA”) which describes this model well) instead of the MEP -which many school districts actually do and call them MEPS- is so often more attractive.

What makes it all the more challenging is that you are dealing with the DOL’s regulatory processes which are geared to MEPs, while also dealing (on the same arrangement) with IRS regulatory processes which deal with each participating employer in a MEP as a single employer plan. The impact on this really falls on the IRS. The Form 5500 epitomizes the problem.

It starts with Form 5500 Section 1 which requires that you identify the plan as a MEP or a single employer plan.  For ERISA purposes, the answer is MEP, and you must check the box. However, for Tax Code purposes, it is not a MEP-do you also check the single employer box?

What flows from that may cause the IRS heartburn. The financial statements are driven by ERISA’s rules and not the Code’s, though the IRS relies upon the DOL information when performing their own compliance activities. This means that such  information is  required to be reported on an aggregated MEP basis, not on a single employer basis. The only “single employer” data provided on the 5500 is each participating employer’s EIN, along with something called “percentage of contributions” for each of those employers. The compliance questions will also be answered on the consolidated MEP basis, not on the “single employer” basis, which may be of a concern to the IRS.

It’s also likely to become messy when the IRS is looking for an individual plan Form 5500 for the 403(b) plan under the EIN of the employer and with a plan number and doesn’t find one. This is exacerbated by the way a plan joins the MEP: it typically merges with the plan of the lead sponsor, but the merger is with a “plan” the IRS doesn’t recognize as a plan.

Let’s not forget about the Form 8955 SSA, either.  This is a tax form, and it is required of all 403(b) plans subject to ERISA’s vesting standards. It is not an ERISA form. With each participating employer under a 403(b) MEP being considered an individual plan, a separate 8955-SSA has to be signed and filed on behalf of each participating employer. The IRS will have difficulty tying this back to the MEP’s Form 5500.

I truly doubt that the IRS’s systems are up to all of this and other process issues I haven’t mentioned, especially given the IRS’s budgetary strain. Even the provisions in RESA which will permit Open MEPS® won’t address these issues. It is likely to exacerbate them. So be prepared to deal with some bizarre regulatory twists in dealing the IRS when working with a 403(b) MEP,  to manage through this lack of regulatory processes, or to seriously consider the aggregation model.



There is a growing public policy concern regarding the ability of workers in the gig economy to access the ability to accrue retirement savings,. Though IRA’s and SEPs are, of course, available, they are nowhere near as effective as worksite retirement plans. Troy Tisue’s testimony at yesterday’s congressional hearing on retirement plans for the gig economy addresses how Open Meps® may help address this issue. In addition to his written testimony, Troy also discussed at the hearing on how there are ways in which aggregation programs-which are currently available- can be used to address this need. To read Troy’s testimony, click here.


The DOL’s proposed regulation permitting Association Health Plans which cover unrelated employers is likely to have a significant impact on the market’s ability to offer Multiple Employer retirement plans to unrelated employers. This is because the regulation permits AHPs through the modification of the ERISA regulatory definition of the term “employer” under Section 3(5). It was this definition that the DOL used to turn down the idea of an Open MEP in Advisory Opinion 2012-04. To be an “employer,” the DOL claimed there must be “commonality” of employment between employers who “controlled” the plan.

The DOL makes a stunning admission in the AHP reg preamble about its historical application of the “commonality” rule. It sounds an awful lot like the argument we made in requesting the MEP advisory opinion:

These definitional terms are ambiguous as applied to a group or association in the context of ERISA section 3(5), and the statute does not specifically refer to or impose the particular historical elements of the “commonality” test on the determination of whether a group or association acts as the “employer” sponsor of an ERISA covered plan within the scope of ERISA section 3(5). Accordingly, that determination may be more broadly guided by ERISA’s purposes and appropriate policy considerations, including the need to expand access to healthcare and to respond to statutory changes and changing market dynamics.

In a nutshell, the DOL is proposing the following:

  • An association can sponsor a health plan which can permit other, unrelated employers to join.
  • The association can be formed for the sole purpose of offering the plan, lifting the long-standing prohibition against associations being formed for the sole purpose of providing the benefit plan.
  • The “commonality” requirement will be met if the members of the association are in the same trade, industry, or state or metropolitan area (even if it crosses state lines).
  • The direct or indirect “control” requirement will be met if the members of the association elect the directors or officers of the association

Interestingly enough, the AHP definition of employer will permit companies with no employees, but with only  “owner-employees,” to participate-which was always a bugaboo with MEPS to begin with. There is also a health care related “non-discrimination” rule which doesn’t apply to retirement plans.

It is highly unlikely that the  DOL will be able to hold the line now against MEPs formed through associations meeting the criteria outlined above, or what I call “AMEPS” (for “Association MEPs), especially considering what the DOL stated in its MEP advisory opinion:

“In your submission, you urge that the Department’s historical interpretation of “employer” under section 3(5) of ERISA regarding multiple employer welfare arrangements (MEWAs) should be restricted to welfare plans and that a less restrictive interpretation be applied to retirement plans. The Department is of the view, however, that the term “employer” should have the same meaning in this context whether applied to the term welfare plan or pension plan. See Sullivan v. Stroop, 496 U.S. 478, 484 (1990).”

 Does this mean that we now run out and establish AMEPs? That, I think, would be ill-advised given that there are likely to be a number of changes to the proposed regs before they become final, and there really are a number of issues related to the proposal which need to be answered. For example, the ERISA language under 3(5) enabling one employer to act for another does not require any sort of trade or geographical link and, given the ways companies morph and move, such a requirement seems to unnecessarily deny modern business realities. Lets see, under the reg, you don’t have to be in same trade as long as you are in a large metro area. But if you have a multi-state, international business in a small town or state (yes, there are many of us like this, given the digital nature of  business),  you can’t belong to an Association plan of an unrelated trade which is based in another state.  There’s sound policy behind this archaic thought? That seems counterintuitive, and restricts small business access to true competitive pricing- particularly where the benefits are not inherently “trade based.” But at the very least,  it does give us a strong basis to advocate the same treatment for association MEPS as AHPs. And for this, no statutory change would be needed.







We have now begun the “season of restatement” for 403(b) plans, which will run until March 2020. This is not a joyous yuletide sort of season; it is instead one of reckoning, I’m afraid. This is because we have been able to effectively ignore a number of difficult issues involving plan documents until now, because it was never critical to address them. But now we must.  It really is going to be a season of learning for us all-for the regulators at the IRS and DOL and practitioners as we try to weave our way to some sort of sensible solutions.

The application of the Spousal Consent and Joint and Survivor Annuity rules is one such issue which we need to consider.  In the 401(a) space, dealing with this issue is pretty straightforward.  The rules apply uniformly to all 401(a) plans through 401(a)(11) and 417, which is well coordinated with ERISA’s requirements under Section 205. As long as the default form of benefit under the plan is not a J&S benefit; and the spouse is entitled to the survivor benefits upon the participant’s death unless the spouse consents otherwise; the rules are met. If, however,  the plan is a money purchase plan, or holds money transferred in from a money purchase plan or a defined benefit plan, or the plan’s default distribution is an annuity payment, the spouse must consent to a lump sum distribution-or even to a loan. Many will remember the efforts we went through years ago cleansing those annuity payment requirements from plans and products, just so no spousal consent would be required for lump sum distributions or loans.

Now, though, we have to deal with documenting these rules as they apply to 403(b) plans. Its not so simple:

  • 401(a)(11) and 417 do NOT apply to 403(b) plans This means the only way the spousal rights rules apply to 403(b)plans  is through ERISA Section 205.  This means that if you have a non-ERISA 403(b) plan, the spousal rights rules DO NOT apply. Be careful to avoid inadvertently applying them in the restatement.
  • HOWEVER, many non-ERISA 403(b) annuity contracts impose the spousal rights rules as a matter of contract language, even though they may not be imposed as a matter of law. Because the underlying contract is technically part of the plan document, this means those spousal rights will apply as a matter of a plan term. Be careful with your documentation of this. There is an interesting problem here: if the central “wrap” plan document states there are no spousal rights, but the underlying contract states that there are, there is a conflict-and  IRS rules says that the wrap plan document language will govern. But the problem is that the insurer is legally bound to honor the spousal rights as a matter of contract, and CANNOT legally ignore such claims. Screwing up the plan document language will inevitably lead to litigation, especially given when large account balances are involved, so be careful.
  • ERISA 403(b) plans are covered by the spousal rights rules, by virtue of ERISA Section 205. If the underlying annuity contract provides that the default payment is an annuity-as many old contracts, particularly individual ones will-ERISA will require spousal consents for lump sum payments and loans. Even if the wrap plan document makes the normal form of payment a lump sum in an attempt to “cleanse” this problem, the DOL has never concurred with the IRS’s position that the centralized plan documents control. The insurer is legally bound to follow the contract terms; and it is a registered security, which can cause security law problems if this language is ignored. So, again, be very careful in your restatement, especially if you have legacy contracts in the plan.
  • Finally, there’s the “money purchase plan” issue. It is typical for a 403(b) plan to provide a set percentage of pay as the employer benefit under the plan, much like a money purchase plan. There is a very real and unresolved issue as whether or not this set formula makes the plan a “money purchase plan” which is subject to the J&S rules, even if you have avoided the “annuity” language problem noted above. There is a sound position that such plans are NOT money purchase plans. [*] This  is a substantial issue and one upon which you will need to make a determination when you restate the 403(b) plan.

[*] A money purchase plan is defined only by Code Section 401(a)(27), as a 401(a) tax qualification requirement-which does not apply to 403(b) plan. Under ERISA 205, an individual account plan becomes subject to the J&S rules only if it is a “money purchase plan” under Section302(a)(2)(B).  ERISA does not define “money purchase plan,” and it would be difficult to apply any other definition than that which is contained in the Code-which, here, does not apply to 403(b) plans.


Any discussion on any tax issue addressed in this blog (including any attachments or links) is not intended to be used, and cannot be used, for purposes of avoiding penalties imposed under the United States Internal Revenue Code or promoting, marketing or recommending to another person any transaction or tax-related position addressed therein. Further, nothing contained herein is intended to provide legal advice, nor to create an attorney client relationship with any party.

On November 14, the Office and Management and Budget received from the DOL a proposed rule which would broaden the definition of employer under ERISA to allow more businesses to join association health plans.

This was in response to the  Oct. 12 executive order which directed the DOL to “consider expanding the conditions that satisfy the commonality‑of-interest requirements under current Department of Labor advisory opinions interpreting the definition of an “employer” under section 3(5) of the Employee Retirement Income Security Act of 1974.”

The DOL’s advisory Opinion on MEPs in 2012 was specifically premised on the DOL’s interpretation of   the definition of “employer,” for health plan (MEWA)  purposes. The DOL ‘s position was that it had no authority to redefine their historical definition of employer it developed for MEWA purposes merely for retirement plan purposes, that it was bound by the statute to apply the same “employer” definition to both health plans and retirement plans. Does this mean that the new proposed definition of “employer” will, necessarily, by operation of statute, be expanded for retirement plan purposes as well? If this is the case, MEPS would be opened up to unrelated employers under the same set of circumstances as they would for health plan purposes.

No one has seen the specific language of the proposed regulation, of course. But it would seem that the DOL (in its own collective mind) really has no authority to expand the definition of “employer” for ACA/Association Health plan purposes without also expanding it for retirement plans.

We will watch closely for this new proposed reg when it is published. But it’s hard to see how this would not also apply to retirement plans, and thereby opening up MEPS. Even if the language of the proposed reg is limited to health plans, there will be an opportunity under the comment process to make the case that there, legally, can be no difference-at least according to the DOL’s own rules….


Anyone who has had the pleasure of dealing with the 403(b) annuity recognizes that it is truly an “odd duck” – especially when looking at it through the lens of the 401(a) experience.  So it should come as no surprise that this “odd-duckiness” shows up in settlements under the EPCRS’s Closing Agreement Program” (or CAP).

The CAP is the portion of EPCRS which is used when the IRS discovers uncorrected plan errors on audit. Ultimately, it involves the plan sponsor negotiating a final settlement with the IRS. This inevitably also involves the agreeing to pay a CAP sanction which, under current guidance, cannot be less than the fee that would be payable had the sponsor corrected under the IRS’s Voluntary Compliance Program.

There are formally 10 different factors under EPCRS which the IRS takes into account when proposing the sanction amount. One of those elements (and what used to be the most significant factor before the 2106 modifications to the program under 2016-51) is the “Maximum Payment Amount” (the MPA). The IRS agent is required to calculate the MPA as part of the CAP process, which is the tax which the IRS could collect because of the failure of the plan to meet the tax compliance rules. This includes the:

  • tax on the “trust;”
  • additional income tax on the employer from the loss of employer deductions (plus interest and penalties arising from the loss);additional income tax on the individual from the failure (including taxability of a loan);and
  • any other tax which may result from the failure that would apply but for the CAP settlement.

This number in the 401(a) world can be a very large number, even for a small plan, especially when taken over a number of years.

But it’s interesting to see the effect on a 403(b) plan going through this process:

  • There is no tax exempt trust here that is enjoying 501(a) tax exempt treatment. An annuity contract or a custodial account are not, themselves, taxpayers-or entities which can be taxed. The individual who owns the contacts can be taxed (so, the third bullet point applies), but that’s it. This number should be zero.
  • The employer is a tax-exempt entity. This means there should be no loss of deduction, except in the very rare case where a tax exempt might have been taxed under the UBIT rules.
  • The individual will suffer taxation, the amount of which should be included in the MPA. But there are a couple of important factors which come into play here:
    • “Inside buildup.” The tax rules which apply to “non-qualified” annuities (which is the terms we use to describe annuities sold outside of 401(a) and 403(b) plans) defer the taxation of earnings of the investments inside that contract (including the “separate account” investments) until those amounts are distributed-and even then, those earnings might only be taxed (if it’s considered a “distribution as an annuity”) using something called the “exclusion ratio.” The exclusion ratio has the effect of back loading the tax into much later years. Because most 403(b) annuity contracts will qualify as annuities for “inside buildup” purposes-even if they fail 403(b) treatment- the earnings should be excluded from the MPA. Note this treatment will not apply to custodial accounts.
    • Finally, virtually all of the operational errors (except for non-discrimination) will only impact the contracts against which the error occurred. So, for example, a 415 violation will only impact the particular contract to which the excess was made, and only to the extent of the excess. It does NOT “blow up” the entire plan. So, the MPA should only include the tax on portion of the excess in the account of the affected individual.

Though the MPA has a much more limited impact on the sanction amount than prior to 2016, it still can affect the ultimate sanction by showing that many errors may not really be all that egregious (especially when compared to a 401(a) sanction amount on the same type of error).


The implementation of the IRS’s 403(b) pre-approved plan program by a number of vendors has been delayed by at least one thorny issue which will cause concern for a number of 403(b) plan sponsors. The IRS seems to have technically resolved the issue with its guidance on “effective date addendums,” (see ).  But at least one of the issues it was meant to address has left many plan sponsors with questions of what to do now.

One of the roots of the problem is what the IRS generally refers to as the “once in, always in” rule of 403(b)’s “universal availability” requirement.  “Once in, always in” refers to the IRS position that once an employee has become eligible to make elective deferrals into the plan by reaching the 1000-hour benchmark in one year, that employee will remain eligible to make elective deferrals in future years-even in years where that employee does not complete 1000 hours of service.  That person can only lose that ability to defer if they become part of one of those limited categories of employees which can be excluded (such as certain students).

Now, this rule is not explicitly stated in the 403(b) regulations, nor is it part of the Code Section 403(b). Though this position has not been widely publicized over the years, the IRS applies it on audit, and the Internal Revenue Manual itself contains this interpretation as well.

It all came to a head in the pre-approved document process. The IRS requires that the “once in, always in” term be included in the pre-approved document.  The practical problem that came to the attention of the IRS is that many employers may not have not applied this rule, over time, consistent with the required plan document term. With the pre-approved document covering practices back to January 1, 2010, plans which adopted this the pre-approved plan with this term automatically included term might immediately have a “form and operation” problem which would force them into EPCRS.

This problem is not limited to the “once-in-always-in”rule, as there can be any number of practices required by the pre-approved document which the employers have not “practiced” since January 1, 2010. The “effective date addendum” addresses this “automatic” form and operation problem, though it will really serve to complicate plan 403(b) plan documents even further. Why I say “even further” is because 403(b)plan documents are complex beasts to begin with. For example, they incorporate by reference the underlying annuity contracts and custodial agreements as plan terms. A number of us have raised the issue that there are going to be inevitable conflicts between the central plan document and those (often complex)  underlying contracts, and the IRS’s position that the centralized plan document must control in the event of such conflicts may not always be tenable.

The “effective date rule” does address the technical issue of adopting terms which were not applicable back to January 1, 2010. What plan sponsors need to be aware of, though, is the need to implement the once in, always in rule going forward. But it does not address the elephant in the room: what happens to the employer which did not use that rule back to January 1, 2010, and admits in their plan document that they didn’t?

A couple of years ago, I noted in a blog the legal challenges of offering to 403(b) plans the type of collective trust investments  are sold to 401(k) plans. This can be very diffiuclt-if not virtually impossible- to do because of the way the technical rues apply.

The University of California made news when it announced that it has found a way to apply the collective trust rules in such a way  to make just such an offering to its 403(b) plan. I was not involved with its development. However, if they did what I think they did, it really does speak to the sophistication of their counsel AND the hoops through which one must jump in order to make it happen. The downside to this is that it still does not “crack the nut” to make CIT’s  available on a bulk basis as 81-100 trusts for unrelated employers. But it still is pretty cool.

It is possible they could have done something like the  following. Make sure you have a copy of your US Code handy (or at least you’ve pulled up the  Cornell’s on-line US Code  tool). Its complicated. Or, as they say, don’t try this at home…..

403(b)(7)(C) requires that a custodial account hold shares of a “regulated investment company” (or a “RIC”) as defined in Code Section 851(a). 851(a) actually has thee different types of RICS. Two of them are under 851(a)(1), which are your run-of-the-mill mutual fund and a unitized investment trust  (for example, a 403(b) annuity contract uses the unitized investment trust version of a RIC). My earlier blog focused on these two types, because that is what is typically offered in the marketplace.

However, 851(a)(2)  permits a third type of RIC. An 851(a)(2) RIC is  a common trust fund “or similar fund” which is otherwise excluded from the definition of a RIC  under 15 USC 80a-3(c)(3) (which is part of the Investment Company Act of 1940, which regulates investment companies). 80a-3(c)(3) excludes a common trust from the definition of RIC (for that section’s purposes, but not for 403(b))  as long as the trust interests are not (i) advertised; (ii) offered for sale to the general public; and (iii) “fees and expenses charged by such fund are not in contravention of fiduciary principles established under applicable Federal or State law.”

In order to qualify use this type of RIC as a 403(b) regulated investment company,  the common trust also cannot be included in the definition of “common trust fund”under Code Section 584(a) (per Code Section 851(a)(2)).  584(a) does NOT include state chartered  trust companies, it only covers national banks and trusts.

What does this all mean? It means that as long as the CIT is offered by a state chartered trust company, and meets the 80a-3(c)(3) limitations (and the other limitations under 851) explained above, the interests in the trust can be purchased by a 403(b) custodian.

Add to this the fact that the UC plan is a governmental plan, the CIT interests would be exempt from registration under the Investment Company 40 Act-meaning no prospectuses, proxies, or other such requirements the typical mutual fund or registered annuity contract must provide. The plan’s 403(b) custodial account could then hold the interests in that CIT.

There’s even a PLR on it; PLR 8003155.

Even if this is not what actually happened, (and there are a number of details and legal issues which needed to be resolved in order to actually make this work), it could be something close to this. For sophisticated state universities with large plans, or even those large  consolidated K-12 plans, this could be a very real option. I suspect we’ll be seeing more of the same.

My hat’s off to the folks who put it together!