Some Additional Reflections on MetLife
Way back in 1989, in the Bruch v. Firestone decision, the Supreme Court told us that a plan decision-maker’s rejection of a benefit application would be subject to de novo judicial review . . . unless the plan included magic words vesting the decision-maker with DISCRETION. If the decision-maker has discretion, the trial court’s review of the decision must be deferential, only to be reversed if “an abuse of discretion,” “unreasonable” or “arbitrary and capricious.”
Within 30 minutes of the decision, every lawyer worth his salt or at least seeing an opportunity to generate some fees, wrote a letter to every client with an employee benefit plan saying, in effect, pay me some money to review your plan to make sure it gives you discretionary authority and, if it does not, pay me a little more money to amend your plan to give you discretionary authority. Any plan today whose benefit rejections are subject to de novo judicial review (for lack of the magic words) probably has a potential malpractice action against its attorney. In the real world we live in, the default principle of de novo review of Firestone is a dead letter.
But the Supreme Court said something else in Firestone: a court, when it is reviewing a benefit rejection by a decision-maker armed with discretionary review, should consider, as a factor, any conflict of interest of the decision-maker. One clause in one sentence is all we got explaining this. We didn’t know, for example, how to tell whether a decision-maker had a Firestone conflict, and we didn’t know what a reviewing court was to make of such a conflict? Does it mean de novo review? Does it mean a little bit less deference? Or something else entirely?
Well now, 19 years later, the Supreme Court has decided to give us an answer to these questions. It was not worth the wait.
Look first at the majority opinion. Here is the two sentence version of the majority opinion: in deciding whether a conflicted plan decision-maker’s rejection of a benefit claim is an abuse of discretion, a district judge should consider and weigh all factors, of which a conflict is one factor. The conflict is not too important, though, if the plan has structural safeguards against decision-maker bias, such as erecting a wall between the “claims administrators and those interested in firm finances.”
Here are three problems with the majority opinion.
1. There are now two levels of discretionary review: first, the plan claims administrator has discretion in deciding whether the plan should pay benefits, and second, the district court has discretion in deciding whether the plan claims administrator’s decision was an abuse of discretion. (When a district court is given authority to “balance” all relevant factors, an appellate court will generally not reverse the district court’s decision unless the district court abused its discretion.)
The cynic in me says that this means that most cases will turn on whether a district court judge is philosophically oriented toward business autonomy or consumer protection. Of course, the cynic in me also says that this will not be a very serious break from the past. What may be a break from the past, though, is that an appellate court will have a bit more trouble reversing a decision it does not like. (Frank Cummings makes these points in a comment he wrote in response to Andrew Oringer’s discussion of MetLife.)
2. The issue that no one discusses about Firestone is the standard against which a claims administrator’s discretion is tested. (And this should be a key issue whether or not there is a conflict.) What does it mean to abuse discretion? Courts talk endlessly around this issue but never focus on exactly on how discretion is evaluated. Oh sure, we know some things. You can’t treat some participants radically differently from other participants: that would be arbitrary. You can't ignore undisputed evidence. That would be wrong. But these are examples of arbitrary decisions. They do not define the boundaries of the discretion that a plan administrator may not abuse.
So when we say that an administrator has discretion, what do we mean? Here are a few possible suggestions:
1) When a plan vests an administrator with discretion, it means that the administrator will resolve all interpretative and evidentiary questions against the claimant. (And if this is what it means, should plans make some sort of disclosure to participants so that they understand this?) Here, the standard against which decisions are evaluated is “defensible even if not correct result.”
2) When a plan vests an administrator with discretion, it means that the administrator will strive to reach a correct legal judgment. Note that the reason for judicial discretion here is not to save money by reducing benefits, but to minimize litigation costs, since the plan’s expenses in litigating a benefit denial would arguably be lower than under de novo judicial review. Here, the standard against which decisions are evaluated would be “the process was fair and the administrator was neutral and fair.”
3) When a plan vests an administrator with discretion, it means that the administrator will apply his special understanding of the employer and its employees in resolving benefit claims. Here the standard against which decisions would be evaluated would be “legally correct unless the special circumstances of the employer and employee make a different approach defensible.”
4) When a plan vests an administrator with discretion, it means that the administrator knows the price point at which the employer will terminate or cutback the plan and will consider this point in deciding close benefit questions. I’m not even going to attempt to construct the standard against which decisions would be evaluated under this understanding of discretion.
5) When a plan vests an administrator with discretion, it means that the administrator will endeavor to make benefit decisions that conform to employee understanding of the plan. Here the standard against which decisions would be evaluated would be “reasonably in accord with the reasonable expectations of plan participants.”
The majority opinion probably best accords to 2), but most plan sponsors, at least at some level, probably would endorse 1), and 1) certainly comes closest to explaining judicial decision-making, at least in the pre-MetLife world. But all of these standards are consistent with the idea of a plan administrator exercising discretion, even though they would sometimes produce markedly different outcomes.
3. The majority opinion does not fully resolve the question of when a plan administrator has a conflict, but certainly hints that conflicts are the rule, not the exception. (Justice Robert’s dissent emphasizes this point.) The question, though, is whether there are always conflicts.
Let me posit two the two classic situations where at least some people would have argued, pre-MetLife, that the claims administrator did not face a conflict: First, an overfunded defined benefit plan. The approval of a claim will not come directly out of the employer’s pocket, but the employer will be affected by the approval, either in the form of increased future contributions or a reduced potential reversion. But does this a conflict make? (Maybe the one situation where there would be no conflict is where a defined benefit plan is overfunded and provides that all surplus assets will be allocated to plan participants.) Second, an employer who contracts with a third party claims decision-maker who decides but does not pay claims. Since the employer will choose the firm, and the employer may want to choose a firm that will reduce its costs, the firm may have a financial interest in rejecting close claims.
(The majority opinion sort of endorses, or almost endorses, a similar argument about employers selecting insurance companies, since the premiums they charge will partly reflect their claim review policies. It says “For one thing, the employer’s own conflict may extend to its selection of an insurance company to administer its plan. An employer choosing an administrator in effect buys insurance for others and consequently (when compared to the marketplace customer who buys for himself) may be more interested n an insurance company with low rates than in one with accurate claims processing.”)
So, will almost every case pose a conflict, as Justice Roberts suggests in his dissent? And will this mean elevated review in all situations? The majority opinion is a bit circular here. The conflict will not count much if the one who makes decisions is walled off from the folks concerned about his or her employer’s bottom line. But if the decision maker is sufficiently walled off so as not to care about the employer’s profitability, why is there a conflict at all? If we say, well, there is still some potential for conflict because you cannot really wall off any employee from concerns about the employer’s bottom line, how can we say with any real confidence that the conflict is really only a small one, as the majority seems to want us to say?
What about the dissents. We can ignore Kennedy’s dissent, which says I agree with the majority decision but think we should give MetLife a chance to show that its conflict was insignificant.
Justice Robert’s decision is the opposite: I agree with the majority that MetLife was really, really wrong when it rejected Glenn’s claim, but I disagree with virtually everything else that the majority opinion says, especially that a district court, in exercising review over a claims denial, must take into account every conflict, no matter how infinitesimal. Justice Roberts would have preferred a standard that said, in effect, a conflict only matters if it affected the claims process; conflicts don’t matter, in this formulation, unless the plaintiff can demonstrate that the claims administrator in fact yielded to the conflict. My guess is that under the Roberts’ formulation, plaintiffs would hardly ever be able to persuade a court to consider a conflict, unless of course there is a whole lot of discovery and quite possibly an involved trial, because credibility of witnesses would sometimes if not often be central to the determination of whether a conflict infected the decision-making.
The difference between Roberts and the majority, then, appears to be on who will bear the burden of proving the importance, or non-importance, of the conflict. Under the majority, the plan must prove that the conflict probably did not affect the decision, while under Roberts, the participant would have to prove that the conflict probably did affect the decision.
The separate dissent of Justice Scalia and Thomas found even Justice Roberts’ dissent unconscionably unfair to the employer. A conflict, in their view, should be irrelevant: all that should matter is whether the decision is reasonable. There are other plausible ways of thinking about the Scalia/Thomas dissent, but exploring them would be a waste of whatever bandwidth would be required.