In an informative post on Balkinization last week, Nelson Tebbe, Richard Schragger, and Micah Schwartzman (TSS) explain how a critical premise of the Supreme Court's decision in Burwell v. Hobby Lobby is being violated, at least temporarily. The premise (which Justice Kennedy's concurrence appeared to make a necessary condition of his providing a fifth vote for the majority, and which is arguably required by the Establishment Clause) is that providing corporate employers with an exemption from the legal obligation to provide employees with health insurance that covers contraception without cost sharing will not impose harm on third parties--i.e., that employees will continue to have access to the same coverage.
As TSS note, because the mandate in the case has issued, Hobby Lobby and Conestoga Wood are currently exempt from the contraception insurance requirement. The Administration is working on a proposed IRS/EBSA/HHS rule that would provide the employees with no-cost-sharing contraception insurance without their employers' participation, but that new rule is not yet in place. However, under the Supreme Court's own 1988 ruling in Bowen v. Georgetown University Hospital (GUH), a federal agency does not have the authority to promulgate retroactive regs unless that authority is expressly delegated by Congress--which has not delegated such authority here. Thus, TSS conclude, the employees will not be reimbursed for their out-of-pocket contraception costs incurred between the issuance of the mandate and the finalization of the new rule. Accordingly, at least during the interim period, the premise of no burden on third parties is false.
I agree with the analysis of TSS but I want to add a wrinkle concerning Bowen v. GUH. Because that case only states a presumption of statutory construction, it is overridable and, in my view, should be overridden here by RFRA.
But first, let's consider a hypothetical constitutional case. Suppose that Hobby Lobby arose not under RFRA but under the pre-Employment Division v. Smith Free Exercise law and suppose that the Court concluded: 1) that the Constitution required the government to accommodate Hobby Lobby if it could do so without imposing significant costs on third parties; and 2) that such an accommodation would indeed be readily available; but 3) only if the relevant federal executive agencies promulgate a reg with retroactive effect for the interim period. Under such circumstances, I think the right answer would pretty clearly be for the Court to order the government to accommodate and to promulgate a partially retroactive reg. After all, if the agency can't promulgate a retroactive reg, then accommodating Hobby Lobby does impose significant costs on third parties, and so Hobby Lobby is not entitled to an accommodation. But an otherwise valid claim of constitutional right (to accommodation) should not have to yield to a mere rule of statutory construction (concerning delegation of retroactive rulemaking authority).
Matters are somewhat more complicated with an actual RFRA claim but I think we reach the same bottom line. Now the question is whether Congress, when it enacted RFRA, intended to delegate to administrative agencies the authority to promulgate retroactive regs if doing so was necessary to providing religious accommodations (because necessary to prevent some class of third parties from being burdened by an otherwise-required accommodation). I think the answer is probably yes. Assuming that the government can accommodate relatively easily (which will be true if the law without the accommodation is not narrowly tailored to serve a compelling interest) then the Congress that wrote RFRA can be presumed to want the relevant agency to promulgate the needed reg, even if it must be retroactive.
The immediately preceding paragraph is subject to three caveats.
First, I would concede that Congress when it enacted RFRA almost certainly did not actually intend to authorize retroactive rulemaking because it's doubtful that anyone in Congress anticipated the problem that has now arisen. So when I say that Congress intended to delegate the relevant authority what I mean is something more like the overall purpose of Congress in enacting RFRA is best facilitated by reading it to encompass subsidiary power in the courts to give effective remedies, even if that means displacing some default rules of statutory construction. This strikes me as unproblematic unless one is a fairly strict textualist or one thinks that the clear statement rule of Bowen v. GHS is constitutionally required. I'm not a strict textualist. Meanwhile, I read Bowen v. GHS as connected to constitutional principles regarding congressional power to delegate authority to agencies, but I do not read it as saying that the clear statement rule is in fact constitutionally required.
Second, in my substantive analysis I asked whether Congress would prefer 1) no accommodation or 2) an accommodation plus retroactive rule. But there could be a third option: 3) an accommodation that becomes effective only after a purely prospective reg has been promulgated. If the SCOTUS thought that this was the preference of Congress, then it should have ordered the agencies to promulgate a rule as soon as possible, with Hobby Lobby and Conestoga Wood to receive their exemptions only after the rule became effective. The Court did not pursue this option (perhaps because it was unaware of the Bowen v. GHS issue) but I think it's probably inferior to option 2) anyway, because it means that plaintiffs like Hobby Lobby must comply with laws that violate their (valid) RFRA religious while they wait for the wheels of the bureaucracy to turn. Option 2) gives the claimants their relief sooner and, so long as substantial harm to third parties can be avoided by retroactive rulemaking, avoids substantially burdening third parties. Win-win.
Third, one might worry that the courts ought not to be in the business of delegating lawmaking power to federal agencies. This strikes me as a legitimate worry but one that does not distinguish between purely prospective rulemaking and rulemaking that has some retroactive effect. There are circumstances in which an accommodation of a valid RFRA claim would require a modification to an existing regulatory scheme even though the scheme, as modified, might fall outside of the apparent scope of any prior delegation to an agency. Nonetheless, RFRA entitles successful claimants to go to court to "obtain appropriate relief against [the] government." That is the language of broad remedial authority, sufficient, in my view, to include a court order to an agency to take action that would otherwise be outside of its authority. The only constraint here might be the non-delegation doctrine, but I would think that the very forgiving requirement of an "intelligible principle" would be satisfied by the initial grant of authority to the agency to make whatever pre-exception rule is at issue, combined with the policy of RFRA.