Opinion for the Court filed by Circuit Judge BROWN.

Concurring opinion filed by Circuit Judge KAVANAUGH, with whom Circuit Judge HENDERSON joins.

Approximately 1,700 U.S. Airways current and retired pilots are in the midst of a lawsuit challenging benefit determinations of the Pension Benefit Guaranty Corporation (“PBGC”). A subset of 111 plaintiffs (“the pilots”) requested a preliminary injunction to prohibit the PBGC from implementing its benefit determinations while the suit is pending. After considering the four factors for a preliminary injunction, the district court denied the motion. Because the pilots show neither a substantial likelihood of success on the merits nor irreparable harm, we affirm.

I. Background

On August 11, 2002, U.S. Airways filed for Chapter 11 reorganization in the bankruptcy court for the Eastern District of Virginia. See generally In re U.S. Airways Group, Inc., 296 B.R. 734 (Bankr. E.D.Va.2003). The bankruptcy court found the reorganization plan would create “a serious funding shortfall for the [company’s] defined benefit pension plan.” Id. at 738. The company moved for judicial findings to permit distress termination of the pilots’ retirement plan and notified the PBGC of its intent to terminate the plan. The bankruptcy court approved distress termination, and U.S. Airways agreed with the PBGC to set March 31, 2003 as the termination date of the plan.

The PBGC is a federal government corporation — created by the Employee Retirement Income Security Act of 1974 (“ERISA”) — that insures private sector defined-benefit pension plans. See 29 U.S.C. § 1302. The PBGC acts as guarantor of underfunded pension plans, paying benefits to participants such as the pilots in this action. See id. § 1322. When a plan is underfunded, the PBGC’s benefit payments are subject to statutory and regulatory limits. See, e.g., id. §§ 1322, 1361;[*1271] 29 C.F.R. § 4044.13. ERISA permits the PBGC to serve as trustee to administer an underfunded plan in addition to its role as guarantor. 29 U.S.C. § 1342(b). The PBGC has applied to serve as trustee in every terminated plan, and courts typically grant its application. Pineiro v. PBGC, 318 F.Supp.2d 67, 72 (S.D.N.Y.2003). Following this pattern, the PBGC was appointed to serve as trustee of the U.S. Airways retirement plan.

Under its usual practice, the PBGC began making initial payments to pilots based on an estimate of what the benefits would be. See Boivin v. U.S. Airways, Inc., 297 F.Supp.2d 110, 114 (D.D.C.2003). The parties agree that it normally takes the PBGC between two and three years to make a final, formal determination of benefits for each participant. See id. Without waiting for the final determination, a group of pilots challenged the estimated benefits. This court dismissed the claim for failure to exhaust, holding the pilots must await the final determination. Boivin v. U.S. Airways, Inc., 446 F.3d 148, 158-59 (D.C.Cir.2006).

When a final benefits determination differs from the estimate, the PBGC either repays the shortfall or collects the surplus. If a participant was initially paid a lump-sum estimate and the PBGC later determined the estimate to be too high, it requests repayment of the surplus amount in a process it calls “recovery.” If a participant is still receiving monthly benefit payments based on an initial estimate but the PBGC has determined that the estimate was too high, it reduces the amount of future monthly payments to recover the surplus in a process it calls “recoupment.”

In February 2008, the PBGC finalized its formal benefit determinations and notified 111 out of the 1,700 plaintiffs that it would be seeking recovery or recoupment, depending on the circumstance. According to the PBGC, of the 111 pilots, 74 were sent recovery notices. Jones Deck at 4. Those 74 pilots, on average, received lump-sum payments of over $680,000. Id. On average, the PBGC is seeking $7,049.67 in recovery, and no pilot’s recovery amount is more than 1.53% of the initial lump-sum amount. Id. The other 37 pilots are still receiving monthly payments; as to them, the PBGC is seeking an average recoupment of $59.86 per month. Id. No pilot’s recoupment is more than 10% of their current monthly payment. Id.

The 111 pilots sought a preliminary injunction against the PBGC’s recovery and recoupment efforts. The district court denied the injunction, Davis v. PBGC, 596 F.Supp.2d 1, 5 (D.D.C.2008), and we affirm.

II. Standard of Review

The denial (or grant) of a preliminary injunction is classified as an immediately appealable interlocutory order. 28 U.S.C. § 1292(a)(1). On a motion for a preliminary injunction, the district court must balance four factors: (1) the movant’s showing of a substantial likelihood of success on the merits, (2) irreparable harm to the movant, (3) substantial harm to the nonmovant, and (4) public interest. CFGC v. England, 454 F.3d 290, 297 (D.C.Cir.2006). This Court “review[s] a district court’s weighing of the four preliminary injunction factors and its ultimate decision to issue or deny such relief for abuse of discretion.” Id. Legal conclusions — including whether the movant has established irreparable harm — are reviewed de novo. Id.

The four factors have typically been evaluated on a “sliding scale.” Davenport v. Int’l Bhd. of Teamsters, 166 F.3d 356, 361 (D.C.Cir.1999). If the movant makes an unusually strong showing on one of the factors, then it does not necessarily have to make as strong a showing on[*1272] another factor. For example, if the movant makes a very strong showing of irreparable harm and there is no substantial harm to the non-movant, then a correspondingly lower standard can be applied for likelihood of success. See, e.g., WMATC v. Holiday Tours, 559 F.2d 841, 843 (D.C.Cir.1977). Alternatively, if substantial harm to the nonmovant is very high and the showing of irreparable harm to the movant very low, the movant must demonstrate a much greater likelihood of success. It is in this sense that all four factors “must be balanced against each other.” Davenport, 166 F.3d at 361. When seeking a preliminary injunction, the movant has the burden to show that all four factors, taken together, weigh in favor of the injunction. CFGC, 454 F.3d at 297.

The pilots, misreading our precedent, insist they need “only establish that serious legal questions are at issue” in order to succeed on appeal. Appellants’ Br. at 6. They reach that conclusion by focusing on language in Holiday Tours, where this court noted a movant need not necessarily show a 51% likelihood of success on the first prong of the preliminary injunction analysis. 559 F.2d at 843. But Holiday Tours did not eliminate the other factors. The court simply acknowledged that a lessor likelihood of success might suffice if each of the other three factors “clearly favors” granting the injunction. Id. The pilots argue that, because the other three factors are “not contrary to” a preliminary injunction, they need only raise serious legal questions. Appellants’ Br. at 7. They are wrong.

The Supreme Court has recently addressed the standard for a preliminary injunction. See Winter v. NRDC, — U.S. -, 129 S.Ct. 365, 375, 172 L.Ed.2d 249 (2009) (holding that irreparable injury must be likely, “not just a possibility”). We note that the analysis in Winter could be read to create a more demanding burden, although the decision does not squarely discuss whether the four factors are to be balanced on a sliding scale. See id. at 392 (Ginsburg, J., dissenting) (“[Cjourts have evaluated claims for equitable relief on a ‘sliding scale,’ sometimes awarding relief based on a lower likelihood of harm when the likelihood of success is very high. This Court has never rejected that formulation, and I do not believe it does so today.”). We need not decide whether a stricter standard applies, because the pilots fail even under the “sliding scale” analysis of Davenport.

III. Discussion

Of the eleven substantive claims in the lawsuit, only three are proffered in support of a preliminary injunction. First, the pilots argue the PBGC incorrectly interprets an ERISA provision prioritizing benefits. The relevant language from ERISA limits Priority Category 3 — the prioritization level relevant here — to benefits “based on the provisions of the plan (as in effect during the 5-year period ending on [the plan’s termination] date) under which such benefit would be the least.” 29 U.S.C. § 1344(a)(3)(A). The PBGC has determined the plan has sufficient assets to cover all benefits in Priority Category 3, but the pilots believe the PBGC has improperly excluded the U.S. Airways Early Retirement Incentive Program from Priority Category 3.

The program authorizes incentives for early retirement. The question is whether the program was “in effect during the 5-year period ending on [the plan’s termination] date.” Id. (emphasis added). The phrase is not defined in § 1344(a)(3)(A). The parties do not dispute the relevant dates: The program was adopted by U.S. Airways on December 4, 1997. The program included a self-defined effective date of January 1, 1998. By the terms of the[*1273] program, no pilots could retire or collect payments under the program until May 1, 1998. The pension plan in this case was terminated on March 31, 2003, so the statutory five-year period began on March 31, 1998. If the early retirement plan was “in effect” before March 31,1998, then it satisfies the statutory requirement of being “in effect” for the full five-year period. If, however, the early retirement plan did not go into effect until after March 31, then it would not satisfy the statute’s five-year requirement and thus would not fall into Priority Category 3.

The PBGC says “in effect” means operationally effective — when pilots could elect to retire under the Early Retirement Incentive Program and when they could begin receiving payments under the program. Thus, according to the PBGC, the program was not “in effect” until May 1, 1998 — after the beginning of the five-year period on March 31, 1998. The pilots argue for a de novo, non-deferential interpretation, but we think Chevron-deference applies.

The pilots acknowledge the PBGC generally receives Chevron-deference for its authoritative interpretations of ambiguous provisions of ERISA. See PBGC v. LTV Corp., 496 U.S. 633, 651-52, 110 S.Ct. 2668, 110 L.Ed.2d 579 (1990); Mead Corp. v. Tilley, 490 U.S. 714, 722, 109 S.Ct. 2156, 104 L.Ed.2d 796 (1989). But deference should not apply, they say, when the PBGC is acting as trustee rather than guarantor, noting that no case or court has addressed the question of whether the PBGC receives Chevron-deference for decisions it makes as trustee. We see no reason to depart from the usual deference we give to an agency interpreting its organic statute. The pilots point out that a private party serving as trustee would not receive Chevron- deference, but this point proves nothing. Unlike a private trustee, the PBGC has unique experience and “practical agency expertise” in interpreting ERISA. LTV Corp., 496 U.S. at 651, 110 S.Ct. 2668. The PBGC is therefore “better equipped” to interpret ERISA than courts, id., and it is for this reason we defer to the PBGC’s authoritative and reasonable interpretations of ambiguous provisions of ERISA.

The PBGC’s interpretation of the relevant ERISA provision is embodied in its regulations limiting benefits in Priority Category 3 to “the lowest annuity benefit payable under the plan provisions at any time during the 5-year period.” 29 C.F.R. 4044.13(b)(3)(i) (emphasis added). Because this language refers to benefits that are “payable,” and because the early retirement program, by its own terms, did not allow the benefits to be payable until May 1, 1998, the PBGC determined that the program was not in effect for the full five-year period. Lacking any statutory definition or guidance, the meaning of the phrase “in effect” is ambiguous. See Chevron v. NRDC, 467 U.S. 837, 843, 104 S.Ct. 2778, 81 L.Ed.2d 694 (1984). And, because the Early Retirement Incentive Program only became operationally effective when it was first possible for pilots to retire under the program — or even collect payments under it — it is reasonable for the PBGC to use May 1, 1998 as the date the program came into effect.

The pilots’ second claim also relates to inclusion of benefits in Priority Category 3. The parties agree that for the entire five-year period prior to termination of the plan, the plan capped maximum benefits, tying the maximum cap to the level established in an IRS statute (26 U.S.C. § 415(b)). Congress amended the IRS statute approximately two years before termination, increasing the maximum cap in the statute. Because the plan tied itself to § 415(b), the maximum under the plan for the final two years was correspondingly raised.

[*1274] Although the PBGC agrees the maximum cap was in effect for the entire five years, it does not agree that the amended increase was in effect for all five years. Indeed, the PBGC is factually correct on this point, as even the pilots concede the amendment only occurred during the last two years of the relevant period. The PBGC therefore based its final determination on the lower cap. Though the pilots prefer the higher cap, the statutory text is plainly against them: Priority Category 3 is “based on the provisions of the plan (as in effect during the 5-year period ending on [the plan’s termination] date) under which such benefit would be the least” 29 U.S.C. § 1344(a)(3)(A) (emphasis added). The question here is not when the maximum-cap provision came into effect, but for which value “such benefit would be the least.” Because the benefit “would be the least” based on the figure applied during the first three years, the PBGC appropriately applied the lower value.

The pilots again argue against Chevron deference. Nothing about this second claim alters our earlier analysis. Since its action on the maximum cap is plainly consistent with the statutory language, the PBGC’s interpretation is sound. The pilots also claim Rettig v. PBGC, 744 F.2d 133 (D.C.Cir.1984), requires automatic increases based on congressional action. But Rettig is unhelpful. That case dealt with vesting standards for a pension plan terminated within the immediate aftermath of ERISA’s enactment. Id. at 135. This court held the PBGC was not permitted to phase-in vesting improvements which were made mandatory by statute without providing reasonable justification for the phase-in. Id. at 156. The case is not relevant here and it adopts none of the sweeping propositions the pilots cite.

The pilots’ third claim also involves a matter of statutory interpretation. ERISA states “the corporation shall guarantee, in accordance with this section, the payment of all nonforfeitable benefits.” 29 U.S.C. § 1322(a). In other words, the statute prescribes that the PBGC “shall guarantee” some defined “amount of monthly benefits,” up to a certain limit. The question is: what does it mean for the defined amount to be “guaranteed”?

The PBGC argues a “guarantee” is the amount a participant is “guaranteed” to receive, either from the plan assets or with help from the PBGC. For example, suppose the maximum amount of monthly benefits to be “guaranteed” is $4,000. If the pension plan is supposed to pay someone $9,000 per month, and if there are enough assets to pay $7,000 per month, then the PBGC has no obligation. The participant received the $4,000 that was “guaranteed” — and more — so the PBGC need not provide any additional payment. The pilots, on the other hand, believe the PBGC is still on the hook for the remaining $2,000 in the above scenario. The pilots see the guaranteed amount not as the amount a participant is guaranteed to receive, but as an amount the PBGC is obligated to pay.

The statutory text only says “the corporation shall guarantee” the specified amount. Id. It does not explain whether the specified amount is the amount the recipient is guaranteed to receive or whether it is the amount the PBGC is obligated to pay. We agree with the PBGC that the better interpretation is that a “guarantee” is the amount that a participant is “guaranteed” to receive.

For each of their three claims, the pilots have failed to demonstrate a substantial likelihood of success and thus make a very poor showing on the first prong. Their arguments on the second prong are similarly weak, as they cannot establish irreparable harm. The only alleged injuries in this case are economic.[*1275] The pilots complain that they are not getting enough money from the PBGC. If they succeed in their suit on the merits, the PBGC will have to give them the money they request. As this court has held, “in the absence of special circumstances, ... recoverable economic losses are not considered irreparable.” Taylor v. Resolution Trust Corp., 56 F.3d 1497, 1507 (D.C.Cir.1995). The Supreme Court has echoed this message, finding that “the temporary loss of income, ultimately to be recovered, does not usually constitute irreparable injury.” Sampson v. Murray, 415 U.S. 61, 90, 94 S.Ct. 937, 39 L.Ed.2d 166 (1974).

The only case the pilots cite to demonstrate irreparable harm is Friends for All Children, Inc. v. Lockheed Aircraft Corp., 746 F.2d 816 (D.C.Cir.1984). In that case, this court affirmed the long-standing principle that an injunction will not be given for monetary relief prior to final judgment on the merits. Id. at 828-29. In fact, in Friends, the court held only that there was an exception to that general rule in cases where liability had already been established on the merits. Id. at 829. This case is distinguishable from the unique procedural setting in Friends, where liability had already been found and a preliminary injunction was sought prior to the complicated determination of monetary remedies. Additionally, although not explicitly stated in Friends, its analysis is consistent with the “sliding scale” on which preliminary injunctions are to be assessed. Davenport, 166 F.3d at 361. In Friends, liability had already been determined on the merits; the showing on prong one was not just a high likelihood of success, but a certainty of success. In such a case, where one factor has been shown to a certainty, it is appropriate to apply a lower threshold on other prongs, such as irreparable injury, in the balance of the four factors. See, e.g., CFGC, 454 F.3d at 297. In this case, however, the pilots do not make a good showing on the first prong, and we see no reason to depart from the general rule that economic harm does not constitute irreparable injury. The pilots argue their case is distinguishable from ordinary economic injury because many of their members are old and may not live to see final judgment. Their assertion that the possibility of death converts mere economic injury into irreparable harm is without citation or precedent, and we do not adopt it here.

Regarding the third and fourth prongs, the district court found minimal harm to the PBGC, noting that its injury was also purely economic, and found “[t]he public interest factor is a wash.” Davis, 596 F.Supp.2d at 5. Both parties devote very little argument to either of these prongs. We see no need to revisit the district court’s analysis. Given that the pilots have made very weak showings on prongs one and two, it is clear they cannot make the corresponding strong showings required to tip the balance in their favor.

IV. Conclusion

Because the pilots have failed to demonstrate either a substantial likelihood of success on the merits or irreparable harm, we affirm the district court’s denial of a preliminary injunction.

So ordered.