OPINION

SCHWARZER, Senior District Judge.

Once more we must address an issue arising out of the protracted litigation between Microsoft Corporation and its freelance workers, this time to decide whether the district court abused its discretion in the amount of attorneys' fees it awarded to class counsel.

FACTUAL AND PROCEDURAL BACKGROUND

Beginning in 1987, Microsoft supplemented its workforce with workers known as "freelancers," who agreed in writing that they would not be eligible for Microsoft employee benefits, including the Employee Stock Purchase Plan ("ESPP") and the Savings Plus Plan ("SPP"). In 1992, eight former freelancers brought this action challenging Microsoft's refusal to provide them with benefits under these plans. The district court certified a class and dismissed the action. After a panel of this court reversed the dismissal of both the ESPP and SPP claims, the full court voted to rehear the case en banc. The en banc court also reversed the district court's dismissal of the ESPP claim, but held that plaintiffs had not exhausted their administrative remedies for the SPP claim, and remanded that claim to the plan administrator for adjudication in the first instance. Vizcaino v. Microsoft Corp., 120 F.3d 1006 (9th Cir. 1997). On remand, the district court substantially narrowed the class. Vizcaino v. Microsoft Corp., 21 Employee Benefits Cas. 2821(BNA), 1998 WL 122084 (W.D.Wash. 1998). This court then granted mandamus, held the class to include persons who had worked for Microsoft after 1990, and identified factors to be applied in determining individual eligibility. Vizcaino v. United States Dist. Ct. for the W. Dist. of Wash., 173 F.3d 713 (9th Cir. 1999). Settlement negotiations followed, and after two years the parties submitted a proposed settlement of all claims to the district court. The agreement required Microsoft to deposit $96,885,000 into a settlement fund, to be distributed to the class members after payment of incentive awards, costs, and fees. Microsoft also changed its staffing and worker classification practices, resulting in the hiring of over 3000 class members as W-2 employees entitled to participate in employee benefit plans and programs.

After receiving written submissions and hearing argument, the district court approved the settlement on extensive findings of fact and conclusions of law. It then received class counsel's application for an award of attorneys' fees of $27,127,800 (28% of the cash settlement fund). Two members of the class objected. After considering the submissions of counsel and the objectors, and hearing argument on the fee award, the court entered an order approving class counsel's fee request. Vizcaino v. Microsoft Corp., 142 F.Supp.2d 1299 (W.D.Wa. 2001). Before us now is the objectors' appeal from that order. We review for abuse of discretion. Class Plaintiffs v. Jaffe Schlesinger, P.A., 19 F.3d 1306, 1308 (9th Cir. 1994).

DISCUSSION

Objectors challenge the district court's order on three grounds: First, and principally, that in awarding a fee of 28% of the settlement fund, it ignored the so-called increase-decrease rule; second, that in applying a lodestar cross-check, it used an improper methodology; and third, that in denying objectors' fee request without explanation, it abused its discretion. We address each contention in turn.

I. THE DISTRICT COURT'S PERCENTAGE CALCULATION

The district court found that the settlement fund was the product of the successful claim for benefits under Microsoft's ESPP. Because Washington law governed the claim, it also governs the award of fees. Mangold v. Calif. Pub. Utils. Comm'n, 67 F.3d 1470, 1478 (9th Cir. 1995). Under Washington law, the percentage-of-recovery approach is used in calculating fees in common fund cases. Bowles v. Dep't of Ret. Sys., 121 Wash.2d 52, 72, 847 P.2d 440 (1993) (holding that in a common fund case, "the size of the recovery constitutes a suitable measure of the attorneys' performance"). The district court followed the Washington practice of looking to federal law for guidance in this area, and we will do the same. See id. Under Ninth Circuit law, the district court has discretion in common fund cases to choose either the percentage-of-the-fund or the lodestar method. In re Wash. Pub. Power Supply Sys. Sec. Litig., 19 F.3d 1291, 1295-96 (9th Cir. 1994) ("WPPSS"). Objectors do not challenge the district court's choice of the percentage method, only its application.

The district court based its percentage award on Bowles, which states that "[i]n common fund cases, the `benchmark' award is 25 percent of the recovery obtained," with 20-30% as the usual range. Bowles, 121 Wash.2d at 72-73, 847 P.2d 440. Ninth Circuit cases echo this approach. Paul, Johnson, Alston Hunt v. Graulty, 886 F.2d 268, 272 (9th Cir. 1989). Objectors contend that the award is nevertheless excessive, arguing that the court erred in failing to take into account that this is a megafund case to which it should have applied what objectors call the increase-decrease rule. They rely principally on WPPSS, in which the district court chose the lodestar rather than the percentage method in awarding fees from a $687 million settlement fund. The district court observed that "in many cases awards fall outside the `typical' range and . . . the percentage of an award generally decreases as the amount of the fund increases." WPPSS, 19 F.3d at 1297. We did not adopt this observation as a principle governing fee awards. Rather, we merely noted that in cases of that magnitude, fund size is one relevant circumstance to which courts must refer, stating:

We agree with the district court that there is no necessary correlation between any particular percentage and a reasonable fee. With a fund this large, picking a percentage without reference to all the circumstances of the case, including the size of the fund, would be like picking a number out of the air. . . . Because a court must consider the fund's size in light of the circumstances of the particular case, we agree with the district court that the 25 percent "benchmark" is of little assistance in a case such as this.

Id. We concluded that the district court had acted within its discretion in considering the size of the fund in adopting the lodestar method.

The 25% benchmark rate, although a starting point for analysis, may be inappropriate in some cases. Selection of the benchmark or any other rate must be supported by findings that take into account all of the circumstances of the case. As we said in WPPSS, in passing on post-settlement fee applications, "courts cannot rationally apply any particular percentage — whether 13.6 percent, 25 percent or any other number — in the abstract, without reference to all the circumstances of the case." Id. at 1298; see also Camden I Condominium Ass'n, Inc. v. Dunkle, 946 F.2d 768, 775 (11th Cir. 1991) (noting with approval that district courts are increasingly "`supporting their percentage awards with particular findings showing factors considered.'" (quoting HERBERT NEWBERG, ATTORNEY FEE AWARDS § 2.07 (1st ed. 1986))). Objectors' argument that the district court erred in not fixing a lower percentage, such as one between 6% and 10%, flies in the face of this reasoning. The question is not whether the district court should have applied some other percentage, but whether in arriving at its percentage it considered all the circumstances of the case and reached a reasonable percentage. We now turn to the court's examination of those circumstances.

First, the court found that counsel "achieved exceptional results for the class." Vizcaino, 142 F.Supp.2d at 1303. The court found that counsel pursued this case in the absence of supporting precedents, in the face of agreements signed by the class members forsaking benefits — a fact that led four judges of this court to dissent from the panel and en banc opinions — and against Microsoft's vigorous opposition throughout the litigation. Exceptional results are a relevant circumstance. See Torrisi v. Tucson Elec. Power Co., 8 F.3d 1370, 1377 (9th Cir. 1993) (considering counsel's "expert handling of the case"); Six (6) Mexican Workers v. Ariz. Citrus Growers, 904 F.2d 1301, 1311 (9th Cir. 1990) (noting plaintiffs' "substantial success"); In re Prudential Ins. Co. Sales Practices Litig., 148 F.3d 283, 339 (3d Cir. 1998) (observing that "results achieved were `nothing short of remarkable'" (quoting In re Prudential Ins. Co. Sales Practices Litig., 962 F.Supp. 572, 585-86 (D.N.J. 1997))).

Second, the court found the case to have been extremely risky for class counsel for the reasons just stated. Twice plaintiffs lost in the district court — once on the merits, once on the class definition — and twice counsel succeeded in reviving their case on appeal. Risk is a relevant circumstance. See In re Pac. Enter. Sec. Litig., 47 F.3d 373, 379 (9th Cir. 1995) (holding fees justified "because of the complexity of the issues and the risks"); Bebchick v. Wash. Metro. Area Transit Comm'n, 805 F.2d 396, 408 (D.C. Cir. 1986) (considering counsel's repeated successes in overturning adverse determinations) (calculating lodestar); cf. WPPSS, 19 F.3d at 1302 (finding district court's failure to apply multiplier to lodestar calculation was abuse of discretion where case was "fraught with risk and recovery was far from certain").

Third, the court found that counsel's performance generated benefits beyond the cash settlement fund. During the litigation, Microsoft agreed to hire roughly 3000 class members as regular employees and to change its personnel classification practices, a benefit counsel valued at $101.48 million during the 1999-2001 period alone. Vizcaino, 142 F.Supp.2d at 1301 n. 1. The court observed that the litigation also benefitted employers and workers nationwide by clarifying the law of temporary worker classification. Moreover, it noted that as a result of this litigation, many workers who otherwise would have been classified as contingent workers received the benefits associated with full time employment. Incidental or non-monetary benefits conferred by the litigation are a relevant circumstance. See In re Pac. Enter., 47 F.3d at 379 (considering "nonmonetary benefits in the derivative settlement"); cf. Bebchick, 805 F.2d at 408 (allowing an upward adjustment to the lodestar "to reflect the benefits to the public flowing from [the] litigation"); Mills v. Elec. Auto-Lite Co., 396 U.S. 375, 395, 90 S.Ct. 616, 24 L.Ed.2d 593 (1970) (stating that a corporation may receive a substantial benefit from a derivative suit justifying a fee award regardless of whether the benefit is pecuniary).

Fourth, the court found the 28% rate to be at or below the market rate. It cited the retainer agreements between counsel and the named plaintiffs promising to pay class counsel 30% of any recovery. The agreements alone, although somewhat probative of a reasonable rate, are not particularly helpful. For instance, the retainer agreements did not involve the class and, because they were made precertification, are not binding on the class. However, the district court did credit class counsel's evidence showing that the retainer agreements reflected the standard contingency fee for similar cases. This finding does not constitute an abuse of the court's discretion.

We note with respect to this factor that we do not adopt the Seventh Circuit's approach in percentage fee award cases, as set forth in In re Continental Illinois Securities Litigation, 962 F.2d 566, 568 (7th Cir. 1992). There, that court stated that in awarding fees in common fund cases, courts should determine a reasonable fee by attempting to replicate the market rate. While an exclusively market-based approach may have superficial appeal, in the context of class action litigation in which attorneys' fees are determined post hoc by the court (without regard to any private arrangement), it may in many cases be illusory. Unlike in cases where lawyers compete for lead counsel status and may even bid in a court-supervised auction, in employment class actions like this one, no ascertainable "market" exists. See, e.g., ALAN HIRSCH AND DIANE SHEEHEY, AWARDING ATTORNEY'S FEES AND MANAGING FEE LITIGATION 99-101 (1994) (describing practice sometimes used in the Northern District of California); In re Auction Houses Antitrust Litig., 2001 Trade Cas. (CCH) ¶ 73,170, 2001 WL 170792 (S.D.N.Y. Feb.22, 2001). The "market" is simply counsel's expectation of court-awarded fees. The Seventh Circuit's effort to construct a market for such cases by determining what counsel "would have received had they handled a similar suit on a contingent fee basis, with a similar outcome, for a paying client" seems to us an unhelpful measure in many cases, and certainly an inappropriate measure to apply to all cases. In re Cont'l Ill., 962 F.2d at 572. Unlike commercial litigation where the fee is determined by application of the negotiated contingency percentage to the amount of the recovery, in class action litigation the fee is determined on the basis of what a court finds to be reasonable. An attempt to "estimate the terms of the contract that private plaintiffs would have negotiated with their lawyers had bargaining occurred at the outset of the case" strikes us as entirely illusory and speculative. In re Synthroid Mktg. Litig., 264 F.3d 712, 718 (7th Cir. 2001). Where evidence exists, such as here, about the percentage fee to which some plaintiffs agreed ex ante, that evidence may be probative of the fee award's reasonableness. But, to the extent that a market analogy is on point, in most cases it may be more appropriate to examine lawyers' reasonable expectations, which are based on the circumstances of the case and the range of fee awards out of common funds of comparable size.

Fifth, the court found that counsel's representation of the class — on a contingency basis — extended over eleven years, entailed hundreds of thousands of dollars of expense, and required counsel to forgo significant other work, resulting in a decline in the firm's annual income. These burdens are relevant circumstances. Six (6) Mexican Workers, 904 F.2d at 1311 (noting that litigation lasted more than thirteen years); Torrisi, F.3d at 1377 (considering counsel's bearing the financial burden of the case); Bebchick, 805 F.2d at 407 (same).

We conclude that the district court considered the relevant circumstances and did not abuse its discretion in finding a 28% fee award to be reasonable under the percentage method.

II. THE DISTRICT COURT'S LODESTAR CROSS-CHECK

The district court applied the lodestar method as a cross-check of the percentage method. Calculation of the lodestar, which measures the lawyers' investment of time in the litigation, provides a check on the reasonableness of the percentage award. Where such investment is minimal, as in the case of an early settlement, the lodestar calculation may convince a court that a lower percentage is reasonable. Similarly, the lodestar calculation can be helpful in suggesting a higher percentage when litigation has been protracted. Thus, while the primary basis of the fee award remains the percentage method, the lodestar may provide a useful perspective on the reasonableness of a given percentage award.

The court found that counsel's fees for work done on this case, if charged at current hourly rates, would amount to $7,386,876. It found nothing in the record to suggest that any of the hours claimed should be disallowed. Objectors quibble about some of the hours and charges, but we find no abuse of discretion. See WPPSS, 19 F.3d at 1298-99. Calculating fees at prevailing rates to compensate for delay in receipt of payment was within the district court's discretion. See Gates v. Deukmejian, 987 F.2d 1392, 1406 (9th Cir. 1992).

Objectors' principal quarrel is with the district court's lode star cross-check, which resulted in a multiplier of 3.65. The court found this number reasonable by considering the factors in Kerr v. Screen Extras Guild, Inc., 526 F.2d 67, 69-70 (9th Cir. 1975), including "the complexity of this case, the risks involved and the length of the litigation." Vizcaino, 142 F.Supp.2d at 1306. The bar against risk multipliers in statutory fee cases does not apply to common fund cases. WPPSS, 19 F.3d at 1299-1300. Indeed, "courts have routinely enhanced the lodestar to reflect the risk of non-payment in common fund cases." Id. at 1300. This mirrors the established practice in the private legal market of rewarding attorneys for taking the risk of nonpayment by paying them a premium over their normal hourly rates for winning contingency cases. Id. at 1299. In common fund cases, "attorneys whose compensation depends on their winning the case must make up in compensation in the cases they win for the lack of compensation in the cases they lose." Id. at 1300-01 (internal citation and quotation omitted). Class counsel here have represented that they would not have taken this case other than on a contingency basis. They perform little work on an hourly basis, and the rates they submitted were what they took to be market rates, in other words, rates that did not already reflect an expectation of excellent results.

Thus, a multiplier was appropriate in this case. The district court's percentage of the fund analysis discussed above addressed the substantial risk class counsel faced, compounded by the litigation's duration and complexity. The court considered these circumstances in arriving at a multiplier which was within the range of multipliers applied in common fund cases. We find no abuse of discretion.

III. THE DENIAL OF OBJECTORS' REQUEST FOR ATTORNEYS' FEES

Objectors contend that the district court abused its discretion in rejecting their request for attorneys' fees, arguing that they caused the district court to require class counsel to submit time records and that they brought about minor procedural changes in the settlement agreement. Because objectors did not increase the fund or otherwise substantially benefit the class members, they were not entitled to fees. Bowles, 121 Wash.2d at 70-71, 847 P.2d 440 (stating that under Washington law, fees may be awarded only if authorized by "contract, statute or recognized ground in equity" (internal citation and quotation omitted)). The equitable common fund/common benefit doctrine "authorizes attorney fees only when the litigants preserve or create a common fund for the benefit of others as well as themselves." Id.; Class Plaintiffs v. Jaffe Schlesinger, P.A., 19 F.3d 1306, 1308 (9th Cir. 1994). In the absence of a showing that objectors substantially enhanced the benefits to the class under the settlement, as a matter of law they were not entitled to fees, and the district court did not abuse its discretion.

CONCLUSION

"Because in common fund cases the relationship between plaintiffs and their attorneys turns adversarial at the fee-setting stage, courts have stressed that when awarding attorneys' fees from a common fund, the district court must assume the role of fiduciary for the class plaintiffs." WPPSS, 19 F.3d at 1302. Accordingly, fee applications must be closely scrutinized. Rubber-stamp approval, even in the absence of objections, is improper. We are satisfied that in this case, the district court subjected the application to the requisite scrutiny and did not abuse its discretion in determining a reasonable fee in light of the relevant circumstances of the case.

AFFIRMED. Case Fund Fee(%) Fee($) Multiplier In re Rite Aid Corp. Sec. Litig., In re Lease Oil Antitrust Litig., 186 F.R.D. 403 In re Merry-Go-Round Enterprise, Inc., In re Copley Farm., Inc., 1 F.Supp.2d 1407 Walco Investments, Inc. v. Thenen, In Re Informix Corp. Sec. Litig., In Re Combustion, Inc., 968 F.Supp. 1116 Kurzweil v. Philip Morris Co., In Re Sumitomo Copper Litig., Local 56, United Food Comm'l Workers Union $115m 2.8% $ 3 2.4 v. Campbell Soup Co., 954 F.Supp. 1000 In Re Ikon Office Sol'ns, Inc., Sec. Litig., In Re Sunbeam Sec. Litig., Bussie v. Allamerica Fin'l Corp., Haynes v. Shoney's, Ingram v. The Coca-Cola Co., 200 F.R.D. 685 Bowling v. Pfizer, Inc., 922 F.Supp. 1261 Fanning v. Acromed Corp., Baird v. Thompson Consumer Elecs., Rosted v. First USA Bank, In Re Sorbates Direct Purchaser Antitrust Litig., In Re Aetna, Inc. Sec. Litig., In Re Paracelsus Corp. Sec. Litig., In Re Commercial Explosives Antitrust Litig., Van Vranken v. ARCO, 901 F.Supp. 294 Ramah Navajo Chapter v. Babbitt, Branch v. F.D.I.C., In Re IDB Communication Group, Inc., In Re 1996 Medaphis Corp. Sec. Litig., In Re MiniScribe Corp., 257 B.R. 56 In Re Nat'l Health Laboratories Sec. Litig., In Re Teletronics Pacing Systems, Inc., In Re Medical Care America, Inc. Sec. Litig., In Re Melridge, Inc., Sec. Litig., In Re Carbon Dioxide Antitrust Litig.,