McLAUGHLIN, Circuit Judge:
Following the settlement for over $54 million of a securities class action, plaintiffs’ counsel sought attorneys’ fees of 25% of the recovery, amounting to $13.5 million. The United States District Court for the Southern District of New York (Kram, J.) declined to award that amount. Instead, the court awarded $2.1 million, amounting to about 4% of the recovery, based on counsel’s “lodestar” of hours actually and reasonably billed. Counsel now appeal, arguing that the district court erred by: (1) refusing to award fees on a percentage [45]*45of the recovery basis; and (2) declining to enhance their lodestar with a multiplier.
We hold that either the lodestar or percentage of the recovery methods may properly be used to calculate fees in common fund cases, and that the district court did not abuse its discretion in choosing the lodestar in this case. Nor do we find any abuse of discretion in the district court’s award of a fee of about 4% of the recovery. Accordingly, we affirm.
BACKGROUND
This case arises from the ashes of what is regarded by some as the most spectacular scam of the 1980s.1 The complaint alleges that early in that decade, Michael Milken of Drexel Burnham Lambert Group, Inc. began to successfully tout high risk, high yield “junk” bonds as a way to finance growth for otherwise under-capitalized companies. For a while the junk market flourished. Eventually, however, it became apparent that the market comprised, not arm’s-length participants, but primarily a group of Milken clients. These people depended on him to sell their high risk bonds to a so-called “daisy chain” of other Milken controlled clients with proceeds from their own (typically over-financed) junk offerings. The whole pyramid fell apart when the market realized that junk debt carried a much higher default rate than had been advertised. The initiation of criminal and civil enforcement proceedings against Milken and Drexel exacerbated the matter. In January 1989, Drexel pleaded guilty to, inter alia, federal securities fraud, and agreed to pay $650 million in fines and restitution. In April 1990, Milken followed suit by pleading guilty to, inter alia, securities fraud, and agreeing to pay $600 million in fines and restitution.
The primary defendant in this case— Integrated Resources, Inc., a diversified financial services company — was allegedly part of Milken’s daisy chain. When financing from Drexel dried up, Integrated found itself unable to fund its current liabilities. On June 15, 1989, Integrated announced that it was defaulting on over $1 billion of its short term debt, and the prices of its publicly traded securities plummeted. Immediately thereafter, a group of plaintiffs’ law firms filed various actions on behalf of a putative class of Integrated securities holders. They alleged violations of, inter alia, section 10(b) of the Securities Exchange Act of 1934, and Rule 10b-5 promulgated thereunder. All the actions were consolidated before Judge Shirley Wohl Kram of the United States District Court for the Southern District of New York, and two law firms— Milberg Weiss Bershad Hines & Lerach LLP and Abbey, Gardy & Squitieri, LLP (then Abbey & Ellis) — were designated co-lead counsel. In their second amended complaint, plaintiffs named as defendants: (1) Integrated and some of its officers and directors; (2) Drexel; (3) Milken; and (4) Touche Ross & Co., Integrated’s outside auditor.
In February 1990, the consolidated action was automatically stayed with respect to both Integrated and Drexel after they sought bankruptcy protection. Another Southern District Judge, the Honorable Milton Pollack, who had been chosen by the Multidistrict Litigation Panel to handle the numerous derivative and class actions brought against Milken and his cohorts, also assumed control over the Drexel bankruptcy proceeding. In that proceeding, Judge Pollack appointed Milberg Weiss to represent the interests of, inter alia, the Integrated shareholders.
Four separate settlements were eventually reached with Integrated Resources and its co-defendants. First, the Integrated class received about $22.3 million from the Drexel bankruptcy proceeding super[46]*46vised by Judge Pollack. Second, the class received about $19.3 million as its pro rata share of the global settlement — also supervised by Judge Pollack — of all claims asserted against Milken and his cohorts. Third, the Integrated class received about $7.6 million from the settlement of the claims against Integrated and its officers and directors. Finally, the class received $4.9 million in settlement of its claims against Touche Ross. Total recovery was thus over $54 million.
Although Judge Pollack had supervised the substantive settlement of several aspects of this case, it fell to Judge Kram to award attorneys' fees. Throughout the fee proceedings before Judge Kram, counsel maintained they should be awarded a simple percentage of the recovery as a fee, rather than having to submit to a review of their billed hours under the so-called “lodestar” method. Specifically, counsel sought 25% of the total recovery, or a total fee of $13.5 million.
The first fee application was made in October 1992 with respect to the funds recovered from the Drexel bankruptcy. Judge Kram appointed Michael D. Hess as a special master to review the application. Special Master Hess’s initial Report and Recommendation recommended that counsel receive their requested 25% fee. Judge Kram, however, directed Mr. Hess to revise his recommendation and to base any fee award on counsel’s lodestar.
Following an exhaustive review of counsel’s billed hours, and after reducing for various charges he found excessive, Mr. Hess submitted a second Report, this time recommending a lodestar award of $1,416,-572.75. Among other things, Mr. Hess noted that counsel sought then-current hourly rates, even though their billings spanned the three and one-half years preceding the fee petition. For example, the Milberg firm requested hourly fees for attorney time ranging from $125 to $500 per hour. Mr. Hess reasoned that allowing such then-current hourly rates was justified in order to compensate counsel for the delay in payment.
By order dated June 10, 1993, Judge Kram adopted Mr. Hess’s recommendation but reduced counsel’s lodestar to $1,284,-704.80, citing, inter alia, “over 80 instances where the time records of Milberg Weiss, indicating meetings and telephone conferences with co-counsel, do not correspond with the time records of other counsel.” In the same order, Judge Kram explained that she had considered but rejected counsel’s contentions that they should be awarded fees on a percentage basis. Judge Kram also declined to award a multiplier.
In October 1996, plaintiffs’ counsel made a second fee application, seeking 25% of the nearly $32 million total recovery from the three Milken, Integrated, and Touche Ross settlements. Alternatively, counsel asked that, if the lodestar method were used, then a multiplier should be applied to their lodestar (which would grant them the same 25% of recovery). In November 1996, Judge Kram appointed David H. Pi-kus as a special. master to review this second fee application.
In 1998, Mr. Pikus issued a Report and Recommendation recommending a lodestar award of $865,326.68. Like Mr. Hess, he reduced the lodestar for excessive charges. In addition, Mr. Pikus saw no need for a multiplier. He reasoned that counsel had benefitted from the information generated by the various parallel federal investigations of Milken and his cohorts, and that “[tjhere was no groundbreaking issue which loomed significant in this case.” He also found that the hourly rates sought by counsel, ranging as high as the $550 per hour charged by Melvyn I. Weiss of the Milberg firm, fell '“clearly at the higher end” of “the prevailing range.” Mr. Pikus concluded, however,, that allowing these high hourly rates was justified to compensate for the risks undertaken by counsel in prosecuting the case, as well as to recognize the quality of the representation rendered. In a January 18,1999 order, Judge [47]*47Kram adopted Mr. Pikus’s recommendation, holding, inter alia, that awarding “an enhancement multiplier for the results achieved and risks born by Plaintiffs’ Counsel would likely result in their overcompensation.”
All in all, the district court awarded counsel over $2.1 million in fees for their efforts in this case. Dissatisfied with that amount, counsel, led by the Milberg firm, now appeal. They do not challenge the reductions in their lodestar, although they carp at what they perceive as nitpicking. Instead, counsel argue that the district court erred by: (1) refusing to award fees on a percentage of recovery basis; and (2) declining to enhance their lodestar with a multiplier.
Through oral argument on this appeal, the class members’ interest in the common fund was unrepresented. Accordingly, following oral argument, this Court invited Special Master Pikus to respond to counsel’s arguments. Mr. Pikus accepted our invitation and submitted a brief seeking affirmance.
DISCUSSION
From time immemorial it has been the rule in this country that litigants are expected to pay their own expenses, including their own attorneys’ fees, to prosecute or defend a lawsuit. See Alyeska Pipeline Serv. Co. v. Wilderness Soc’y, 421 U.S. 240, 247, 95 S.Ct. 1612, 44 L.Ed.2d 141 (1975). There is a salient exception to this general rule that applies where an attorney succeeds in creating a common fund from which members of a class are compensated for a common injury inflicted on the class. See Trustees v. Greenough, 105 U.S. 527, 533, 26 L.Ed. 1157 (1881). In that situation, the attorneys whose efforts created the fund are entitled to a reasonable fee — set by the court — to be taken from the fund. See Boeing Co. v. Van Gemert, 444 U.S. 472, 478, 100 S.Ct. 745, 62 L.Ed.2d 676 (1980) (citing cases). The rationale for the doctrine is an equitable one: it prevents unjust enrichment of those benefiting from a lawsuit without contributing to its cost. See id. (citing Mills v. Electric Auto-Lite Co., 896 U.S. 375, 392, 90 S.Ct. 616, 24 L.Ed.2d 593 (1970)). Courts have used two distinct methods to determine what is a reasonable attorneys’ fee.
The first is the lodestar, under which the district court scrutinizes the fee petition to ascertain the number of hours reasonably billed to the class and then multiplies that figure by an appropriate hourly rate. See Savoie v. Merchants Bank, 166 F.3d 456, 460 (2d Cir.1999). Once that initial computation has been made, the district court may, in its discretion, increase the lodestar by applying a multiplier based on “other less objective factors,” such as the risk of the litigation and the performance of the attorneys. Id. (internal quotation marks omitted).
The second method is simpler. The court sets some percentage of the recovery as a fee. See id. In determining what percentage to award, courts have looked to the same “less objective” factors that are used to determine the multiplier for the lodestar. See Brown v. Phillips Petroleum Co., 838 F.2d 451, 454-55 (10th Cir.1988).
It bears emphasis that whether calculated pursuant to the lodestar or the percentage method, the fees awarded in common fund cases may not exceed what is “reasonable” under the circumstances. Brown, 838 F.2d at 455; see Savoie, 166 F.3d at 460. What constitutes a reasonable fee is properly committed to the sound discretion of the district court, see Kirsch v. Fleet Street, Ltd., 148 F.3d 149, 172 (2d Cir.1998), and will not be overturned absent an abuse of discretion, such as a mistake of law or a clearly erroneous factual finding. See Herman v. Davis Acoustical Corp., 196 F.3d 354, 356 (2d Cir.1999). Indeed “abuse of discretion”— already one of the most deferential standards of review — takes on special significance when reviewing fee decisions. See [48]*48In re Bolar Pharm. Co. See. Litig., 966 F.2d 731, 732 (2d Cir.1992) (per curiam). “[T]he district court, which is intimately familiar with the nuances of the case, is in a far better position to make [such] decisions than is an appellate court, which must work from a cold record.” Id.
I. The District Couri’s Refusal to Aivard a Percentage Fee
Counsel complain that there is confusion in this Circuit as to whether district courts may use the percentage method. They maintain that this lack of clarity caused the district court to labor under the misapprehension that, as a matter of law, it could not award a percentage fee. That error, counsel continue, caused them to be “wrenched through” the lengthy and costly “nitpicking” of the lodestar examination conducted by Special Masters Hess and Pikus. As succinctly stated at oral argument, counsel now urge us to “junk” the lodestar altogether in common fund cases, and to remand for an award of an appropriate percentage fee.
Concededly, this Circuit’s approach to the alternative methods of calculating fees has evolved in a somewhat circuitous fashion. For much of the 20th century, the percentage approach prevailed. See, e.g., Winkelman v. General Motors Corp., 48 F.Supp. 504 (S.D.N.Y.1942), aff'd sub nom. Singer v. General Motors Corp., 136 F.2d 905 (2d Cir.1943). By the 1970s, however, the routine award of fees in the 20% to 30% range, see Rosenfeld v. Black, 56 F.R.D. 604, 605-606 (S.D.N.Y.1972) (collecting cases), had led to a perception that the percentage approach increasingly tended to yield too little for the client-class, and an unjustified “golden harvest of fees” for the lawyer. City of Detroit v. Grinnell Corp., 495 F.2d 448, 468, 469 (2d Cir.1974) (“Grinnell I”) (internal quotation marks omitted).
In Grinnell I, we responded to this perception by reversing and remanding a 15% fee award with instructions to base future awards on counsel’s lodestar. The “mathematical exercise” of calculating the lodestar, we reasoned, was “the only legitimate starting point for analysis,” id. at 471, because it alone could “claim objectivity.” Id. at 470. We drove the point home when the case returned several years later, by again reversing the fee award which, though based on lodestar, amounted to about 10% of recovery and was still excessive. See City of Detroit v. Grinnell Corp., 560 F.2d 1093, 1095, 1103 (2d Cir.1977) (“Grinnell II ”). Motivated by similar concerns of over-compensation, and recognizing that not every lawyer will fairly subordinate his own commercial interests to those of his client,2 other circuits also adopted the lodestar method as the only appropriate means of calculating fees. See Copeland v. Marshall, 641 F.2d 880, 890 (D.C.Cir.1980) (en banc); Furiado v. Bishop, 635 F.2d 915, 920 (1st Cir.1980); Grunin v. International House of Pancakes, 513 F.2d 114, 128 (8th Cir.1975); Lindy Bros. Builders, Inc. v. American Radiator & Standard Sanitary Corp., 487 F.2d 161, 167 (3d Cir.1973) (“Lindy /”); Lindy Bros. Builders, Inc. v. American Radiator & Standard Sanitary Corp., 540 F.2d 102, 112 (3d Cir.1976) (“Lindy II”).
As so often happens with simple nostrums, experience with the lodestar method proved vexing. Our district courts found that it created a temptation for lawyers to run up the number of hours for which they could be paid. See In re Union Carbide Corp. Consumer Prod. Bus. Sec. Litig., 724 F.Supp. 160, 167-68 (S.D.N.Y.1989). For the same reason, the lodestar created an unanticipated disincentive to early settlements. See Savoie, 166 F.3d at 461 (citation omitted). But the primary source [49]*49of dissatisfaction was that it resurrected the ghost of Ebenezer Scrooge, compelling district courts to engage in a gimlet-eyed review of line-item fee audits. See Union Carbide, 724 F.Supp. at 167-168. There was an inevitable waste of judicial resources.
In 1984, in Blum v. Stenson, the Supreme Court observed that “under the ‘common fund doctrine,’ ... a reasonable fee is based on a percentage of the fund bestowed on the class.” 465 U.S. 886, 900 n. 16, 104 S.Ct. 1541, 79 L.Ed.2d 891 (1984). This statement provided all the impetus needed for a rejuvenation of the percentage method. A year after Blum, the Third Circuit, which had been a vigorous champion of the lodestar method, see Lindy, 487 F.2d at 167, commissioned a blue ribbon task force to review the matter. See Third Circuit Task Force, Court Awarded Attorney Fees, 108 F.R.D. 237 (1985). The Task Force unequivocally recommended a return to the percentage method in common fund cases, see id. at 258, a recommendation which the Third Circuit ultimately followed a decade later, see In re General Motors Corp. Pick-Up Truck Fuel Tank Products Liab. Litig., 55 F.3d 768, 821-22 (3d Cir.1995).
Since Blum, six other circuits have reaffirmed that district courts enjoy the discretion to use either the lodestar or the percentage method. See Johnston v. Comerica Mortgage Corp., 83 F.3d 241, 244-46 (8th Cir.1996); In re Thirteen Appeals Arising Out Of The San Juan Dupont Plaza Hotel Fire Litig., 56 F.3d 295, 305 (1st Cir.1995); Rawlings v. Prudential-Bache Properties, Inc., 9 F.3d 513, 516 (6th Cir.1993); Harman v. Lyphomed, Inc., 945 F.2d 969, 975 (7th Cir.1991); Paul, Johnson, Alston & Hunt v. Graulty, 886 F.2d 268, 272 (9th Cir.1989); Brown v. Phillips Petroleum Co., 838 F.2d 451, 454 (10th Cir.1988); but see Longden v. Sunderman, 979 F.2d 1095, 1099 n. 9 (5th Cir.1992). We ourselves have recognized that Blum indicates that “the pereentage-of-the-fund method is a viable alternative,” Savoie, 166 F.3d at 460. And, though we have had occasion to do so, we have not meddled with the numerous district courts in this Circuit who have chosen the percentage method. See, e.g., Gwozdzinnsky v. Sandler Assoc., No. 95 Civ. 1400(SHS) (S.D.N.Y. Feb. 11, 1997) (awarding fees on a percentage basis), aff'd 159 F.3d 1346 (2d Cir.1998) (Table); see also In re Nasdaq Market-Makers Antitrust Litig., 187 F.R.D. 465, 484-85 (S.D.N.Y.1998) (collecting cases).
Despite the overwhelming weight of these authorities, some uncertainty seems to persist over whether this Court clings to the blanket prohibition announced in the Grinnell opinions against percentage fees. See, e.g., Polar Int’l Brokerage Corp. v. Reeve, 187 F.R.D. 108, 120 (S.D.N.Y.1999) (noting that “[t]he Second Circuit has not yet expressed its view on the propriety of using [the percentage method]”). Uncertainty is no longer warranted. It is true that in In re “Agent Orange” Prod. Liab. Litig., 818 F.2d 226, 232 (2d Cir.1987), we cited the Grinnell opinions for the proposition that this Circuit has “adopted a lodestar formula for calculating fees” in common fund cases. But Agent Orange did not appraise the propriety of the percentage approach in light of Blum, because that issue was neither contested by the parties, nor addressed by the panel. See Agent Orange, 818 F.2d at 229. As such, Agent Orange does not bar a return to this Circuit’s earlier practice of setting fees on a percentage of the fund basis. See Getty Petroleum Corp. v. Bartco Petroleum Corp., 858 F.2d 103, 113 (2d Cir.1988) (an earlier panel’s “sub silentio ... resolution of [an] issue” with which it was not confronted is “ ‘not binding precedent’ ”) (quoting Korwek v. Hunt, 827 F.2d 874, 877 (2d Cir.1987)).
The express goal of the Grinnell opinions was to prevent unwarranted windfalls for attorneys. See Grinnell II, 560 F.2d at 1098-1099; Grinnell I, 495 F.2d at 469-71. So long as that object is achieved, we see no need to compel district courts to under[50]*50take the “cumbersome, enervating, and often surrealistic process” of lodestar computation. Savoie, 166 F.3d at 461 n. 4 (quoting Court Awarded Attorney Fees, 108 F.R.D. at 258).
That said, we reject counsel’s arguments to “junk” the lodestar altogether. They note that the District of Columbia and Eleventh Circuits mandate the exclusive use of the percentage approach in common fund cases, reasoning that it more closely aligns the interests of client and attorney, and more faithfully adheres to market practice. See Swedish Hosp. Corp. v. Shalala, 1 F.3d 1261, 1271 (D.C.Cir.1993); Camden I Condominium Ass’n v. Dunkle, 946 F.2d 768, 774 (11th Cir.1991). We choose to follow our own precedents, which clearly establish that the lodestar approach is an accepted but not exclusive methodology in- common fund cases. See Savoie, 166 F.3d at 461; Agent Orange, 818 F.2d at 232.
Apart from precedent, we question the wisdom of abandoning the lodestar entirely. To be sure, there are cases “where [the district court] can calculate the relevant parameters (hours expended and hourly rate) more easily than it can determine a suitable percentage to award.” General Motors, 55 F.3d at 821; see, e.g., Charles v. Goodyear Tire & Rubber Co., 976 F.Supp. 321, 324 (D.N.J.1997). And the lodestar remains useful as a baseline even if the percentage method is eventually chosen. Indeed, we encourage the practice of requiring documentation of hours as a “cross check” on the reasonableness of the requested percentage. General Motors, 55 F.3d at 820. Of course, where used as a mere cross-check, the hours documented by counsel need not be exhaustively scrutinized by the district court. See In re Prudential Ins. Co. Am. Sales Litig., 148 F.3d 283, 342 (3d Cir.1998) (approving this practice). Instead, the reasonableness of the claimed lodestar can be tested by the court’s familiarity with the case (as well as encouraged by the strictures of Rule 11).
In sum, we hold that both the lodestar and the percentage of the fund methods are available to district judges in calculating attorneys’ fees in common fund cases. Of course, no matter which method is chosen, district courts should continue to be guided by the traditional criteria in determining a reasonable common fund fee, including: “(1) the time and labor expended by counsel; (2) the magnitude and complexities of the litigation; (3) the risk of the litigation ...; (4) the quality of representation; (5) the requested fee in relation to the settlement; and (6) public policy considerations.” Union Carbide, 724 F.Supp. at 163 (summarizing Grinnell opinions).
The district court’s use of the lodestar method in this case was a permissible exercise of its discretion. Counsel argue, however, not that Judge Kram abused her discretion in choosing the lodestar method, but that she committed an error of law by assuming that she had no authority to award a percentage fee. A remand is required, counsel maintain, to allow the district court to consider the fee uninfected by that erroneous assumption.
We disagree. Counsel’s assertion that the district court erroneously refused even to consider using the percentage approach is squarely contradicted by the record. First of all, Special Master Hess’s original Report and Recommendation explicitly spelled out the percentage option. Then, in a June 10, 1993 order, the district court made clear that it had chosen the lodestar approach only after considering but “rejecting] counsels’ implicit assumption that attorneys’ fees will be awarded based on a percentage of the fund obtained.” And in a November 13, 1996 order, the court reiterated that it had rejected the percentage approach after having “determined that application of the lodestar method was appropriate.'” (emphasis added). As these explanations reveal, the court clearly understood that it could have awarded a per[51]*51centage fee in this case. It simply chose not to do so.
Moreover, we question whether counsel could have suffered any prejudicial error arising from the district court’s use of the lodestar. True, Special Master Hess’s initial Report and Recommendation recommended that counsel receive their requested 25% fee. But there is no real indication in the record that Judge Kram herself would have awarded a more generous fee had she used the percentage, rather than the lodestar, method. And, of course, we are in no position to redress counsel’s agonies in being subjected to lodestar review. As counsel themselves tacitly concede, they have already incurred the costs of being “wrenched” through the lengthy lodestar process in the district court. We cannot undo such miseries on this appeal; but even if we could, we would not be sympathetic. Today, all lawyers, even those in the more traditional corporate practice, must submit to the “nitpicking” of fee review. Our concern is that district judges should not have to participate in the same tedium.
II. The District Court’s Refusal to Award a Multiplier
Counsel assert that even though the district court insisted on basing their fee on the lodestar, it should still have applied a multiplier of 6 to that lodestar. To put it more bluntly, counsel continue to assert that they were entitled to a 25% fee. They claim that by refusing to award such a fee, the district court: (1) departed from the “parameters of proper fee jurisprudence,” which they read as establishing 25% of the recovery as the “benchmark” for fee awards in common fund cases; and (2) failed to provide reasonable compensation for the risks assumed and the quality of representation rendered. We are not persuaded.
A. Deviation from 25% “Benchmark ”
According to counsel, 25% of the recovery — whether reached by application of a multiplier, or as a straight percentage — is an established “benchmark” in common fund fees. Counsel argue that the award of a fee amounting to less than 4% of recovery in this case is so far removed from this “benchmark” as to constitute an abuse of discretion. We do not agree.
We are not unaware of the assumption that “[t]wenty-five percent is the ‘benchmark’ that district courts should award in common fund cases.” In re Pacific Enters. Sec. Litig., 47 F.3d 373, 379 (9th Cir.1995) (quotation omitted). Indeed, the Ninth Circuit has cautioned that district courts must justify departure from the 25% benchmark by pointing to unusual circumstances. See Paul, Johnson, 886 F.2d at 272. (It should be noted, however, that even under this approach, the 25% figure remains subject to adjustment to fit a particular case, see id.). And we concede that district courts across the nation have apparently eased into a practice of “systematically” awarding fees in the 25% range, “regardless of type of case, benefits to the class, numbers of hours billed, size of fund, size of plaintiff class, or any other relevant factor.” Court Awarded Attorney Fees, 108 F.R.D. at 274 n. 32 (collecting cases).
This routine largesse has been justified on the theory that a reasonable fee should reflect prevailing rates in the relevant market, see In re Continental Illinois Sec. Litig., 962 F.2d 566, 568 (7th Cir.1992), and that 25% — as a “formula born from judicial practice and experience” — reflects the best estimate of what the market practice would be in a typical common fund case. 1 Alba Conte, Attorney Fee Awards § 2.02 (2d ed.1993); see 1 id. at § 2.08; Continental Illinois, 962 F.2d at 572. There is also commendable sentiment in favor of providing lawyers with sufficient incentive to bring common fund eases that serve the public interest. See Union Car-hide, 724 F.Supp. at 168.
We are nonetheless disturbed by the essential notion of a benchmark. We [52]*52agree that many class actions serve a useful purpose, that lawyers who successfully prosecute them deserve reasonable compensation, and that market rates, where available, are the ideal proxy for their compensation. The problem is that we cannot know precisely what fees common fund plaintiffs in an efficient market for legal services would agree to, given an understanding of the particular case and the ability to engage in collective arm’s-length negotiation with counsel. And “hard data” on analogous situations — such as the fees sophisticated corporate plaintiffs typically agree to pay their attorneys — are “sketchy.” William J. Lynk, The Courts and the Plaintiff’s Bar: Awarding the Attorney’s Fee in Class-Action Litigation, 23 J. Legal Stud. 185, 187 (1994).
Moreover, even a theoretical construct as flexible as a “benchmark” seems to offer an all too tempting substitute for the searching assessment that should properly be performed in each case. Starting an analysis with a benchmark could easily lead to routine windfalls where the recovered fund runs into the multi-millions. “Obviously, it is not ten times as difficult to prepare, and try or settle a 10 million dollar case as it is to try a 1 million dollar case.” Union Carbide, 724 F.Supp. at 166. Indeed, empirical analyses demonstrate that in cases like this one, with recoveries of between $50 and $75 million, courts have traditionally accounted for these economies of scale by awarding fees in the lower range of about 11% to 19%. See Lynk, 23 J. Legal Stud, at 202; see also 1 Conte, Attorney Fee Awards § 2.09 (putting range at 13% to 20%).
But the principal analytical flaw in counsel’s argument lies in their assumption that there is a substantial contingency risk in every common fund case. We harbor some doubt that this assumption is justified in cases such as this. At least one empirical study has concluded that “there appears to be no appreciable risk of non-recovery” in securities class actions, because “virtually all cases are settled.” Janet Cooper Alexander, Do the Merits Matter ? A Study of Settlements in Securities Class Actions, 43 Stan. L.Rev. 497, 578 (1991). Anecdotal evidence tends to confirm this conclusion. Indeed, Mr. Weiss and his partner William S. Lerach of the Milberg firm have stated that losses in these cases are “few and far between,” and that they achieve “a significant settlement although not always a big legal fee, in 90% of the cases we file.” In re Quantum Health Resources, Inc. Sec. Litig., 962 F.Supp. 1254, 1258 (C.D.Cal.1997) (internal quotation marks and citations omitted). Of course, we are not suggesting that compensation for risk is never permitted, merely that it is not — under either the percentage or lodestar methods — an appropriate starting assumption. Even where there is some contingency risk but recovery remains virtually certain, we question whether a fully informed group of plaintiffs able to negotiate collectively would routinely agree to pay their lawyers a fee of 25% of a multi-million dollar settlement.
The point is that plaintiffs in common fund cases typically are not fully informed. Nor are they able to negotiate collectively, or at arm’s length. This is why we emphasized in Grinnell I, as we rejected a 15% fee, that awards in these cases are proper only “‘if made with moderation.’” 495 F.2d at 469 (emphasis added) (quoting Greenough, 105 U.S. at 536). As Grinnell II instructs, the court is to act “as a fiduciary who must serve as a guardian of the rights of absent class members.” 560 F.2d at 1099 (internal quotation marks omitted).
We appreciate that fixing a reasonable fee becomes even more difficult because the adversary system is typically diluted— indeed, suspended — during fee proceedings. Defendants, once the settlement amount has been agreed to, have little interest in how it is distributed and thus no incentive to oppose the fee. See Continental Illinois, 962 F.2d at 572. Indeed, the same dynamic creates incentives for [53]*53collusion — the temptation for the lawyers to agree to a less than optimal settlement “in exchange for red-carpet treatment on fees.” Weinberger v. Great N. Nekoosa Corp., 925 F.2d 518, 524 (1st Cir.1991) (citing John C. Coffee, Jr. The Unfaithful Champion: The Plaintiff as Monitor in Shareholder Litigation, 48 Law & Con-temp. Probs. 5, 26-83 (1985)). And the class members — the intended beneficiaries of the suit — rarely object. See Federal Judicial Center, Empirical Study of Class Action in Four Federal District Courts 76 (1996). Why should they? They have no real incentive to mount a challenge that would result in only a “minuscule” pro rata gain from a fee reduction. Continental Illinois, 962 F.2d at 573. It is not without significance that when Mr. Weiss, lead counsel on this appeal, stood up at oral argument to petition for a bigger slice of his clients’ recovery, no one sat adjacent to him at opposing counsel’s table.
All these considerations have fed the perception among both commentators and the Congress that plaintiffs in common fund cases are mere “figureheads,” and that the real reason for bringing such actions is “the quest for attorney’s fees.” Ralph K. Winter. Paying Lawyers, Em-poivering Prosecutors, and Protecting Managers: Raising the Cost of Capital in America, 42 Duke L.J. 945, 984 (1993); see Private Securities Litigation Reform Act of 1995, H.R.Rep. No. 104-369 (1995) passim, reprinted in 1995 U.S.C.C.A.N. 730, passim (criticizing abusive lawyer-driven securities class actions). This is why we continue to approach fee awards “with an eye to moderation.” Grinnell II, 560 F.2d at 1099 (quoting Grinnell I, 495 F.2d at 470).
Applying such principles of moderation here, we cannot say that the district court abused its discretion merely because the fee awarded is at odds with the 25% “benchmark” embraced by counsel. Nor does the award of a fee of about 4% constitute an abuse of discretion simply because it deviates materially from the 11% to 19% usually awarded in similar cases. Instead, we adhere to our prior practice that a fee award should be assessed based on scrutiny of the unique circumstances of each case, and “a jealous regard to the rights of those who are interested in the fund.” Grinnell I, 495 F.2d at 469 (internal quotation marks omitted); see Grinnell II, 560 F.2d at 1098-99.
B. Assessment of Risk and Quality of Representation
Counsel suggest that the district court failed to provide reasonable, or indeed any, compensation for the risks assumed and the quality of representation rendered in this case. We cannot agree.
Of course contingency risk and quality of representation must be considered in setting a reasonable fee. See Agent Orange, 818 F.2d at 234-36; Union Carbide, 724 F.Supp. at 163. These factors, however, do not always compel enhanced fees. Indeed, consistent with our overarching concern for moderation, we have not hesitated to reverse where we felt an improper appraisal of these factors led to overcompensation. See, e.g., Kaplan v. Rand, 192 F.3d 60, 72 (2d Cir.1999); Grinnell II, 560 F.2d at 1099; Grinnell I, 495 F.2d at 469-71. Conversely, as counsel conceded at oral argument, this Court has never found that a district court abused its discretion by awarding in a common fund case a fee that counsel assailed as too stingy. Though a close issue, we decline to take that unprecedented step here.
Adopting the reports of the special masters, and conducting its own independent review, the district court found that, from counsel’s perspective, this was a “promising” case, with almost certain prospects of a large recovery from solvent defendants. The court reasonably concluded that enhancing fees above the already generous rates included in the lodestar “would likely result in [counsel’s] overcompensation.” This conclusion rested on the specific findings that: (1) coun[54]*54sel benefited from the spadework done by federal authorities during the criminal and civil actions brought against Drexel and Milken; (2) “[t]here was no groundbreaking issue which loomed significant in this case;” (S) despite the Drexel and Integrated bankruptcies, the likelihood of non-payment was slim, because “[m]ost of the defendants ... were solvent, well established individuals or entities,” and the Integrated directors and officers were the beneficiaries of an insurance policy; (4) use of current hourly billing rates compensated counsel for delay in payment; and (5) use of high hourly billing rates compensated counsel for the quality of their efforts, and what risk there was in the case.
Counsel unleash a host of challenges to these findings, some of which do not require analysis.3 Those that do, we discuss below.
1. Risk Multiplier
We have historically labeled the risk of success as “perhaps the foremost” factor to be considered in determining whether to award an enhancement. Agent Orange, 818 F.2d at 236 (quoting Grinnell I, 495 F.2d at 471). Counsel claim that the district court bungled this crucial factor in several respects.
First, counsel point to the general hurdles — such as the defenses available to defendants, including lack of scienter— that they overcame in achieving a settlement in this case. They argue that because their fee was entirely contingent on their ability to overcome such hurdles, they must, as a matter of law, be compensated by a fee enhancement.
That is not law in this Circuit. Risk falls along a spectrum, and should be accounted for accordingly. For example, we have held that public policy considerations justified the award of no contingency allowance in a case that was risky simply because it was of “highly questionable merit.” Agent Orange, 818 F.2d at 235-36. Similarly, there are cases where the risk is “so slight” that any enhancement for the contingent nature of the fee must be “minimal.” Lindy I, 487 F.2d at 168; see Agent Orange, 818 F.2d at 236 (“As the chance of success on the merits or by settlement increases, the justification for using a risk multiplier decreases.”).
Here, the district court clearly found that this case fell in the low range of the risk continuum. We would be hard pressed to overturn that finding. After all, not only is there evidence that the contingency risk in these cases is generally low, this case in particular arose from one of the most notorious financial frauds of the 1980s. Counsel do not deny that the government’s prior efforts against Drexel and Milken dramatically increased their chances of success; nor do they dispute Mr. Pikus’s specific finding that such success did not hinge on any “novel” legal issue. See Grinnell I, 495 F.2d at 471 (noting that parallel government action and legal novelty are among the best indi-cia of contingency risk).
[55]*55In sum, while defendants in this case obviously enjoyed the potential defenses routinely available in securities class actions, we find no clear error in the district court’s essential finding that this was a generally “promising” matter for the plaintiffs right from the start.
Nor are we persuaded by counsel’s argument that the district court’s findings on the risk of non-payment were clearly erroneous. Counsel do not even attempt to argue that Michael Milken was anything other than a deep-pocket defendant. Nor do they suggest that Touche Ross was apparently impecunious. True, Integrated and Drexel did go bankrupt. But as Mr. Weiss himself conceded in an affidavit filed in the district court, Integrated’s directors and officers had an insurance policy providing $15 million in coverage. And the fact that Drexel went bankrupt shortly after this litigation started does not mean that its estate was so apparently insufficient as to foreclose a common fund recovery.
Counsel’s final point on this issue is that the “daisy chain” claims against Drexel and Milken were not predicated on primary violations of Section 10(b), but on aiding and abetting liability. They claim that in light of the Supreme Court’s holding in Central Bank of Denver, N.A v. First Interstate Bank of Denver, N.A., 511 U.S. 164, 114 S.Ct. 1439, 128 L.Ed.2d 119 (1994) (abolishing Section 10(b) aiding and abetting liability), success against Milken and Drexel was “far from assured” despite the notorious nature of their fraud.
Again, we are unpersuaded. It is well-established that litigation risk must be measured as of when the case is filed. See DiFilippo v. Morizio, 759 F.2d 231, 234 (2d Cir.1985); In re Fine Paper Antitrust Litig., 751 F.2d 562, 583 (3d Cir.1984). While Central Bank did indeed abolish aiding and abetting liability, that case was decided in 1994. This action was filed five years earlier when aiding and abetting liability was universally recognized in securities litigation. See Central Bank, 511 U.S. at 192, 114 S.Ct. 1439 (Stevens, J., dissenting); see, e.g., Ross v. Bolton, 904 F.2d 819, 824 (2d Cir.1990); IIT, An International Inv. Trust v. Cornfeld, 619 F.2d 909, 922 (2d Cir.1980). As such, Central Bank simply could not, and indeed did not, pose a risk to recovery in this action.
2. Quality Multiplier
Counsel claim that the district court also improperly assessed the quality of representation rendered in this case. They argue that their performance is reflected by the extraordinary results achieved for the class. According to counsel, the recovery of $54 million represents nearly 90% of what would have been “provable” class damages of $62 million.
Clearly, the results achieved in this case were commendable. Indeed, following the Drexel and Milken settlements, Judge Pollack praised counsel as “the cream of the profession,” whose “genius and dedication” were vital in resolving the complexities of the litigation. In re Michael R. Milken & Assoc. Sec. Litig., MDL No. 924, 1993 WL 413673, at *9 (S.D.N.Y. Oct. 7, 1993).
That said, the district court’s refusal to award a quality multiplier does not warrant reversal. We agree with counsel that the quality of representation is best measured by results, and that such results may be calculated by comparing “the extent of possible recovery with the amount of actual verdict or settlement.” Lindy II, 540 F.2d at 118.
We question, however, counsel’s bald assertion that they recovered nearly 90% of class damages. That percentage is based on counsel’s estimate that only $62 million of class damages would have been “provable” as due to defendants’ fraud. But actual market losses in this case were $90 million. Concededly, counsel’s $62 million figure is based on an expert report commissioned by them. We are hesitant to accept that report unquestioningly, however, because it has not been tested through [56]*56the adversarial process. We are especially reluctant to do so since the valuation of damages in securities class actions is not a “hard science,” and all such reports are singularly susceptible to attack. In re Executive Telecard, Ltd. Sec. Litig., 979 F.Supp. 1021, 1024-26 (S.D.N.Y.1997) (excluding expert report as insufficiently reliable to be admitted at trial).
In any event, a big recovery does not necessarily justify a quality multiplier. As cautioned in Grinnell II: “a large settlement can as much reflect the number of potential class members or the scope of the defendant’s past acts as it can indicate the prestige, skill, and vigor of the class’s counsel.” 560 F.2d at 1099. Here, the class was extensive — over 4,000 proofs of claim were filed. More importantly, the scope of defendants’ misconduct was unprecedented. Again, two of the defendants — Drexel and Milken — were convicted of criminal conduct bearing directly on the claims advanced in thté case. There can be no doubt that this was a proper basis for declining to award a formal quality multiplier. Indeed, a good portion of counsel’s lodestar was based on hours spent scouring the records developed during the parallel criminal proceedings. While this case may not quite be characterized as “piggy back” in nature, counsel’s performance was certainly helped enormously by the prior actions against Drexel and Milken. Given these factors, the district court did not abuse its broad discretion in declining to award a quality multiplier.
3. The District Court’s Chosen Method of Compensating for Risk and Quality of Representation
Even assuming that counsel deserved some enhanced compensation for the risk they took on and the representation they rendered in this case, we would find no basis for reversal. While the district court declined to award formal multipliers for risk and quality of representation, the court did consider those factors by allowing counsel to recover generous hourly fees. We conclude that this approach was a permissible exercise of discretion.
Unsurprisingly, counsel find flaws with the district court’s approach. In particular, they take issue with Special Master Pikus’s finding that the hourly rates allowed for their lodestar were high. But relying on his own experience as a practitioner, as well as on empirical data, see Altman Weil Pensa Inc., Survey of Law Firm, Economics (1997), Special Master Pikus properly measured counsel’s fees against the prevailing market rates “for comparable attorneys of comparable skill and standing in the pertinent legal community.” Kirsch, 148 F.3d at 172. Based on that comparison, Mr. Pikus found that counsel’s requested billing rates — ranging as high as the $550 per hour charged by Mr. Weiss of the Milberg firm — fell “clearly at the higher end” of “the prevailing range.” This was not clearly erroneous. See Altman Weil Pensa at II — 5 (average billing rate for top decile of partners in Northeast was $295 per hour in 1997). Indeed, the sources relied upon by counsel merely support the accuracy of Mr. Pikus’s finding. See Cities At A Glance: High And Low Hourly Billing Rates, Nat’L L.J. (Dec. 21, 1998) (listing hourly rates in cities across the country, including New York).4
Counsel obviously would have preferred an enhanced fee in this case. We are not unsympathetic. This litigation was low risk, precisely because it was of apparent merit. Such cases are to be encouraged, especially when prosecuted by the caliber [57]*57of counsel representing plaintiffs here. See Agent Orange, 818 F.2d at 236.
On the other side of the ledger, however, is our longstanding concern for moderation. That concern is amplified by our nagging suspicion that attorneys in these cases are routinely overcompensated for such things as contingency, risk. There is also the very broad discretion enjoyed by district judges in determining a reasonable fee. When the exercise of that discretion is supported by adequate findings and is consistent with our preference for moderation, as it was here, we will not substitute our own predilections for the judgment of the district court.
CONCLUSION
The orders appealed from are AFFIRMED.