Full opinion text
Opinion for the Court filed by Circuit Judge WILKEY. Dissenting opinion filed by Circuit Judge J. SKELLY WRIGHT. WILKEY, Circuit Judge: Northwest Airlines, Inc., appeals from an award of $3,453,779.49 in attorneys fees and costs to counsel for plaintiffs. We find that the district court abused its discretion in calculating the fee award. Accordingly, we reverse and remand. I. Facts This case involves a dispute over the amount of attorneys fees due the plaintiffs as prevailing party in a complex and longstanding employment discrimination suit. The suit on the merits began in 1970. Acting on behalf of more than 3,300 women employed by Northwest Airlines, Inc., the plaintiffs challenged various personnel management practices of Northwest under Title VII and the Equal Pay Act. In 1973 the trial court essentially adopted the legal arguments urged by the plaintiffs, and issued an order in their favor. In 1976 a panel of this court sustained the trial court’s statutory interpretation, but remanded for reconsideration of the proper remedy. This court denied rehearing en banc in 1977, and in 1978 the Supreme Court declined to issue a writ of certiorari. Upon remand, the district court awarded $52 million in compensatory relief, as well as significant injunctive relief. That ruling has been affirmed in principal part by another panel of this court. Following the close of litigation on the merits, the fee application process began in earnest. In June of 1982 the district court ordered counsel to commence negotiations on the attorneys fees. Bredhoff & Kaiser, the attorneys for the plaintiffs throughout the merits phase of this litigation, retained Daniel Rezneck of Arnold & Porter to negotiate and, if necessary, litigate the attorneys fee award. In November 1982 the parties reported to the court that their negotiations had been fruitless. Litigation over the size of the fees began. Pursuant to a timetable presented by the parties and approved by the trial court, the parties entered several months of simultaneous discovery. During this period the attorneys for the plaintiffs sought to discover the fees received by the counsel for the defense; the court refused to compel discovery of this information. The plaintiffs subsequently requested just over $5 million to cover all attorneys fees and expenses from 15 July 1970 to 15 July 1983; the defendants proposed an award of just over $1 million. In reaching these rather divergent figures, both plaintiffs and defendants relied on the “lodestar” approach adopted by this Court in Copeland v. Marshall, setting separate lodestar figures for the litigation on the merits and the litigation on the attorneys fees. Both sides supplemented their proposals with copious appendices, exhibits, affidavits, charts, and summaries. A. Proposed Fee for Litigating the Merits Both plaintiffs and defendants substantially agreed on the number of hours “reasonably expended” in the litigation on the merits. The defendants raised two objections to the quantity of hours — a request for 160.5 hours logged by an attorney who was not a member of the firm but who did attend the trial and assist the Bredhoff & Kaiser attorneys, and an objection to 254.-25 attorney hours spent on allegedly noncompensable issues. Both sides agreed to the validity of the remaining 10,929.50 hours expended on the merits. The parties differed radically, however, on how the reasonable hourly rates should be set. The defendants proposed figures based on the hourly rates actually charged by Bredhoff & Kaiser attorneys in similar cases. The defendants then invited the court to examine whether certain work done by partners could have been done by associates or even paralegals, and to compensate that work at lower rates if “partner-level” expertise had not been required. The plaintiffs proposed rates intended to reflect a “prevailing community rate ” — a rate, as it happened, substantially exceeding that ordinarily charged by Bredhoff & Kaiser attorneys. The plaintiffs substantiated the validity of this rate with voluminous affidavits from their own and other area attorneys. The parties thus proposed significantly different lodestar figures for the merits. The plaintiffs sought a merits lodestar of $1,494,450.80. The defendants proposed a merits lodestar of $903,875.15. The plaintiffs then proposed two separate 100 percent multipliers of the merits lodestar — one to account for the quality of the work performed, and another to account for the contingency of payment. The defendants proposed no multiplier at all. With these multipliers taken into account, the plaintiffs sought $4,483,352.40 as attorneys fees for the litigation on the merits. B. Proposed Fee for Litigating the Attorneys Fees Award The parties also differed radically on the proper award for litigating the attorneys fee issue. Laffey sought compensation for 2,579 hours spent by Bredhoff & Kaiser and Arnold & Porter attorneys in compiling the fee award. Northwest argued that only 930.25 attorney hours were compensable. In objecting to the bulk of the hours sought by Laffey’s attorneys, Northwest raised three objections: (1) that needless hours were spent compiling affidavits, some of which detailed the history of the case for a trial judge who had presided over it from its inception, and others of which described the billing practices of other area attorneys; (2) that needless hours were spent in interfirm meetings which would not have occurred had Bredhoff & Kaiser not retained Arnold & Porter; and (3) that needless hours were spent seeking discovery of irrelevant facts. With the variation of hours and hourly rates taken into account, Laffey proposed a lodestar for attorneys fees of $358,375.02; Northwest proposed $64,311.39. C. Expenses The parties also differed significantly on whether plaintiffs could be reimbursed for various expenses incurred in the litigation. Northwest first objected to many of the hours spent by paralegals and law clerks on the case. It then objected to compensating any out-of-pocket expenses in excess of those includable in taxable costs. Laffey sought a total of $256,650.14; Northwest offered slightly less than half of that, $127,959.23. D. Proposed Total Fees Laffey ultimately requested $5,098,-377.56 in attorneys fees. This request reflected a lodestar of $1,494,450.80 on the merits, two 100 percent multipliers of that lodestar, an attorneys fee lodestar of $358,-375.02, and expenses of $256,650.14. Northwest proposed a total fee of $1,096,145.77. This reflected a lodestar on the merits of $903,875.15, no multipliers, an attorneys fee lodestar of $64,311.39, and costs of $127,959.23. E. District Court Award The district court awarded plaintiffs $3,469,829.49. This award reflected a lodestar of $1,471,241.25 on the merits, a doubling of that lodestar to account for the “risk premium,” an attorneys fee lodestar of $294,324.99 and costs and expenses of $216,972.00. The district court carefully scrutinized the number of hours claimed by Laffey. With regard to the merits, the court rejected Northwest’s claim that certain issues were non-compensable as “lost,” but did exclude a portion of those hours as not being reasonably expended. The court also allowed compensation for hours expended by a lawyer not affiliated with either of the firms representing Laffey, but at a reduced hourly rate. The court found compensable the great majority of the 2,579 hours claimed for litigating the attorneys fee, but excluded 467.75 hours as being unreasonable or excessive. The deductions made by the trial court reflected adjustments for preparation of excessive affidavits, redundant work caused by the introduction of a new law firm into the case, and other inefficient practices. It did not exclude the hours expended in offensive discovery, finding that the discovery requests, while ultimately unsuccessful, were “not frivolous, but were bona fide efforts to advance Plaintiffs’ legitimate litigation interests.” In the award the district court rejected Northwest’s claim that the plaintiffs’ attorneys should be restricted to their own market rates, awarding them instead “prevailing market rates” based on a matrix drawn from the rates charged by other attorneys. While evidence of Bredhoff & Kaiser’s normal rates was filed with the court, the trial judge apparently relied not at all on the firm’s normal billing rates. Instead, the trial court found that it was “not limited to counsels’ historic fees but must frame an award that reflects the true value of the services rendered.” The court then fixed “generous” rates at the highest end of rates charged by firms in the community. The court refused to adjust the hourly rates to account for alleged inefficiencies in the utilization of attorneys. It found the defense objection to the staffing patterns employed by Laffey’s attorneys to be based on “the erroneous assumption that there is a single, correct staffing pattern for every lawsuit”; it concluded that allocation of responsibility within a firm depended on many variables and found the allocation chosen by Laffey’s attorneys to be reasonable. The court granted a 100 percent multiplier to account for contingency of payment. In doubling the merits award to account for the contingency of payment, the court first found it beyond dispute that plaintiffs were “entitled” to a risk adjustment of some sort simply because receipt of fees depended on the successful termination of the litigation. The court assessed the initial likelihood of success at approximately 50 percent; taking into account the chance of failure and the number of hours placed at risk, the court found the plaintiffs “fully deserving” of a 100 percent contingency adjustment. The court rejected a requested 100 percent multiplier for excellence of results, finding that the acknowledged excellence of the firms involved was reflected in the lodestar rates. The court found that the plaintiffs were entitled to compensation for costs which were not taxable, and held the documentation to be adequate. It also found compensable the lost income suffered by Laffey because of her involvement in the case. The trial court carefully scrutinized the costs claimed by both Laffey and her attorneys, however, and disallowed some as unreasonable. F. Issues on Appeal The defendant appeals the award to this court. This appeal presents five major issues: (1) whether the “reasonable hourly rate” for private, for-profit law firms should be based on the rates the firm charges in the marketplace; (2) whether the district court abused its discretion in finding that the plaintiffs’ counsel’s allocation of tasks among partners and associates was reasonable on the facts of this case; (3) whether the district court abused its discretion in doubling the lodestar fee on the merits to account for the risk of not prevailing; (4) whether the district court abused its discretion by awarding compensation for litigating the attorneys fee issue; and (5) whether plaintiffs in Title VII and Equal Pay Act cases may be awarded compensation for reasonable expenses which are not taxable costs under 28 U.S.C. § 1920 (1982). II. Analysis Fair Labor Standards Act of 1938 and Title VII of the Civil Rights Act of 1964 authorize district courts to award a reasonable attorneys fee to prevailing civil rights litigants. The purpose of such provisions is to encourage private litigants to act as “private attorneys general” on behalf of enforcement of the civil rights laws. Congress clearly hoped to provide an adequate economic incentive for private attorneys to take employment discrimination cases, and thereby to ensure that plaintiffs would be able to obtain competent legal representation for the prosecution of legitimate claims. Despite the clarity of Congress’s objective, “[this] new field of law ... has grown so fast and become so complex that it has baffled the efforts of courts and lawyers to comprehend and apply it.” Before 1975, federal courts had awarded attorneys fees on their own authority on the theory that private attorneys general should be compensated for enforcing certain important rights. In Alyeska Pipeline Service Co. v. Wilderness Society, however, the Supreme Court criticized this practice and warned lower federal courts that they lacked power to award fees to private attorneys general in the absence of express authorization from Congress. Congress enacted a law authorizing such awards to prevailing civil rights litigants the following year. Throughout this turbulent period, confusion reigned. One scholar’s review of the case law led him to conclude that “[t]he only truly consistent thread that runs throughout federal court decisions on attorneys’ fees is their almost complete inconsistency.” He elaborated: Given the frequency with which courts are confronted with the task of fee setting and the impact that it has upon the allocation of legal resources, one would expect a general consensus to have emerged on the manner in which reasonable attorneys’ fees should be determined. On the contrary, there are nearly as many approaches to the issue as there are judges____ [M]any lower courts have confronted the problem with little or no analysis; those courts that have been more analytical have adopted widely varying approaches. To a great extent the outcome to these cases has depended upon “the roll of the dice” — from court to court and from case to case. Examples of inconsistent fee awards to prevailing parties are not hard to find. Suggestive of the lack of clear guidelines is a report on the “U.S. Court of Appeals for the Second Circuit, ... [which] recently upheld Section 1988 awards to Wall Street’s Cravath, Swain & Moore at the hourly rate of $60 in McCann v. Coughlin, 698 F.2d 112 (2d Cir.1983), and to the New York Legal Aid Society at rates exceeding $150 per hour in the Blum [v. Stenson, — U.S. -, 104 S.Ct. 1541, 79 L.Ed.2d 891 (1984) ] case.” More generally, a National Association of Attorneys General study released just this year found that: With different approaches being applied by the different circuits and even by various courts within each circuit, parties ... are subject to different approaches and hence different results. Courts disagree on what factors should be applied, how they should be applied, and even what they mean... As a result, in cases decided between 1974 and 1979, hourly rates awarded to civil rights attorneys varied by 685 percent. “At present, the enormous variation of fee awards cannot be explained in terms of the differing facts and circumstances from case to case____ As a result, from court to court and from case to case, attorneys and litigants who are similarly situated are subjected to widely differing treatment.” To reduce the arbitrariness characteristic of court awards of attorneys fees, the Supreme Court recently intervened and approved a version of the lodestar method of setting rates. In Hensley v. Eckerhart, the Court set forth the elements of this method for arriving at a fair attorneys fee: “The most useful starting point for determining the amount of a reasonable fee is the number of hours reasonably expended on the litigation multiplied by a reasonable hourly rate. This calculation provides an objective basis on which to make an initial estimate of the value of a lawyer’s services.” The product of the reasonable hourly rate times the hours reasonably expended is termed the “lodestar” figure. In Hensley the Court specified the basic methodology for setting the “hours reasonably expended.” The district court must exclude hours not reasonably worked. The parties are first called upon to exclude hours which are “excessive, redundant, or otherwise unnecessary”; should they fail to exercise such “ ‘billing judgment,’ ” the court should exclude the hours. Also eliminated from the calculus are hours spent on “unsuccessful claims.” More recently the Court has provided guidance on how to derive the second half of the equation, the “reasonable hourly rate.” In Blum v. Stenson, the Court faced a situation in which a non-profit law firm had represented the prevailing plaintiffs. The court held that the attorneys should receive fees based on the “prevailing market rates in the relevant community,” rather than on the cost of providing the legal service. As the Court noted in Hensley, setting the lodestar figure “does not end the inquiry.” The district court may adjust the fee upward or downward due to factors not reflected in the lodestar, with the most important factor supporting adjustment being the degree of success obtained. In Hensley the Supreme Court observed that the trial judge “may consider other factors” identified in Johnson v. Georgia Highway Express, Inc., though noting that “many of these factors usually are subsumed within the initial calculation of hours reasonably expended at a reasonable hourly rate.” Among the factors identified in Johnson is “[wjhether the fee is fixed or contingent.” In adopting the lodestar method of setting attorneys fees, the Supreme Court has placed great emphasis on the need for an efficient, objective system of awarding fees. One stated reason for choosing the lodestar approach is that it “provides an objective basis” on which to base the more subjective “enhancements” of the fee. The Court clearly hoped that this objective basis would enable parties and courts to avoid “a second major litigation” over the size of attorneys fees; ideally, the Court posited, the parties could settle the attorneys fee issue out of court. A. Reasonable Hourly Rates All parties agree that the attorneys for the plaintiffs in this case ought to be paid a “reasonable hourly rate”; all agree that the ultimate measure of the reasonableness of the hourly rate should be the “market rate” for Bredhoff & Kaiser’s services. The rather narrow task before this court is to determine how that “reasonable market rate” should be set. Laffey argues that the court must apply the same procedures to “for-profit” firms that it applies to nonprofit legal organizations. In each case the court must assess the “true value” of attorneys with similar skills, and pay them according to that true value. The rates charged by a firm have some probative value, but may be discounted or ignored if the court finds that other attorneys of similar caliber charge more or less. The trial court enjoys broad discretion to set rates within the parameters of the rates charged by other lawyers in the community. Northwest argues that the complex and burdensome task of establishing a “prevailing market rate” is a necessary evil which cannot be avoided where non-profit firms are involved, but which should be dispensed with where private, for-profit law firms have established market rates for similar services. Under this approach the rates customarily charged by a firm would presumptively stand as the reasonable rates. The positions of the parties are no doubt affected by the fact that in this case the “prevailing market rate” established by the district court exceeds “any hourly rate at which [Bredhoff & Kaiser] ... has ever billed its regular fee-paying clients.” However, this discrepancy need not exist in every case — under both approaches the touchstone remains the “market value” of the services rendered; any deviation between the court’s and the market’s assessment suggests that one or the other has misvalued the attorneys’ services. The choice between the two methods thus does not involve a categorical choice between higher rates or lower rates, but questions of judicial administration. 1. Congressional Intent As the Supreme Court observed in Blum, “[r]esolution of these ... arguments begins and ends with an interpretation of the attorney’s fee statute.” “The legislative history,” the Court pointed out, “explains that ‘a reasonable attorney’s fee’ is one that is ‘adequate to attract competent counsel, but ... [that does] not produce windfalls to attorneys.’ ” The “windfall” Congress sought to avoid is the awarding of fees in excess of the rate at which qualified counsel would be willing to represent civil rights claimants who have legitimate grievances. The congressionally-mandated inquiry is thus not into the “true value” or worth of an attorney’s services. Instead, the trial court must ascertain the fee at which competent counsel would be willing to accept meritorious civil rights eases. As this court recently stated in Murray v. Weinberger, “the purpose of the statute [authorizing fee shifting in Title VII cases] is to benefit meritorious claimants — not to subsidize the legal profession.” Although the Supreme Court has not set out the method by which district courts are to determine the hourly rates of attorneys working for profit, the Court in Blum took pains to note that the “prevailing market rate” — that is, the rate “prevailing in the community for similar services by lawyers of reasonably comparable skill, experience and reputation” — was a term of convenience, rather than some ideal rate which all firms should or could charge. The Court recognized that setting a market rate for legal services is inherently difficult, since in that “traditional sense there is no such thing ais a prevailing market rate for the service of lawyers in a particular community.” But the Court observed that although the fee setting process in section 1988 cases differed from most private sector cases in that the losing party paid his opponent’s legal fees, the “critical inquiry in determining reasonableness is now generally recognized as the appropriate hourly rate. And the rates charged in private representations may afford relevant comparisons.” Because the Legal Aid Society lawyers in Blum had no established billing rates, the Court required the parties to submit evidence of the “prevailing community rate.” Rates consistent with “those prevailing in the community” were deemed to be reasonable and were referred to — “for convenience” — as the “prevailing market rate.” This court’s cases are consistent with the Supreme Court’s recent pronouncement in Blum. In Copeland v. Marshall, this court approved fees based on the hourly rates customarily charged by the lawyers in a firm. Again, in National Association of Concerned Veterans v. Secretary of Defense, we stated that “[t]he best evidence would be the hourly rate customarily charged by the affiant himself or by his law firm.” And in Concerned Veterans, while the applicants attempted to support their claim by referring to evidence of fees received in similar cases, the government argued that “there was no evidence ... supporting the billing rates assigned to the associates who worked on this case”; this court remanded, holding that “[a]s private counsel with fee-paying clients, the applicants should have provided the Court with more informative data about the value of their services in the market.” Most recently, in Murray v. Weinberger, this court approved the district court’s use of attorneys’ customary billing rates in awarding fees for litigation brought under Title VII. Other circuits have also determined the reasonableness of attorneys fees by reference to the firm’s own established rates. The district court apparently believed that setting the “market rate” required precisely the same methodology employed to set the “prevailing market rate” in Jordan v. United States Department of Justice, in which this court accepted an affidavit recounting the rate “prevailing in the local legal community” for similar legal services. But this court in Jordan —like the Supreme Court in Blum —did not discuss the manner in which reasonable hourly rates for private attorneys should be set; because the attorneys involved had no regular billing rates, the court was required to set a hypothetical rate where none otherwise existed. The approach intimated by the Supreme Court and routinely employed by this and other circuit courts best effectuates Congress’s objective of attracting qualified counsel to bona fide civil rights cases. By setting the fee award at the attorney’s customary billing rate, the opportunity cost of foregone representations is precisely offset by a fee award in the same amount. As one academician has explained: The court must determine a value for the attorney’s time that will place statutory fee cases on a competitive economic basis and that will compensate attorneys in equitable fee cases for the loss sustained in creating the appropriated benefit. For lawyers engaged in customary private practice, who at least in part charge their clients on an hourly basis regardless of the outcome, the marketplace has set that value. For these attorneys, the best evidence of the value of their time is the hourly rate which they most commonly charge their fee-paying clients for similar legal services. This rate reflects the training, background, experience, and previously demonstrated skill of the individual attorney in relation to other lawyers in that community. ... [A] somewhat different situation is presented when the attorney does not have a customary hourly rate set by the competitive marketplace. 2. Administrability, Equity, and Efficiency Several strong policy reasons support tying the “reasonable hourly rate” to the firm’s own billing rates. Proceeding presumptively with the firm’s own rates allows the court to avoid the essentially impossible task of selecting one rate over another from a wide range of “market” rates, it limits the power of the trial judge arbitrarily to reward or punish attorneys by setting rates virtually at will, and it allows the parties and the court to avoid a “second major litigation” over the ratemaking process. The marketplace best measures “market value”; appraisal by no other method has as much claim to veracity and objectivity. a. The difficulty of judicial ratemaking No satisfactory method has been devised for “reinventing” the market for attorneys’ services. Setting the market yalue of any good or service can be a treacherous business. The volatility of the stock and commodities markets speaks eloquently of the difficulty of setting prices even for fungible goods in an efficient market; the complexity of the ratemaking procedures administered by federal regulatory agencies suggests the difficulty of constructing appropriate economic models from which to derive fair rates. The ratemaking process must necessarily prove even more difficult where attorneys are involved. As the Supreme Court has recognized, because attorneys are not fungible no one rate would be appropriate for all. Nor can the appropriate rate be set by reference to easily established factors, such as academic credentials. Some factors which make one attorney worth more than another — such as academic credentials and years of experience — are either reducible to paper or apparent to a trial judge. But others — such as the ability to communicate with clients, efficiency, judgment, and personality — are not so easily reduced or quantified. Given the complexity of the market for legal services, setting the “true value” for an individual attorney’s services promises to be neither a science nor an art, but a largely arbitrary divination. Nor do references to the overall market provide an infallible guide to the value of a particular commodity. An investor in International Business Machines would not be content to track the price of his stock by reference only to the overall average of the New York Stock Exchange; he would want the quotation for his particular stock. Similarly, he would not likely wish to sell his stock for a price picked from among the prices of all stocks on the exchange; he would prefer to sell at the price for which his stock sold. If there were a set market price or if the rates clustered in a narrow range, reference to other attorneys would provide an adequate guide. But, as this case indicates and as the dissent acknowledges, the spread in “market rates” charged by firms offering similar services is a broad one. In a case such as this one, which involves thousands of hours, a choice from either extreme of the spectrum would affect the ultimate fee award by hundreds of thousands of dollars. b. Arbitrariness If no basis exists for the figures used by the district court to set the lodestar, the entire elaborate process becomes a charade. The process merely mimics — rather than provides — mathematical objectivity. When the numbers fed into the lodestar are fundamentally arbitrary, no amount of calculation can restore objectivity. The fee setting approach becomes a complex and expensive overlay of delusive mathematical form over a process fundamentally grounded in an arbitrary assessment. So long as the result is arbitrary, it would be cheaper, simpler and easier to return to the days when the judge simply awarded an undivided figure that seemed fair. An examination of the district court’s opinion in this case underscores how arbitrary the picking of a fee could be. The record reflected a broad range of rates for attorneys of similar experience. In picking within this broad range, the district judge offered no explanation other than that he felt the Bredhoff & Kaiser lawyers to be worth it. There was no analysis of the market for legal services; there was no explanation as to why the Bredhoff & Kaiser attorneys were worth more in this case than they were in similar cases litigated at the same time. The dissent does not defend the district court’s judgment on this score. Rather than fleshing out its “true value” test for the benefit of trial courts in the future, the dissent is content to score debater’s points by “nickel and diming” the guidelines we offer. In accordance with its limited ambition, the dissent does not refute the relative superiority of customary billing rates to determine the “reasonable hourly rate.” The dissent concedes, in fact, that “in some, but by no means all, instances the applicant’s customary billing practice will provide the ‘best evidence’ of the market rate.” The dissent could have profitably stopped there, for the criticisms it offers of historical billing rates apply with even greater force to the components of the “true value” test it has chosen to defend. For example, the dissent quite correctly remarks that private sector billing rates often “depend on the nature of the service rendered, the relationship with the client, its ability to pay, or myriad other considerations.” But the trial court should be able to choose the billing rate for the most closely analogous service, which logically would be found among the rates charged by the claimant firm, and the range of billing rates for different clients should not be too wide (especially if the court — as it should — disregards abnormally high or low rates which may reflect the particularly attractive or distasteful nature of a matter, extraordinary time pressures, and other extraneous considerations). Whether the dissent accepts these reassurances or not, the •district court’s task is certainly easier when it has only one firm’s rate structure to grapple with rather than that of an entire legal community. If the customary billing rates of an individual lawyer or law firm cannot provide “predictability and ease of administration,” the dissent’s recommended open-ended inquiry into the going rate for legal services over an entire community must, a fortiori, lead to whimsical and capricious awards. c. Avoiding a “second major litigation” The approach taken by the district court also conflicts with the Supreme Court’s admonition to avoid engendering a “second major litigation” over the scope of attorneys fees. If the court takes seriously its self-imposed mandate to fix the “true value” of attorney services, it must draw on the rango of techniques used by other ratemaking agencies — voluminous records, expert testimony, elaborate economic models, and so forth. This elaborate procedure, of course, will be just the sort of protracted proceeding this court and the Supreme Court have inveighed against. Protracted proceedings cannot be avoided by vesting broad, unreviewable discretion in the trial judge to fix rates. Vesting absolute discretion in the district court to value an attorney’s services not only permits arbitrary fee awards, but also necessarily hampers accurate forecasting of what rate ultimately will prove “reasonable.” So long as so much flex remains in the hourly rate, the parties — especially those whose every hour will be compensated by the other side — have little reason to settle the amount of the fee award; sheer economic self-interest mandates litigating the “reasonable hourly rate” to the bitter end. The uncertainty would diminish, and the chance of settlement increase, if the hourly rates were set according to the established rates of those seeking compensation. It is easy to see that the incentives to settle are much stronger when fees are based on the prevailing firm’s historical rates. The combination of documented billable hours and a predictable hourly rate yields a projected lodestar figure that the parties can expect will withstand scrutiny. With specific enhancements limited to the exceptional case, the size of the fee award can be predicted. The losing party has every incentive to pay a predictable fee — any reluctance to pay only generates litigation over the size of the fee award, without much change of decreasing the final award. Indeed, the losing party pays a double penalty for litigating the size of the attorneys fees, since it must pay both its own and the prevailing party’s attorneys for litigating the fee award. With the magnitude of the fee award more or less apparent from the start, it is in the interest of the losing party to settle promptly and fully. So long as the losing party behaves rationally, the winning party has little incentive to press for excessive fees. The prevailing party will not be likely to collect higher fees for the merits, and if unreasonable in pressing its claims will not be compensated for the time spent litigating attorneys fees. The prevailing party would thus be well advised to litigate only when the losing party is being unreasonable — collecting both what is due on the merits and additional fees for the attorneys fee litigation — but settle when a reasonable settlement is offered. Such settlements depend on the predictability of the size of the ultimate award. Such a fee can be predictable only when the components — the hourly rate, the hours worked, and the specified enhancements— are also predictable. By basing the hourly rate on a predictable figure, today's opinion takes a significant step toward a system where fee awards routinely will be settled. By providing a meaningful benchmark for use by trial courts, we also hope to minimize appellate intervention into disputes over fee awards. The dissent argues, rather counterintuitively, that by providing guidelines to district judges we will encourage litigants to challenge the fee setting methodology on appeal in this court. We encourage no such conduct. First, by acknowledging our routine use of an attorney’s customary billing rates to determine the reasonable hourly rate, we reaffirm past practice. Our adherence to precedent will not raise false hopes of successful appeals. In contrast, it is the approach employed by the district court and defended by the dissent that gives rise to frequent appeals. Because no litigant can be sure that the fee award represents the prevailing attorney’s “true value,” he will have an almost irresistable incentive to seek an opportunity to convince this court that the district judge abused his discretion. Indeed, in many instances he will have good cause for his ire: the inevitable variability and inconsistency of the “true value” approach creates an ever growing pool of litigants with genuine grievances. Rather than review an essentially unguided albeit conscientiously made guess as to the true worth of an attorney, it is preferable to articulate objective, congressionally-inspired standards in advance. Proper application by the district courts in the first instance should minimize the need for appellate review. In the words of one scholar, “a more rational and consistent approach among the courts would greatly reduce the existing necessity of relitigating the ground rules in each case.” 3. The Dissent Although we have fully answered the dissent’s objections at the appropriate points in our opinion, we feel compelled to point out two conceptual flaws which, we respectfully submit, permeate the dissent’s analysis. First, in claiming that our approach contravenes the “unambiguous intent of Congress,” the “express admonition” of the Supreme Court, and “an unbroken line of cases in this circuit,” the dissent vastly overstates its case. Because our task is one of statutory construction, and because legislative history is a uniquely valuable tool in that endeavor, a closer look at the dissent’s argumentation on this point provides a useful illustration. In claiming that our construction is foreclosed by the legislative history, the dissent accurately points out that the Senate Committee Report cited three district court cases as correctly applying the “appropriate standards.” But the dissent’s quotation from the first case, Stanford Daily v. Zurcher, to the effect that “[t]his court does not accept the attorneys’ usual billing rates as definitively fixing their billing rates for this litigation,” is beside the point because we do not consider it definitive either. Indeed, the district court in the very next sentence suggested, consistent with our analysis, why such rates should be only presumptively binding: “[ T ]he simple fact [is] that attorneys may be leaving the area of their professional expertise in taking on pro bono publico litigation and that, as a result, their billing rates should reflect this fact.” Because the attorney’s rates “compare[d] favorably with the rates charged by other attorneys in this area ... [and because] these rates reflected the attorneys’ expertise,” the judge granted the fee request with only a minor modification. A clearer example of the methodology we outline today — reference to the customary billing rate followed by comparison to the prevailing community rate to ensure that the attorney’s customary rate is reasonable — could hardly be hoped for. The dissent also cites Swann v. Charlotte-Mecklenburg Board of Education for the proposition that “Congress did not intend to use historical billing rates as a cap on fee awards.” The district judge may have based his award on the “[f]ees paid to opposing counsel,” as the dissent suggests. But we will never know, because he also listed eight other factors, including such surrogates for historical billing rates as “[l]oss of other business” and “[f]ees customarily charged for similar services.” The mystery must remain unsolved because, after commenting casually on the eight factors, the judge concluded by saying: “Based on all the factors noted above ... I find that plaintiffs’ counsel are entitled to a fee of $175,000.00 for their services.” Given the conclusory manner of this congressionally-approved fee calculation, it is not surprising that the Supreme Court has rejected the dissent’s implicit assumption that Congress intended the methodology of these cases to be followed closely; otherwise, the Court could not have mandated the' lodestar method of fee-setting in the face of Swann. The dissent therefore surely overstates its case to say that this decision forbids the use of customary billing rates as presumptively reasonable hourly rates. (The third case cited in the Senate Report, Davis v. County of Los Angeles, concerned a “privately funded non-profit public interest law firm” for which past billing rates were unavailable. That court did not reach the question we decide today.) More fundamentally, the dissent’s reasoning — that because the legislative history does not explicitly endorse the approach we adopt today that it therefore precludes it— is difficult to fathom. To begin with, our approach, by in effect setting the reasonable hourly rate at the opportunity cost of litigating civil rights claims, fully satisfies Congress’s objective of encouraging meritorious litigation without awarding “windfall” fees. Even if not explicitly endorsed by the text of the statute and the history of its passage, the dissent cannot logically leap to the conclusion that our construction is forbidden. The committee reports and floor debate do not explicitly endorse the “true value” formula that the dissent would affirm either. Rather than hiding behind a nonexistent congressional intent, the dissent should defend its preferred fee setting approach on the basis of its consistency with the statutory scheme and with recent pronouncements of the Supreme Court. Second, the dissent’s approach is schizophrenic. It seeks to distance itself from use of customary billing rates because it feels incompetent to “plumb the mysteries of economic analysis,” yet at the same time it endorses an approach that requires district judges to duplicate the free market and ascertain an attorney’s “true value” by comparing his skill, experience, and reputation to lawyers who bill paying clients. The dissent is willing to compare fee applicants to private sector attorneys, but it is curiously unwilling to use the market rates of the attorneys themselves. The dissent’s attempt to unhook the customary billing rate from both private and nonprofit law firms must founder, because the entire lodestar framework mandated by the Supreme Court makes private attorneys’ hourly rates the baseline for comparison. After all, “[i]n order to establish a market value for the services of a lawyer who does not have an hourly rate set by the marketplace, it is necessary to look to the rate charged by comparable attorneys who do.” The implication of this point is important. Contrary to the dissent’s accusations, the approach set out above does not discriminate between private and nonprofit law firms — the inquiry is the same. For private law firms, the prevailing market rate for the firm’s services is presumptively found in the firm’s customary billing rates. For nonprofit law firms, the objective is still to determine the prevailing market rate for that firm’s services; but because such firms have no past billing history, a proxy for the actual market rate of the firm’s services must be found by examining the rates prevailing in the community for similar services by lawyers of reasonably comparable skill, experience, and reputation. 4. Summary When the costs of litigating attorneys fees are unavoidable, and they may be when firms with no relevant billing histories are involved, they must be tolerated. But when fixed market rates already exist, there is no good reason to tolerate the substantial costs of turning every attorneys fee case into a major ratemaking proceeding. In almost every case, the firms’ established billing rates will provide fair compensation. The established rates represent the opportunity cost of what the firm turned away in order to take the litigation; they represent the lawyers’ own assessment of the value of their time. To the extent unusual circumstances exist, those exceptional circumstances are best taken into account in adjustments of the lodestar. The trial court here improperly based the hourly rates on its own unguided assessment of the “true value” of the attorneys’ services. The proper approach is simpler and less arbitrary, as we have set forth above, and which may be summarized: The party seeking compensation should provide evidence of the rates charged by the firm for performing similar work in private representations. As the Supreme Court said in Blum, “the critical inquiry in determining reasonableness is ... the appropriate hourly rate. And the rates charged in private representations may afford relevant comparisons.” When the firm has been engaged “in private representation,” as Bredhoff & Kaiser has been, we can think of no more “relevant comparison” that “the rates charged in private representation” by Bredhoff & Kaiser itself. Next, the court may then “bracket” this rate by establishing that it falls within the rates charged by other firms for similar work in the same community. Again, as Justice Powell said in Blum (in the context of determining a non-commercial rate), “the burden is on the fee applicant to produce satisfactory evidence ... that the requested rates are in line with those prevailing in the community for similar services by lawyers of reasonably comparable skill, experience and reputation.” So long as the firm's own rate falls within the rate brackets, it is the market rate for the purposes of calculating the lodestar. Any adjustments or enhancements should be made during the “subjective” phase of the fee setting procedure, using the guidelines set forth in Hensley and Blum. Such adjustments should be made only for exceptional circumstances, and should be carefully and precisely explained. The above approach is in complete accord with the congressional objective as interpreted by the Supreme Court. The hourly rates Bredhoff & Kaiser lawyers were happily receiving from other clients were surely “adequate to attract” them to this litigation; substantially more, as was granted by the trial court here, would produce the very windfall Congress and the Supreme Court have said should be avoided. Because the district court applied an incorrect legal standard in that it relied not at all on the past billing histories of the law firms representing the prevailing plaintiffs, and instead attempted to assess the “true value” of their services, we must reverse and remand. B. Allotment of Tasks The appellants also object to the allotment of tasks between partners, associates, and paralegals. They object that the fees have artificially been elevated because partners have performed many tasks that would have been performed by associates at other law firms. The crux of their argument is that setting the “true value” of a firm’s services requires an analysis somewhat more probing than looking merely to the hourly rates; it requires an analysis of the firm’s staffing patterns. The analysis requested is not far different from that which a sophisticated client — who recognizes that the lowest hourly rate will not always generate the lowest overall bill— would make in an arm’s length negotiation. If this court were to adopt the approach followed by the district court, there would be some merit to this argument. Hourly rates — whether for lawyers or for steelworkers — are to some degree a function of productivity. Productivity, for lawyers and steelworkers alike, reflects both individual efficiency and capital investment. Some lawyers are simply more efficient than others; others are more productive because sophisticated support systems allow them to produce more in less time than other lawyers without such support. Given that the level of support could affect the productivity of two lawyers of equal ability, a “true value” analysis of a firm’s worth would have to include examination of the factors affecting productivity in the firm’s structure. The same lawyer might have an equal value on a per case basis whether as a single practitioner with one secretary or in a large firm supported by legal associates, paralegals, a battery of secretaries, word processing equipment and automated legal research tools. However, that same lawyer would obviously be worth much more per hour if he were working in a large firm supported by every conceivable human and electronic aid than as a solitary practitioner with minimum assistance. Experienced partners, for example, are generally considered most productive when their time is spent supervising less experienced attorneys. In this case, however, nearly two-thirds of the hours billed were billed by partners. This indicates that the lower priced — if equally talented — partners here did work that would not have been done by partners at a more highly leveraged firm. A private client would take such factors into account in comparing fees. A court setting a “true value” on a firm’s services would be obligated to consider such factors — and they are almost infinite in their variability — in constructing the ideal fee package for the firm. However, by deferring to the market we have obviated the need to set a “true”staffing pattern to go along with a “true value” set on the firm’s services. The record indicates that Bredhoff & Kaiser staffed this case much as it would a case for a paying client. The market has found Bredhoff & Kaiser’s staffing patterns — combined with their normal hourly rates — to produce an overall bill in accord with their market value. This market-tested allotment of both tasks and fees is not to be questioned by this court so long as it is reasonable. C. Contingency Adjustment The trial court doubled the lodestar on the merits to account for a “risk factor.” Northwest argues that this adjustment was unwarranted. 1. Contingency Multiplier The appellants characterize the contingency award as reflecting only the odds at the outset of the case. They claim that “[t]he district judge reasoned that, because he believed that plaintiffs’ chance of success at the outset of the litigation was 50%, a doubling of the lodestar was mathematically necessary to account for the counsel’s risk.” Under this approach, the contingency multiplier represents simply the inverse of the risk factor. Such an approach is presumably based on a goal of making “fee award cases as attractive to lawyers as their ‘usual’ cases.” Taken to its logical extreme, such an approach would engender results which would be at odds with current practice. For if the plaintiff’s chance of success was less than one-half, the fee should be multiplied by a factor greater than two. In theory, there should be no upper limit to this process: when the chance of success was one in fifty, the fee should be multi- ■ plied by fifty, and so forth. But commentators and courts have not gone this far. Other consequences of this approach lead to graver criticisms. The award would be greatest when the losing party was most unreasonable in litigating its case, creating a perverse penalty for those least culpable. In most cases the chances of winning could not be set with anything approaching mathematical precision, and so vast increases in attorneys would derive from a spurious mathematical base. Even if the chances of winning could be precisely assessed, moreover, the net effect of such a system would be to make a marginal case as attractive to bring as a very strong case — leading, one commentator has speculated, to a situation in which every conceivable claim would be litigated, subject only to the ability of the courts to handle the burden. If the capacity of the judicial system remained constant, these marginal claims would displace civil rights cases with a greater probability of success. The net effect would be a dilution in the deterrent force of the civil rights statutes. Had the trial judge applied such a crude multiplier, he clearly would have misapplied the controlling law. The statute allows awards only to prevailing parties. To multiply all awards to account for the risk of losing would ultimately generate a pool of legal fees sufficient to compensate all attorneys bringing all civil rights suits; the fund generated would differ from insurance only in the manner in which it was collected and distributed. While such a scheme is arguably desirable, it clearly is not the one adopted by Congress. The Supreme Court has emphasized the congressional intent to award fees only to prevailing plaintiffs. In Hensley the Court ruled that no attorneys fees could be granted for claims unrelated to those for which relief was granted. The Court reasoned that “[t]he congressional intent to limit awards to prevailing parties requires that these unrelated claims be treated as if they had been raised in separate lawsuits, and therefore no fee may be awarded for services on the unsuccessful claim.” The Court later added that “Congress has not authorized an award of fees whenever it was reasonable for a plaintiff to bring a lawsuit or whenever conscientious counsel tried the case with devotion and skill.” The same logic which restricts compensation to those portions of a lawsuit directly related to the relief procured also forbids multiplying attorneys fees so as effectively to compensate counsel for other, losing claims which may be brought. The prevailing party may expect full compensation for prevailing claims; there is no provision for compensating losing, unrelated claims in the same case, or other losing eases which might or might not involve the same parties. Any crude multiplier derived simply from the plaintiffs chance of success must be rejected as contrary to the congressional scheme. 2. Contingency Enhancement Laffey argues, however, that the district court employed a more sophisticated analysis than simply assessing the plaintiffs’ initial chance of success. • Laffey characterizes the relevant inquiry as looking not only to the initial outlook for success, but also to the magnitude of what was placed at risk. Under this more complex analysis, Laffey argues, the trial judge found that a doubling of the merits lodestar was required to compensate the plaintiffs fully in this particular case. In effect, Laffey argues, given the uncertainty of prevailing (and hence receiving payment) and the prospect of large outlays of time and money, a rational lawyer would expect twice the going rate to account for that uncertainty. Although a contingency multiplier would fully subsidize unsuccessful suits and encourage the filing of nonmeritorious claims, both effects in contravention of the statute’s purpose and the Supreme Court’s decision in Hensley, the propriety of an enhancement to the award is less clear. In Stanford Daily v. Zurcher — which was cited by the Senate Committee Report as “correctly applying]” the “appropriate standards” — the court held that it “[cjlearly ... must increase the fees award ... to reflect the fact that the attorneys’ compensation, at least in part, was contingent in nature.” But the court’s reason for making this adjustment was that “the contingent fee insures that counsel are compensated not only for their successful efforts but also for unsuccessful litigation____ [OJver the long run, substantial fees awards in successful cases will provide full and fair compensation for all legal services rendered to all clients.” Obviously, the California district court’s reasoning flatly contradicts the Supreme Court’s holding in Hensley that unsuccessful counts (and suits) do not qualify for an award of attorneys fees under the Act. In Blum, decided this Term, the Court made the propriety of contingency enhancements ever more suspect by reserving the question in such a way as to indicate that it had grave doubts as to whether such adjustments should ever be made: “We have no occasion in this case to consider whether the risk of not being the prevailing party ... may ever justify an upward fee adjustment.” Regardless whether contingency enhancements are ever warranted, it is clear in this case that a showing sufficient to qualify plaintiffs’ counsel for an enhancement cannot be made. The Supreme Court and this court have made it clear that “enhancements” are to be awarded only in the exceptional case. The Blum Court, immediately after refraining from deciding whether contingency adjustments would be appropriate because the issue was not raised in the fee request, did “reiterate what was said in Hensley,” that “in some cases of exceptional success an enhancement award may be justified.” Following the Supreme Court’s suggestion, this court applied “[t]he principle that only rare and exceptional circumstances can justify an upward adjustment of the lodestar” to an enhancement made to reflect “the contingency of payment due to the risk of losing on the merits.” In this case, the district court applied an erroneous legal standard when it assumed that plaintiffs’ counsel were “entitled ” to a contingency enhancement. As we stated only last month in Murray v. Weinberger, “an upward fee adjustment to compensate for the risk of losing — if ever appropriate (and Blum did not decide that it ever would be appropriate) — is permissible only in an exceptional case.” Whatever the Supreme Court meant by the term “exceptional,” this case does not fall within its embrace. The district court found that the plaintiffs’ initial chance of success was 50 percent. That chance of winning reflects the norm for litigated cases —in the average case, one side will win, one will lose. The level of risk was not unusual. Since this case did not present an exceptional level of risk, no risk enhancement should be awarded. D. Attorneys Fee Litigation Northwest objects to the magnitude of the award for litigating the attorneys fee issue. In addition to various specific objections, Northwest notes that more attorneys worked on the fee request than ever worked on the merits, and that more hours were billed on the fee request than were billed in any given year on the merits. We are, quite frankly, aghast at the hours devoted to the fee request. The award of fees for 3,400 hours of attorney time shows clearly that this fee request did turn into just the sort of “second major litigation” feared by both this and the Supreme Court; the prodigious consumption of lawyers’ hours and judicial time should give pause to anyone concerned with the allocation of legal resources. And it is for this reason, aside from the fact that it is basically fair to all concerned, that we have settled on the fee per hour customarily charged for similar work by the law firm making the claim as the most objective market rate and simplest for the court to administer. At the same time, we cannot say that the district court erred. Northwest objects to three major components of the fee award— the hours spent preparing an affidavit reciting the history of the case, time spent on interfirm meetings between the two firms involved in the fee request, and time spent in pre-application discovery. The trial court adequately addressed each of these issues. The judge specifically noted that the lengthy affidavit recounting the case was useful in his disposition of the fee request; he excluded those hours he felt were the result of unnecessary duplication caused by the entry of another firm into the case; and he examined the hours spent in discovery for reasonableness. His assessment of what constituted reasonable litigation of the attorneys fee issue, while perhaps more generous than this court might find appropriate on de novo review, does not constitute an abuse of discretion. E. Out-of-Pocket Expenses Appellants also challenge the award of certain litigation costs to the appellees. The gist of the appellant’s argument is that only taxable costs may be awarded in addition to fees calculated with regard to the hourly rate. This interpretation is overly narrow. The compensation structure employed by most firms encompasses more than hours billed; various expenses incurred by the firm are passed through to clients. As several other courts have held, “[t]he authority granted in section 1988 to award a ‘reasonable attorney’s fee’ included the authority to award those reasonable out-of-pocket expenses incurred by the attorney which are normally charged to a fee-paying client, in the course of providing legal services.” Since private, for-profit attorneys are involved, this court need not attempt to trace an unwavering line between those out-of-pocket expenses which are compensable and those which are not. The line of division — as with the hou