These appeals arise from a dispute over attorney's fees sought by Mabel A. King, the prevailing party in a Title VII lawsuit against the District of Columbia and James F. Palmer, Director of the D.C. Department of Corrections (collectively, the "District" or "District of Columbia"). King was represented throughout her lawsuit by counsel on a contingent fee basis. Following judgment on the merits, the district court awarded King attorney's fees and costs, as well as an enhancement for the risk of nonpayment equal to fifty percent of the fees for which counsel was at risk. The District of Columbia contends that no enhancement was proper, or alternatively that the level of enhancement should have been only twenty-five percent or less. On the basis of prevailing practice in the Washington, D.C. legal market, we hold that King was entitled to a 100 percent enhancement of the amount of attorney's fees at risk. That enhancement, however, may not be awarded on time devoted to the litigation of fee awards.

Mabel King initiated this suit against her employer, the District of Columbia, in July 1983 in order to obtain relief from gender-based discrimination. In September 1984, the district court entered judgment for defendants. On appeal, this court reversed and remanded, instructing the district court to enter judgment for King and to determine an appropriate remedy. King v. Palmer, 778 F.2d 878, 882 (D.C.Cir.1985). We noted that the appropriate remedy should include, at a minimum, the promotion of King, an award of back pay, and "a full consideration of any further relief." Id. at 882 n. 7. We remanded for further consideration King's related claims that she was the victim of a discriminatory work environment and of reprisals for having filed an EEOC complaint. Id. at 883. On remand, the district court awarded King a retroactive promotion and back pay, as instructed. The parties settled the claims of the discriminatory work environment and retaliation; the merits of the case are not at issue.

King was represented throughout the proceedings by Robert M. Adler, a sole practitioner, who was assisted on occasion by an associate and a law clerk or paralegal. The fee agreement between Adler and King provided that she would be responsible for billings of up to $5,000, as well as for costs and expenses. The agreement also provided that in the event King were to prevail and recover attorney's fees, she could recover any amounts paid, while Adler would receive whatever additional attorney's fees might be awarded. In effect, then, $5,000 of Adler's total fee was noncontingent.

As a result of this court's decision on the merits, King became entitled to an award of attorney's fees and costs pursuant to the fee-shifting provisions of Title VII. See 42 U.S.C. Secs. 2000e-5(k), 2000e-16(d) (1982). Accordingly, in March 1986 Adler submitted an application for an interim award of attorney's fees and costs incurred through February 28, 1986. He requested a "lodestar" fee (the presumptive fee arrived at by multiplying a reasonable hourly rate by the number of hours reasonably expended on the litigation) of $210,160 (based on his historic billing rates), as well as a contingency fee enhancement of thirty-five percent.

By order dated June 11, 1986, the district court awarded interim lodestar fees of $95,937.76. In October 1986, Adler submitted a supplemental application for lodestar fees incurred from March 1 to October 20, 1986, in the amount of $34,485. In addition, he again requested a risk enhancement of thirty-five percent. By order dated February 27, 1987, the court awarded further interim fees of $7,938.

On June 10, 1987, the district court considered the balance of the fee requests and entered a memorandum and order awarding Adler additional lodestar fees and costs of $137,953.28. King v. Palmer, Civ. No. 83-1980, Revised Memorandum, 1987 WL 12794 (D.D.C. June 10, 1987) ("Mem. Op. I"). The resulting total lodestar for the case to that point was $232,707.62, based on then-current billing rates. These fees represented compensation for all phases of the litigation, including time spent in pursuit of attorney's fees and time spent on various other matters following this court's remand, such as the defendants' petition for rehearing. As to the risk enhancement, the court retained jurisdiction of the matter pending the Supreme Court's decision in Pennsylvania v. Delaware Valley Citizens' Council for Clean Air, 483 U.S. 711, 107 S. Ct. 3078, 97 L. Ed. 2d 585 (1987). Mem. Op. I at 12. The court noted, however, that a "15 [percent] bonus for the risk of not prevailing would ... be appropriate in the event that such an award is authorized by the Supreme Court." Mem. Op. I at 13. The court rejected Adler's claim for a bonus based on exceptional success, and that decision is not appealed here.

Following the decision in Delaware Valley, Adler again applied for a contingency enhancement, this time requesting an augmentation of 100 percent. On September 20, 1988, the district court awarded plaintiff a fifty percent enhancement for the risk of nonpayment, or a total of $113,858.31, based on the entire contingent portion of the previously awarded lodestar fee ($232,707.62 - $5,000 = $227,707.62). King v. Palmer, Civ. No. 83-1980, Memorandum at 5, 1988 WL 104970 (D.D.C. Sept. 20, 1988) ("Mem. Op. II").

The district court considered the request for risk enhancement under the standards established by Justice O'Connor's tie-breaking concurrence in Delaware Valley. The court found that King had offered evidence that "many lawyers in the civil rights employment discrimination market would not accept employment in a case like this on a purely contingency fee basis." Mem. Op. II at 3. The court asserted, however, that because the fee arrangement in the instant case was only partially contingent, it was distinguishable from the case of McKenzie v. Kennickell, 684 F. Supp. 1097 (D.D.C.1988), aff'd, 875 F.2d 330 (D.C.Cir.1989). Mem. Op. II at 3. Instead, the court relied on Palmer v. Shultz, 679 F. Supp. 68 (D.D.C.1988), which also involved a partially contingent arrangement. Mem. Op. II at 3-4. There, the court had determined that "attorneys in the Washington D.C. community will only accept a fully contingent case if their recovery will be at least double their normal hourly billing rate and will only accept a partially contingent case if their recovery is enhanced by at least 50 percent." Mem. Op. II at 4 (quoting Palmer, 679 F. Supp. at 74). Adhering to Justice O'Connor's suggestion, in Delaware Valley, that "each court let a determination such as Judge Smith's [in Palmer ] control future cases such as this one," the court adopted a fifty percent enhancement in the instant case. Mem. Op. II at 4.

Both King and the District appealed the enhancement award.

The District contends that no risk enhancement should have been awarded or, in the alternative, that the level of the enhancement should have been twenty-five percent or less. King maintains that it should have been 100 percent of the amount at risk. The District also challenges certain elements of the risk enhancement award, claiming that it was excessive because it was based, in part, on a lodestar that included fees that were not at risk or for which an enhancement was unnecessary to attract competent counsel. We address these issues in turn.

A. The Availability and Level of Risk Enhancement

In McKenzie v. Kennickell, 875 F.2d 330, 332-33 (D.C.Cir.1989), we adopted Justice O'Connor's concurring opinion in Delaware Valley, 483 U.S. at 731-34, 107 S.Ct. at 3089-91, as the basis for determining the availability and level of contingency fee enhancements in this circuit. See also Weisberg v. U.S. Dep't of Justice, 848 F.2d 1265, 1272 (D.C.Cir.1988); Thompson v. Kennickell, 836 F.2d 616, 621 (D.C.Cir.1988). We followed the two-part test suggested by Justice O'Connor: first, whether the relevant legal market adds a premium for contingency, and second, whether in the absence of an enhancement for risk the plaintiff would have faced substantial difficulty in finding counsel. McKenzie, 875 F.2d at 332.

To secure a contingency enhancement, a fee applicant must first show the degree to which the relevant legal market compensates for contingency. Delaware Valley, 483 U.S. at 733, 107 S.Ct. at 3090. The "relevant market," for purposes of this analysis, is composed of all contingency claims in the District of Columbia, with special reference to those involving complex federal litigation such as (but not limited to) Title VII cases. McKenzie, 875 F.2d at 334-35. In addition, the relevant market is to be determined as of the time the case was undertaken by counsel, rather than at the time fees were awarded. Id. In making such an assessment, "district courts and courts of appeals should treat a determination of how a particular market compensates for contingency as controlling future cases involving the same market" in order to avoid "[h]aphazard and widely divergent compensation." Delaware Valley, 483 U.S. at 733, 107 S.Ct. at 3090. We thus rejected an individualized approach to fee awards. McKenzie, 875 F.2d at 332, 336, 338.

In awarding a fifty percent enhancement in this case, the district court correctly followed the principle of consistency in fee determinations discussed above, as it adhered to the previous contingency enhancement findings of Palmer v. Shultz, 679 F. Supp. at 74. Its error, however, was in applying different levels of enhancement to fully and partially contingent cases. The court asserted, without explanation, that there was a "crucial distinction" between such cases and thereby relied on the finding in Palmer that "plaintiffs 'have shown that attorneys in the Washington D.C. community will only accept a ... partially contingent case if their recovery is enhanced by at least 50 percent.' " Mem. Op. II at 4 (quoting Palmer, 679 F. Supp. at 74).

Although the Palmer court's calculations of attorneys' fees are difficult to discern, it appears to us that it actually awarded an enhancement of 100 percent of the contingent portion of the fees--precisely the enhancement urged by King--even though the court referred to a fifty percent enhancement for partially contingent cases. Palmer was obligated to pay her counsel approximately fifty-three percent of their fees; thus, the case was about forty-seven percent contingent. See 679 F. Supp. at 73. Counsel charged plaintiff rates that were considerably below market. Id. at 71-72. The total lodestar award was $291,160.43, which was based on historic market rates. Id. at 76. For purposes of the enhancement, however, the court used a lodestar figure of $143,612.34, which was based on the historical rates actually charged by counsel. See id. at 73-76; Affidavit of Ellen Kabcenell Wayne, at 4 (counsel in Palmer describing calculation of fee enhancement). The enhancement award of $71,806.17, therefore, was exactly fifty percent of the lesser amount, and slightly over 100 percent of the contingent portion (forty-seven percent). See Palmer, 679 F. Supp. at 76.

Be that as it may, we conclude that the same percentage enhancement should be applied to the contingent portion of a fee regardless of whether the case is fully or partially contingent. None of the affidavits in the record reveals a practice of charging a lower percentage enhancement of the amount at risk in partially contingent arrangements, nor do previous decisions in our circuit reflect such a practice. As an economic matter, such an approach makes little sense: we see no reason why Adler, whose fee was ninety-eight percent contingent, should not receive the same rate of enhancement as an attorney who accepted the case on a 100 percent contingent basis. In either case, the attorney would face the same risk of nonpayment for services rendered on a contingent basis, and should therefore be entitled to receive the same percentage enhancement on whatever portion of the lodestar that is at risk.

The question to be determined, then, is the level of enhancement prevailing in the relevant Washington market in 1983. King submitted declarations of over seventy practitioners in support of her contention that a 100 percent risk enhancement was required to ensure contingency representation in the Washington market. Although the District correctly observes that many of these affidavits refer only to a "reasonable" enhancement while others range from significantly below to significantly above 100 percent, we find that the bulk of the evidence supports a 100 percent enhancement as to both the current and the 1983 markets. See, e.g., Declaration and Supp. Declaration of John R. Erickson; Declaration and Supp. Declaration of Joel P. Bennett.

The District nevertheless argues that as Adler himself initially requested only a thirty-five percent enhancement in this case and as he had been awarded only a ten percent enhancement in an earlier Title VII case, see Bundy v. Jackson, Civ. No. 77-1359 (D.D.C. Sept. 14, 1982), Adler could not reasonably have expected an enhancement of fifty percent, much less 100 percent, when he agreed to represent King. However true that may be, the District's approach would "contravene the settled doctrine ... that fees must be awarded regardless of the identity of the fee petitioner." McKenzie, 875 F.2d at 338. In McKenzie we "reject[ed] the notion that applicants must demonstrate that they ... were specifically attracted by the possibility of a contingency enhancement." Id. Rather, as Justice O'Connor counseled in Delaware Valley, we treat contingency cases as a class, based on the prevailing practices within the relevant legal market.

That practice has now been defined, in a number of recent decisions within this circuit, as requiring a contingency enhancement of 100 percent for the class of cases relevant here. See, e.g., Thompson v. Kennickell, 710 F. Supp. 1, 6, 9 (D.D.C.1989) (awarding enhancement of 100 percent and finding "that the attorney market in the District of Columbia requires contingency enhancements of 100% to 200%"); Palmer v. Shultz, 679 F. Supp. at 74 ("at least 100 percent" for fully contingent cases); Broderick v. Ruder, Civ. No. 86-1834, Mem. Op. at 4 (D.D.C. Sept. 13, 1989) (awarding 100 percent enhancement); see also McKenzie, 875 F.2d at 336 (upholding a 50 percent enhancement, which plaintiff had not appealed, as permissible but describing it as "below the Title VII contingency enhancements typically awarded in this circuit").

We note that although the 100 percent enhancement established by the Washington legal market is substantially higher than the upper limit of one-third urged by Justice White in his plurality opinion in Delaware Valley, 483 U.S. at 730, 107 S.Ct. at 3089, it does not provide attorneys with the incentive to take on cases "in which [they] believe[ ] there is less than a 50-50 chance of prevailing," which the plurality would discourage. Id. This is so because with a 100 percent enhancement, attorneys who calculate the risks of the cases they agree to litigate will only realize their non-contingent rates of compensation if they succeed in at least fifty percent of those taken on a contingency basis. Of course, some attorneys will undoubtedly choose to accept higher risk cases because of their personal belief in the cause, the novelty of the issue, or the potential importance of the case. Under our ruling, however, attorneys who accept such cases must bear the cost of the added risk.

To recapitulate, we hold that the district court erred as a matter of law in setting the enhancement at fifty percent. Based on the standards of McKenzie, the record before us, and prior decisions within this circuit, we find that the appropriate enhancement to be awarded Adler for his risk of nonpayment is 100 percent of the amount of lodestar fees at risk. Moreover, in conformity with the goal urged by Justice O'Connor, and adopted by this court, that contingency cases within a given market be treated consistently, see Delaware Valley, 483 U.S. at 733, 107 S.Ct. at 3090; McKenzie, 875 F.2d at 332, we hold 100 percent to be the appropriate contingency enhancement to be awarded in the District of Columbia until new evidence indicates a significant change in the legal market--or until we receive new guidance for the award of contingency fees from either Congress or the Supreme Court.

To prevail on her claim for fee enhancement, King must also establish, pursuant to the second prong of the McKenzie test, that in the absence of a risk enhancement, she would have faced substantial difficulty in securing the services of a competent lawyer at the time she sought one. McKenzie, 875 F.2d at 332, 336-38. This inquiry, we explained, is "necessarily a counterfactual [one]; [attorneys seeking enhancements] are not required to show that plaintiffs actually did face difficulty." Id. at 337. Nor is it necessary for them to demonstrate that they were "specifically attracted by the possibility of a contingency enhancement." Id. at 338.

The District contends that the award should be reversed because the district court made no finding that King would have faced substantial difficulties in securing counsel in the absence of risk enhancement. In addition, the District attacks the sufficiency of the evidence proffered by King, arguing that the affidavits submitted are not conclusive proof that there were no lawyers who would have taken King's case on a contingent basis without the possibility of enhancement. The District speculates, for instance, that a public interest group or possibly one of several private attorneys might have agreed to represent her, although its argument rests not on any affirmative evidence but merely on supposed ambiguities in certain of the affidavits submitted by King.

It is true that the district court did not explicitly find that King would have faced substantial difficulty in obtaining counsel in the absence of risk enhancement. That finding, however, was implicit in the court's citation to the Palmer court's findings that attorneys in the District would not accept contingent cases without some risk enhancement. Mem. Op. II at 2, 4. Moreover, the affidavits filed with the court make it abundantly clear that without this inducement, King would have faced substantial difficulty retaining counsel. See, e.g., Supp. Declaration of George M. Chuzi ("During 1983, I was personally familiar with most of the attorneys regularly bringing Title VII suits in the District of Columbia on behalf of plaintiffs. Had Mr. Adler not agreed to represent Mrs. King in this case, I am unaware of any other Title VII attorney who would have agreed in 1983 to represent her on a contingency fee basis [without some risk enhancement].").

The District's attacks on the sufficiency of King's evidence are unavailing. The test adopted by this court does not require a showing that it would have been impossible to secure counsel on other than an enhanced contingency fee basis, as the District's arguments seem to assume; it requires only a showing of "substantial difficulty." Moreover, the evidence submitted need not be scientifically or statistically unassailable; "[e]vidence in the form of affidavits from practitioners is acceptable." McKenzie, 875 F.2d at 337-38. Here, King has met her burden.

Finally, we are unimpressed by the District's argument that Adler's willing representation is relevant evidence that King has failed to satisfy the "substantial difficulty" test. Adler himself declared in an affidavit that he would not have accepted the case without the possibility of risk enhancement. For its part, the District offered no evidence that Adler would have taken the case without risk enhancement; it merely points to the fee agreement with King, in which Adler states the obvious fact that "there is no assurance" as to the amount of an attorney's fee recovery, if any. More fundamentally, however, it is irrelevant that Adler might have accepted King's case without an enhancement, so long as she was able to demonstrate (as she has) that hers is in a class of cases in which "plaintiffs would have faced 'substantial difficulties' [in securing counsel] in the absence of contingency enhancements." McKenzie, 875 F.2d at 337 (emphasis in original).

B. Challenged Elements of the Risk Enhancement Award

The District argues that even if some enhancement is appropriate, certain portions of the district court's risk enhancement award should be set aside. The District asserts, first, that the lodestar on which the enhancement was calculated should not have included fees for any time spent after December 13, 1985, when this court issued its decision on the merits, because such fees were no longer truly at risk.

We agree with the District that compensation for time devoted to securing an award of attorney's fees may not be enhanced. It is not disputed that the time spent litigating a fee award is itself compensable. Copeland v. Marshall, 641 F.2d 880, 896 (D.C.Cir.1980) (en banc). Thus such compensation is not subject to a genuine risk of loss. Furthermore, as Adler himself admitted at oral argument, we doubt that lawyers require special incentives to pursue their own compensation. We therefore hold that contingency enhancements may not be applied to fees accrued in petitioning for, or litigating, fee awards. See Baughman v. Wilson Freight Forwarding Co., 583 F.2d 1208, 1219 (3d Cir.1978) ("[H]ours devoted to the fee petition should not be augmented ... because of the contingent nature of the case....").

We affirm the district court, however, as to all the remaining elements of the lodestar. In originally calculating the lodestar, the district court fully addressed and discussed the compensability of Adler's time following this court's remand. See Mem. Op. I at 8-9, 15-18. From the district court's opinion, it is clear that King was not assured of victory on the merits at the time she incurred the post-remand fees. Rather, these fees related to the defendants' petition for rehearing, the United States' motion to file an amicus brief, and the retaliation and hostile work environment claims remanded by this court. Accordingly, "in view of the district court's superior understanding of the litigation," we conclude that the court properly based enhancement on all the contingent elements of the lodestar fee except that attributable to time spent in pursuit of attorney's fees. See Hensley v. Eckerhart, 461 U.S. 424, 437, 103 S. Ct. 1933, 1941, 76 L. Ed. 2d 40 (1983).

Lastly, the District argues that the risk enhancement award was erroneously calculated on a lodestar that was computed using Adler's then-current, as opposed to historic, billing rates. The result, claims the District, was to magnify the amount appropriate to compensate for risk of loss by adding to the contingency enhancement an "interest" factor designed to compensate for delay in payment. Both parties concede that the use of then-current rates in calculating the lodestar was appropriate as compensation for delay in receiving fees--in other words, as the equivalent of interest. As the economic cost of the risk of nonpayment for which he is to be compensated began at the commencement of representation, we see no reason why counsel should not be compensated for the delay in receiving the enhancement award as well.

We hold that King is entitled to a risk enhancement of 100 percent of the amount of fees at risk. We also hold that the enhancement may not be applied to fees awarded for time spent on attorney's fees matters. We therefore reverse the order of the district court and remand for a calculation of fee enhancement and entry of judgment in accordance with this opinion.

Reversed and Remanded.

STEPHEN F. WILLIAMS, Circuit Judge, concurring:

I concur in the opinion of the court except for the passage at 766-67 that rests approval of the 100% enhancement on affidavits as to market conditions. I write separately to explain why I regard the selection of a risk enhancement figure as in reality a policy choice, one which we as courts are not qualified to make but are required to by controlling precedent. I conclude with a few thoughts on the implications of contingency enhancements with one-way fee-shifting.

* The opinion's sifting of evidence as to the local market for work on a contingent fee in my view veils a policy judgment: a decision on what the minimum chances of victory must be for a case to be financed by the combination of fee-shifting with contingency enhancements. The answer is, in effect, to allow such financing for any case where the plaintiff's prospects of success are 50% or better.

Unless we impute irrationality to lawyers as a class, we must assume that they are ready to work on contingency at a whole range of premium levels that will, adjusting for the riskiness of the case and for their own risk aversion, make the contingent work the equivalent of noncontingent. Thus, momentarily putting aside risk aversion, if they can get double the basic fee, they will be ready to handle cases with a 50-50 chance of victory or better (the fee in half the cases, doubled, equals noncontingent reimbursement in 100% of cases); if they can recover triple the basic fee, they will be ready to take cases with a 33% chance of victory or better (the fee in one third of the cases, tripled, equals noncontingent reimbursement in all); and so on.

Risk aversion--the conventional preference for an assured $1000 over a 50-50 chance of $2000, or for a bird in the hand over two in the bush--may change the numbers slightly, but probably not a great deal; a large law office, at least, can diversify the risks over a large range of cases. Of course to keep the risks satisfactory, lawyers must make sure to take pools of somewhat homogeneous cases; winning three $1000 cases will not balance losing three $100,000 ones.

We have here a slew of affidavits from local lawyers, many asserting an unwillingness to take contingent work on a basis of less than 100% enhancement, others a vaguer insistence on "reasonable" enhancement. I have some skepticism about the alleged insistence on 100% enhancement; if a lawyer can identify a class of cases with a two-out-of-three chance of success, he will need only a 50% enhancement to attain the equivalent of noncontingent pay. Justice White plainly recognized the point in his opinion in Pennsylvania v. Delaware Valley Citizens' Council for Clean Air, 483 U.S. 711, 107 S. Ct. 3078, 97 L. Ed. 2d 585 (1987), when he suggested a percentage limit to deter lawyers from bringing suits "in which the attorney believes there is less than a 50-50 chance of prevailing." Id. at 730, 107 S. Ct. at 3089. (Justice White set that percentage at 33 1/3%. In fact, this would require lawyers to insist on cases with a three-out-of-four prospect of success.)

If I am right as to the relation between riskiness and premium, the assertion of many lawyers that they will take a contingent fee case only at a 100% enhancement or better is puzzling. One possibility is simply that plaintiff has been much more assiduous than defendant at ferreting out supporting statements. Another is that fine-tuning the risk level in terms of a line between 50-50 and two-out-of-three is very difficult--clients do not tell the whole story, juries are a bit of a gamble, judges perhaps no better. Nonetheless, lawyers and clients do make judgments of this sort when they evaluate settlement proposals, so I suspect that if Congress set 50% as the maximum enhancement for a shifted contingent fee, a very considerable bar would develop to handle such business. In sum, I view causation as running in the opposite direction from that supposed by the controlling precedents; I see the judicial judgment as defining the market, not vice versa.

Thus the decision seems to me properly Congress's. For when we look beneath the veneer of market analysis, the allowance of a 100% enhancement is clearly legislative--making the policy judgment that it is suitable to allow use of enhanced contingent fee-shifting for cases with a 50-50 chance of success or better. I know of no basis on which a court would be competent to select that level--or any other. But given the various precedents aptly summarized in Judge Buckley's opinion, I believe we have no choice but to make the judgment. The only guidance that we have on the subject is section V of Justice White's opinion in Delaware Valley, a section joined by three other Justices. As I've already noted, that discussion points in two directions: the proposed 33 1/3% enhancement implies a minimum-prospect case with a .75 chance of success, but the proposed 50-50 minimum-prospect case implies a 100% enhancement. As my colleagues have chosen the 100% (and thus 50-50), and the entire legislative judgment is one on which I have no claim to authority at all, I certainly have no basis for dissent.

The implications of contingency enhancements with one-way fee-shifting are curious. In the normal contingency fee arrangement (i.e., without fee-shifting), the plaintiff's possible gain sets a limit on the investment of legal effort; it usually represents the sole source of any return on that investment. There will of course be cases of principle, but they will not feed or clothe much of the bar much of the time. To introduce one-way fee-shifting, as Congress has for Title VII of the 1964 Civil Rights Act; see Christiansburg Garment Co. v. EEOC, 434 U.S. 412, 422, 98 S. Ct. 694, 701, 54 L. Ed. 2d 648 (1978) (construing fee-shifting section, which allows fees for the "prevailing party," as allowing awards to prevailing defendants only when plaintiff's claim was "frivolous, unreasonable, or groundless"), weakens but does not eliminate that barrier. It survives to a degree, because without enhancements for contingency, counsel must contract for enough of plaintiffs' possible recoveries to compensate themselves for the time spent on unsuccessful suits. The enhancement for contingency lowers the barrier still further; where it is set at 100%, as here, all cases with a 50-50 chance or better become promising spots for investment of legal effort. (Plaintiff's potential recovery has a residual relevance even here, as agreements entitling the lawyer to part of the recovery provide an additional source of compensation for time spent on losers.)

But this does not mean that all 50-50 cases will be litigated to the hilt; a defendant can just give up. If one assumed that the stakes were symmetrical (a dollar gain for plaintiff being matched by a dollar loss for defendant), the same forces that in conventional contingent-fee litigation limit the plaintiff's lawyer's investment would operate indirectly, through the defendant (though with the terms of the likely settlement affected by the one-way characteristic of the fee-shifting, a matter addressed below). Often, however, they will not be. For example, in a case by an employee claiming wrongful dismissal, there may be only a modest gain to the plaintiff from securing reinstatement; the next-best employment opportunity may be quite similar. But the employer may have come to the view that plaintiff's presence imposes substantial costs--salary (less the employee's positive contribution), plus negative effects on output. The present discounted value of these costs will represent a potential downside risk of the litigation for defendant.

In considering settlement, defendant will look at the following (at least) as the costs of persisting: (1) his additional legal fees; and (2) the costs of losing, including (a) monetary judgments, (b) the sort of costs suggested in the paragraph above, and (c) the costs of plaintiff's fee (adjusted for the risk enhancement), all adjusted for the likelihood of losing. Cf. George L. Priest, Selective Characteristics of Litigation, 9 J. of Leg.Stud. 399, 401 (1980). The rule we here adopt makes 2(c) a hefty item.

Two consequences emerge, then: First, adding contingent-fee enhancement to one-way fee-shifting enlarges the class of cases that will be brought free of a conventional limit on the investment of legal resources in litigation, namely, the value of the litigation to the plaintiff. It thus implies an increase in the portion of the country's wealth to be devoted to litigation (mainly to lawyers). Second, adding enhancement for contingency greatly improves the plaintiff's bargaining power in settlement negotiations. As several opinions in Delaware Valley observed, enhancement of a shifted fee has the effect of making losing defendants pay not only their own litigation costs, and those of the plaintiffs to whom they lose, but also, indirectly, the fees of losing plaintiffs. See 483 U.S. at 719-20, 722, 107 S. Ct. at 3083-84, 3085 (section III-A of Justice White's opinion and thus the opinion of the Court); see also id. at 732, 107 S.Ct. at 3090 (O'Connor, J., concurring in part, and concurring in the judgment). But cf. id. at 752-53, 107 S.Ct. at 3100-101 (Blackmun, J., dissenting). As the fee-shifting is one-way only, this is not offset by any possibility of defendants' recovering their own legal expenses in the event of victory. The resulting enfeeblement of defendants' bargaining position alters the character of the settlements that will emerge, shifting the real-world impact of the substantive law toward plaintiffs. While one might argue that this shift merely counterbalances factors artificially impeding plaintiff recoveries and thus makes the net impact of the statute correspond to its language, it may in fact carry the balance far beyond that point.