EASTERBROOK, Circuit Judge.

A prior appeal in this nationwide class action ended with two principal conclusions: first, the district court did not abuse its discretion in approving the settlement; second, the court erred in capping attorneys' fees at 10% of any "megafund" recovery. In re Synthroid Marketing Litigation, 264 F.3d 712 (7th Cir. 2001). A court must give counsel the market rate for legal services, we held. Although the market rate, as a percentage of recovery, likely falls as the stakes increase, whether it exceeds 10% for recoveries above $100 million must be answered by reference to arrangements that satisfy willing buyers and sellers rather than the compensation that a judge thinks appropriate as a matter of first principles.

On remand, the district court awarded the third-party payor (TPP) class counsel 22% of that class's recovery, plus litigation costs and expenses. Consumer class counsel received 30% of the first $10 million recovered by that class, 25% of the next $10 million, 20% of the third $10 million, 15% of the fourth $10 million, and 10% of the remaining $48 million — an average of 15.45% — without separate compensation for litigation costs and expenses. In re Synthroid Marketing Litigation, 201 F.Supp.2d 861 (N.D.Ill. 2002). Consumer class counsel appeal in quest of a higher award; a group of TPPs calling itself the Health Benefit Payers appeals to seek a reduction in the award made to the TPP counsel.

Readers seeking the back story can find it in our prior opinion and the district court's decision on remand. For now it is enough to say that the district court divided the plaintiffs (who contend that Knoll Pharmaceuticals and its successors defrauded purchasers into paying too much for Synthroid) into two classes: consumers who used the drug to treat their hypothyroidism, and third-party payors (insurers and other health-care vehicles) that reimbursed some of the consumers or paid directly for the Synthroid. The consumer class receives $88 million, and the TPP class $46 million. Under the district court's formula, total disbursements to consumer class counsel are $13.6 million, while TPP class counsel receive fees of $10.12 million plus about $621,000 in reimbursements.

Fixing the market rate for the legal services of the TPP counsel is simple, the district judge found, because attorneys and clients set it themselves through arms'-length negotiations. All of the TPPs are sophisticated financial intermediaries with in-house counsel who can (and do) shop for legal services in a national market. Many of the TPPs hired law firms to conduct this litigation, and these TPPs agreed on which of these lawyers would take the laboring oar for the class. Almost all of the TPPs that hired lawyers for this case did so on contingent rather than hourly fees, and the average rate that the TPPs agreed to pay their lawyers is 22% of any recovery. As we remarked the last time around, the outcome of this competitive process among informed buyers and sellers defines the market rate for legal services, given the risks and investments of time that the lawyers expected to encounter in this case. 264 F.3d at 720. The district judge therefore adopted it as the measure of TPP class counsel's compensation.

According to the Health Benefit Payers, however, a better market measure is available: the deal the parties made in July 1999 when negotiating toward settlement. Under this agreement, many TPPs that had their own lawyers would pay them from their share of the kitty; counsel representing the TPP class as a whole would receive about 22% of the portion of the fund attributable to the remaining TPPs. Class counsel's compensation would have been close to 10% of the whole fund. Yet the district judge awarded class counsel more than twice what counsel had agreed to accept. How can that be a market rate?, the Health Benefit Payers ask. The district court did not answer this question. Instead it ruled that, because the Health Benefit Payers are class members rather than class representatives, they lack standing to protest the amount of fees. This decision is hard to fathom. The Health Benefit Payers stand to receive more from the settlement fund if they win on this appeal than if they lose; payments to counsel come at their expense, and this loss is redressable by a favorable judicial decision. What more is required for standing? See Lujan v. Defenders of Wildlife, 504 U.S. 555, 112 S.Ct. 2130, 119 L.Ed.2d 351 (1992). The Supreme Court has held that class members may obtain appellate resolution of their objections without formally intervening, see Devlin v. Scardelletti, 536 U.S. 1, 122 S.Ct. 2005, 153 L.Ed.2d 27 (2002), so we take up the Health Benefit Payers' arguments.

One fundamental problem is that these arguments should have been made three years ago, as reasons to affirm the district court's initial award, which at 10% of the TPP class fund was exactly what the Health Benefit Payers now say is right. Yet the Health Benefit Payers did not bother to file a brief in that appeal. We analyzed the situation on the assumption, held by all who provided us with their views, that the TPP class counsel would receive a share of the entire settlement fund, not a fund reduced (as the July 1999 document contemplated) by the entitlements of TPPs that had engaged separate counsel. Our remand instructed the district court to proceed exactly as it did. The earlier appeal concerning the TPP counsel, and the remand, were pointless if the only proper outcome is an award equaling 10% of the whole fund. That is the very award we reversed, and the law of the case is against reviving it now.

What is more, although we used the word "agreement" two paragraphs ago to describe what transpired in July 1999, no definitive agreement was reached and signed. The negotiations came unglued. True, the stumbling block was an issue other than attorneys' fees, but the fact remains that no deal resulted. Contracts are enforced or not as a whole; negotiators cannot insist on enforcement of one provision unless all loose ends have been tied up. Negotiations such as those that occurred in mid-1999 are designed to bring closure; TPP class counsel may have been willing to accept less than their legal entitlement in order to increase the chance that they would be paid then and there. They were not paid in 1999. Now, four years later, the case is on its second appeal. For four years the TPP class counsel have had to fight for their compensation, incurring costs (and facing risks) that they would not have borne had the case been wrapped up and payment made in 1999. Perhaps, too, concessions made on attorneys' fees were related to some of the issues on which agreement could not be achieved. Moreover, according to TPP class counsel, the 1999 bargain on attorneys' fees was a response to the Health Benefit Payers' representation that they would opt out, a step that would have prevented class counsel from recovering any fees on their account. After the arrangement broke down, however, the Health Benefit Payers remained in the class and thus must bear their portion of the legal expense. At all events, until a contract is signed — and, in class litigation, approved by the court under Fed.R.Civ.P. 23(e) — no one is bound by any of the proposed terms. If the lawyers representing the Health Benefit Payers contributed toward the success of this litigation, they could have sought a distribution from the fund. They did not do so and are not entitled to have the class counsel's compensation for their work on behalf of the entire class (including the Health Benefit Payers) cut down.

Having awarded the TPP class counsel 22% of that fund, the district court awarded only 15% of the separate, and larger, fund to consumer class counsel. The district judge derived the 15% figure from the average bid by plaintiffs' law firms in the handful of securities class actions in which other courts have held auctions to choose lead counsel. The judge thought that the 22% contingent fee actually negotiated in this very case was inferior to these auctions as a benchmark for two principal reasons: first, the judge wrote that the market in legal services is not competitive; second, the judge opined that class counsel had handled this suit inefficiently. They obtained an excellent result for the class, the judge thought, but took too long and spun too many wheels in the process.

The upshot is that the TPP class lawyers recover at a higher rate (and almost as much in absolute dollars), even though the consumer class counsel bore the principal risk of outright loss. As our first opinion explained, by the time the TPPs appeared as parties, the consumer class counsel had done the entrepreneurial work and the defendants had agreed to a substantial settlement. Insurers and other intermediaries paid their lawyers to secure a portion of that settlement for themselves (and to enlarge the total pot if possible). The risk in that venture was limited, for normal rules of subrogation entitled the insurers to compensation for their outlays. Consumer class counsel, by contrast, took the risk that they would come away with nothing — and, as our initial opinion observed, that was a significant risk, for the consumer class did not have an easy road. If the suit had a 50% chance of ending in defendants' favor, then an award equal to 15% of the pot (given that the class has prevailed) is equivalent to an ex ante offer of 7.5% of the amount that will be recovered if the class should prevail. Yet we know that the sophisticated TPPs agreed to pay their lawyers 22% of whatever could be diverted from an offer that was already on the table. It is most unlikely that the market rate for law firms negotiating to represent the consumer class ex ante (the right time, for reasons our initial opinion explained) would have been 15% of any eventual winnings, when we know that even after a good deal of the risk had been dissipated the TPPs had to offer 22% to sign up lawyers on contingent fee.

Average rates differ from marginal rates, so it is possible for one set of lawyers to receive extra compensation for risk even though its average rate is lower. What is required is that the marginal rate be higher throughout but that the lawyers for the riskier class generate a higher recovery. Suppose that the TPPs had agreed to pay their lawyers 25% of the first $40 million and 10% of everything over that, averaging out to 23% of the actual $46 million settlement. It is easy to see how consumer class counsel could receive a greater marginal award on every increment and still take home a lower average. For example, the (hypothetical) ex ante contract with consumer class counsel might grant 35% of the first $20 million, 25% of the next $20 million, and 10% of the residue. Class counsel would receive more than TPP counsel on the initial $40 million, to compensate for the greater risk of loss, and then the same marginal rate on higher increments. The hypothetical schedule we have given would work out to an average fee of 19.1% on the actual $88 million recovery, and the average fee would fall to 15.3% if the recovery were $150 million. These averages would lie below the average for the TPP counsel — but the important thing for purposes of compensating risk-bearing is that, for each band of recovery, the consumer class counsel would recover at least as much as TPP counsel, and for the initial bands of recovery consumer counsel would recover more.

Unfortunately, however, the district judge did not adopt such a structure. Only for the lowest two bands does the compensation of consumer class counsel exceed that of TPP class counsel: the first $10 million (30% for consumer counsel, 22% for TPP counsel) and the next $10 million (25% for consumer counsel, 22% for TPP counsel). By the third $10 million, consumer class counsel is down to 20%, and for everything over $40 million consumer class counsel receives 10% while TPP class counsel receives 22%. That fails to compensate risk-bearing activities, though voluntary market transactions would be certain to provide such compensation.

The district court's reasons for thinking 22% too high as a benchmark are not persuasive — or at least not supported on this record, which does not contain any evidence that the market in legal services is uncompetitive or that the TPPs are victims of a cartel. No law firm supplies more than a tiny fraction of the nation's legal services (even of the specialized submarket in big-stakes litigation). The Herfindahl-Hirschmann Index in this market is minuscule, and no evidence in this record implies that law firms have conspired to reduce competition. It is true, as the district court emphasized, that the bids in auctions for the right to represent classes in some securities cases fall below 22% of the recovery. Yet if securities suits present less risk to the plaintiff class than does a fraud suit against a drug manufacturer, it is unsound to use a contingent fee appropriate to the former as the measure in the latter. The record does not show how risky securities suits are (something that may have been affected by the Securities Litigation Reform Act of 1995 and any ensuing changes to suit selection by the securities — plaintiffs' bar) and therefore does not permit a reliable comparison between fees in securities cases and the fees in a suit such as this one.

There is, moreover, considerable question just what is being auctioned in bidding to represent a class. Normally an auction specifies the precise product to be sold (a particular painting, a share of stock in a named corporation, or 5,000 cubic yards of concrete having defined attributes). For legal services, however, it is hard if not impossible to hold the quality dimension constant. Contingent-fee arrangements are used when it is difficult to monitor counsel closely; otherwise some different arrangement, such as hourly rates, is superior. See Kirchoff v. Flynn, 786 F.2d 320, 324 (7th Cir. 1986); A. Mitchell Polinsky Daniel L. Rubinfeld, Aligning the Interests of Lawyers and Clients, 5 Am. L. Econ. Rev. 165 (2003). When it is hard to monitor counsel's effort and other elements of quality, it is also hard to know what the bid represents. Maybe it shows that less work will be invested, and that less compensation then is required. See Jill E. Fisch, Lawyers on the Auction Block: Evaluating the Selection of Class Counsel by Auction, 102 Colum. L.Rev. 650 (2002); Lucian Arye Bebchuk, The Questionable Case for Using Auctions to Select Lead Counsel, 80 Wash. U.L.Q. 889 (2002). Lawyers will earn a competitive return even at the lower level of compensation, but the class may be worse off. Large and sophisticated purchasers of legal services, such as Exxon/Mobile and General Motors, do not acquire legal services at auction; even clients able to monitor lawyers closely may be worried about the effect of the auction process on quality. So it is not possible to say, without other evidence of a kind missing in this record, that the outcome of auctions for the right to represent other classes in other litigation shows that the 22% contingent fee agreed to in arms'-length transactions between well informed parties in this case is "too high."

As for the possibility that consumer class counsel litigated inefficiently: using too many hours to achieve a given result is a problem with hourly billing, but the district court did not employ the lodestar method of compensation. If consumer class counsel invested too many hours, dallied when preparing the settlement, or otherwise ran the meter, the loss falls on counsel themselves. One advantage of the contingent fee is that the client (or the judge as protector of the class's interests) need not monitor how many hours the lawyers prudently devoted to the case. The client cares about the outcome alone. Here, the district judge found (and we agree), the outcome is excellent from the class's perspective. Inefficient conduct of the litigation therefore does not afford any reason to reduce class counsel's percentage of the fund that their work produced.

Instead of remanding for still a third calculation, we think it best to set the fees ourselves, as we have done in other class actions that have necessitated multiple appeals, so that the class members may at last receive their awards (something that is not possible until the attorneys' stakes have been determined). See, e.g., Florin v. Nationsbank of Georgia, N.A., 60 F.3d 1245, 1248 (7th Cir. 1995); In re Continental Illinois Securities Litigation, 985 F.2d 867, 869 (7th Cir. 1993). See also Divane v. Krull Electric Co., 319 F.3d 307, 318 n. 2 (7th Cir. 2003) (collecting authority). We stick as close as possible to the district court's approach and thus give consumer class counsel 30% of the first $10 million and 25% of the next $10 million. Because consumer class counsel bore at least as much risk as TPP class counsel for the band from $20 million to $46 million, consumer class counsel is entitled to 22% of that portion of the recovery. And we think that 15% of all amounts over that is a decent estimate of the fee that would have been established in ex ante arms'-length negotiations. Because the consumer class recovered a total of $88 million, the fee comes to $17.52 million, or 19.9% of the fund. (The award in absolute dollars must be adjusted to reflect the interest that the fund has been accumulating; the parties should be able to agree on this mechanical calculation.) Because we have used the TPP award as the benchmark, and the TPP class counsel recovered costs and expenses on top of their 22% (as their contracts provided), consumer class counsel also are entitled to a separate award on this score. We hope that this sum can be liquidated quickly in the district court so that the fund may be distributed promptly to the consumers.

The decision of the district court with respect to the TPP class is affirmed on Appeal No. 02-2519. The decision of the district court with respect to the consumer class is vacated on Appeal No. 02-2398, and the case is remanded for entry of the fee award we have described and for further proceedings consistent with this opinion.