This is the second time we visit the fairness of a proposed class settlement stemming from Fleet Mortgage Corporation's ("Fleet") alleged violations of various state consumer protection laws, as well as various other federal and state laws. We rejected the first proposed settlement because, among other reasons, the district court failed to consider adequately the value of the claims of the so-called "information-sharing class" (a class of consumers whose privacy interests were purportedly intruded upon, but who did not suffer any out-of-pocket damages). See generally Mirfasihi v. Fleet Mortgage Corp., 356 F.3d 781 (7th Cir.2004). Following remand, the district court did not adequately consider the potential value of the information-sharing class's claims. We therefore reverse and remand the case for further consideration of this limited issue.

This lawsuit was brought on behalf of approximately 1.6 million people whose home mortgages were owned by Fleet. Plaintiff's core allegations are that Fleet sold mortgage information (including loan amounts, the type of loan, and repayment histories) to third-party telemarketing companies for the purpose of selling certain financial products to the class (which they purportedly did not want). The complaint alleges (and Fleet does not deny) that Fleet was an active collaborator in drafting the script that the telemarketers used and allowed direct billing of the fees for the telemarketers' products onto the mortgage bill of its customers, without obtaining pre-approval from customers. Plaintiff alleges, among other things, that Fleet violated the Truth-in-Lending Act and the Fair Credit Reporting Act, along with various state consumer protection laws.

In 2004, the parties entered into the initial settlement agreement (the "First Settlement"), which provided for two plaintiff classes: a "telemarketing class" and an "information-sharing class," comprised of approximately 190,000 and 1.4 million members, respectively. The telemarketing class consisted of Fleet customers (outside of certain customers in Minnesota and California, who were the subject of other pending lawsuits, since settled) who purchased, and were later billed on their mortgage account statements, for some of the financial products pitched to them by telemarketing firms. The information-sharing class consisted of all Fleet customers whose mortgage information was transmitted to telemarketers in the prior six years, but who did not purchase any services from the telemarketers.

As detailed more fully in our prior opinion, see Mirfasihi, 356 F.3d at 783-84, the First Settlement provided that Fleet would set aside $2.4 million to pay the telemarketing class's claims, in addition to $243,000 in disgorgement of profits to this class, which would have resulted in payments ranging from $10 to $135 to each of the telemarketing class claimants. The First Settlement provided $750,000 in fees for the class lawyers. The information-sharing class, however, was left out in the cold and received nothing.

After the district court approved the settlement, Intervenors Michael Green and Angela Perry appealed, arguing that the settlement was unfair and unreasonable on a number of grounds. See Mirfasihi, 356 F.3d at 783-85. And we agreed. Id. The First Settlement contained a number of troubling "warning signs," including: (1) the payment of disgorged profits to the telemarketing class (despite the fact that the profits appeared to come from members of the information-sharing class); (2) the reversion of unclaimed funds to Fleet; (3) a "handsome fee for class lawyers," despite the meagerness of the relief provided in the settlement agreement to class members; and (4) the fact that the information-sharing class did not have separate counsel from the telemarketing class. See id. The reversion feature was particularly favorable to Fleet because, based upon the number of telemarketing claimants who filed claims, Fleet stood to recoup approximately half of the total $2.4 million payout. Furthermore, the lack of separate counsel for the two classes (or, more accurately, the information-sharing class and the telemarketing subclass), coupled with the lack of any recovery for the information-sharing class members, created the impression that class counsel had not properly considered the interests of these 1.4 million claimants. Id. at 785. But our core concern—and one that, as will be apparent below, continues to trouble us—is that the district court failed to consider with adequate specificity the reasonableness of an entire class receiving a "big fat zero" in the settlement. Id. at 785. Specifically, the district court did not canvass all potential avenues of recovery to determine whether the information-sharing class's claims were indeed essentially hopeless (and thus worthless) under the pertinent controlling law.

Following this court's disapproval of the First Settlement, the parties engaged in additional settlement talks, as well as mediation, and ultimately reached a "new" settlement agreement (the "Second Settlement"). Under the terms of the Second Settlement, the information-sharing class will not receive any direct payments. Fleet agrees to send the $243,000 in disgorged funds, as well as any excess funds not claimed by the telemarketing class claimants, to unaffiliated third-party public interest groups and/or charitable institutions devoted to consumer privacy concerns. Fleet's maximum payment under the various provisions is $2.4 million. The class lawyers continue to receive $750,000 in fees.

After the case was reassigned to another district court judge, the parties presented the Second Settlement for approval. The district court approved the settlement, holding that valuing the information-sharing class's claims at zero was fair and reasonable because, among other reasons, they suffered no actual damages, and, as a result, probably have no recovery under various states' consumer protection statutes (and common law causes of action), and moreover, the claimants faced significant hurdles in obtaining class certification. The district court also approved of the valuation of the telemarketing claims and the attorneys' fees, and generally found that the Second Settlement properly addressed the "warning signs" that we had discussed in our prior opinion.

It is well-settled that district judges presiding over proposed class settlements are "expected to give careful scrutiny to the terms of proposed settlements in order to make sure that class counsel are behaving as honest fiduciaries for the class as a whole" because "class actions are rife with potential conflicts of interest between class counsel and class members." Mirfasihi, 356 F.3d at 785 (collecting and citing cases). District judges must therefore "exercise the highest degree of vigilance in scrutinizing proposed settlements of class actions" to consider whether the settlement is "fair, adequate, and reasonable, and not a product of collusion." Reynolds v. Beneficial Nat'l Bank, 288 F.3d 277, 279 (7th Cir.2002). Indeed, the district court judge functions as "a fiduciary of the class, who is subject therefore to the high duty of care that the law requires of fiduciaries." Id. at 280. As a general principle, a district court should evaluate, among other things, the probability of plaintiff prevailing on its various claims, the expected costs of future litigation, and hints of collusion. See, e.g., Mars Steel Corp. v. Cont'l Illinois Nat'l Bank & Trust Co., 834 F.2d 677, 681-82 (7th Cir.1987); Reynolds, 288 F.3d at 283-85. Our review of the district court's scrutiny of a proposed class settlement is more limited: we review the district court's decision for abuse of discretion. Isby v. Bayh, 75 F.3d 1191, 1196-97 (7th Cir. 1996).

As an initial observation, there is not much that is substantially "new" in the Second Settlement. For instance, Fleet's total out-of-pocket expenses remain capped at $2.4 million, the telemarketing class members remain eligible for the same range of payments, plaintiff's counsel retains its $750,000 fee, and the information-sharing class continues to receive nothing. There are, however, some minimal improvements in the Second Settlement. For example, in negotiating the Second Settlement, the information-sharing class received separate counsel from the telemarketing class. But the reconfiguration of the representation was counter-intuitive: it was the telemarketing class-not the information-sharing class—that received brand new counsel, Clifford Cantor, while the attorneys who had negotiated the unimpressive First Settlement slid over to the information-sharing class. This is an odd result, given our concerns over the possibility that the attorneys who negotiated the First Settlement had "sold [the information-sharing class] down the river." Mirfasihi, 356 F.3d at 785. It seems clear to us that it was the information-sharing, not the telemarketing, class that needed a fresh look at the settlement value of their claims. Wouldn't the proposition that their claims were worthless be bolstered if independent counsel, unclouded by prior positions and assumptions, concurred with the assessment reached in the First Settlement? Whatever the reason for this particular reconfiguration of representation, the information-sharing class appears to have received no tangible benefit.

Another minor improvement in the Second Settlement is that it eliminates the broad reversion allowance in the First Settlement, which allowed Fleet to recoup directly any unclaimed funds from the telemarketing class. The unclaimed "balance" (the telemarketing class only filed $276,000 in claims out of an approximately $2 million-plus pot) is now targeted for the Electronic Privacy Information Center ("EPIC") or other charitable institution devoted to consumer privacy concerns. But the devil is in the details, and EPIC likely should not be eagerly awaiting a large-sum check because before any unclaimed balance can be distributed to third-party charities, Fleet is allowed to subtract (1) the amounts it paid to settle a similar, but separate, lawsuit in California (the "Koluncich" suit) and (2) "any amounts paid to satisfy the claims of any class members, including class counsels' attorneys' fees, any class notice, and costs, whether in this Litigation or otherwise." (Settlement Agreement at par. 9). This provision allows Fleet to deduct approximately $1 million ($750,000 in attorneys' fees and $270,000 class notice costs) incurred in the present litigation, plus another $300,000 in costs associated with the settlement of the separate Koluncich litigation in California, as well as an unspecified (and undefined) amount of other "costs" incurred. In reviewing this portion of the settlement agreement, the district court commented that "[t]here will be no reversion of any portion of the $2.4 million" to Fleet. Although this may be technically correct, there is a negligible practical distinction between a reversionary provision that sends cash directly back to Fleet and line-item deductions right off the $2.4 million pot that allow Fleet to pay its other outstanding bills, including ones unrelated to the present litigation.

But these wrinkles, standing alone, are not enough for us to find that the district court abused its discretion in approving the settlement. Our concern remains the proper valuation of the information-sharing class's claims, which is the true sticking point in the district court's otherwise thoughtful and well-reasoned analysis of the objectors' other protests.1 In our prior opinion, "we emphasized the district judge's duty in a class action settlement situation to estimate the litigation value of the claims of the class and determine whether the settlement is a reasonable approximation of that value." Mirfasihi, 356 F.3d at 786 (citing Reynolds, 288 F.3d at 284-85 (other citations omitted)). Here, in evaluating the strength of the information-sharing class's claims, the district court reviewed the consumer protection statutes in Illinois, Maryland, Massachusetts, and Wisconsin, which were the jurisdictions in which suits have been brought against Fleet. But it is not clear why the district court limited its analysis to just these four jurisdictions in light of the fact that the information-sharing class was not limited to residents from these four states. The analysis should have considered, at a minimum, the pertinent consumer protection statutes (and other potential bases for claims) in the states where the information-sharing class members resided. The likely reason for this truncated analysis was that the parties did not draw the district court's attention to a compiled fifty-state survey of consumer protection statutes, which was one in a large series of exhibits to the parties' joint motion for approval of the Second Settlement. Not surprisingly, the district court's opinion makes no mention of the reliability of this fifty-state survey, nor does it apply the survey to its claims evaluation.

The district court's analysis of the information-sharing class's claims primarily focused on Massachusetts's consumer protection statute (Chapter 93A), which is where the objectors place their greatest faith in potential recovery. Setting aside the fact that a proper estimate of the claims here should include a broader canvass of the potentially applicable law, the district court's analysis of Chapter 93A's applicability is nonetheless curious in that it seems to suggest that Chapter 93A provides a potential toehold for recovery. Specifically, the district court noted that in Leardi v. Brown, 394 Mass. 151, 474 N.E.2d 1094 (Mass.1985), the Supreme Judicial Court of Massachusetts held that Chapter 93A "create[s] a legal right, the invasion of which, without more, constitutes injury," and "under circumstances where there has been an invasion of a legally protected interest, but no harm for which actual damages can be awarded ... the statute provides for the recovery of minimum [statutory] damages" in the amount of twenty-five dollars. Id. at 1102. On its face, this language suggests that the objectors' position that the information-sharing class members may be entitled to statutory damages of $25 per claimant (which places the maximum potential value of their claims at approximately $35 million) is not frivolous—and thus should have some settlement value. The district court summarily dismissed the significance of Leardi, concluding, without analysis, that it was likely that there was no basis for applying Massachusetts law to the information-sharing class's claims.2

Although the record is not clear on this issue, it appears that the information-sharing class in this action was a nationwide class that did not necessarily exclude Massachusetts residents. In addition, Fleet's principal place of business is located in Massachusetts and, although the Complaint is not (and need not be) crystal clear on this, Fleet may process many of the information-sharing class's mortgages in Massachusetts. See First Am. Cmpl. at ¶ 9. As a result, the objectors' argument that Massachusetts law could apply to at least a subset of the claims does not appear so outlandish as to deprive their claims of all settlement value on that basis alone, and the district court should revisit this issue with fresh eyes. Moreover, the choice-of-law issues in nationwide class actions are rarely so uncomplicated that one can delineate clear winning and losing arguments at an early stage in the litigation. See In re Rhone-Poulenc Rorer Inc., 51 F.3d 1293, 1300-03 (7th Cir.1995) (recognizing difficulty of class certification created by differences in the laws of 50 states); see also Jeremy T. Grabill, Comment, Multistate Class Actions Properly Frustrated by Choice-of-Law Complexities: The Role of Parallel Litigation in the Courts, 80 TUL. L. REV. 299 (2005); Ryan Patrick Phair, Comment, Resolving the "Choice-of-Law Problem" in Rule 23(b)(3) Nationwide Class Actions, 67 U. CHI. L. REV. 835 (2000). And it would seem that the legal uncertainty resulting from the complicated choice-of-law issues in this case should not cut solely against the value of plaintiff's claims here, but should also be a factor in considering the value of Fleet's defenses.

In any event, in basic terms, a claim analysis under these circumstances would require consideration of (1) the probability of the information-sharing class having grounds of recovery under any applicable law; (2) the probability of such favorable law applying to the entire information-sharing class (rather than differing subsets); and (3) the probability of winning on the merits. We cited cases describing this methodology in the our prior opinion, and do so again. See Mirfasihi, 356 F.3d at 786, citing Mars Steel Corp., 834 F.2d at 682; In re General Motors Corp. Engine Interchange Litigation, 594 F.2d 1106, 1132 n. 44 (7th Cir.1979); In re General Motors Pick-Up Truck Fuel Tank Products Liability Litigation, 55 F.3d 768, 806 (3d Cir.1995); In re Traffic Executive Association-Eastern Railroads, 627 F.2d 631, 633 (2d Cir.1980); see also Reynolds, 288 F.3d at 284-85. The calculations in this case are simplified by the fact that the upper value of the information-sharing claims are likely capped by the maximum of any applicable statutory damages, and therefore quantifying the highest potential damages for the information-sharing class is not subject to the wider range of imprecision that accompanies the prediction of jury verdicts on compensatory or punitive damages. Moreover, the district court may use its discretion in requesting additional briefing or even an evidentiary hearing (if necessary) to obtain greater information and analyses pertinent to claims evaluation in this case. It is understood that mathematical certainty and precision is not possible in valuing claims, particularly at this early stage in the litigation, but more is needed here to ensure that the information-sharing class's interests are properly considered. See Reynolds, 288 F.3d at 285. Thus, on remand, the district court should consider and analyze the full cross-section of potentially applicable state law and arrive at a clearer estimate of the potential value of the information-sharing class's claims to allow proper evaluation of the reasonableness of the proposed settlement.

For the foregoing reasons, the judgment of the district court is REVERSED and we REMAND for further proceedings consistent with this opinion.