concurring in part and dissenting in part:
I largely agree with the majority’s ultimate disposition of most of this case, but this agreement in result cannot paper over my significant disagreement with the majority’s interpretation of the Privacy Act. Most fundamentally, I respectfully dissent from the majority’s conclusion in part II of its opinion that only a plaintiff who can prove actual damages is entitled to recover statutory damages under 5 U.S.C. § 552a(g)(4). Even if the statute must be narrowly construed as a waiver of sovereign immunity, it will not bear the interpretation the majority adopts today. I would hold that a plaintiff can recover statutory damages of $1,000 upon proof that he has suffered an adverse effect as a result of an intentional or willful violation of the Privacy Act. § 552a(g)(l)(D), (g)(4). Because Buck Doe has offered sufficient evidence that he was adversely affected by the Department of Labor’s wrongful dissemination of his social security number, I would remand his case for trial. I would affirm the district court’s grant of summary judgment to the Secretary against the other six Doe plaintiffs on the alternative ground that those plaintiffs failed to present any evidence of adverse effects resulting from the Department’s violation of the Privacy Act. I join the majority’s treatment of the class certification issue in part V of its opinion except to the extent that the majority’s discussion depends on its view that proof of actual damages is a prerequisite for the recovery of statutory damages.1 I join parts IV and VI of the majority opinion in full.
I.
The principal issue in this case is whether proof .of actual damages is a prerequisite to the recovery of statutory damages under the Privacy Act. Although I dissent from the majority’s holding on that issue, I begin my analysis by briefly covering the ground the majority and I share. Specifically, I explain why I agree with the majority that the district court properly entered summary judgment against all the plaintiffs except Buck Doe because none of those plaintiffs satisfied the Privacy Act’s adverse effect requirement.
Subsections (A) through (C) of the Privacy Act’s civil remedies provision, 5 U.S.C. § 552a(g)(l), authorize suit for certain specified violations of the Act. Subsection (D) is a catchall provision that authorizes suits against any agency that “fails to comply 'with any other provision of this section-, or any rule promulgated thereun*186der, in such a way as to have an adverse effect on an individual.” 5 U.S.C. § 552a(g)(l)(D). Because the Does’ Privacy Act claims fall under subsection (D), each of them must prove that he suffered an adverse effect as a result of the Department of Labor’s wrongful disclosure of his social security number in order to have standing to sue.
A plaintiff can satisfy the adverse effect requirement in § 552a(g)(l)(D) even if he has sustained no actual damages, for the meaning of “adverse effect” in subsection (g)(1)(D) is both distinct from and broader than the meaning of “actual damages” in subsection (g)(4)(A). See Quinn v. Stone, 978 F.2d 126, 135 n. 15 (3d Cir.1992).2 The district court, after adopting an extremely loose reading of the adverse effect requirement, held that all of the Does satisfied the requirement. In essence, the court' read the adverse effect language in § 552a(g)(l)(D) as requiring a plaintiff to demonstrate only that the government intentionally and willfully mishandled the plaintiffs own personal information (as opposed to that of some third party). Most courts, however, closely scrutinize a plaintiffs claim for evidence of adverse effects in a way that conflicts with the district court’s permissive approach to the requirement. See, e.g., Quinn, 978 F.2d at 135—37; Albright v. United States, 732 F.2d 181, 186-88 (D.C.Cir.1984). As the Third Circuit has explained, the adverse effect requirement actually contains two prongs: a standing requirement (the plaintiff must have suffered a cognizable injury) and a requirement that the adverse effect be caused by the agency’s violation of the Act. See Quinn, 978 F.2d at 135. The causal prong makes it especially clear that an adverse effect must be something distinct from the intentional and willful violation itself. For if a violation of the Privacy Act was sufficient to constitute an adverse effect, there could be no question of whether the violation caused the adverse effect, and hence the causal prong would be superfluous. As a result, the district court set the bar of the adverse effect requirement too low by not insisting that the Does offer evidence of adverse effects separate from, and caused by, the Department of Labor’s willful and intentional violation of their Privacy Act rights.
That said, an adverse effect can take many different forms depending on which provisions of the Privacy Act have been violated. Financial harm is the most tangible adverse effect that can result from-an agency’s wrongful disclosure of a social security number. We have observed that an “unscrupulous individual” armed with a person’s social security number “could obtain a person’s welfare benefits or Social Security benefits, order new checks at a new address on that person’s checking account, obtain credit cards, or even obtain the person’s paycheck.” Greidinger v. Davis, 988 F.2d 1344, 1353 (4th Cir.1993). Obviously, a person suffers an adverse effect when his social security number is used by another to commit some crime of identity fraud, but a plaintiff can also be adversely affected before any crime occurs by distress about possible misuse of his illegally disclosed social security number. Most courts have held that mental distress or emotional harm of this kind is sufficient to constitute an adverse effect. See, e.g., Quinn, 978 F.2d at 135-36 (holding that stress and emotional anguish are sufficient *187to establish adverse effect); Albright, 732 F.2d at 186 (stating that “emotional trauma alone is sufficient to qualify as an ‘adverse effect’ ”); Parks v. IRS, 618 F.2d 677, 682-83 (10th Cir.1980) (stating that psychological harm can constitute an adverse effect). It follows that, at minimum, Buck Doe proffered evidence that he had suffered adverse effects sufficient to survive summary judgment when he declared in his affidavit that “no amount of money could compensate [him] for worry and fear of not knowing when someone would use [his] name and social security number to ... use [his] checking account or even get credit cards.” If found credible by the trier of fact, Buck Doe’s testimony that he worried about the possible misuse of his social security would satisfy the adverse effect requirement.
The harder question is what to do with the remaining Doe plaintiffs. Their social security numbers, like Buck Doe’s, were wrongfully disclosed to strangers. Given the many troubling consequences that could flow from this disclosure, persons in their position could be expected to experience at least some degree of distress at the prospect that their social security numbers might be misused. Yet, aside from Buck Doe, none of the plaintiffs offered any evidence of such distress. The question, then, is whether we can simply presume that they have suffered emotional harms sufficient to qualify as adverse effects under § 552a(g)(l)(D). The question is not likely to arise with any frequency because serious plaintiffs will not ordinarily fail to offer evidence that they have been harmed by agency violations of their privacy. Still, I believe that adverse effects must be proven rather than merely presumed, and the plaintiffs other than Buck Doe have failed to present evidence of adverse effects sufficient to survive summary judgment. Accordingly, I would affirm the district court's grant of summary judgment to the Secretary against the remaining Doe plaintiffs on the alternative ground that those plaintiffs failed to present any evidence that they suffered adverse effects from the agency’s wrongful disclosure of their social security numbers. I now turn to the more difficult task of discussing the fate of Buck Doe’s Privacy Act claim.
II.
Recovery in subsection (D) Privacy Act suits is governed by § 552a(g)(4), which provides:
In any suit brought under the provisions of subsection (g)(1)(C) or (D) of this section in which the court determines that the agency acted in a manner which was intentional or willful, the United States shall be liable to the individual in an amount equal to the sum of-
(A) actual damages sustained by the individual as a result of the refusal or failure, but in no case shall a person entitled to recovery receive less than the sum of $1,000; and
(B) the costs of the action together with reasonable attorney fees as determined by the court.
5 U.S.C. § 552a(g)(4).
The majority and I agree that to recover statutory damages under the Privacy Act, an individual must prove that he has suffered an adverse effect as a result of an agency’s intentional or willful violation of the Act. We also agree that emotional distress can qualify as an adverse effect and that Buck Doe’s statements about his emotional distress are sufficient to allow a finding that he suffered an adverse effect as a result of the Department of Labor’s wrongful disclosure of his social security number. Finally, we agree that proof sufficient to establish an adverse effect may be insufficient to establish actual damages. *188We disagree, however, on the question of whether proof of actual damages is a prerequisite to the recovery of statutory damages.
The majority holds that, purely as a matter of textual analysis, statutory damages under the Privacy Act may be recovered only by a plaintiff who can prove that he has sustained actual damages. It refrains from deciding whether the term “actual damages” can include emotional distress or is limited to out-of-pocket losses. Instead, the majority assumes for the sake of argument that proof of emotional distress could entitle a plaintiff to actual damages, but holds that Buck Doe’s conclusory claims of emotional distress are insufficient. Though the question is somewhat close, the majority’s textual arguments do not convince me that a “person entitled to recovery” in subsection (g)(4)(A) must be someone who has sustained actual damages. I believe the phrase “person entitled to recovery” is more naturally read to mean anyone to whom “the United States shall be liable ... in an amount equal to the sum of’ the amounts specified in sub-paragraphs (A) and (B). Such a person is entitled to recovery because he has proven that he suffered an adverse effect as a result of an agency’s intentional or willful violation of the Privacy act. I think it quite relevant, though the majority apparently does not, that the language in § 552a(g)(4) and similar language in other statutes has been widely read to mean exactly what I believe it means. My biggest problem with the majority’s textual analysis, however, is the majority’s assumption that we can decide whether proof of actual daipages is a prerequisite to the recovery of statutory damages without any attention either to the question of what the term “actual damages” means or to the policies that led Congress to create a statutory damages provision as part of the Privacy Act in the first place. In my view, expanding the scope of inquiry to include these considerations makes it clear that the statute should not be read to condition the recovery of statutory damages on proof of actual damages. Nevertheless, I will first attempt to meet the majority’s analysis on its own terms by assmning with the majority that we can properly decide whether proof of actual damages is a prerequisite to the recovery of statutory damages solely by analyzing the meaning and placement of the phrase “person entitled to recovery” in subsection (g)(4)(A).
A.
1.
Section 552a(g)(4)(A) unambiguously provides that a “person entitled to recovery” shall receive at least statutory damages of $1,000. The critical textual question, of course, is who counts as a “person entitled to recovery.” The majority argues that a person entitled to recovery must be someone who has sustained actual damages. In my opinion, the majority relies too heavily on where the phrase is placed in subparagraph (A) while giving insufficient emphasis to the ordinary meaning of the word “recovery.” As a matter of ordinary language usage, an entitlement to “recovery” is surely broader than an entitlement to actual damages. Thus, the statute is more naturally read to say that anyone to whom the United States “shall be liable” in the amounts specified in subparagraphs (A) and (B) is a “person entitled to recovery.” The majority’s cramped reading of “recovery” would make more sense if actual damages were the only remedy provided by the statute, but they are not. Instead, the text of the statute allows for a plaintiff who has proven an intentional or willful violation of the Privacy Act to recover costs and reasonable attorney fees even if the plaintiff has *189suffered no actual damages at all.3 On the majority’s reading, such a person is not “entitled to recovery” even though the United States is liable to that person for costs and attorney fees. A reading that so strongly conflicts with the ordinary meaning of “recovery” ought to be rejected.4
Three additional reasons support my reading of the statute. First, the majority of courts to interpret § 552a(g)(4) have allowed recovery of statutory damages without proof of actual damages. In fact, disagreement about this issue arose only when courts began to interpret the statute in the light of the principle that waivers of the federal government’s sovereign immunity must be narrowly construed. Second, my interpretation is consistent with that adopted by the Office of Management and Budget (OMB), the agency charged with implementing the Privacy Act. Third, statutory damages provisions that are similar in structure, language, and purpose to § 552a(g)(4) have been uniformly understood to allow recovery of statutory damages without proof of actual damages. Indeed, the legislative history of these provisions clearly evinces Congress’s intent to allow recovery of minimum statutory damages without proof of actual damages. I will expand on each of these points.
First, most circuit courts have read the Privacy Act to allow recovery of statutory damages without proof of actual damages. See Wilborn v. HHS, 49 F.3d 597, 608 (9th Cir.1995) (stating that statutory damages may be awarded to plaintiffs with no provable damages); Quinn, 978 F.2d at 131 (listing adverse effect, but not actual damages, as an element that must be proved in order to recover damages under the Privacy Act for wrongful disclosure of personal records); Waters v. Thornburgh, 888 F.2d 870, 872 (D.C.Cir.1989) (stating that if a plaintiff can establish that he was adversely affected by an intentional or willful violation of the Act, he “is entitled to the greater of $1,000 or the actual damages sustained”); Johnson v. Dep’t of Treasury, 700 F.2d 971, 972 & n. 12 (5th Cir.1983) (stating that the statutory minimum of $1,000 is “of course” recoverable even without proof of actual damages); Fitzpatrick v. IRS, 665 F.2d 327, 330-31 (11th Cir.1982) (stating that plaintiffs who failed to prove actual damages could recover statutory damages); Parks, 618 F.2d at 682-83 (holding that plaintiffs who sought minimum statutory damages had stated a claim for relief when they alleged facts showing an intentional or willful violation of the Act that caused psychological harm constituting an adverse effect); but see Hudson v. Reno, 130 F.3d 1193, 1207 & n. 11 (6th Cir.1997) (holding that proof of *190actual damages is a prerequisite to recovery of statutory damages on the ground that waivers of the federal government’s sovereign immunity must be narrowly construed in favor of the government). Until today, we had assumed, albeit in dicta, that the prevailing view was correct. See Thurston v. United States, 810 F.2d 438, 447 (4th Cir.1987).
The majority appears to regard the decisions favoring my reading of the Privacy Act as irrelevant because none undertakes the sort of textual analysis favored by the majority. That point is true enough, but it is worth observing that sometimes courts fail to engage in textual analysis simply because they think a point is too obvious to require discussion. Although I acknowledge that judicial head counting is not an infallible method of determining the meaning of a statute, the prevalence of my reading in the circuits is at least some evidence against the plausibility of the majority’s reading. Indeed, I think we ought to be troubled that, with one possible exception, the only circuit ever to rule that a plaintiff must prove actual damages in order to recover statutory damages based its conclusion not on its reading of the text, but on the principle that waivers of sovereign immunity must be narrowly construed in favor of the government. See Hudson, 130 F.3d at 1207 & n. II.5
Second, my reading is consistent with the guidelines for implementing the Privacy Act developed by OMB. Privacy Act Guidelines, 40 Fed.Reg. 28949 (July 9, 1975). The OMB guidelines explain that when a court
finds that an agency has acted willfully or intentionally in violation of the Act in such a manner as to have an adverse effect upon the individual, the United States will be required to pay
Actual damages or $1,000, whichever is greater
Court costs and attorney fees.
Id. at 28970. These guidelines “are due the deference accorded to the interpretation of an agency charged with ‘oversight’ of implementation.” Quinn, 978 F.2d at 133. See also Privacy Act § 6, 88 Stat. 1896, 1909 (1974) (authorizing the OMB guidelines); Albright, 631 F.2d at 919-20 n. 5 (explaining that the OMB guidelines are particularly worthy of deference because they “ ‘involve[ ] a contemporaneous construction of a statute by the [persons] charged with the responsibility of setting its machinery in motion, or making the parts work efficiently and smoothly while they are yet untried and new’ ” (quoting Zenith Radio Corp. v. United States, 437 U.S. 443, 450, 98 S.Ct. 2441, 57 L.Ed.2d 337 (1978))).
Third, my interpretation is supported by both the legislative history and the judicial interpretation of statutory damages provisions similar in language, structure, and purpose to § 552a(g)(4). In 1976, only two years after passage of the Privacy Act, Congress created a civil remedy for unauthorized disclosure of tax returns and tax return information. As originally enacted, the statute read:
In any suit brought under the provisions of subsection (a), upon a finding of liability on the part of the defendant, the *191defendant shall be liable to the plaintiff in an amount equal to the sum of—
(1) actual damages sustained by the plaintiff as a result of the unauthorized disclosure of the return or return information and, in the case of a willful disclosure or a disclosure which is the result of gross negligence, punitive damages, but in no case shall a plaintiff entitled to recovery receive less than the sum of $1,000 with respect to each instance of such unauthorized disclosure; and
(2) the costs of the action.
26 U.S.C. § 7217(c) (Supp.1981) (emphasis added), repealed 1982. While somewhat more complex than § 552a(g)(4), this statute poses essentially the same question of interpretation presented by this case, namely, is a plaintiff “entitled to recovery” someone who has sustained actual damages or is he simply someone to whom the defendant “shall be liable” in a specified sum. Congress gave its answer in the statute’s legislative history, explaining its creation of a statutory damages remedy for unlawful disclosure of tax return information as follows:
Because of the difficulty in establishing in monetary terms the damages sustained by a taxpayer as the result of the invasion , of his privacy caused by an unlawful disclosure of his returns or return information, the amendment provides that these damages would, in no event, be less than liquidated damages of $1,000 for each disclosure.
S.Rep. No. 94-938, at 348 (1976), reprinted in 1976 U.S.C.C.A.N. 2897, 3778. Not surprisingly, courts uniformly interpreted § 7217(c) to allow recovery of statutory damages without requiring proof of actual damages. See, e.g., Johnson v. Sawyer, 120 F.3d 1307, 1313 (5th Cir.1997) (stating that under § 7217 “a plaintiff is entitled to his actual damages sustained as a result of an unauthorized disclosure ... or to liquidated damages of $1,000 per such disclosure”); Rorex v. Traynor, 771 F.2d 383, 387-88 (8th Cir.1985) (holding that plaintiffs were entitled to minimum statutory damages under § 7217(c) despite their failure to prove any actual damages). Section 7217 was repealed in 1982 and replaced by 26 U.S.C. § 7431(c), which has also been read to authorize the award of statutory damages without requiring the plaintiff to prove actual damages. See, e.g., Chandler v. United States, 687 F.Supp. 1515, 1521 (D.Utah 1988).6
*192Congress passed another provision with language similar to § 552a(g)(4)(A) as part of the Electronic Communications Privacy Act of 1986 (ECPA). Pub.L. No. 99-508, 100 Stat. 1848. The ECPA creates a civil damages remedy for persons aggrieved by knowing or intentional violations of the Act:
The court may assess as damages in a civil action under this section the sum of the actual damages suffered by the plaintiff and any profits made by the violator as a result of the violation, but in no case shall a person entitled to recover receive less than the sum of $1,000.
18 U.S.C. § 2707(c) (emphasis added). As with the statutory damages provisions in the tax code and the Privacy Act, it could be argued that the placement of the phrase “person entitled to recover” identifies such a person as one who has suffered actual damages. Yet Congress explained that appropriate remedies under this section include “damages ... including the sum of actual damages suffered by the plaintiff and any profits made by the violator as the result of the violation ... with minimum statutory damages of $1,000.” S.Rep. No. 99-541, at 43 (1986), reprinted in 1986 U.S.C.C.A.N. 3555, 3597. And, as with the tax statute and (with a few exceptions) the Privacy Act, this language has not led courts to condition the recovery of minimum statutory damages on proof of actual damages. See, e.g., United Laboratories, Inc. v. Rukin, 1999 WL 608712, at *5 (N.D.Ill.1999) (stating that under § 2707(c) a plaintiff is entitled to the greater of actual damages or $1,000 because of “the difficulty in proving actual damages caused by any particular [violation of the Act].”). Indeed, one district court in this circuit has drawn an analogy between the statutory damages provisions in the Privacy Act and the ECPA. See Dawson v. Leewood Nursing Home, Inc., 14 F.Supp.2d 828, 832 (E.D.Va.1998) (describing both acts as rewarding plaintiffs with minimum statutory damages when they achieve “symbolic victories” by proving that the defendant violated the statute.).
The statutory damages provisions in the Internal Revenue Code and the ECPA confirm that it is not unusual for Congress to use “entitled to recovery” language to identify those plaintiffs who are entitled to minimum statutory damages. Nor is it unusual for Congress to do so in the very same sentence authorizing recovery of actual damages. What is unusual is the majority’s decision to treat this language as conditioning recovery of statutory damages upon proof of actual damages.7
*193In sum, the case for my reading of the phrase “person entitled to recovery” is a simple one. First, the ordinary meaning of the word “recovery” suggests that any person to whom the United States shall be liable in specified amounts is a person entitled to recovery. Second, Congress and other courts have indicated that they understand the language of this statute and others like it to mean exactly what I believe it means.
2.
The majority offers four reasons for its contrary view that a person entitled to recovery must be one who has sustained actual damages. First, it argues that as a matter of statutory interpretation and grammar, only a person who has sustained actual damages can be a “person entitled to recovery” because the phrase appears in the same sentence and subparagraph that authorizes the recovery of actual damages. Second, the majority argues that the separation of the two clauses in subsection (g)(4)(A) by the conjunction “but” “reinforced the conclusion that the second clause (‘in no case shall a person entitled to recovery receive less than the sum of $1,000’) adds a qualification with respect to the class of plaintiffs defined by the first (those who have suffered actual damages) — namely, that in no event shall such a person recover less than $1,000.” Ante at 178. Third, the majority suggests that it would be odd for Congress to limit the liability of the United States to actual damages and then in the same sentence authorize recovery in the complete absence of actual damages. Finally, the majority suggests that Congress’s failure to clearly authorize the award of statutory damages without proof of actual damages when it could easily have done so underscores the correctness of the majority’s interpretation. I find none of these reasons convincing.
The majority’s first argument turns on the placement of the phrase “person entitled to recovery.” Although the majority is aware that the word “recovery” would normally include more than actual damages, it contends that the ordinary meaning of the word must give way because the statute implicitly redefines “recovery” to mean “actual damages” simply by placing the word “recovery” in the same sentence and subparagraph that authorizes an award of “actual damages sustained.”8 *194According to the majority, this placement establishes as a matter of statutory interpretation and grammar that a person entitled to recovery must be one who has sustained actual damages. The majority ought to find this claim puzzling. I would think it highly unusual for Congress to narrow the meaning of ordinary words simply through syntax. If Congress intended for “recovery” to mean “actual damages,” why didn’t it just say “actual damages” instead of using a word with a different (and broader) meaning and then implicitly redefining it through its placement in subparagraph (A)? Such a procedure is hardly “simple” or “straightforward.” Ante at 178. In any event, the majority’s argument fails because its intuitions about the effect of placing the phrases “actual damages sustained” and “person entitled to recovery” in the same subpara-graph do not seem to be shared by other courts or by Congress. If the majority’s argument worked in this case, it would be equally effective when applied to the analogous statutory damages provisions in the tax code and the ECPA. Yet, as I have just pointed out, courts have repeatedly read the language of these statutes to allow the recovery of statutory damages without proof of actual damages. More powerfully still, Congress indicated in legislative history that it intended for the very similar language in those statutes to allow recovery of statutory damages without proof of actual damages. It is unsurprising, .then, that most courts have taken it for granted that § 552a(g)(4) permits the recovery of statutory damages without proof of actual damages.
The majority’s second argument, which is based on the use of the conjunction “but” to connect the two clauses of subpar-agraph (A), fares no better. For a start, it has the same difficulty as the first argument because neither courts nor Congress appear to recognize the rule of grammar proposed by the majority. Again, the statutory damages provisions in both the tax code and the ECPA contain the same language relied upon by the majority, yet courts have read those provisions' — with Congress’s blessing — to authorize the recovery of minimum statutory damages without proof of actual damages. Further, I am unconvinced that the use of the conjunction “but” must mean that the second clause in subparagraph (A) refers to the same class of persons identified in the first clause, namely, individuals who have sustained actual damages. If that had been Congress’s intent, it would likely have used the same language in both clauses by saying, “but in no case shall an individual who has sustained actual damages receive less than the sum of $1,000.” Instead, Congress used the phrase “person entitled to recovery” and began the clause containing that phrase with the words “but in no case.” This suggests discontinuity between the two clauses. The explanation of the discontinuity is that Congress meant to signal that the phrase “person entitled to recovery” is intended to refer not to an individual who has sustained actual damages, but to anyone to whom the United States “shall be liable” in specified amounts.
The majority’s third argument is that it makes no sense to believe that Congress would limit recovery to actual damages in subparagraph (A) and then, in the very same sentence, authorize the award of $1,000 to plaintiffs who have proven no *195actual damages at all. I have two responses to this argument. First, my reading of the statute does not, of course, authorize the award of statutory damages to a plaintiff totally unharmed by an agency’s violation of the Privacy Act. A plaintiff must still satisfy the adverse effect requirement. Second, it is not unprecedented for Congress to provide for an award of statutory damages in the absence of provable actual damages. As I have already explained, Congress did exactly that when it created a civil remedy for unauthorized disclosure of tax return information “[bjecause of the difficulty in establishing in monetary terms the damages sustained” as a result of such disclosures. S.Rep. No. 94-938, at 348. The same rationale explains why Congress would create a statutory damages remedy in the Privacy Act for those failing to prove actual damages. Indeed, the prevalence of my interpretation of § 552a(g)(4) in the courts partly reflects the fact that the remedy of minimum statutory damages is a fairly common feature of federal legislation. See 26 U.S.C. 7217(c) (Supp.1981) (current version at 26 U.S.C. § 7431(c)) (providing minimum statutory damages for unlawful disclosures of tax return information); 18 U.S.C. § 2707(c) (providing minimum statutory damages for violations of the ECPA); 15 U.S.C. § 1640(a) (providing minimum statutory damages remedy for violations of the Truth in Lending Act). In contrast, I am not aware of any statute in which Congress has “provide[d] for a ‘statutory minimum’ to actual damages.” Ante at 177.
The majority also offers a fourth reason in support of its interpretation of § 552a(g)(4)(A), though it does not regard the argument as necessary to its conclusion. According to the majority, my reading of the statute is unpersuasive because if Congress had intended to allow recovery of statutory damages without proof of actual damages, it could easily have written that the United States shall be liable “in an amount equal to the sum of (A) whichever is greater: actual damages sustained by the individual as a result of the refusal or failure, or $1,000.” Ante at 178. It is equally true, however, that if Congress had intended to condition the recovery of statutory damages on proof of actual damages, it could have easily drafted subpara-graph (A) to say that “in no case shall an individual who has sustained actual damages receive less than the sum of $1,000.” In short, the majority’s fourth argument simply underlines the point that show that Congress intended to condition the recovery of minimum statutory damages upon proof of actual damages.
To sum up, I believe that on balance the better reading of the statute is that any person to whom the United States shall be liable for its intentional or willful violations of the Privacy Act is a person entitled to recovery. Even if the majority is correct that the use of “person entitled to recovery” to mean any person to whom the United States shall be liable is poor statutory draftsmanship, Congress signaled in the legislative history of the statutory damages provisions in the tax code and the ECPA that it thinks otherwise. It is not our place to say that language Congress has repeatedly used to authorize the recovery of statutory damages without proof of actual damages is inadequate to that task.9
3.
Like the majority, however, I must acknowledge that my reading of the statute is not inevitable. This brings me to the majority’s fall-back position. It argues that, at worst, the statute is ambiguous and that its interpretation should prevail by virtue of the principle that waivers of *196the federal government’s sovereign immunity must be narrowly construed, a principle I will refer to as the “sovereign immunity canon.” This argument raises several potentially complicated questions. First, I have some doubts about whether the sovereign immunity canon should even apply in the circumstances of this case.10 Even conceding that the canon applies, I am not sure what it means to say that a statute must be narrowly construed in favor of the sovereign. It is, of course, clear that we must accept the narrower of two equally convincing interpretations; but it is far from clear exactly how much better one interpretation must be before it can prevail over another interpretation. Cf. Anto-nin Scalia, Common-Law Courts in a Civil Law System: The Role of United, States Federal Courts in Interpreting the Constitution and Laws, in A Matter of Interpretation: Federal Courts and the Law 28 (Amy Gutmann ed., 1997) (stating that rules that weight the scales of interpretation in favor of a particular result make it “virtually impossible to expect uniformity and objectivity” in statutory interpretation because such rules “add [ ], on one or the other side of the balance, a thumb of indeterminate weight”). I need not explore these questions in any detail, however, because I believe that policy considerations decisively tip the scales in favor of my interpretation. Specifically, the majority’s position undermines Congress’s purposes in creating a statutory damages remedy as part of the Privacy Act by making that remedy unavailable to many of its intended beneficiaries. For the reasons I explain below, I conclude that my reading must be preferred to the majority’s even if the sovereign immunity canon applies.
B.
I noted above that aside from my objections to the majority’s textual arguments, I *197disagree with the majority’s apparent assumption that we can sensibly decide whether proof of actual damages is a prerequisite to the recovery of statutory damages without any consideration of what “actual damages” means or what policies led Congress to create a statutory damages remedy as part of the Privacy Act in the first place. In my view, we cannot properly interpret the statute without identifying the policy consequences of the majority’s position and asking whether Congress could have intended these consequences. The consequences of conditioning the recovery of statutory damages upon proof of actual damages vary depending on the meaning of “actual damages.” If “actual damages” includes only pecuniary losses, then statutory damages are unavailable to plaintiffs who have suffered only emotional distress even though such distress is the typical injury caused by invasions of privacy. If “actual damages” also includes emotional distress that is both “demonstrable” and “sufficiently articulated” under the standards in Price v. City of Charlotte, 98 F.3d 1241, 1254 (4th Cir.1996), then statutory damages are unavailable to plaintiffs who have suffered minor emotional distress that could support only an award of nominal damages under 42 U.S.C. § 1983.11 I believe the majority’s argument must be wrong because, on either of these interpretations of “actual damages,” conditioning the availability of the statutory damages remedy upon proof of actual damages limits the availability of that remedy more sharply than Congress intended. To see why this is so, it is necessary to say more about Congress’s reasons for creating a statutory damages remedy as part of the Privacy Act.
Section 552a(g), the civil remedies section of the Privacy Act, protects citizens against government invasions of their privacy by creating a “self-help enforcement program ... [designed to] encourage[ ] the filing of lawsuits.” Parks, 618 F.2d at 685. See also Privacy Act § 2(b)(6), 88 Stat. at 1896; S.Rep. No. 93-1183, at 83 (1974), reprinted in 1974 U.S.C.C.A.N. 6916, 6997 (stating that the civil remedies provision in the Privacy Act was “designed to encourage the widest possible citizen enforcement through the judicial process ... [which was necessary because] the Act does not give any administrative body authority to ensure compliance with the Act”). To understand why Congress chose to create a statutory damages floor as part of this private enforcement scheme, we need only look to Congress’s purpose in enacting similar statutory damages provisions. We explained this purpose in Scrimgeour v. IRS, 149 F.3d 318 (4th Cir.1998):
[T]he $1,000.00 statutory damage award specified in I.R.C. § 7431(c)(1)(A) is included for the benefit of taxpayers. Actual damages for the invasion of privacy that occurs when tax returns are wrongfully disclosed can be hard to quantify. In order to encourage taxpayers to act as “private attorneys general” and pursue suits against the IRS for violations of I.R.C. § 6103, Congress enacted the statutory damages provision to ensure that in meritorious cases of wrongful release a taxpayer would not walk away *198from the courthouse empty handed for failure of proving actual damages.
Id. at 327 n. 11. See also supra at 190, 192 (citing the legislative history of the statutory damages provisions in the Internal Revenue Code and the ECPA). In short, Congress creates statutory damages remedies because it wants to encourage civil enforcement suits in situations where actual damages are difficult to prove.12 As we observed in Scrimgeour, the danger that plaintiffs with meritorious cases will “walk away from the courthouse empty handed” is especially great when the wrong to be remedied is the invasion of privacy because damages arising from that wrong can be hard to quantify. This is because the typical injury caused by invasions of privacy is mental distress. See Time, Inc. v. Hill, 385 U.S. 374, 384-85 n. 9, 87 S.Ct. 534, 17 L.Ed.2d 456 (1967); Johnson, 700 F.2d at 977. If these are the reasons why Congress created the statutory damages remedy in § 552a(g)(4)(A), it is difficult to understand why Congress would have conditioned the availability of that remedy on the plaintiffs ability to prove either economic harm or the kind of emotional harm that could support a substantial emotional damages award under the standards articulated in Price v. City of Charlotte.
The implausibility of the majority’s interpretation of the Privacy Act is espeeially clear if the term “actual damages” includes only pecuniary harms. Even the Secretary acknowledges that Congress created a self-help enforcement scheme to protect citizens against government invasions of privacy and that Congress included a statutory damages remedy within that scheme to ensure that citizens with little or no provable damages would have an incentive to sue. Because mental or emotional distress is the typical injury caused by invasions of privacy, it makes no sense to believe that Congress would undercut the force of its statutory damages remedy by refusing to make statutory damages available to citizens who suffered only emotional injuries. I conclude that if actual damages under the Act are limited to out-of-pocket losses, the majority’s claim that proof of actual damages is a prerequisite to the recovery of statutory damages must be wrong.13
Though the question is closer, I also believe that the majority’s reading of the Act is untenable even if “actual damages” includes emotional distress sufficient to satisfy the standards in Price v. City of Charlotte. According to the majority, a plaintiffs “conclusory” testimony about his emotional distress, even if believed by the trier of fact, is not sufficient to recover statutory damages. The plaintiffs testimony must establish that he “suffered de*199monstrable emotional distress, which must be sufficiently articulated.” Price, 98 F.3d at 1254.
I appreciate the need for the sort of inquiry Price demands in assessing substantial damages awards based solely on the plaintiffs own testimony regarding his emotional distress. In the context of the Privacy Act, I have no difficulty with the claim that a person who seeks to recover emotional distress damages in excess of the $1,000 statutory minimum should be required to bring forward the sort of proof that Pnce demands. A reviewing court must be able to satisfy itself both that some actual injury has occurred and that the jury has not simply been left to speculate about the proper amount of damages. In addition, the legislative history of the Privacy Act suggests that Congress worried about the possibility that juries would simply pull damages numbers out of the air based not on the actual harms suffered by the plaintiff, but on the jury’s judgment about the gravity of the government’s violation of the Privacy Act. See Johnson, 700 F.2d at 983-85 (suggesting that Congress used the term “actual damages” in the Privacy Act to emphasize that a plaintiff could not recover presumed damages based simply on the fact that his privacy had been invaded). This worry is appropriately addressed by the Price framework.14
It is an altogether different matter, however, to make the Price framework the standard for deciding when a Privacy Act plaintiff may recover minimum statutory damages. The problem with this approach is that, even on the most restrictive reading, the statute plainly awards $1,000 in statutory damages to a plaintiff who can prove $1, or even one penny, of actual damages. The Price framework was simply not designed to help courts decide whether a plaintiffs testimony adequately supports damages awards for such small sums. Consider the various factors that Price regards as adequate corroboration of emotional distress: “medical, psychological, or psychiatric treatment, the presence of physical symptoms, loss of income, and impact on the plaintiffs conduct and lifestyle.” Ante at 182. I think it highly unlikely that any plaintiff would seek medical, psychological, or psychiatric treatment, lose income, or undergo a change of lifestyle because of one dollar’s worth of emotional distress. Physical symptoms (sleeplessness, for example) might be present even in very minor cases of emotional distress, but then again they might not. All these factors are of course relevant to deciding whether a plaintiff experienced significant emotional distress — the kind that could support a substantial compensatory damages award. But they have little to say about whether claims of very minor emotional distress are genuine, and the majority agrees that if emotional distress counts as “actual damages” at all, proof of very minor emotional distress entitles the plaintiff to statutory damages under the Privacy Act. See ante at 177 (stating that statutory damages are available under the Act “where actual damages are greater than $0 but less than $1,000”). The Price framework is unhelpful in evaluating claims of very minor emotional distress for a simple reason: no plaintiff in a lawsuit *200under § 1983 (or any other cause of action without a statutory damages provision) ever sets out to prove that he suffered one dollar’s worth, let alone one penny’s worth, of emotional distress. When a plaintiff winds up with nominal damages, this means either that he conceded the absence of actual injury and sought to vindicate an abstract right or that he sought substantial damages and lost. A nominal damages award surely does not mean that the plaintiff has tried, but failed, to prove that he suffered one dollar’s worth of emotional distress.15
This point is not undercut by the majority’s appeal to Carey v. Piphus, 435 U.S. 247, 98 S.Ct. 1042, 55 L.Ed.2d 252 (1978). According to the majority, a plaintiff who would be entitled only to nominal damages has sustained no actual damages because nominal damages “are designed to vindicate legal rights ‘without proof of actual injury.’ ” Ante at 181 (quoting Carey, 435 U.S. at 266, 98 S.Ct. 1042 (1978)). The majority accurately states one purpose of awarding nominal damages, but it does not follow that an award of nominal damages signals the complete absence of actual injury. Nominal damages are also awarded “in cases where, although there has been a real injury, the plaintiffs evidence entirely fails to show its amount.” Black’s Law Dictionary 392 (6th ed.1990). See also Restatement (Second) of Torts § 907 cmt. c (1979) (“Nominal damages are properly awarded when, although the claimant shows significant harm, its amount is not proved with sufficient certainty to entitle him to an award of compensatory damages.”). Here, the fact that Buck Doe would be entitled to only nominal damages in a § 1983 case simply indicates that a jury could not reasonably make a substantial damages award on the basis of his somewhat limited testimony regarding his emotional distress because any attempt to do so would be unduly speculative. It does not mean that he suffered no injury at all.
The problems with the majority’s approach are well illustrated by its treatment of Buck Doe’s Privacy Act claim. In holding that Buck Doe has produced evidence of adverse effects sufficient to survive summary judgment, the majority acknowledges that a reasonable trier of fact could credit Buck Doe’s testimony about his emotional distress and could determine that he really did experience some degree of anxiety about the possibility that others would misuse his social security number. It follows that Buck Doe has presented evidence that he suffered an “injury in fact,” ante at 181 n. 6,1 and that a reasonable factfinder could conclude that he has established more than a merely technical invasion of his rights. This point makes it *201very difficult to understand the majority’s conclusion that Buck Doe nonetheless failed to present evidence of “actual damages” sufficient to survive summary judgment. The majority appears to hold that even though a reasonable factfinder could conclude that Buck Doe suffered some genuine emotional distress (that he was telling the truth when he claimed that he worried about possible misuse of his social security number), no factfinder could reasonably conclude that this emotional distress warranted even one penny of compensatory damages. I do not see how this position can be correct. In other words, I do not see any room to draw a line between proof of emotional distress sufficient to constitute an adverse effect and proof of emotional distress sufficient to establish a penny’s worth of actual damages. Certainly Price cannot help courts to make this sort of distinction. As a result, I conclude that the majority’s adoption of Price as the measure of when a plaintiffs emotional distress proves “actual damages” simply raises the bar for recovering statutory damages higher than Congress meant to put it by placing that remedy beyond the reach of plaintiffs who suffer very minor emotional harms.
In responding to the foregoing argument, the majority appears to acknowledge, even to embrace, this consequence of its position. It suggests that the denial of statutory damages to plaintiffs who have suffered only emotional injuries that are “very minor” or not “meaningful,” ante at 181 n. 6, is simply the price of the actual damages requirement’s “gatekeeping function of avoiding tremendous over-compensation of plaintiffs,” id. In response, my first point is that I do not see how the idea that a plaintiff must prove “meaningful” or more than “very minor” emotional distress can be squared with the majority’s recognition that, even on its own reading of the Act, proof of any amount of actual damages greater than zero is enough to trigger the statutory damages remedy. See ante at 177. New would describe emotional harms warranting one dollar’s worth of compensatory damages as “meaningful,” yet such harms would be sufficient to merit statutory damages under § 552a(g)(4)(A) even if the majority was correct that proof of actual damages is a prerequisite to the recovery of statutory damages. Second, I think that any purpose Congress might have had of avoiding “tremendous overcompensation” of plaintiffs is adequately served by requiring plaintiffs to prove they have suffered an adverse effect as a result of the government’s invasion of their privacy. Cf. Fitzpatrick, 665 F.2d at 330 (explaining that the $1,000 minimum damages floor is meant to “avoid a situation in which persons suffering injury had no provable damages and hence no incentive to sue” (emphasis added)).
The majority has a further rejoinder to these points, however. It argues that because the “ ‘actual damages’ requirement” is both distinct from and more demanding than the adverse effect requirement, very minor emotional harms sufficient to count as adverse effects must be insufficient to count as actual damages. Ante at 180-81 n. 6. I agree with the majority that a convincing interpretation of the statute must explain why Congress chose to use two terms, “adverse effect” and “actual damages,” when it could have used one. I also agree that if the term “actual damages” includes emotional distress, the showing of emotional distress necessary to satisfy any actual damages requirement imposed by the Privacy Act would have to be greater than the showing necessary to satisfy the adverse effect requirement. To my mind, however, these points cut the other way and simply reinforce my argument that the Privacy Act does not contain *202an actual damages requirement at all. Rather, it awards minimum statutory damages upon proof of an adverse effect suffered as a result of an agency’s willful or intentional violation of the Privacy Act. In addition, it creates an actual damages remedy by allowing recovery of proven actual damages in excess of the $1,000 statutory minimum. On my reading, then, the showing needed to recover proven actual damages for emotional distress is of course greater than that needed to satisfy the adverse effect requirement.16
To put the point in another way, the majority cannot produce a satisfying interpretation of the Act by claiming both that the Privacy Act imposes an actual damages requirement and that the term “actual damages” can include emotional distress. If even very minor emotional distress counts as “actual damages,” there seems to be no distinction between the adverse effect requirement and the actual damages requirement. As the majority points out, this result seems wrong because it fails to explain why Congress created two requirements rather than one. The majority avoids this problem by holding that only emotional distress satisfying the Price criteria counts as actual damages. This creates a gap between the adverse effect requirement and the actual damages requirement, but it does so only by making the statutory damages remedy unavailable to plaintiffs who have suffered very minor emotional harms. I do not think this result can be squared with Congress’s intent in creating the statutory damages remedy, for the text of the Privacy Act indicates that if Congress created an actual damages requirement at all, that requirement should be satisfiable by proof of one dollar’s worth (or even one penny’s worth) of actual damages. The better course is simply to abandon the idea that proof of actual damages is a prerequisite to the recovery of statutory damages.
In sum, the majority’s claim that proof of actual damages is a prerequisite to the recovery of statutory damages is inconsistent with Congress’s intent in creating a statutory damages remedy for agency violations of citizens’ privacy, and this is so regardless of whether the term “actual damages” is limited to out-of-pocket losses or includes emotional distress. This conclusion follows from three basic points about the Privacy Act. First, Congress created the statutory damages remedy as an incentive to suit because it recognized that damages from government invasions of privacy are hard to prove. Second, Congress recognized that the typical injury caused by the invasion of privacy is emotional distress. Third, Congress intended for the statutory damages remedy to be available to plaintiffs who suffered even very minor harms as a result of the government’s intentional or willful invasion of their privacy. If these three points are accepted, as I think they must be, Congress could not have intended to condition the recovery of statutory damages upon proof of actual damages. The point is especially clear if “actual damages” means out-of-pocket losses, for it would be senseless for Congress to create a statutory damages remedy in order to encourage *203citizen enforcement of the Privacy Act and then condition the availability of that remedy upon proof of economic losses when the typical injury caused by invasions of privacy is emotional distress. Though the question is closer, I think the claim that proof of “demonstrable” emotional distress under Price is needed to recover statutory damages is also untenable. The problem with this claim is that the Price framework is intended to help courts decide when a plaintiff has suffered the kind of emotional distress that can support a substantial damages award. When that framework is used to decide whether a plaintiff is entitled to statutory damages, the inevitable result is that plaintiffs who suffer very minor harms will go uncompensated. This cannot have been Congress’s intent, for if Congress had intended to impose an actual damages requirement at all, the text of § 552a(g)(4)(A) indicates that any amount of actual damages would be sufficient to trigger the statutory damages remedy.
Accordingly, I would hold that proof of actual damages is not a prerequisite for the recovery of statutory damages under § 552a(g)(4)(A). A plaintiff who proves that he has suffered an adverse effect as a result of an agency’s willful or intentional violation of the Act is entitled to the statutory damages remedy. A plaintiff may, of course, recover any proven actual damages in excess of $1,000.
III.
In conclusion, I respectfully dissent from the majority’s holding in part II of its opinion that actual damages are a prerequisite for the recovery of statutory damages under the Privacy Act and from its consequent decision to grant summary judgment to the Secretary against Buck Doe. I concur in the majority’s affirmance of the entry of summary judgment against the other Doe plaintiffs because they failed to show they had suffered any adverse effects as a result of the Department of Labor’s wrongful disclosure of their social security numbers. Finally, I would order a remand for a trial in Buck Doe’s case.
. I agree with the majority that the plaintiffs’ proposed class definition was sufficiently overbroad to fail the typicality requirement. See ante at 183-84. Though the district court could have narrowed the class definition on its own initiative, its failure to do so was not an abuse of discretion. See Lundquist v. Sec. Pac. Auto. Fin. Servs. Corp., 993 F.2d 11, 14-15 (2nd Cir.1993).
. A few courts appear to ignore the distinction between adverse effects and actual damages by holding that only proof of economic loss qualifies as an adverse effect. See, e.g., Risch v. Henderson, 128 F.Supp.2d 437, 441 (E.D.Mich.1999). This cannot be correct. Indeed, the Secretary concedes in her brief that the terms "adverse effect” and "actual damages” have different meanings. See Brief of Appellee/Cross Appellant Chao at 22 n.3.
. This would still be true even if the majority was correct in holding that a plaintiff who cannot prove actual damages will recover no damages at all. It may seem odd that a plaintiff could be entitled to costs and attorney fees without being entitled to any damages, but any apparent oddity in this result only reinforces my point that statutory damages should be available under the Act without proof of actual damages.
. The majority finds this argument unpersuasive. After observing that "person entitled to recovery” appears before costs and attorney fees are provided for, the majority reasons that Congress would not have made the definition of "person entitled to recovery” "dependent upon a subsequent statutory provision.” Ante at 179. The problem with the majority's argument, however, is that on my reading the definition of "person entitled to recovery” is not dependent on a subsequent provision. Rather, a person entitled to recovery is someone to whom "the United States shall be liable ... in an amount equal to the sum of” the amounts referred to in subpara-graphs (A) and (B). That language appears prior to the use of "person entitled to recovery” in subparagraph (A).
. The D.C. Circuit’s decision in Molerio v. FBI, 749 F.2d 815, 826 (D.C.Cir.1984), a case that did not appeal to sovereign immunity principles, is sometimes cited as evidence that the D.C. Circuit has adopted the position that proof of actual damages is a prerequisite for die recovery of statutory damages. The prec-edential force of Molerio on this point is unclear because the D.C. Circuit has subsequently said that proof of an adverse effect is sufficient to recover statutory damages. See Waters, 888 F.2d at 872.
. The Secretary argues that § 7431(c) does not help the Does' case because that section more clearly expresses Congress's intent to allow recovery of statutory damages without proof of statutory damages than does § 522(g)(4)(A). Section 7431(c) provides that
the defendant shall be liable to the plaintiff in an amount equal to the sum of—
(1) the greater of—
(A) $1,000 for each act of unauthorized disclosure of a return or return information with respect to which such defendant is found liable, or
(B) the sum of—
(i) the actual damages sustained by the plaintiff as a result of such unauthorized disclosure, plus
(ii) in the case of a willful disclosure or disclosure which is the result of gross negligence, punitive damages, plus
(2) the costs of the action.
This section contains exactly the sort of language that the majority regards as adequate to authorize the award of statutory damages without proof of actual damages. See ante at 178. The majority fails to recognize, however, that both Congress and the courts apparently regard the language in the new § 7431 as interchangeable with that of its predecessor. Section 7217 was replaced by § 7431 as part of a comprehensive tax reform bill in 1982, see Tax Equity and Fiscal Responsibility Act of 1982, Pub.L. No. 97-248, 96 Stat. 324, and the legislative history provides no indication that the new section was meant to change the operation of the statutory damages remedy. *192Courts have treated the two statutes interchangeably, see, e.g., Flippo v. United States, 670 F.Supp. 638, 642 (W.D.N.C.1987) (using committee report regarding § 7217 to interpret § 7431), and have uniformly interpreted both versions of the statute to award minimum statutory damages without proof of actual damages.
. The majority criticizes my use of the legislative history of the statutory provisions in the Internal Revenue Code and the ECPA to illuminate the meaning of § 552a(g)(4)(A). According to the majority, see ante at 178 n. 2, our decision in Sigmon Coal Co. v. Apfel, 226 F.3d 291, 308 n. 7 (4th Cir.2001), indicates that it is "dubious” to rely on Congress's intent in enacting one statute to discover "the intent of another Congress in enacting a separate, unrelated statute." This observation made sense in the context of Sigmon, but it has little bite here. In Sigmon we rejected an argument that unambiguous language in one provision of the Coal Act should be interpreted in a manner consistent with our earlier interpretation of a differently worded provision in the Comprehensive Environmental Response, Compensation, and Liability Act (CERCLA). Compare 26 U.S.C. § 9701(c)(2)(A) (a provision of the Coal Act at issue in Sigmon) with 42 U.S.C. § 9601 (a provision of CERCLA at issue in United States v. Carolina Transformer Co., 978 F.2d 832, *193837-38 (4th Cir.1992)). Here, the language of the Privacy Act’s statutory damages provision is strikingly similar to that of the analogous provisions in the tax code and the ECPA. Consequently, we ought, if at all possible, to give the language of all three statutes a consistent interpretation. Cf. Green v. Bock Laundry Mach. Co., 490 U.S. 504, 528, 109 S.Ct. 1981, 104 L.Ed.2d 557 (1989) (Scalia, J„ concurring in the judgment) (stating that a statute should be interpreted in a manner "most compatible with the surrounding body of law into which the provision must be integrated — a compatibility which, by a benign fiction, we assume Congress always has in mind”). This means that the statutory damages provisions in the tax code and the ECPA — including their legislative history— have significant implications for our interpretation of the Privacy Act. The point is espe-dally dear in the case of 26 U.S.C. § 7217(c), the original statutory damages provision in the tax code. If, as the majority claims, Congress had really intended to require proof of actual damages as a prerequisite to the recovery of statutory damages in § 552a(g)(4)(A), I find it difficult to believe that Congress would have chosen to use essentially the same language only two years later in the tax code to authorize the recovery of statutory damages without proof of actual damages.
. The majority's characterization of subpara-graph (A) as having the "sole and entire purpose” of "limiting] the liability of the United States to actual damages sustained,” ante at 177, is simply a case of oversell. The subpar-agraph’s purpose is more neutrally described as specifying the amount of damages to be recovered by a plaintiff to whom the United *194States is liable. Further, the “sole and entire purpose” of the subparagraph cannot be to limit recovery to actual damages sustained because the statutory minimum means that some plaintiffs will recover more than their actual damages. That is the whole point of the statutory minimum.
. I would quickly add that, as I noted above, Congress is at least as guilty of poor draftsmanship on the majority's reading of the statute as it is on mine.
. The Supreme Court has primarily used the sovereign immunity canon to answer questions about whether a lawsuit may be brought at all or about the availability of certain forms of relief, not to decide what a plaintiff must prove to obtain relief that has been unequivocally authorized. See, e.g., Lane v. Pena, 518 U.S. 187, 196-97, 116 S.Ct. 2092, 135 L.Ed.2d 486 (1996) (using canon to decide whether plaintiff could recover monetary damages from the federal government); Irwin v. Dep’t of Veterans Affairs, 498 U.S. 89, 94-96, 111 S.Ct. 453, 112 L.Ed.2d 435 (1990) (using canon to decide whether statute of limitations may be equitably tolled in action against the federal government); Library of Congress v. Shaw, 478 U.S. 310, 318-21, 106 S.Ct. 2957, 92 L.Ed.2d 250 (1986) (using canon to decide whether plaintiff could recover interest award in action against the federal government). Indeed, the Court recently suggested that when Congress has clearly provided for a damages remedy, some subsidiary questions about the proper administration of that remedy might be properly answered without reference to the sovereign immunity canon. See West v. Gibson, 527 U.S. 212, 222, 119 S.Ct. 1906, 144 L.Ed.2d 196 (1999). In short, it is not clear that the Supreme Court intends for the sovereign immunity canon to be used to answer every question of statutory interpretation that could affect the federal government's legal liability. Consequently, it is not surprising that most of the courts to interpret § 552a(g)(4)(A) have done so without reference to the sovereign immunity canon. Moreover, it is simply bad law to apply the sovereign immunity canon to questions of statutory interpretation that determine the availability of relief to a particular plaintiff in a particular circumstance. To insist on a clear statement at this level of detail places unrealistic demands on congressional drafting and threatens to frustrate congressional intent. See John Copeland Nagle, Waiving Sovereign Immunity in an Age of Clear Statement Rules, 1995 Wis. L.Rev. 771, 819-22 (1995). These considerations make me doubt the wisdom of applying the sovereign immunity canon in this case, and I fear that after today's decision it is only a matter of time before the government begins asking us to apply the canon to such questions as whether a particular set of consequences counts as an adverse effect under § 552a(g)(l)(D) or whether a particular manner of releasing personal information counts as an agency disclosure under § 552a(b).
. There is a third possibility that I need not discuss in detail. If "actual damages” includes any evidence of emotional distress credited by a trier of fact, statutory damages are unavailable only to a plaintiff who cannot show that he suffered even minor emotional distress as a result of an agency's violation of his privacy. Under this option, any proof of emotional distress sufficient to qualify as an adverse effect will count as actual damages. Thus, this position is indistinguishable in practice from my position that proof of an adverse effect is sufficient to recover statutory damages.
. Because the legislative history of the Privacy Act contains no explicit discussion of Congress’s decision to include a statutory damages remedy in § 552a(g)(4), see Johnson, 700 F.2d at 981 n. 25, I am admittedly assuming that Congress’s purpose here was the same as its purpose in creating similar statutory damages remedies. Because the statutory damages provisions in the Privacy Act and the tax code both aim at the same harm — the invasion of privacy caused by government disclosures of personal information — I believe this assumption is warranted.
. Like the majority, I need not reach the difficult question of the meaning of "actual damages,” a question that has been extensively analyzed by the Fifth and Eleventh Circuits. See Johnson, 700 F.2d at 974-86; Fitzpatrick, 665 F.2d at 329-31. I simply point out that if the Secretary is correct in her contention that "actual damages” includes only out-of-pocket losses, proof of actual damages cannot be a prerequisite to the recovery of statutory damages. Conversely, I believe that the majority's holding commits this circuit to the position that the term "actual damages” includes at least emotional distress that would qualify as "demonstrable” under Pnce.
. In many jurisdictions plaintiffs may recover presumed damages (substantial damages awards without proof of actual pecuniary or emotional harm) for dignitary torts such as the invasion of privacy. See 2 Dan B. Dobbs, Law of Remedies § 7.1 (2d ed.1993). Dobbs suggests that § 552a(g)(4)(A) of the Privacy Act can best be understood as the statutory equivalent of presumed damages. Id. at § 7.3(4). In other words, the suggestion is that Congress wished to incorporate some of the force of the common law doctrine of presumed damages for dignitary torts while placing limits on jury discretion.
. The majority faults me for characterizing the Price framework as a tool for reviewing damages awards that exceed "some relatively high threshold of 'substantiality,' ” ante at 180 n. 6, and insists that Price is an appropriate framework for determining whether any award of more than nominal damages for emotional distress is warranted. I acknowledge that the notion of a "substantial” damages award is somewhat vague, so perhaps some clarification is in order. Any award over the statutory minimum of $1,000 qualifies as "substantial,” and perhaps the criteria in Price might also be useful in evaluating awards of considerably less than $1,000. (It bears noting, however, that no case of which I am aware has used Price to vacate a jury award for emotional distress of less than $1,000.) In contrast, an award of $1.00 in actual damages for emotional distress is not "substantial" on any account. In saying that the Price framework is designed to evaluate ' "substantial” damages awards, my primary point is that Price is not helpful in deciding when a plaintiff has suffered the one dollar's worth (or even one penny's worth) of actual damages that should warrant the recovery of statutory damages even on the majority’s reading of the Privacy Act.
. There is, of course, another way to explain why Congress chose to use the term "adverse effect” as well as the term "actual damages.” The Act could be read to say that emotional distress is an adverse effect, that proof of an adverse effect is sufficient to recover statutory damages, and that "actual damages” includes only pecuniary harms. On this reading, plaintiffs who suffer only emotional distress as a result of willful and intentional violations of the Privacy Act recover $1,000 — no more and no less — while plaintiffs who can prove pecuniary harms in excess of $1,000 recover the full amount of these "actual damages.” This seems to be the reading adopted by the Eleventh Circuit in Fitzpatrick.