Before this court are two consolidated appeals and a cross-appeal. Anne Stover and Rita Chandi (together defendants) appeal from separate and final judgments entered in the United States District Court for the District of Minnesota after each pled guilty to one count of mail fraud, in violation 18 U.S.C. § 1341. Stover was sentenced to twenty-four months imprisonment, three years supervised release, a special assessment of fifty dollars, and restitution in the amount of $40,-000 to be paid jointly and severally with Chandi. United States v. Stover, No. CR 3-94-98(1) (D.Minn. Aug. 21, 1995) (judgment). Chandi was sentenced to eighteen months imprisonment, three years supervised release, a special assessment of fifty dollars, and restitution in the amount of $40,000 to be paid jointly and severally with Stover. United States v. Chandi No. CR 3-94-98(2) (D.Minn. Aug. 21, 1995) (judgment). For reversal, defendants argue that the district court clearly erred in finding that they knowingly targeted unusually vulnerable victims for purposes of imposing a two-level sentencing enhancement pursuant to U.S.S.G. § 3A1.1 (Nov. 1994).1 Chandi separately argues that the district court clearly erred in finding that she was accountable for losses in excess of $500,000 as part of her offense conduct under U.S.S.G. § 2Fl.l(b)(l)(K). The government, on cross-appeal, argues that the district court abused its discretion in failing to identify the victims to whom restitution is owed and in ordering Stover to pay only $40,000 in restitution when the undisputed evidence at sentencing showed a loss to victims in the amount of $643,617. For the reasons discussed below, we affirm in part and reverse in part, and remand the case to the district court for resentencing.
Background
In December 1988, Stover founded Families for Children (FFC), an adoption agency based in St. Paul, Minnesota.2 FFC was licensed by the Minnesota Department of Human Services as a nonprofit corporation in 1990 and registered with the Minnesota Attorney General as a charity in 1992. According to Stover, FFC initially concentrated on placing for adoption children born in foreign countries. In 1991, however, FFC shifted its focus to the placement of American-born infants. During the relevant period of time, Stover had the title Executive Director of FFC and Chandi had the title Associate Director of FFC.
FFC operated by entering into contracts with prospective adoptive individuals, couples, or families (hereinafter referred to as “clients”), requiring them to pay an up-front fee ranging up to $11,500. Meanwhile, FFC would seek out pregnant women who were willing to give up their babies for adoption (referred to as “birth mothers”), by offering them financial and other forms of support. FFC held itself out as specializing in open adoptions, in which the birth mother would take part in the selection of the adoptive parent or parents and would be permitted periodic contact with the child.
*1382By mid-1992, the demand for children from FFC clients exceeded FFC’s available and anticipated supply. Stover and Chandi began .lying to their clients. According to the facts admitted in their plea agreements, defendants
knowingly made intentional misrepresentations to prospective and existing clients about the ratio of birth mothers to adoptive parents that were clients at FFC. These intentional misrepresentations included, but were not limited to, statements at various times that FFC was working with approximately 30 birth mothers, and had only approximately 30 waiting clients. In reality, at the time these statements were made, the true number of birth mothers was much lower than stated, and the true number of waiting clients was much higher than stated.
Designated Record at 24-25 (Stover’s Plea Agreement ¶ 1); id. at 31-32 (Chandi’s Plea Agreement ¶ 1).
On October 4, 1993, the Minnesota Attorney General’s Office filed a complaint against FFC and also applied for and obtained a temporary injunction closing down FFC and appointing a receiver to wind up FFC’s affairs. According to the government, upon examining FFC’s records, authorities discovered that FFC had approximately seventy-five clients on its waiting list and was working with only one or two birth mothers as of October 1993. Id. at 9 (Indictment ¶ 18).
The government states that Stover’s salary was $88,000 per year as of June 1992 and, after further raises (which she gave herself), was up to $95,150 per year as of October 1993. The government further alleges that Stover paid herself $2,500 per month in “rent” for FFC’s use of the basement of her house, even though her monthly mortgage payment for the whole house was only approximately $1,500; she also used FFC funds to pay for a ear for herself, day care for her children, and domestic services. Chandi’s salary during the same time period increased from approximately $21,000 to $42,500 per year.
On August 24, 1994, defendants were charged in an eighteen-count indictment. They each pled guilty to count eight of the indictment, alleging mail fraud in violation of 18 U.S.C. § 1341. The plea agreement included a provision acknowledging the district court’s authority, pursuant to 18 U.S.C. § 3663(a)(3), to order restitution in any amount up to and including the amount of loss deemed to be relevant conduct under U.S.S.G. § 2Fl.l(b)(l).
The presentence investigation report (PSR) recommended that each defendant receive a two-level upward adjustment for targeting unusually vulnerable victims. U.S.S.G. § 3A1.1 (Nov. 1994) (vulnerable victim enhancement). Defendants objected to the recommendation. The government supported it and submitted extensive documentation, including victim impact statements, to illustrate the manner in which defendants allegedly preyed upon people who were particularly desperate to adopt. At sentencing, the district court applied the two-level victim-related enhancement to each defendant’s guidelines calculation.
The PSR also set forth the precise amount of loss suffered by each of seventy-two FFC clients, which totalled $643,617. Consequently, defendants’ base offense level of 6 was increased by 10 pursuant to U.S.S.G. § 2Fl.l(b)(l)(K) (offense conduct provision requiring a 10-level increase if loss is more than $500,000). The PSR also recommended exact restitution to each victim.
The district court calculated Stover’s and Chandi’s total offense levels as 17 and 15, respectively. Each was assigned a criminal history category I. Stover was sentenced to twenty-four months imprisonment and Chan-di was sentenced to eighteen months, each receiving the lowest sentence within them respective ranges. As to restitution, the district court initially ordered defendants to pay restitution as ordered by the probation office, without specifying the amount of restitution owed or the names of the victims. The government moved for modification of the restitution order on grounds that the order lacked sufficient specificity. In its written judgments and commitment orders, the district court ordered defendants each jointly and severally accountable to pay $40,000 in restitution, but still did not specify the names *1383of the victims or the amounts owed to them individually. These appeals and cross-appeal followed.
Discussion
Application of vulnerable victim enhancement
Defendants argue that the district court erred in giving them each a two-level upward adjustment under the vulnerable victim provision, U.S.S.G. § 3A1.1 (Nov. 1994), which provides: “[i]f the defendant knew or should have known that a victim of the offense was unusually vulnerable due to age, physical or mental condition, or that a victim was otherwise particularly susceptible to the criminal conduct, increase by 2 levels.” The district court explained its reasons for imposing the two-level adjustment under § 3A1.1 as follows.
This crime involved fraud that preyed upon the victims’ willingness to spend large amounts of money in order to adopt children. This situation is similar to the scenarios described in Application Note 1 to § 3A1.1, which indicates that the enhancement should apply “where the defendant marketed an ineffective cancer cure or in a robbery where the defendant selected a handicapped victim.”[3] The victims’ strong desire to adopt made them financially more vulnerable than other individuals and particularly susceptible to [defendants’] criminal conduct.
Designated Record at 180-81 (statement of reasons for imposing sentence (Stover) at 1-2); id. at 185-86 (statement of reasons for imposing sentence (Chandi) at 1-2). Defendants argue that the district court based its decision upon the financial vulnerability of their victims. Therefore, they argue, the district court erred because, according to this court’s holding in United States v. Ravoy, 994 F.2d 1332 (8th Cir.1993), financial vulnerability is generally not a ground for finding the existence of a vulnerable victim.
Defendants further argue that, under applicable case law, the vulnerable victim en-haneement applies only where there is evidence of both an unusual vulnerability or particular susceptibility of the victim and targeted exploitation of that weakness. Defendants maintain that the mere fact that their clients had the desire to adopt children made the fraud possible, but did not make their clients unusually vulnerable or particularly susceptible to the crime. See United States v. Paige, 923 F.2d 112, 113-14 (8th Cir.1991) (reversing application of enhancement even though defendant targeted young store clerks because he considered them inexperienced and naive; such clerks were not unusually vulnerable). Even if their clients’ desire to adopt was powerful, defendants argue, there is no evidence that the clients’ judgment was particularly impaired. Second, defendants argue that there was no evidence that they targeted their victims’ vulnerabilities, as the law requires. See, e.g., United States v. Callaway, 943 F.2d 29, 31 (8th Cir.1991) (although victim was young and handicapped, record does not support a finding that the defendant chose her victim because of those factors); United States v. Cree, 915 F.2d 352, 354 (8th Cir.1990) (enhancement justified only when defendant’s actions in some way exploited or took advantage of the victim’s vulnerability). On this point, defendants maintain that they had every intention of successfully placing a child with each of their clients and, therefore, this case is materially different from those in which the fraud is based upon deliberately false promises. Defendants conclude that, because neither of the two requirements exists in the present case, the district court erred in imposing the vulnerable victim enhancement under U.S.S.G. § 3A1.1.
In response, the government argues that the district court did not clearly err in making its vulnerable victim finding. The government argues that this case is analogous to the example in the -commentary, to which the district court referred, concerning the defendant who markets an ineffective cure for cancer. U.S.S.G. § 3A1.1, comment, (n. 2) (Nov. 1995); id. comment, (n. 1) (Nov. 1994). The government maintains that many *1384of defendants’ victims had problems with infertility and suffered the attendant emotional effects of that condition. Moreover, the government argues, defendants knew about these problems from the written forms filled out by some of their prospective clients. The government claims that defendants targeted those clients’ emotional vulnerability by promising a “quick fix.” The government maintains that, notwithstanding the district court’s failure to cite the infertility of some of the victims as a ground for finding unusual vulnerability, this court should uphold the sentencing enhancement on that basis.
The government further suggests that the district court’s application of § 3A1.1 is supported by cases in which the defendants were given the enhancement because they victimized a specific group or class of people. The government cites, for example, United States v. McDermott, 29 F.3d 404, 411 (8th Cir.1994), in which this court upheld the application of § 3A1.1 to the sentences of two defendants convicted of racially-motivated hate crimes where the victims included geographically-isolated, African-American youth and a physically-disabled fourteen-year-old Caucasian girl. The government notes that, in McDermott, this court rejected the argument that the victims’ race could not be the basis for applying a § 3A1.1 enhancement even though the victims of such civil rights violations typically are racial minorities. Id. at 411.4 Similarly, the government argues, the victims in the present case shared a desire to adopt children and that desire may have made the fraud possible, but it certainly did not diminish the victims’ particularized emotional vulnerability. The government also argues that defendants’ claim of innocent intentions goes to the question of whether there was intentional fraud and is irrelevant to this sentencing issue. Moreover, the government emphasizes the wording of § 3A1.1 which broadly states that the two-level adjustment applies if the defendant “knew or should have known” that the victim was unusually vulnerable or particularly susceptible to the criminal conduct. The government maintains that the requirement in Eighth Circuit case law, that the defendant must “target” an unusually vulnerable victim, should not eclipse the plain meaning of the guideline provision. See, e.g., United States v. O’Brien, 50 F.3d 751, 757 (9th Cir.1995) (emphasizing “knew or should have known” language as basis for affirming application of § 3A1.1 enhancement).
We begin by noting that our analysis is complicated by the fact that after defendants’ sentencing, but before defendants’ appeals were submitted to this court, the sentencing guidelines were amended, effective November 1, 1995. As a consequence, what was formerly U.S.S.G. § 3A1.1 became § 3Al.l(b). The exact wording of the guideline provision has not changed, nor have the hypothetical examples in the accompanying commentary. However, the commentary has been changed in one important respect. The pre-November 1995 commentary included the sentence: “This adjustment applies to offenses where an unusually vulnerable victim is made a target of criminal activity by the defendant.” U.S.S.G. § 3A1.1 comment, (n. 1) (Nov. 1994). That commentary language was the foundation for the targeting requirement relied upon in decisions such as Cree, 915 F.2d at 358-54 (quoting commentary). However, in the current version of the Sentencing Commission’s commentary, that key sentence has been replaced by the statement: “Subsection (b) applies to offenses involving an unusually vulnerable victim in which the defendant knows or should have known of the victim’s unusual vulnerability.” U.S.S.G. § 3Al.l(b) comment, (n. 2) (Nov. 1995). In making this change, the Sentencing Commission specifically stated “[tjhis amendment revises the Commentary of § 3A1.1 to clarify application with respect to [the targeted victim] issue.” U.S.S.G.App. C, Amend. 521, at 430 (Nov.1995) (emphasis added).
*1385In the aftermath of this “clarification” by the Sentencing Commission, we will be required to consider carefully the continuing vitality of our previously well-established holding that “enhancing a defendant’s sentence based on victim vulnerability is justified only when a defendant’s actions in some way exploited or took advantage of that vulnerability.” Cree, 915 F.2d at 354. We note, for example, that in United States v. Feldman, 83 F.3d 9, 16 (1st Cir.1996), the First Circuit observed that under the pre-Novem-ber 1995 guidelines
the “target” language split the circuits on the issue of whether the government had to prove that the defendant was motivated by the victim’s special vulnerability in order to lay a foundation for the upward adjustment, see, e.g., United States v. Smith, 39 F.3d 119, 124 (6th Cir.1994); United States v. Cree, [915 F.2d at 354], or whether the government merely had to show that the defendant targeted his [or her] victim with the knowledge (actual or constructive) that the victim was unusually vulnerable, see, e.g., United States v. O’Brien, [50 F.3d at 754-55],
The First Circuit then concluded:
[a]s for future eases, the Sentencing Commission has removed all reasonable doubt by amending the commentary to § 3A1.1. In an effort to resolve “some inconsistency in the application of § 3A1.1 regarding whether this adjustment required proof that the defendant had ‘targeted the victim on account of the victim’s vulnerability,’ ” U.S.S.G.App. C, Amend. 521, at 430 (Nov. 1995), the Commission deleted the “target” language.
Feldman, 83 F.3d at 16. Thus, the First Circuit held that cases such as Cree no longer represented the prevailing rule. Id.
In the present case, after defendants’ appeals were submitted to the court, the parties supplemented their briefs with letters to the court concerning the applicability of the guideline amendments to defendants’ sentences. Defendants argue that we should apply the guidelines in effect at the time of sentencing, notwithstanding the November 1995 amendments. The government argues that Amendment 521 governs the present case because it is expressly a “clarification” of the guidelines.
The government’s position appears to be supported by United States v. Stinson, 30 F.3d 121,122 (11th Cir.1994) (per curiam), in which the Eleventh Circuit, on remand from the Supreme Court, instructed the district court to resentence the defendant in accordance with an amendment to the commentary of the guidelines, even though that amendment had become effective after the defendant’s original sentencing. The Eleventh Circuit’s decision to apply the guidelines amendment retroactively was based upon the fact that the amendment was purportedly a “clarification” and not a substantive change in the law. Id. (“[although we have doubts that this amendment just clarifies the pertinent guidelines (as opposed to making a substantive change in the law), we cannot conclude that the Commission’s viewing of the amendment as a clarification is plainly wrong”). However, a determination that an amendment is a “clarification” does not necessarily end the inquiry. In Stinson v. United States, 508 U.S. 36, 42-43, 113 S.Ct. 1913, 123 L.Ed.2d 598 (1993) (remanding the ease to the Eleventh Circuit), the Supreme Court held that, even though the commentary to the guidelines is generally authoritative and binding on the courts, “[i]t does not follow that commentary is binding in all instances.” The Supreme Court went on to explain that “the guidelines are the equivalent of legislative rules adopted by federal agencies. The functional purpose of commentary (of the kind at issue here) is to assist in the interpretation and application of those rules.” Id. at 45, 113 S.Ct. at 1919. “[T]his type of commentary is akin to an agency’s interpretation of its own legislative rules.” Id. “/PJrovided an agency’s interpretation of its own regulations does not violate the Constitution or a federal statute, it must be given ‘controlling weight unless it is plainly erroneous or inconsistent with the regulation.’” Id. (emphasis added) (citations omitted).
In United States v. Stinson, it was clearly understood that the defendant’s sentence would be decreased if the amendment were applied. 30 F.3d at 122 (defendant’s felon-in-possession offense was basis for es*1386tablishing his career offender status and amendment specifically excluded that offense from the category of predicate crimes of violence). Consequently, no issue existed as to whether resentencing under the amended commentary would violate the ex post facto clause of the Constitution. In the present case, by contrast, Amendment 521, affecting the commentary to § 3A1.1, would have the effect, if any, of increasing defendants’ sentences. “Article I of the United States Constitution provides that neither Congress nor any State shall pass any ‘ex post facto Law.’ ” Miller v. Florida, 482 U.S. 423, 429, 107 S.Ct. 2446, 2450, 96 L.Ed.2d 351 (1987). As a general rule of constitutional law, a violation of the ex post facto prohibition occurs where (1) the law applied is retrospective, that is, it applies to events occurring before its enactment, and (2) it is disadvantageous to the defendant to whom it is applied, provided, however, that it does not involve merely a procedural change and its onerous effects are not offset by other ameliorative provisions. Id. at 430-32, 107 S.Ct. at 2451-52. In the present case, the commentary to U.S.S.G. § 3A1.1 is a “law” for purposes of engaging in ex post facto analysis. See, e.g., United States v. Levi, 2 F.3d 842, 844-45 (8th Cir.1993) (holding that, while the sentencing guidelines and commentary are “laws” for purposes of ex post facto analysis, some policy statements are not). Therefore, notwithstanding the Sentencing Commission’s description of Amendment 521 as a “clarification,” we hold that applying the new language set forth in U.S.S.G. § 3A1.1 comment. (n. 2) (Nov. 1995), as opposed to the language set forth in U.S.S.G. § 3A1.1 comment. (n. 1) (Nov. 1994), would in this case violate the Constitution’s prohibition against ex post facto laws because: the application would be retrospective; it would, if anything, increase defendants’ sentences; it would not merely involve a procedural change; and it would not be offset by other ameliorative provisions. We therefore decline to apply the new commentary. We analyze this particular case according to the law as it existed at the time of defendants’ criminal conduct, which, for all intents and purposes, is the same as the law which existed at the time of sentencing and at the time the parties initially briefed this case. We now turn to the merits of the vulnerable victim issue.
We review for clear error the district court’s finding that there was a vulnerable victim in the present case. United States v. Cron, 71 F.3d 312, 314 (8th Cir.1995); United States v. Boult, 905 F.2d 1137, 1138-39 (8th Cir.1990) (existence of a vulnerable victim is a factual determination reviewable under the clearly erroneous standard). In the present case, we are not dealing with one of the types of victim vulnerability expressly enumerated in § 3A1.1 (i.e., age, physical or mental condition). Rather, we are faced with the difficult question of whether defendants’ victims were “otherwise particularly susceptible to the criminal conduct,” within the meaning of § 3A1.1. In United States v. Castellanos, 81 F.3d 108, 110-11 (9th Cir.1996) (emphasis added), the Ninth Circuit provided the following interpretation of that elusive language, based upon a survey of numerous decisions of the federal courts of appeals.
[I]t is not enough to support a finding of particular susceptibility under § 3A1.1 that the victims are more likely than other members of the general population to become a victim to the particular crime at issue. The reason for this is that criminals will always tend to target their victims with an eye toward success in the criminal endeavor. Thus, the chosen victims are usually more susceptible than the general population to the criminal conduct.
The appellate courts have consistently refused to find a class of victims to be particularly susceptible to criminal conduct simply because they were statistically more likely to fall prey to the defendant’s crime....
Especially in eases involving some kind of scheme to defraud, the criminal will typically direct his [or her] activities toward those persons most likely to fall victim to the scheme. But all defendants targeting such victims do not necessarily merit a sentence enhancement under § 3A1.1. Otherwise, all but the most unthinking of criminal defendants would be candidates for upward adjustment under *1387§ 3A1.1. Instead, the victims to whom § 3A1.1 applies are those who are in need of greater societal protection.... They are the persons who, when targeted by a defendant, render the defendant’s conduct more criminally depraved. Paige, 923 F.2d at 113.
We agree with the above-quoted interpretation of the phrase “otherwise particularly susceptible to the criminal conduct,” as used in § 3A1.1. With that in mind, we now turn to the arguments presented in these appeals. To begin, we agree with defendants’ argument that persons who desire to adopt, when victimized by a scheme to defraud in the adoption business, are “[a]s a general class, ... not the type of class as a whole for which § 3A1.1 was designed.” Brief for Appellant Stover at 12. Cf. United States v. Morrill, 984 F.2d 1136, 1137 (11th Cir.1993) (en bane) (per curiam) (on remand from the Supreme Court, holding that bank tellers, as a group, are not “otherwise particularly susceptible” to bank robbery within the meaning of the guidelines); accord U.S.S.G. § 3Al.l(b), comment. (n. 2) (Nov. 1995); id. comment, (n. 1) (Nov. 1994) (“a bank teller is not an unusually vulnerable victim solely by virtue of the teller’s position in a bank”).
However, the district court determined that the enhancement was appropriate in the present ease because defendants preyed upon their victims’ “strong desire to adopt” and their seemingly blind willingness to spend large amounts of money toward that end. The district court compared the victims in the present case to the hypothetical cancer patient looking for a cure, used as an example in the commentary to § 3A1.1. In light of that comparison, we understand the district court’s reasoning to be that defendants preyed on victims whose particular susceptibility derived from their “strong desire to adopt,” not their financial circumstances. Therefore, defendants’ reliance on Ravoy is misplaced. See Brief for Appellant Stover at 9 (citing Ravoy, 994 F.2d at 1335-36 (vulnerable victim finding based upon “distressed financial circumstances” was clearly erroneous because the victims, although mostly low-income people who had fallen on hard economic times, lacked a sufficiently particularized vulnerability)).
The case before us is similar in many respects to the example in the commentary referring to the defendant who markets an ineffective cure for cancer. In both cases, it appears that the victims’ vulnerability results from a sense of desire or desperation, presumably created by circumstances beyond their control. However, in the commentary example, we can also presume that the hypothetical victim of the fraud is a person afflicted with a potentially fatal medical condition. Recognizing that there are no bright lines in this analysis, we think the need for societal protection, and the inference of heightened criminal depravity, is greater in the cancer patient context than in the adoption setting. In our opinion, the cancer patient’s inherent desperation for a life-saving cure is sufficiently different from the “strong desire to adopt” felt by the victims in the present case that a distinction should be drawn for purposes of applying § 3A1.1. Moreover, the mere fact that many of defendants’ victims were readily disposed to spend thousands of dollars in order to pursue their heartfelt dreams of having a child still does not, in our opinion, create the type of particular susceptibility contemplated by § 3A1.1. The clients’ willingness to spend such large sums of money made them more likely to fall victim to defendants’ fraud, but did not create the extra need for societal protection which § 3A1.1 is designed to address. Therefore, without intending to discount the pain and disappointment suffered by defendants’ victims, we hold that the district court’s vulnerable victim finding was clearly erroneous.
We next turn to the government’s main argument on this issue — that it was, more specifically, the infertility of some of defendants’ victims that made those victims particularly susceptible to the fraud. As a threshold matter, we recognize that we may consider this proposed basis to affirm, which was presented by the government to the district court but not expressly mentioned by the district court in its statement of reasons for making its vulnerable victim finding. See United States v. Garrido, 995 F.2d 808, 813 *1388(8th Cir.) (the court of appeals may affirm on any ground supported by the record), cert. denied, 510 U.S. 926, 114 S.Ct. 330, 331, 126 L.Ed.2d 276 (1993). We further recognize that, given the proper set of facts, a person’s infertility, if known to the defendant, might support a finding of particular susceptibility to adoption-related fraud. Cf. United States v. Malone, 78 F.3d 518, 522-23 (11th Cir.1996) (affirming application of § 3A1.1 adjustment to carjacking offenders who targeted eabdrivers whom the defendants knew were required to take certain unusual risks, as opposed to other less easily victimized drivers). However, upon careful review of the record in the present case, we find no evidence that defendants offered their services selectively rather than to the general public at-large. Moreover, the evidence indicates that defendants vigorously pursued the business of anyone who was willing to pay their fees, without any genuine regard for how unfortunate the clients’ particular circumstances were. In the present case, the fact that defendants were able to bait and hook clients who had difficulties with infertility, and were possibly emotionally distraught as a result, was inherent in the nature of their ruthless crime, but does not suggest that those victims were “targets” within the pre-November 1995 meaning of U.S.S.G. § 3A1.1. In other words, the government did not demonstrate that defendants’ actions in some way exploited or took advantage of the victims’ infertility, or that any of the victims were chosen for that reason. Therefore, we hold that, under the law applicable to the present case, the vulnerable victim enhancement was not justified. See, e.g., Callaway, 943 F.2d at 31; Cree, 915 F.2d at 354 (enhancement justified only when defendant’s actions in some way exploited or took advantage of the victim’s vulnerability); U.S.S.G. § 3A1.1, comment, (n. 1) (Nov. 1994) (“[t]his adjustment applies to offenses where an unusually vulnerable victim is made a target of criminal activity by the defendant”).
Amount of loss attributable to Chandi under U.S.S.G. § 2F1.1
Chandi additionally argues that the district court imposed a ten-level increase based upon a clearly erroneous finding that the amount of loss for which she was responsible exceeded $500,000. See U.S.S.G. § 2Fl.l(b)(l)(K) (offense conduct provision requiring a 10-level increase if loss is more than $500,000). Chandi argues that the $643,617 loss figure should have been reduced in her case because (1) thirteen couples, whose fees accounted for $107,705 of the $643,617 total figure, were FFC clients before she joined the agency and (2) eight additional couples from New York, whose fees accounted for $37,275, had no contact with her and only paid fees pursuant to a New York law which limits fees to services rendered. Chandi argues that, subtracting these amounts, the total loss attributable to her is less than $500,000, thereby reducing her total offense level by one.
Upon review, we hold that, as to the New York clients, the evidence supports an inference that Chandi did in fact have contact with those clients. In any case, under § 2F1.1, Chandi is responsible for those clients’ losses because they were a foreseeable consequence of defendants’ fraudulent scheme. See U.S.S.G. § 1B1.3 (specific offense characteristics such as amount of loss shall be based upon all reasonably foreseeable acts and omissions of others in furtherance of jointly undertaken criminal activity). Chandi’s contention that New York law prohibits agencies from charging for services not yet rendered is irrelevant because the payments made by the New York clients were all based upon FFC’s misrepresentations, and none of those clients successfully adopted a child through FFC. Thus, the amount of loss attributable to Chandi exceeds $500,000, regardless of whether she may be held accountable for losses suffered by clients whom FCC acquired prior to June 1992. Accordingly, we hold that the district court’s finding as to the total amount of loss attributable to Chandi was not clearly erroneous.
Identification of victims in the restitution order
The government argues, on cross-appeal, that the district court abused its discretion in fashioning its restitution order. First, the government argues that, despite its re*1389peated and specific requests, the district court failed to identify by name the victims of the ordered restitution (i.e., the payees). The government maintains that it is implicit throughout the language of 18 U.S.C. § 3663 and U.S.S.G. § 5E1.1 (restitution), that, when restitution is ordered, the victim must be specifically named. For example, the government points out that, under 18 U.S.C. § 3663(h)(2), a victim named in a restitution order may enforce that order as a civil judgment; however, when no victim is specifically named in a restitution order, it is “problematic,” if not impossible, for victims to enforce the restitution order in the event that the defendant fails to pay. Moreover, the government points out, U.S.S.G. § 5E1.1 (emphasis added) provides that the restitution order “should specify the manner in which, and the persons to whom, payment is to be made.”
In the present case, the restitution section of each defendant’s written judgment contains the following statement directly beneath the heading “Name of Payee”: “Information to be submitted by the probation office.” Designated Record at 169 (Stover’s judgment); id. at 174 (Chandi’s judgment). Upon careful review of the law governing the district court’s authority to order restitution, we hold that the district court lacked authority to leave the designation of the payee or payees entirely to the discretion of the probation office, as indicated in the written judgments. As a general rule, the district courts should designate the recipient or recipients when ordering restitution pursuant to 18 U.S.C. § 3663. See United States v. Miller, 900 F.2d 919, 922-24 (6th Cir.1990) (vacating sentences and remanding to the district court for, among other things, clarification of “whom it has found to be a victim entitled to restitution payments and the amount of restitution each victim is to be paid”); accord United States v. Seligsohn, 981 F.2d 1418, 1423 (3d Cir.1992) (“the court should designate recipients of the restitution”). The Sixth Circuit’s decision in Miller also raises the important point that a lack of clarity with respect to victim identity and the amount of restitution owed to each victim may complicate other matters — for example, determinations of whether restitution has been offset by payment of a civil judgment, or vice versa. 900 F.2d at 922. In Seligsohn, 981 F.2d at 1424, the Third Circuit held that
the unguided discretion to determine who are “victims” should not be entrusted to either the United States Attorney or the Probation Office....
... [T]he designation of those eligible should be made by name where that is possible. Where names are unknown, designations can be made by a description specific enough to provide appropriate guidance for the United States Attorney [or the Probation Office] in determining those entitled to share in the proceeds.... When the total available funds will be insufficient to pay all victims, a court should also devise a system of equitable priority or pro ration.
In the present case, we direct the district court, on remand, to identify the payees in the restitution order and to specify either the amounts to be paid each victim or an appropriate method of equitable distribution.
Amount of restitution as to Stover
The government separately argues that the amount of restitution ordered by the district court, $40,000, is inadequate as to Stover, particularly in light of the parties’ acknowledgement in the plea agreement of the district court’s authority to order restitution up to and including the full amount of the loss, in accordance with 18 U.S.C. § 3663(a)(3) (the court may order restitution in any criminal case to the extent agreed to by the parties in a plea agreement). The government notes that it twice asked the district court to order restitution in the full amount of the loss (which was undisputed), or, in the alternative, to order restitution in the amount of defendants’ ill-gotten gains. As to Stover, the government maintains, this latter amount was nearly $250,000, based upon the salary and expense money she received dining the relevant time period. The government concludes that the amount of restitution which Stover was ordered to pay, $40,000, was so inadequate as to constitute an abuse of discretion.
We agree with the government that the amount of restitution Stover was ordered to *1390pay is low. However, in light of the wide discretion ordinarily afforded the district courts in determining the amount of restitution, United States v. Berndt, 86 F.3d 803, 809 (8th Cir.1996), we hold that the district court did not abuse its discretion in ordering Stover to pay $40,000, jointly and severally with Chandi.
Conclusion
For the foregoing reasons, we reverse the district court’s imposition of a two-level upward adjustment under U.S.S.G. § 3A1.1 for each defendant; affirm the ten-level increase to Chandi’s base offense level under U.S.S.G. § 2F1.1; and affirm the amount of restitution ordered to be paid by Stover. We vacate each defendant’s sentence with directions to the district court to resentence defendants without the vulnerable victim enhancement and to amend the restitution orders in accordance with this opinion.
. Pursuant to guideline amendments effective November 1, 1995, this guideline provision now appears in U.S.S.G. § 3Al.l(b).
. There is some disagreement over whether Chandi participated in founding FFC. She claims that she did not become involved in FFC until 1992. The indictment alleges that Stover and Chandi founded FFC together in 1988, and Stover refers to herself and Chandi as “co-founders” of FFC.
. In the current version of the Guidelines Manual, reflecting amendments that became effective November 1, 1995, these examples are contained in application note'2. U.S.S.G. § 3A1.1, comment. (n. 2) (Nov. 1995).
. Under the current version of the guidelines, reflecting the amendments effective November 1, 1995, a defendant who intentionally selects a victim because of actual or perceived race, color, religion, national origin, ethnicity, gender, disability, or sexual orientation is now subject to a 3-level enhancement under a separate subsection of § 3A1.1, entitled “Hate Crime Motivation.” U.S.S.G. § 3Al.l(a) (Nov. 1995).