OPINION OF THE COURT
STAPLETON, Circuit Judge:The City of Philadelphia (City) and intervening defendant United Minority Enterprise Associates (UMEA) appeal from the *591district court’s judgment declaring that the City’s set-aside program for black construction contractors, Phila. Code § 17-500 et. seq. (Chapter 17-500 or Ordinance), violates the Equal Protection rights of the Contractors Association of Eastern Pennsylvania (CAEP) and eight other contracting associations (Contractors). We agree with the district court that Chapter 17-500 is not narrowly tailored to serve a compelling state interest, and we will affirm.
I.
Chapter 17-500 was adopted by the Philadelphia City Council in November of 1982. As originally enacted, it contained a sunset provision calling for it to expire after seven years. In May of 1987, Council extended the Ordinance until January 1, 1998. In 1988, the program was amended, inter alia, to include the set-aside for handicapped persons.
The preamble of Chapter 17-500 set forth only the following general legislative findings:
WHEREAS, ... in the City of Philadelphia Minorities constitute approximately forty-six percent of the population including thirty-eight percent Black, five percent Spanish surname, two percent Oriental and one percent Indian and women constitute approximately fifty percent of the population; and
WHEREAS, A pattern of past and present racial, sexual and economic discrimination have unfairly limited the ability of Minority and Female Owned Businesses to compete for an equitable share of such contracts with the City of Philadelphia;- and
WHEREAS, The citizens of Philadelphia share a commitment to the eradication of present manifestations of such discrimination; and
•I» sfc
WHEREAS, A series of goals for Minority and Female Owned Business participation would increase such participation and eradicate such manifestations and so that they will reach economic parity with majority businesses at large....
The preamble of the 1987 Ordinance extending the program includes the following additional legislative findings:
WHEREAS, Economic parity with majority businesses at large has not yet been reached, and based on the progress already demonstrated, cannot be reached within the remaining effective period of this Chapter; and
WHEREAS, The Council desires to extend the effective period of this Chapter in order to more fully implement its goals program in the economic development of Minority and Female Owned Businesses, and to help eliminate the large amount of unemployment among the City’s population — particularly its minority population. ...
No additional legislative finding relevant to the issues before us has been made.
Chapter 17-500, in its current form, seeks to increase the participation of “disadvantaged business enterprises” (DBEs) in City contracting. DBEs are businesses defined as those at least 51% owned by “socially and economically disadvantaged” persons. “Socially and economically disadvantaged” persons are, in turn, defined as “individuals who have ... been subjected to racial, sexual or ethnic prejudice because of their identity as a member of a group or differential treatment because of their handicap without regard to their individual qualities, and whose ability to compete in the free enterprise system has been impaired due to diminished capital and credit opportunities as compared to others in the same business area who are not socially disadvantaged.” Phila. Code § 17-501(11). As we found in Contractors Ass’n of Eastern Pa. v. City of Philadelphia, 6 F.3d 990, 999 (3d Cir.1993) (Contractors II),1 this definition “includes only individuals who are both victims of prejudice based on status and economically deprived.” Businesses majority-*592owned by racial minorities (minority business enterprises or MBEs) and women are rebutt-ably presumed to be DBEs, see § 17-501(11)(a), but businesses that would otherwise qualify as DBEs are rebuttably presumed not to be DBEs if they have received more than $5 million in City contracts.
Chapter 17-500 sets participation “goals” for different categories of DBEs: racial minorities (15%), women (10%) and handicapped (2%). These percentage goals are percentages of the total dollar amount spent by the City in each of the three contract categories: vending contracts, construction contracts, and personal and professional service contracts. Dollars received by DBE subcontractors in connection with City financed prime contracts are counted towards the goals as well as dollars received by DBE prime contractors.2
The 1982 Ordinance created the Minority Business Enterprise Council (MBEC) to oversee the set-aside program, promulgate regulations, and certify the eligibility of contractors to participate.3 Under the program, each City agency is expected to meet the participation goal in each type of contract. The MBEC or an agency may recommend exempting individual contracts or groups of contracts from Chapter 17-500’s requirements when there are insufficient DBEs in the Philadelphia Standard Metropolitan Statistical Area to ensure adequate competition and reasonable prices for the contracts. § 17-505(1) & (2).
Chapter 17-500 and its implementing regulations call for City agencies and the MBEC to formulate an annual plan for achieving the established goal in the construction area. See MBEC Reg. §§ 5, 11. Two different strategies are authorized. When there are sufficient DBEs qualified to perform a City contract to ensure competitive bidding, a contract can be let on a sheltered market basis — i.e., only DBEs will be permitted to bid. In other instances, the contract will be let on a non-sheltered basis — i.e., any firm may bid — with the goals requirements being met through subcontracting. The sheltered market strategy has seen little use. It was attempted on a trial basis, but there were too few DBEs in any given area of expertise to ensure reasonable prices, and the program was abandoned.4 See App. at 262-63. Evidence submitted by the City indicates that no construction contract was let on a sheltered market basis from 1988 to 1990, see App. at 4493, and there is no evidence that the City has since pursued that approach. Consequently, the Ordinance’s participation goals have been achieved almost entirely by insisting that bidding prime contractors subcontract work to DBEs in accordance with the goals.
When the goals are to be achieved by imposing subcontracting requirements, each would-be prime contractor must submit a “Schedule for Participation” (Schedule) of DBEs or a “Request for Waiver.” See MBEC Reg. § 6.1. A Schedule details the names and addresses of participating DBE subcontractors, the type and amount of work they are to perform, and the dollar value of their services. A Request for Waiver consists of a statement that the contractor has made a good faith effort to utilize DBEs but has failed to meet the goals for the contract, along with documentation of the good faith effort and a list of those DBEs with whom the contractor was able to make commitments. § 6.1(C).
*593Compliance with Chapter 17-500’s goals is to be considered “an element of responsiveness” of the bid when an agency awards a contract. § 6.1(A). The significance of complying with the goals is determined by a series of presumptions. Where at least one bidding contractor submits a satisfactory Schedule for Participation, it is presumed that all contractors who did not submit a satisfactory Schedule did not exert good faith efforts to meet the program goals, and the “lowest responsible, responsive contractor” gets the contract. § 6.1(D). Where none of the bidders submits a satisfactory Schedule, it is presumed that all but the bidder who proposes “the highest goals” of DBE participation at a “reasonable price” did not exert good faith efforts, and the contract is awarded to the “lowest, responsible, responsive contractor” who is granted a Waiver and proposes the highest level of DBE participation at a reasonable price. § 6.1(E). Non-complying bidders in either situation must rebut the presumption in order to secure a waiver.
The district court found that the practical effect of the regulations and the system of presumptions is to create a protected segment of City construction work for which non-DBE contractors could not compete. See Contractors Ass’n v. City of Philadelphia, 893 F.Supp. 419, 426 (E.D.Pa.1995): During the late 1970s, the contractors who obtained the contracts for the bulk of all public works contracts in the Philadelphia SMSA tended to use their own workforces and equipment to perform the necessary work, and subcontractors were used only when special skills were required. Id. After the enactment of Chapter 17-500, non-MBE contractors had to engage in “creative subcontracting,” carving out portions of the contract for DBEs that they would otherwise have done themselves. Id. at 426 n. 5.
The City’s own records substantiate the effect of Chapter 17-500. MBEs have participated in nearly every public works contract let by the City, usually at or near the participation goal.5 Prime contracts awarded to MBEs constituted only a small fraction of the contract amounts counted towards the participation goal.6 Chapter 17-500 thus is essentially a subcontracting set-aside program.
II.
This appeal is our third occasion to consider this challenge to Chapter 17-500. On the first appeal, we affirmed the district court’s ruling that the Contractors had standing to challenge the set-aside program, but reversed the grant of summary judgment in their favor because UMEA had not been afforded a fair opportunity to develop the record. See Contractors Ass’n of Eastern Pa. v. City of Philadelphia, 945 F.2d 1260 (3d Cir.1991) (Contractors I). On the second appeal, we reviewed a second grant of summary judgment for the Contractors. See Contractors II, 6 F.3d 990. We first concluded that the Contractors had standing to challenge the program only as it applied to the award of construction contracts. We then held that the pre-enactment evidence available to the City Council in 1982 did “not provide a sufficient evidentiary basis” for a conclusion that there had been discrimination against women and minorities in the construction industry. 6 F.3d at 1003. We further held, however, that evidence of discrimination obtained after 1982 could be considered in determining whether there was a sufficient evidentiary basis for Chapter 17-500.
After evaluating both the pre-enactment and post-enactment evidence in the summary judgment record, we affirmed the grant of summary judgment insofar as it declared to *594be unconstitutional those portions of the program requiring set-asides for women and non-black minority contractors. We further held that the two percent set-aside for the handicapped passed rational basis review and ordered the court to enter summary judgment for the City with respect to that portion of the program. Finally, we concluded that the portions of the program requiring a set-aside for black contractors could stand only if they met the “strict scrutiny” standard of Equal Protection review and that the record reflected a genuine issue of material fact as to whether they were narrowly tailored to serve a compelling interest of the City as required under that standard. The present appeal follows a nine-day bench trial and a resolution by the district court of the issues thus presented. That trial and this appeal thus concern only the constitutionality of Chapter 17-500’s preferences for black contractors.
III.
At trial, the City presented the work of Dr. Andrew F. Brimmer.7 That work, which was done in 1992 after the filing of this suit, is reflected in two pretrial affidavits (Brimmer Affidavits I & II),8 and his trial testimony. The core of his analysis concerning discrimination by the City centered on disparity indi-ces prepared using data from fiscal years 1979-81. The disparity indices were calculated by dividing the percentage of all City construction dollars received by black construction firms by their percentage representation among all area construction firms, multiplied by 100.9 Dr. Brimmer testified *595that the disparity index for black construction firms in the Philadelphia metropolitan area for the period studied was about 22.5. According to Dr. Brimmer, the smaller the resulting figure was, the greater the inference of discrimination, and he believed that 22.5 was a disparity attributable to discrimination.10 A number of witnesses testified to discrimination in City contracting before the City Council, prior to the enactment of Chapter 17-500. Dr. Brimmer believed that his statistical evidence was corroborated by their testimony.
Based on information provided in an affidavit by John Macldin, a former City employee, Dr. Brimmer also concluded that black representation in contractor associations was disproportionately low in 1981 and that between 1979 and 1981 black firms had received no subcontracts on City-financed construction projects.
The City also offered evidence concerning two programs instituted by others prior to 1982 which were intended to remedy the effects of discrimination in the construction industry but which, according to the City, had been unsuccessful. The first was the Philadelphia Plan, a program initiated in the late 1960s to increase the hiring of minorities on public construction sites. The second was a series of programs implemented by the Philadelphia Urban Coalition, a non-profit organization (Urban Coalition programs). These programs were established around 1970, and offered loans, loan guarantees, bonding assistance, training, and various forms of non-financial assistance concerning the management of a construction firm and the procurement of public contracts. According to a former City Council member, Joseph Coleman, and the former General Counsel of the Urban Coalition, Oscar Gas-kins, although both programs had some successes, neither program succeeded in eradicating the effects of discrimination.
The City pointed to the waiver and exemption sections of the Ordinance as proof that there was adequate flexibility in its program. Finally, the City contended that its fifteen percent goal was appropriate. Although greater than the percentage of minority contractors (2.4%) and black contractors (0.7%) in the Philadelphia SMSA, it is not tied directly to the proportion of minorities (41.8%) or blacks (37.9%) in the local population. The City maintained that the goal of fifteen percent may be required to account for waivers and exemptions allowed by the City, is a flexible goal rather than a rigid quota in light of the waivers and exemptions allowed by the Ordinance, and is justified in light of the discrimination in the construction industry.
The Contractors presented testimony from an expert witness, Dr. LaNoue,11 challenging the validity and reliability of Dr. Brimmer’s work and conclusions. Dr. LaNoue testified to many problems in the design of Dr. Brim-mer’s study, including, inter alia, the data used, the assumptions underlying the study, and the failure to include federally-funded contracts let through the City Procurement Department.
The Contractors relied heavily on the legislative history of Chapter 17-500, pointing out that it reflects no identification of any specific discrimination against black contractors and, indeed, no data from which a Council person could find that specific discrimination against black contractors existed or that Chapter 17-500 was an appropriate remedy for any such discrimination. They pointed as well to the absence of any consideration of race-neutral alternatives by the City Council *596prior to enacting the Ordinance. On cross-examination of Oscar Gaskins, the Contractors elicited testimony that indicated that the Urban Coalition programs were relatively successful, undermining the contention that race-based preferences were needed. Finally, the Contractors argued that the fifteen percent figure must have been simply picked from the air and had no relationship to any legitimate remedial goal because the City Council had had no evidence of identified discrimination before it.
At the conclusion of the trial, the district court made extensive findings of fact and conclusions of law. It determined that the record reflected no “strong basis in evidence” for a conclusion that discrimination against black contractors was practiced by the City, non-minority prime contractors, or contractors associations during any relevant period. 893 F.Supp. at 447. The court also determined that Chapter 17-500 was “not ‘narrowly tailored’ to even the perceived objective declared by City Council as the reason for the Ordinance.” Id. at 441.
IV.
We review the district court’s findings of fact for clear error. Sheet Metal Workers, Local 19 v. 2300 Group, Inc., 949 F.2d 1274, 1278 (3d Cir.1991). This includes the district court’s determination regarding the accuracy of the facts or assumptions on which the expert testimony was based. Concrete Works of Colorado v. City & Cty. of Denver, 36 F.3d 1513 (10th Cir.1994), cert. denied, — U.S. -, 115 S.Ct. 1315, 131 L.Ed.2d 196 (1995). However, “[u]ltimately, whether a strong basis in evidence of past or present discrimination exists, thereby establishing a compelling state interest for the municipality to enact a race-conscious ordinance, is a question of law,” subject to plenary review. Id. at 1522. The same is true of the issue of whether there is a strong basis in evidence for concluding that the scope of the ordinance is narrowly tailored to remedy the identified past or present discrimination.
V.
The threshold issue raised by the parties concerns the burden of persuasion with respect to the issues tried by the district court. The City insists that the district court placed that burden upon it and that, in doing so, the district court ignored our holding in Contractors II. We disagree. To explain our disagreement, we need to begin with a brief summary of the governing Equal Protection Law.
The current law of the land concerning the Equal Protection Clause and municipal, race-based, affirmative action programs is set forth in Justice O’Connor’s opinion in City of Richmond v. J. A. Croson Co., 488 U.S. 469, 109 S.Ct. 706, 102 L.Ed.2d 854 (1989). All such affirmative action programs, when challenged, must be subjected to “strict scrutiny” review.12 Id. at 493, 109 S.Ct. at 721 (O’Connor, J., for four justices); id. at 520, 109 S.Ct. at 735 (Scalia, J., concurring). Accordingly, a program can withstand a challenge only if it is narrowly tailored to serve a compelling state interest. The municipality has a compelling state interest that can justify race-based preferences only when it has acted to remedy identified present or past discrimination in which it engaged or was a “passive participant;” race-based preferences cannot be justified by reference to past “societal” discrimination in which the municipality played no material role. Moreover, the remedy must be tailored to the discrimination identified. To require less as a
basis for rigid racial preferences would be to open the door to competing claims for ‘remedial relief for every disadvantaged group. The dream of a Nation of equal citizens in a society where race is irrelevant to personal opportunity and achievement would be lost in a mosaic of shifting *597preferences based on inherently unmeasurable claims of past wrongs.
Id. at 505-06, 109 S.Ct. at 728.
“Racial classifications are suspect, and that means that simple legislative assurances of good intention cannot suffice.” Id. at 500, 109 S.Ct. at 725. The municipality must justify its conclusions regarding discrimination in connection with the award of its construction contracts and the necessity for a remedy of the scope chosen. While this does not mean that the municipality must convince the court of the accuracy of its conclusions, it does mean that the program cannot be sustained unless there is a strong basis in evidence for those conclusions. The party challenging the race-based preferences can succeed by showing either (1) that the subjective intent of the legislative body was not to remedy race discrimination in which the municipality played a role, or (2) that there is no “strong basis in evidence” for the conclusions that race-based discrimination existed and that the remedy chosen was necessary. Id. at 500, 510, 109 S.Ct. at 725, 730 (quoting Wygant v. Jackson Bd. of Educ., 476 U.S. 267, 277, 106 S.Ct. 1842, 1849, 90 L.Ed.2d 260 (1986)).
The purpose and importance of this “strict scrutiny” review are spelled out in Croson:
Absent searching judicial inquiry into the justification for such race-based measures, there is simply no way of determining what classifications are “benign” or “remedial” and what classifications are in fact motivated by illegitimate notions of racial inferiority or simple racial politics. Indeed, the purpose of strict scrutiny is to “smoke out” illegitimate uses of race by assuring that the legislative body is pursuing a goal important enough to warrant use of a highly suspect tool. The test also ensures that the means chosen “fit” this compelling goal so closely that there is little or no possibility that the motive for the classification was illegitimate racial prejudice or stereotype.
Classifications based on race carry a danger of stigmatie harm. Unless they are strictly reserved for remedial settings, they may in fact promote notions of racial inferiority and lead to a politics of racial hostility.
Id. at 493, 109 S.Ct. at 722.
Although the Court did not speak directly in Croson to the allocation of the burdens of proof and persuasion, this subject was addressed by this Court in Contractors II, 6 F.3d at 1005-07, and, more recently, by the Tenth Circuit Court of Appeals in Concrete Works of Colorado, 36 F.3d at 1521-23. Both courts concluded that when the race-based classifications of an affirmative action plan are challenged, the proponents of the plan have the burden of coming forward with evidence providing a firm basis for inferring that the legislatively identified discrimination in fact exists or existed and that the race-based classifications are necessary to remedy the effects of the identified discrimination. Once the proponents of the program meet this burden of production, the opponents of the program must be permitted to attack the tendered evidence and offer evidence of their own tending to show that the identified discrimination did or does not exist and/or that the means chosen as a remedy do not “fit” the identified discrimination.
Ultimately, however, the plaintiffs challenging the program retain the burden of persuading the district court that a violation of the Equal Protection Clause has occurred. This means that the plaintiffs bear the burden of persuading the court that the race-based preferences were not intended to serve the identified compelling interest or that there is no strong basis in the evidence as a whole for the conclusions the municipality needed to have reached with respect to the identified discrimination and the necessity of the remedy chosen.
The significance of the allocation of the burden of persuasion differs depending on the theory of constitutional invalidity that is being considered. If the theory is that the race-based preferences were adopted by the municipality with an intent unrelated to remedying its past discrimination, the plaintiff has the burden of convincing the court that the identified remedial motivation is a pretext and that the real motivation was something else. As we noted in Contractors II, *598the burden of persuasion here is analogous to the burden of persuasion in Title VII cases. 6 F.3d at 1006. The ultimate issue under this theory is one of fact, and the burden of persuasion on that ultimate issue can be very important.
The situation is different when the plaintiffs theory of constitutional invalidity is that, although the municipality may have been thinking of past discrimination and a remedy therefor, its conclusions with respect to the existence of discrimination and the necessity of the remedy chosen have no strong basis in evidence. In such a situation, when the municipality comes forward with evidence of facts alleged to justify its conclusions, the plaintiff has the burden of persuading the court that those facts are.not accurate. The ultimate issue as to whether a strong basis in evidence exists is an issue of law, however. The burden of persuasion in the traditional sense plays no role in the court’s resolution of that ultimate issue.
In Contractors II, we were presented with a summary judgment issue. We sustained the district court’s grant of summary judgment with respect to some of plaintiffs’ claims and reversed it as to others. We concluded that the City, in response to the plaintiffs’ motion for summary judgment, had come forward with no evidence that, if credited after testing in the crucible of trial, would support an inference that there had been discrimination by the City in the contracting industry against Hispanic, Asian-American, or Native American persons or women. We held that “plausible hypotheses are not enough to satisfy strict scrutiny, even at the summary judgment stage.” Id. at 1008. Accordingly, summary judgment was sustained with respect to the set-asides for those groups. On the other hand, we held that Dr. Brimmer’s expert analysis along with the City’s anecdotal evidence, if credited, could provide a firm basis for the inference that there had been discrimination against blacks in connection with City-financed construction contracts and that the program was narrowly tailored to remedy the effects of such discrimination. We therefore reversed the contractors’ summary judgment regarding the set-aside for black contractors and remanded for a trial.
We agree with the City that Contractors II established as the law of the case the admissibility of Dr. Brimmer’s analysis. Neither this nor our observations regarding the burden of persuasion, however, required the district court to accept that analysis and Dr. Brimmer’s opinions uncritically. Like any other trier of fact called upon to evaluate expert testimony, it was entitled to accept or reject the facts and assumptions upon which the expert testimony was based and to evaluate the methodology that underlay it. And, of course, the district court was required in the final analysis to determine whether Dr. Brimmer’s analysis and the City’s anecdotal evidence, when viewed in the context of the Contractors’ evidence, constituted the “strong basis in evidence” required by the Equal Protection Clause.
The district court’s opinion explicitly demonstrates its recognition that the plaintiffs bore the burden of persuading it that an equal protection violation occurred.13 As we read the opinion, it applied the appropriate burdens of production and persuasion and conducted the required evaluation of the evidence. As we have indicated, it examined the credited record evidence as a whole and concluded that the “strong basis in evidence” for the City’s position did not exist.14
*599YL
As we have indicated, “[w]hile the States and their subdivisions may take remedial action when they possess evidence that their own spending practices are exacerbating a pattern of prior discrimination, they must identify that discrimination, public or private, with some specificity before they may use race-conscious relief.” Croson, 488 U.S. at 504, 109 S.Ct. at 727. It necessarily follows that a court cannot conduct the strict scrutiny review required by Croson without first identifying with specificity the discrimination allegedly giving rise to the compelling state interest.
In November of 1982, the City Council found only in general terms that a pattern of discrimination against minority contractors existed that had resulted in their receiving less than their “equitable share” of City construction dollars. Subsequently, in the course of this litigation, however, several distinct forms of racial discrimination have been advanced as producing that result. The first is discrimination by prime contractors in the awarding of subcontracts. The second is discrimination by contractor associations in admitting members. The third is discrimination by the City in the awarding of prime contracts. The City would, of course, be directly responsible for the last form of discrimination. The City and UMEA insist that the City may have “passively participated” in the other two forms of discrimination.
Since these three forms of discrimination involve discrimination by different decisionmakers in different markets, we must independently determine whether there is a strong basis in the evidence for believing that each particular form of discrimination existed. In determining whether Chapter 17-500 is narrowly tailored, we can look only to the form or forms of discrimination for which a firm evidentiary basis exists.15
A. The Evidence of Discrimination by Private Prime Contractors
One of the City’s theories is that discrimination by prime contractors in the selection of subcontractors existed and may be remedied by the City.16 As Justice O’Connor observed in Croson:
if the city could show that it had essentially become a “passive participant” in a system of racial exclusion practiced by elements of the local construction industry, ... the city could take affirmative steps to dismantle such a system. It is beyond dispute that any public entity ... has a compelling government interest in assuring that public dollars ... do not serve to finance the evil of private prejudice.
488 U.S. at 492, 109 S.Ct. at 721.
Dr. Brimmer’s disparity study focuses on just one aspect of the Philadelphia construc*600tion industry — the award of prime contracts by the City. He acknowledged that the only information he had about subcontracting came from an affidavit of John W. Macklin supplied to him in the course of his study. As he stated on cross-examination, “I have made no presentation to the Court as to participation by black minorities or blacks in subcontracting. I couldn’t possibly — either misleading or misinforming the Court because I didn’t do it.” App. at 1422.
The only record evidence with respect to black participation in the subcontracting market comes from Mr. Macklin. Based on a review of City records, found by the district court to be “cursory,” Mr. Macklin reported that not a single subcontract was awarded to minority subcontractors in connection with City-financed construction contracts during fiscal years 1979 through 1981. The district court did not credit this assertion.
Mr. Macklin was a member of the MBEC staff and a proponent of Chapter 17-600. He had earlier served within the City’s Office of Minority Opportunity (OMO) from 1981 to 1983.17 Prior to 1982, for solely City-financed projects, the City did not require subcontractors to prequalify, did not keep consolidated records of the subcontractors working on prime contracts let by the City, and did not record whether a particular contractor was an MBE. To prepare a report concerning the participation of minority businesses in public works, Mr. Macklin examined the records at the City’s Procurement Department. The department kept procurement logs, project engineer logs, and contract folders. The subcontractors involved in a project were only listed in the engineer’s log. Mr. Macklin’s testimony concerning his methodology was hesitant and unclear, but it does appear that he examined only 25 to 30 percent of the project engineer logs and that his only basis for identifying a name in that segment of the logs as an MBE was his personal memory of the information he had received in the course of approximately a year of work with the OMO. The district court found the following from Mr. Macklin’s testimony:
[Macklin] went to the contract files and looked for contracts in excess of $30,000.00 that in his view appeared to provide opportunities for subcontracting. (Id. at 13.) With that information, Macklin examined some of the project engineer logs for those projects to determine whether minority subcontractors were used by the prime contractors. (Id.) Macklin did not look at every available project engineer log. (Id.) Rather, he looked at a random 25 to 30 percent of all the project engineer logs. (Id.) As with his review of the Procurement Department log, Macklin determined that a minority subcontractor was used on the project only if he personally recognized the firm to be a minority. (Id.) Quite plainly, Macklin was unable to determine whether minorities were used on the remaining 65 to 70 percent of the projects that he did not review. When questioned whether it was possible that minority subcontractors did perform work on some City public works projects during fiscal years 1979 to 1981, and that he just did not see them in the project logs that he looked at, Macklin answered “it is a very good possibility.”
893 F.Supp. at 434.
The district court found two other portions of the record significant on this point. First, during the trial, the City presented Oscar Gaskins (“Gaskins”), former general counsel to the General and Specialty Contractors Association of Philadelphia (“GASCAP”) and the Philadelphia Urban Coalition, to testify about minority participation in the Philadelphia construction industry during the 1970s and early 1980s. Gaskins testified that, in his opinion, black contractors are still being subjected to racial discrimination in the private construction industry, and in subcontracting within the City limits. However, when Gaskins was asked by the court to identify even one instance where a minority contractor was denied a private contract or *601subcontract after submitting the lowest bid, Gaskins was unable to do so.
Second, the district court noted that since 1979 the City’s “standard requirements warn [would-be prime contractors] that discrimination will be deemed a ‘substantial breach’ of the public works contract which could subject the prime contractor to an investigation by the Commission and, if warranted, fines, penalties, termination of the contract and forfeiture of all money due.” 893 F.Supp. at 438-39. Like the Court in Croson, the district court found significant the City’s inability to point to any allegations that this requirement was being violated. See id. (citing Croson, 488 U.S. at 503, n. 3, 109 S.Ct. at 726 n. 3).
The district court did not err by declining to accept Mr. Maeldin’s conclusion that there were no subcontracts awarded to black contractors in connection with City-financed construction contracts in fiscal years 1979 to 1981. Accepting that refusal, as we do, we agree with the district court’s conclusion that the record provides no firm basis for inferring discrimination by prime contractors in the subcontracting market during that period.
B. The Evidence of Discrimination by Contractor Associations
A city may seek to remedy discrimination by local trade associations to prevent its passive participation in a system of private discrimination. See Croson, 488 U.S. at 503, 109 S.Ct. at 726-27. Evidence of “extremely low” membership by MBEs, standing by itself, however, is not sufficient to support remedial action; the city must “link [low MBE membership] to the number of local MBEs eligible for membership.” Id.
Dr. Brimmer opined that there was statistically low representation of eligible MBEs in the local trade associations. He testified that, while numerous MBEs were eligible to join these associations, three such associations had only one MBE member, and one had only three MBEs. In concluding that there were many eligible MBEs not in the associations, however, Dr. Brimmer again relied entirely upon the work of Mr. Macklin. The district court rejected Dr. Brimmer’s conclusions because it found his reliance on Mr. Macklin’s work misplaced.
The information relied upon by Dr. Brim-mer is found only in an affidavit of Mr. Macklin that was admitted into evidence. In this affidavit, Mr. Macklin asserts that he “reviewed the membership provisions of the bylaws for each of the plaintiff Associations [and] the 1982 OMO Directory of Certified MBE and WBE firms to determine which of the firms were eligible to be members of plaintiff Associations.” App. at 4486. “Based on those reviews,” Mr. Macklin formed an opinion that a listed number of MBE and WBE firms were eligible to be members of the plaintiff Associations. Id.
Because Mr. Macklin did not set forth the criteria for association membership and because the OMO certification list did not provide any information about the MBEs and WBEs other than their names and the fact that they were such, the district court found itself without a basis for evaluating Mr. Macklin’s opinions.
On the other hand, the district court credited “the uncontroverted testimony of John Smith [a former general manager of the CAEP and member of the MBEC] that no black contractor who has ever applied for membership in the CAEP has been denied.” 893 F.Supp. at 440. The court noted as well that the City had not “identified even a single black contractor who was eligible for membership in any of the plaintiffs’ associations, who applied for membership, and was denied.” Id. at 441.
The district court was required to ascertain whether there is a strong basis in the evidence for concluding that there had been discrimination against black contractors by contractor associations. Given the City’s failure to present more than the essentially unexplained opinion of Mr. Macklin, the opposing, uncontradicted testimony of Mr. Smith, and the failure of anyone to identify a single victim of the alleged discrimination, it was appropriate for the court to conclude that a constitutionally sufficient basis was not established in the evidence.
Even if we were to accept Mr. Macklin’s opinions, however, we could not *602hold that Chapter 17-500 is justified by that discrimination. As we have explained, racial discrimination can justify a race-based remedy only if the City has somehow participated in or supported that discrimination. This record would not support a finding that this occurred. Contrary to the City’s argument, nothing in Croson suggests that awarding contracts pursuant to a competitive bidding scheme and without reference to association membership could alone constitute passive participation by the City in membership discrimination by contractor associations. Prior to 1982, the City let construction contracts on a competitive bid basis. It did not require bidders to be association members, and nothing in the record suggests that it otherwise favored the associations or their members.18 While City dollars went to low bidding contractors who, in many instances, paid dues to the Associations, this would not appear to us to constitute support for the membership practices of associations any more than the payment of City dollars to low bidding contractors who do business with discriminatory labor unions constitutes support for those unions. We know from Croson that the latter situation does not involve “passive participation” that will support a system of race-based preferences.19 See 488 U.S. at 498-99, 109 S.Ct. at 724-25.
C. The Evidence of Discrimination by the City
The record provides substantially more support for the proposition that there was discrimination on the basis of race in the award of prime contracts by the City in the fiscal 1979-1981 period. Moreover, we find the Contractors’ critique of that evidence less cogent than did the district court.
The centerpiece of the City’s evidence was Dr. Brimmer’s calculation of disparity indices which gauge the disparity in the award of prime contracts by the City. Following Cow-tractors II, Dr. Brimmer calculated a disparity index for black construction firms of 11.4, based on a figure of 114 such firms available to perform City contracts. At trial, he recognized that the 114 figure included black engineering and architecture firms, so he recalculated the index, using only black construction firms (i.e., 57 firms). This produced a disparity index of 22.5. Thus, based on Dr. Brimmer’s analysis, black construction firms would have to have received approximately 4.5 times more public works dollars than they did receive in order to have achieved an amount proportionate to their representation among all construction firms. Dr. Brimmer found the disparity sufficiently large to be attributable to discrimination against black contractors.
There are circumstances in which a disparity index of 22.5 can constitute a strong basis in evidence for inferring the existence of discrimination. See Associated General Contractors of Cal. v. Coalition for Economic Equity, 950 F.2d 1401, 1414 (9th Cir.1991) (finding disparity equivalent to index of 22.4 sufficient to raise inference of discrimination), cert. denied, 503 U.S. 985, 112 S.Ct. 1670, 118 L.Ed.2d 390 (1992); Cone Corp. v. Hillsborough County, 908 F.2d 908, 916 (11th Cir.) (finding disparity equivalent to index of approximately 50 sufficient to make prima facie case), cert. denied, 498 U.S. 983, 111 S.Ct. 516, 112 L.Ed.2d 528 (1990). We therefore turn to evaluate whether the record as a whole undermines the probative value of Dr. Brimmer’s disparity index analysis.
The district court concluded that Dr. Brim-mer’s analysis, when viewed in the context of the entire record, did not provide a strong basis in evidence for an inference of discrimination in the prime contract market. It reached this conclusion primarily for three reasons. Dr. Brimmer’s study, in the district court’s view, (1) did not take into account whether the black construction firms were *603qualified and willing to perform City contracts; (2) mixed statistical data from different sources; and (8) did not account for the “neutral” explanation that qualified black firms were too preoccupied with large, federally-assisted projects to perform City projects.
The district court was, of course, correct in concluding that a statistical analysis of the kind conducted by Dr. Brimmer should focus on the minority population capable of performing the relevant work. As Croson indicates, “[w]hen special qualifications are required to fill particular jobs, comparisons to the general population (rather than to the smaller group of individuals who possess the necessary qualifications) may have little probative value.” 488 U.S. at 501, 109 S.Ct. at 726 (quoting from Hazelwood School Dist. v. United States, 433 U.S. 299, 309 n. 13, 97 S.Ct. 2736, 2741 n. 13, 53 L.Ed.2d 768 (1977)). In both Croson and Hazelwood, however, the discussion concerning qualifications came in the context of a rejection of an analysis using the percentage of a particular minority in the general population. The issue of qualifications can be approached at different levels of specificity, however, and some consideration of the practicality of various approaches is required. An analysis is not devoid of probative value simply because it may theoretically be possible to adopt a more refined approach.
To the extent the district court found fault with Dr. Brimmer’s analysis for failing to limit its consideration to those black contractors “willing” to undertake City work, we find its criticism more problematic. In the absence of some reason to believe otherwise, one can normally assume that participants in a market with the ability to undertake gain-fid work will be “willing” to undertake it. Moreover, past discrimination in a marketplace may provide reason to believe the minorities who would otherwise be willing are discouraged from trying to secure the work.
It seems to us a substantial overstatement to assert that Dr. Brimmer failed to take into account the qualifications and willingness of black contractors to participate in public works. During the time period in question, fiscal years 1979-81, those firms seeking to bid on City contracts had to prequalify for each and every contract they bid on, and the criteria could be set differently from contract to contract. It would be highly impractical to review the hundreds of contracts awarded each year and compare them to each and every MBE. Dr. Brimmer chose instead to use as the relevant minority population the black firms listed in the 1982 OMO Directory. This would appear to be a reasonable choice that, if anything, may have been on the conservative side.
Edward Robinson, the director of the OMO in 1982, testified that although the federal set-aside program for the Urban Mass Transit Authority (“UMTA”) projects only required it to certify that a particular firm was a bona fide MBE, the OMO wanted to insure that the best possible MBEs participated in the federal projects. These projects were some of the largest public works in the area at the time. When a firm applied to be certified, the OMO required it to detail its bonding experience, prior experience, the size of prior contracts, number of employees, financial integrity, and equipment owned. The OMO visited each firm to substantiate its claims. Although this additional information did not go into the final directory, the OMO was confident that those firms on the list were capable of doing the work required on large scale construction projects. Of about 680 MBE and WBE firms that applied, only 241 were certified. Although Dr. Brim-mer did not go over the qualification of each and every firm in the OMO Directory, the process by which the firms were certified appears to suggest that, as a general proposition, those firms were both qualified and willing to participate in public works projects. Conversely, since City-funded contracts ran the gamut from small to large projects, the OMO list may have been under-inclusive in terms of firms capable of performing some portion of City projects.
The Contractors point to the small number of black firms that sought to prequalify for City-funded contracts as evidence that black firms were unwilling to work on projects funded solely by the City. During the time period in question, City records show that only seven black firms sought to prequalify, *604and only three succeeded in prequalifying. We think it would be inappropriate, however, to conclude that this evidence undermines Dr. Brimmer’s inference of discrimination. As Dr. Brimmer indicated in his testimony, if there has been discrimination in City contracting, it is to be expected that black firms may be discouraged from applying, and the low numbers may tend to corroborate the existence of discrimination rather than belie it. In a sense, to weigh this evidence for or against either party requires us to presume the conclusion to be proved.
While it is true that Dr. Brimmer “mixed data,” the weight given that fact by the district court seems excessive. Dr. Brimmer used data from only two sources in calculating the disparity index of 22.5. He used data that originated from the City to determine the total amount of contract dollars awarded by the City, the amount that went to MBEs, and the number of black construction firms. He “mixed” this with data from the Bureau of the Census concerning the number of total construction firms (8050) in the Philadelphia Standard Metropolitan Statistical Area (PSMSA). The data from the City is not geographically bounded to the same extent that the Census information is. Any firm could bid on City work, and any firm could seek certification from the OMO. Nevertheless, due to the burdens of conducting construction at a distant location, the vast majority of the firms were from the Philadelphia region and the Census data offers a reasonable approximation of the total number of firms that might vie for City contracts. Although there is a minor mismatch in the geographic scope of the data, given the size of the disparity index calculated by Dr. Brimmer, we are not persuaded that it is significant.
Considering Dr. Brimmer’s use of the OMO Directory and the Census data, it seems to us that the index of 22.5 may be a conservative estimate of the actual disparity. While Dr. Brimmer used a figure for black firms that took into account qualifications and willingness, he used a figure for total firms that did not. If he under-counted the number of black firms qualified and willing to undertake City construction contracts or over-counted the total number of firms qualified and willing to undertake City construction contracts, the actual disparity would be greater than 22.5. Further, while Dr. Brim-mer limited the index to black firms, he did not similarly reduce the dollars awarded to minority firms. He continued to use the figure of $667,501, which represented the total amount going to all MBEs. If minorities other than blacks received some of that amount, the actual disparity would again be greater.
Finally, we turn to the district court’s suggestion that the extensive participation of black firms in federally-assisted projects, which were also procured through the City’s Procurement Office, accounts for their low participation in the other construction contracts awarded by the City. In the fiscal years 1979-81, at least 22 black construction firms from the Philadelphia SMSA earned more than $18 million as subcontractors on 17 public works contracts let by the City. At least 15 of these contracts were federally funded.20 In fiscal year 1982, 20 to 25 minority firms, mostly black, won $50 million in subcontracts on the UMTA federally-assisted programs. The Contractors advanced these statistics to make three points: (a) that the City did not discriminate in procurement; (b) that black firms were unwilling to work on City projects because of their preoccupation with federal projects, and Dr. Brimmer’s work is flawed because it does not take this factor into account; and (e) that even if Dr. Brimmer’s analysis is probative, the preoccupation of black firms with federally assisted projects and their resulting unavailability for other City work projects provides a neutral explanation for his disparity.
The first contention may be readily dispensed with. Most, if not all, federally-funded projects were subject to a federal ten-percent set-aside, and the amount awarded to minorities was very near ten percent of the total amount awarded to all contractors. The evidence does not show, therefore, that the City did not discriminate in connection with contracts where its choice was unre*605strained; rather, it simply shows that the City complied with federal requirements on federally-assisted projects.
The second and third contentions are essentially the same point, and raise the most difficult issues for the City. It would not be surprising that some of the largest and most experienced black firms would gravitate towards the large federally-assisted projects, where the subcontracts were likely to be sizable and there was a federal set-aside. Indeed, the record suggests that of the 57 black contractors considered qualified for City contracts by Dr. Brimmer, 20 to 22 succeeded in winning federally funded subcontracts and devoted their resources to federally funded projects. This work undoubtedly affected, to some degree, their ability to perform other City contracts. The court was thus undoubtedly right in suggesting that the availability of substantial amounts of federally funded work and the federal set-aside undoubtedly had an impact on the number of black contractors available to bid on other City contracts.
The extent of that impact is more difficult to gauge, however. That such an impact existed does not necessarily mean that Dr. Brimmer’s analysis is without probative force. If, for example, one reduces the 57 available black contractors by the 20 to 22 that participated in federally assisted projects in fiscal years 1979-81 and takes 35 as a fair approximation of the black contractors available to bid on the remaining City work, Dr. Brimmer’s analysis produces a disparity index of 37, a disparity that still suggests a substantial under-participation of black contractors among the successful bidders on City prime contracts. On the other hand, the City’s difficulty in the 1980s in getting sufficient numbers of black contractors to bid even on sheltered market prime contracts gives credence to the hypothesis that that under-participation was attributable to a dearth, however it came about, of black contracting capacity available for prime city contracts.21
Whether this record provides a strong basis in evidence for an inference of discrimination in the prime contract market is a close call. In the final analysis, however, it is a call that we find unnecessary to make, and we chose not to make it.22 Even assuming that the record presents an adequately firm basis for that inference, the judgment of the district court must be affirmed because Chapter 17-500 is clearly not narrowly tailored to remedy that discrimination.
VIL
As we have indicated, strict scrutiny review requires us to examine the “fit” between the identified discrimination and the remedy chosen in an affirmative action plan. Croson teaches that there must be a strong basis in evidence not only for a conclusion that there is, or has been, discrimination but also for a conclusion that the particular remedy chosen is made “necessary” by that discrimination. 488 U.S. at 510, 109 S.Ct. at 730.
Croson tells us that the degree of “fit” is important in this context for three reasons. First, the absence of a close fit renders suspect the claim that remedying the identified discrimination and resulting injury was the objective of the legislative measure. The fit must provide assurance with respect to purpose; the means chosen must fit the *606“compelling goal so closely that there is little or no possibility that the motive for the classification was illegitimate racial prejudice or stereotype.” Id. at 493, 109 S.Ct. at 721.
Second, the necessary effect of a race-based preference is the imposition of a race-based burden on citizens who have “personal rights” under the Equal Protection Clause “to be treated with equal dignity and respect.” Id. That burden can be justified, if at all, only if it is close to the minimum necessary to redress the discrimination justifying the preference.
Finally, Croson recognizes that those in the disfavored class are not the only ones that can be burdened by a race-based preference. The favored class and society may also be burdened if the remedy extends significantly beyond the justification for the preference. “Classifications based on race carry a danger of stigmatic harm. Unless they are strictly reserved for remedial settings, they may in fact promote notions of racial inferiority and lead to a politics of racial hostility.” Id.
With this background, we turn to the issue of whether the Ordinance is narrowly tailored to serve a compelling state interest. That issue is, of course, shaped by our prior conclusions regarding the absence of a strong basis in evidence reflecting discrimination by prime contractors in selecting subcontractors and by contractor associations in admitting members. This leaves as a possible justification for Chapter 17-500 only our assumption that the record provides a strong basis in evidence for believing the City discriminated against black contractors in the award of prime contracts during fiscal years 1979 to 1981. If the remedy reflected in the Ordinance cannot fairly be said to be necessary in light of the assumed discrimination in awarding prime construction projects, the Ordinance cannot stand. After careful scrutiny of the record, we hold, as did the district court, that the Ordinance is not narrowly tailored.
A. Inclusion of Preferences in the Subcontracting Market
The primary focus of the City’s Chapter 17-500 program is the market for subcontracts to perform work included in prime contracts awarded by the City. While the program includes authorization for the award of prime contracts on a “sheltered market” basis, that authorization has been sparsely invoked by the City. Its goal with respect to dollars for black contractors has been pursued primarily through requiring that bidding prime contractors subcontract to black contractors in stipulated percentages. As we have explained, the 15 percent participation goal and the system of presumptions, which in practice require non-black contractors to meet the goal on virtually every contract, result in a 15% set-aside for black contractors in the subcontracting market.
Here, as in Croson, “[t]o a large extent, the set aside of subcontracting dollars seems to rest on the unsupported assumption that white contractors simply will not hire minority firms.” 488 U.S. at 502, 109 S.Ct. at 726. Here, as in Croson, there is no firm eviden-tiary basis for believing that non-minority contractors will not hire black subcontractors. Rather, the evidence, to the extent it suggests that racial discrimination has occurred, suggests discrimination by the City’s Procurement Department against black contractors who were capable of bidding on prime City construction contracts. To the considerable extent that Chapter 17-500’s program seeks to constrain decision making by private contractors and favor black participation in the subcontracting market, it is ill-suited as a remedy for the discrimination identified.
We do not suggest that an appropriate remedial program for discrimination by a municipality in the award of primary contracts could never include a component that affects the subcontracting market in some way. We hold, however, that a program, like Philadelphia’s current one, which focuses almost exclusively on the subcontracting market, is not narrowly tailored to address discrimination by the City in the market for prime contracts.
B. The Amount of the Set-Aside in the Prime Contract Market
Having decided that the Ordinance is over-broad in its inclusion of subcontracting, we next consider whether the 15 percent goal is narrowly tailored to address discrimination in prime contracting.
*607The district court found as a fact that, prior to the adoption of Chapter 17-500, the City Council “never attempted to determine the scope of the injury it allegedly sought to remedy with the Ordinance.” 893 F.Supp. at 444 n. 23. As one of the drafters acknowledged, “there was not sufficient data available to [the] Council that would indicate the number of minority businesses that currently exist[ed] in Philadelphia and within the surrounding area that would be capable of performing the contracts that the City lets on an annual basis.” App. at 4687. As a Council person admitted, with admirable candor, in the course of the debate on the 1988 amendment creating a set-aside for the handicapped:
We didn’t do a study to see whether or not ten percent was proper for women or the percentage for minorities was correct. We knew that there were a lot of women out there and a lot of minorities out there and they weren’t getting their share. And we set a goal. And it was pretty arbitrary. It was pretty arbitrary.
App. at 5409-10.
The district court also found significant the fact that the preamble to the Ordinance offered only one reference point for the percentages selected for the various set-asides— the percentages of minorities and women in the general population. In addition to the above-quoted legislative findings, the court pointed to the following portions of the legislative history:
When Bill No. 1477 was introduced on September 16, 1982, the sponsors of the Ordinance stated in the preamble:
Goals. Our goals are reasonable and especially so in light of the actual proportion of minorities and females in the local population.
(emphasis in original). The sponsors further stated that, in light of the City’s 53.65 percent female population and 41.8 percent minority population, the Ordinance’s participation goals “reflect a very modest attempt to change what is by any measure a deplorable status quo.”
893 F.Supp. at 445 (citations omitted).
Finally, the district court noted that, until it invalidated Chapter 17-500 in 1990, the 15 percent preference for MBEs was shared by an “amalgam of minorities” including Hispanics, American Indians, Aleuts, Eskimos, Asians and Native Hawaiians. Id. Since there was no evidence that any of these groups suffered from discrimination, the court concluded that the 15 percent preference for minorities was “not a ‘narrowly tailored’ remedy for past discrimination.” Id.23
The record supports the district court’s findings that the Council’s attention at the time of the original enactment and at the time of the subsequent extension was focused solely on the percentage of minorities and women in the general population and that Council made no effort at either time to determine how the Ordinance might be drafted to remedy particular discrimination — to achieve, for example, the approximate market share for black contractors that would have existed, had the purported discrimination not occurred. While the City Council obviously did not tie the 15% participation goal directly to the proportion of minorities in the local population, the goal was either arbitrarily chosen or, at least, the Council’s sole reference point was the minority percentage in the local population.
More importantly, it is equally clear that the City, in the entire course of this litigation, has been unable to provide an evidentia-ry basis from which to conclude that a 15% set-aside is necessary to remedy discrimination against black contractors in the market for prime contracts. Dr. Brimmer’s data indicates that, at most, only 0.7% of the construction firms qualified to perform City-financed prime contracts in the 1979-1981 period were black construction firms.24 This indicates that the 15 percent figure chosen is an impermissible one.
We do not suggest that the percentage of the preferred group in the universe of qualified contractors is necessarily the ceiling for *608all set-asides. It well may be that some premium could be justified under some circumstances. It is nevertheless true that the only evidentiary basis in this record that appears at all relevant to fashioning a remedy for discrimination in the prime contracting market is Dr. Brimmer’s 0.7% figure. That figure does not provide a strong basis in evidence for concluding that a 15% set-aside was necessary to remedy discrimination against black contractors in the prime contract market.
Nor are we able to accept the City’s assertion that a 15% set-aside is necessary to provide an effective remedy because of the waiver and exemption provisions of the Ordinance. As the district court found, “very few contractors have ever received waivers of Chapter 17-500’s set-aside requirements, and even fewer City contracts have been exempted from the set-aside program.” 893 F.Supp. at 444. Indeed, the waivers and exemptions granted from July 1, 1987, to June 30,1990, constituted less than 2% of the City’s public works contracts. The legislative record evidences no expectation on the part of the Council that a material portion of the 15 percent goal chosen would be necessary to offset waivers and exemptions.
C. Program Alternatives That Are Either Race-Neutral Or Less Burdensome To Non-Minority Contractors.
In holding that the Richmond plan was not narrowly tailored, the Court in Croson considered it significant that race-neutral remedial alternatives were available and that the City had not considered the use of these means to increase minority business participation in City contracting. It noted, in particular, that barriers to entry like capital and bonding requirements could be addressed by a race-neutral program of city financing for small firms and could be expected to lead to greater minority participation. Nevertheless, such alternatives were not pursued or even considered in connection with the Richmond’s efforts to remedy past discrimination.
The district court here found that the City’s procurement practices created significant barriers to entering the market for City-awarded construction contracts. Small contractors, in particular, were deterred by the City’s prequalification and bonding requirements from competing in that market. Relaxation of those requirements, the district court found, was an available race-neutral alternative that would be likely to lead to greater participation by black contractors. No effort was made by the City, however, to identify barriers to entry in its procurement process and that process was not altered before or in conjunction with the adoption of Chapter 17-500.
The district court also found that the City could have implemented training and financial assistance programs to assist disadvantaged contractors of all races. As an example of the efficacy of such a program, it pointed to the Philadelphia Urban Coalition’s Minority Contractors Training and Assistance Program implemented in 1970 by the Urban Coalition. This program was designed to train minority contractors and help them obtain financing and bonding so that they could proceed “from one level to the next in the construction industry.” 893 F.Supp. at 441. The record established that this program had achieved substantial success in fulfilling its goals. The district court found, however, that the City had not supported this program and had not considered emulating and/or expanding this program in conjunction with the adoption of the Ordinance.25 Id. at 442.
In 1980, the City did establish the OMO to certify the credentials of minority contractors who were seeking construction opportunities from the UMTA, a federally funded project to build the Center City Commuter Rail Tunnel and the Airport High Speed Line. The OMO would conduct an investigation of a minority contractor and certify that it was minority owned. Although not required by the UMTA to do so, the OMO also scrutinized the qualifications of the MBE to ensure that it was capable of performing projects of the complexity and size involved with the UMTA. During fiscal 1982, the OMO succeeded in obtaining 20 to 25 construction contracts worth $50 million for certified MBEs. The district court found that despite *609this success the City abandoned the “OMO’s strict, but valuable, certification procedures when Chapter 17-500 was enacted in November 1982.” Id. at 443 n. 21. While we recognize that the federal set-aside played a very important role in the OMO achieving the level of minority participation that it did in the UMEA program in 1982, this does not undermine the district court’s conclusion regarding the value of a list of minority contractors with certified credentials.
The record provides ample support for the finding of the district court that alternatives to race-based preferences were available in November of 1982 which would have been either race neutral or, at least, less burdensome to non-minority contractors. The City could have lowered administrative barriers to entry, instituted a training and financial assistance program, and carried forward the OMO’s certification of minority contractor qualifications. The record likewise provides ample support for the court’s conclusion that the “City Council was not interested in considering race-neutral measures, and it did not do so.”26 Id. at 442. To the extent the City failed to consider or adopt these alternatives, it failed to narrowly tailor its remedy to prior or existing discrimination against black contractors.
We find it particularly noteworthy that Chapter 17-500, since its extension, in 1987, for an additional 12 years, has been targeted exclusively toward benefiting only minority and women contractors “whose ability to compete in the free enterprise system has been impaired due to diminished capital and credit opportunities as compared to others in the same business area who are not socially disadvantaged.” Phila. Code § 17-501(11). The City’s failure to consider a race-neutral program designed to encourage investment in and/or credit extension to small contractors or minority contractors seems particularly telling in light of the limited classification of victims of discrimination that Chapter 17-500 seeks to favor.
VIII.
In 1982, the Council of the City of Philadelphia passed an affirmative action program with nothing more than “a generalized assertion that there has been past discrimination in [the] entire [construction] industry.” Croson, 488 U.S. at 498, 109 S.Ct. at 724. At the time it enacted the program, and five years later when it extended the program until 1998, nine years past the original sunset date, the Council had no information before it which provided a firm basis for an inference that there had been discrimination by anyone or against any group, much less an inference that the remedy provided by the program was necessary to redress any particular form of discrimination. In 1989, after being challenged in this suit, the Council undertook a study of whether, prior to the adoption of the program, it had discriminated against women and minorities in the award of City contracts or passively supported such discrimination by others. That study arguably produced a firm basis for believing that the City had discriminated against black contractors in the award of prime contracts. It did not provide a firm basis for believing that any other form of discrimination had occurred.
The City’s affirmative action program has been substantially circumscribed by judicial decrees in this case. The preferences for women and non-black minorities have been stricken. Still, however, the remedy provided by the program substantially exceeds the limited justification that the record provides. The program provides race-based prefer-*610enees for blacks in the market for subcontracts where there is no strong basis in the evidence for concluding that discrimination occurred. It authorizes a 15% set-aside applicable to all prime City contracts for black contractors when there is no basis in the record for believing that such a set-aside of that magnitude is necessary to remedy discrimination by the City in that market. Finally, the City’s program fails to include race-neutral or less burdensome remedial steps to encourage and facilitate greater participation of black contractors, measures that the record shows to be available.
A city may adopt race-based preferences only when there is a “strong basis in evidence for its conclusion that [the] remedial action was necessary.” Id. at 500, 109 S.Ct. at 725. Only when such a basis exists is there sufficient assurance that the racial classification is not “merely the product of unthinking stereotypes or a form of racial politics.” Id. at 510, 109 S.Ct. at 731. That assurance is lacking here, and, accordingly, the race-based preferences provided by Chapter 17-500 cannot stand.
The judgment of the district court will be affirmed.27
Before: SLOVITER, Chief Judge, BECKER, STAPLETON, MANSMANN, GREENBERG, SCIRICA, COWEN, NYGAARD, ALITO, ROTH, LEWIS, McKEE, and ROSENN,* Circuit Judges.. We shall refer to our prior opinions as Contractors I and Contractors II. Since it is the only opinion of the district court we shall discuss, any references to the district court's opinion will refer to the opinion filed in the present appeal, Contractors Ass’n of Eastern Pa. v. City of Philadelphia, 893 F.Supp. 419 (E.D.Pa.1995).
. When a DBE wins a prime contract the entire contract price of the prime contract is counted towards the goals, assuming that the DBE performs 51% of the work and takes responsibility for all of it. When DBEs are subcontractors, only the dollars paid by the prime contractor to the DBEs are counted towards the goals.
. The MBEC was charged with ascertaining "the total number of Minority and Female Owned Businesses in the Philadelphia Standard Metropolitan Statistical Area” (the Philadelphia SMSA) within ninety days of the original staffing of the MBEC. Thereafter, it was to “[s]urvey [those] businesses ... to establish past and current participation levels.” The district court found the Philadelphia SMSA to be the relevant geographic area for the purposes of its inquiry.
.Typically, the City receives at least ten bids for a contract, a number that assures competitive bidding and reasonable prices. Minutes from MBEC meetings indicate that, when the sheltered market approach was attempted, the City sometimes received as few as four bids and, as a result, the lowest bid substantially exceeded the City's estimated cost for the project.
. The MBEC’s 1986 Mid-Year Participation Report, for instance, states that the City achieved approximately 19 percent MBE participation in public works. App. at 2884. The report lists every public works contract let in a six-month period, and on nearly every contract, MBE participation was near or above 15 percent. Reports for other years reflected MBE participation on nearly all contracts, although the aggregate participation level achieved has not always exceeded 15 percent.
. In 1993, for instance, of 452 public works contracts let by the City, only 2 were awarded to MBEs. During the same fiscal year, the City achieved a 11.5 percent participation rate. Other than the 2 prime contracts, the rate was achieved entirely from subcontracting by MBEs.
. Dr. Brimmer is the President of Brimmer & Company, Inc., a financial and economic consulting firm located in Washington. D.C.App. at 3936. Brimmer & Company conducts research and advises clients concerning, among other things, trends in economic activity and prospects. Id. Brimmer & Company has consulted on three studies of statistical disparities and discrimination in local markets on behalf of Atlanta, Georgia; Dade County, Florida; and St. Louis, Missouri. Id. at 3938-39.
. The first affidavit examined all minority business enterprises. See App. at 3936. Subsequent to Contractors II, Dr. Brimmer tailored his study in a second affidavit to examine only black construction contractors because we had narrowed the inquiry to such contractors (Brimmer Affidavit II). See App. at 4501. Both affidavits drew on the same original study of disparity in minority participation in City contracting (Brimmer Study), and the critical statistical evidence offered by Dr. Brimmer in both were disparity indices.
. The disparity indices were calculated by "dividing the percentage participation in dollars of minority groups in the public works contracts awarded by the City of Philadelphia by their percentage availability or composition in the 'population' of Philadelphia area construction firms and multiplying the results by 100.” Brimmer Affidavit I, App. at 3941. $ to MBEs -t- $ on all prime _contracts_ X 100 = Index # MBEs -t- total # of contractors
To calculate an index only for black contractors, Dr. Brimmer used data for black contractors in place of data for MBEs. The Brimmer Study was based on data from: (1) the federal government's 1982 Census of Construction Industries, which included the Philadelphia construction industry; and (2) data from the City consisting of (a) a directory and one-page summary of minority business enterprises (MBEs) in the Philadelphia area in 1982, prepared by the City, and (b) a report that the City submitted to the U.S. House of Representatives Committee on Small Business, dated March 1982, which contained the total dollar value of City-financed prime contracts awarded to minorities in fiscal years 1979, 1980, and 1981.
For Brimmer Affidavit II, Dr. Brimmer constructed the disparity index for black businesses using data from 1979-81. There were five prime contracts awarded to MBEs for a total of $667,-501, which was divided by the total for ah public works contracts awarded, $419,779,641, or 0.1 percent. This figure was the "participation rate.” The availability of black businesses was calculated using the City-supplied figure of 114 available black construction firms (BCFs) in the Philadelphia SMSA in 1982, divided by the federal census figure of 8050 total construction firms with employees in the Philadelphia SMSA at the same time. This yielded an “availability rate” of 1.4%. Dividing the participation rate by the availability rate and multiplying by 100, he calculated a disparity index of 11.4.
$667,501.00 -h $419,779,641.00 x 100 = 11.4 114 BCFs -t- 8050 total firms
On direct examination, Dr. Brimmer recognized that the ci1y figure of 114 had included black architectural and engineering firms, and such firms were not included in the 8050 total firm number. The correct number was 57 BCFs. For the record, he recalculated the appropriate disparity index for BCFs using the 57 figure, which yielded a disparity index of 22.5. See App. at 1198. Whether 11.4 or 22.5, disparities of this magnitude, according to Dr. Brimmer, were attributable to racial discrimination.
. Dr. Brimmer described the import of the disparity indices as follows:
As participation approaches availability, the disparity index approaches 100, and therefore fully equitable participation would achieve an index of 100, whereas complete exclusion would produce a score of 0. Thus, the nearer the measure is to 0 or the lower the value of the disparity index, the greater the extent of racial disparity and the stronger the inference of discrimination.
Brimmer Affidavit I, App. at 3941.
. Dr. LaNoue is the Director of the Project on Civil Rights and Public Contracts at the University of Maryland in Baltimore. App. at 1739-41. He is also a Professor of Political Science and the Director of its Policy Sciences Graduate Program. He has recently researched and written on the appropriate legal and social service standards for post-Croson disparity studies. See, e.g., The Local Officials Guide to Minority Business Programs and Disparity Studies: Responding to the Supreme Court’s Mandate in City of Richmond v. Croson, National League of Cities (1994).
. Portions of Justice O’Connor's opinion were joined by a majority of the Justices. Other portions were not. However, it appears from the opinions of all of the justices that an affirmative action program that would survive strict scrutiny review as perceived in Justice O’Connor’s opinion would be sustained by a majority of the justices and one that would not survive that review would not be sustained by a majority of the justices. As we explained in Planned Parenthood v. Casey, 947 F.2d 682 (3d Cir.1991), that makes Justice O’Connor's opinion the current law of the land.
. For example, the court states:
Once a challenge to race-conscious legislation has been raised, that party bears the ultimate burden of proving that the legislation is unconstitutional. Contractors Ass'n of Eastern Pennsylvania, Inc. v. Philadelphia, 6 F.3d 990, 1005 (3d Cir.1993). The local government, in resisting such a challenge, must show a "strong basis in evidence” to support its racial classifications. Croson, 488 U.S. at 492, 109 S.Ct. at 721.
893 F.Supp. at 424.
. As the Contractors stress, the district court also found as follows in a footnote to its findings of fact:
The legislative history of Chapter 17-500 is replete with references which prove that City Council's enactment of Chapter 17-500 was motivated by racial and gender politics, and not the remediation of any specifically identified discrimination in the Philadelphia construction industry.
893 F.Supp. at 445 n. 24. We do not reach this issue.
. In adopting this approach, we disagree with the UMEA’s insistence that two other forms of racial discrimination are relevant here. In the summary of its argument, the UMEA states:
The district court's finding of fact that there is no credible evidence of racial discrimination in the Philadelphia construction industry prior to the enactment of Chapter 17-500 is clearly erroneous. The City and UMEA proffered sufficient statistical, anecdotal and historical evidence of such discrimination in the Philadelphia construction industry in general, and in Ci1y public works prime contracting in particular.
Appellant UMEA’s brief at 4-5. Later, in its brief, the UMEA relies heavily on evidence of past discrimination by local unions. We read Croson to teach that, while evidence of racial discrimination in the construction industry as a whole and evidence of prior racial discrimination by unions may be probative of past societal discrimination, neither will alone justify a municipality in adopting a race-based remedy. We discuss below in text the relevant evidence of racial discrimination by the City in the market for prime contractors.
. Prior to the Ordinance, the selection of subcontractors was not a part of the bidding process for prime contracts. After a contract was awarded, the prime contractor would get approval to use a particular subcontractor from the project engineer assigned to the contract by the City. At trial, John Smith, general manager of the Contractors Association of Eastern Pennsylvania and former member of the MBEC, testified on cross-examination by the City that ultimately, the prime contractor selected the subcontractors. App. at 363. His testimony was not contradicted, and we conclude, therefore, that if there was discrimination in subcontracting, it was largely done by prime contractors. Even if project engineers played some role in the selection, subcontracting presents a sufficiently different nexus of decision making that we should consider the evidence of its existence separately.
. The OMO was created in 1980 by the City to certify the credentials and capabilities of minority contractors who were interested in contracting on the Urban Mass Transit Authority (UMTA) project. The UMTA was a federally-funded, $530 million project, and due to federal requirements, had a 10 percent minority set-aside requirement.
. The UMEA points to testimony from Mr. Smith that the CAEP was advised by City agencies of the construction contracts they hoped to be awarding over the next six months or so. The record does not disclose, however, whether this information was or was not available to others who inquired.
. Even if it be assumed that the payment of City dollars to low bidding contractors constitutes support for Association membership practices when the receiving contractor is an Association member, Chapter 17-500 does not fashion a remedy narrowly tailored to Association membership discrimination.
. The amounts earned by black contractors in connection with the contracts were not included in Dr. Brimmer’s disparity indices because he looked only to City-funded projects which were not subject to the ten-percent set-aside on federally assisted programs.
. Were we reviewing the district court's finding of fact regarding the Council’s intent in 1982 at the time of the adoption of Chapter 17-500 (See footnote 14, supra ), it would, of course, be inappropriate to consult the City’s experience after that point. However, when the issue for decision is whether there is a strong basis in evidence for believing that City-sponsored discrimination existed and the City is relying on after-acquired evidence to support that position, it is appropriate to consider any evidence, pre- or post-enactment, relevant to the issue of whether such discrimination, or the effects thereof, existed prior to 1982. Contractors II, 6 F.3d at 1003-04; Concrete Works of Colorado, 36 F.3d at 1521.
. We have considered the possibility that the City Council may choose to address in the future the need for a more narrowly tailored affirmative action program to remedy past discrimination. In that context, it may wish to consult Dr. Brim-mer's work. However, any ruling that we might make regarding the probative value of that work in the context of any affirmative action program initiated to address conditions in 1979-1981 would be of limited, if any, value to a legislator assessing the necessity of such a program addressed to conditions more than fifteen years later.
. Nor has the City, since 1990, considered whether the 15% goal should be adjusted to reflect the fact that the preference is now for the sole benefit of black contractors.
. Dr. Brimmer testified that there were 57 black construction firms on the OMO Directory and 8050 total construction firms in the Philadelphia SMSA. Dividing 57 by 8050 yields the 0.7% representation of black firms in the Philadelphia construction market.
. As mentioned, supra, the City alleged that when it considered the Ordinance, it also weighed the effectiveness of the Philadelphia Plan as a race-neutral alternative. All the record indicates is that such a plan was initiated in the 1960s and may have involved the hiring of mi*609norities on construction sites. While the scope and experience of the Philadelphia Plan is not clear, it does not appear to have been addressed to discrimination in the award of construction contracts.
. The legislative history of the Ordinance is generally devoid of any consideration of race-neutral alternatives. Joseph Coleman was a Council member when the Ordinance was passed, and was the only such Council member to testify. He could testify only that he had been aware of the Philadelphia Plan and Urban Coalition program and could not recall any details of them. The only reference in the legislative history to race-neutral measures was a reference in the preamble to the failure of the federal Small Business Administration to increase the number of minority and women businesses. Examining the success or failure of a national program directed to businesses of all types, however, is not constitutionally adequate consideration of the potential effectiveness of race-neutral measures for a particular industry in a particular locality. Cf. Croson, 488 U.S. at 504, 109 S.Ct. at 727 (rejecting city’s reliance on Congress’ determination that discrimination affected construction industry nationally).
. We have considered and found to be without merit the City’s challenges to the district court's judgment based upon (1) Judge Bechtle's failure to recuse, (2) the denial of the motion in limine to exclude the testimony of John Smith, and (3) the alleged failure to the contractors to establish standing.